The residual dark unit of the most recent eastern Mediterranean sapropel (S1) is usually overlain by sediments with enhanced concentrations of MnOx in two separated layers. The variability and magnitude of the Mn enrichment at different locations and water depths indicate that Mn must have been added preferentially to sediments at intermediate (1–2 km) water depths. We propose a two-stage mechanism for the Mn enrichment that involves decreasing oxygenation with increasing water depth. This mechanism involves the loss of reduced Mn2+ from the deepest sediments (>2 km water depth) into overlying anoxic waters and a variable gain of MnOx in sediments in contact with oxygenated waters at shallower depth. In the S1 unit that receives the extra MnOx input, an upper higher Mn-enriched zone (>3 wt %) is maintained continuously at the top of the accumulating S1 unit because the pore waters are anoxic at shallow sediment depth while bottom waters are oxic to some degree. In a reactive-transport model, the Mn enrichment in the upper zone could not be supported by normal sediment diagenesis. Thus the MnOx in the upper Mn horizon must have formed mainly in the water column. The MnOx in the upper Mn-enriched zone adsorbed Mo and Li from seawater in a similar manner as other Mn-enriched oxic sediments, nodules, and crusts, with a Mn:Mo ratio of ∼600:1, a Mn:Li ratio of ∼750:1, and a δ98/95MoMOMO of −2.5 ‰.
 Eastern Mediterranean sediments are generally Corg-poor, but discrete Corg- and sulfide-rich units known as sapropels formed repetitively in the sediment sequence over at least the past 3 Myr [Rohling, 1994]. A close relationship between sapropel formation and orbital precession-driven solar insolation variations has been interpreted to imply that an enhanced flux of monsoonal fresh water to the surface ocean must control sapropel formation in this semi-isolated basin [Rossignol-Strick, 1985; Hilgen, 1991; Lourens et al., 1996; Tuenter et al., 2003]. At present, Corg–poor sediments form [Van Santvoort et al., 2002] because the basin's deep waters are ventilated regularly so that high O2 contents are maintained throughout the water column [Béthoux, 1993; Roether and Well, 2001]. The scenario proposed for sapropel formation envisages that increased runoff leads to water column stabilization and probably to enhanced primary productivity [Rohling, 1994]. Deep-water dysoxia or anoxia then develops in the poorly ventilated deep waters to allow formation of sapropel with enhanced Corg preservation and diagenetic sulfide production in the sediments.
 Managense is a mobile element in sediments because of the ready interconversion of its different chemical species in response to early diagenetic redox changes that occur in sediment pore waters. The bacterial catalyzed remineralization of reactive organic matter in sediments reduces manganese oxide (Mn(III,IV)Ox) to soluble Mn(II) when sediment pore waters become suboxic or anoxic at depth. This Mn2+ diffuses in the pore waters until suboxic/oxic conditions preferentially in the presence of oxidizing bacteria are reencountered. At this boundary the main fraction of Mn2+ is again precipitated as MnOx, either in the surficial sediments or in the overlying seawater. This behavior means that Mn is best suited to studies of changes in water column oxidation status during formation of the most recent sapropel (S1) between 9.5 and 6 14C kyr ago [Mercone et al., 2000]. Two distinctly separated Mn-enriched zones are generally found in the sediments immediately above the visual expression of sapropel S1. The upper zone whose host sediment is about 5.7 kyr 14C convention B.P. (interpolated [Reitz et al., 2006]) has been related to changes in water column oxygenation that occurred at the end of this most recent sapropel episode. It has been important in quantifying the postdepositional oxidation that most S1 units have experienced since their formation [De Lange et al., 1989; Higgs et al., 1994; Thomson et al., 1995, 1999; Van Santvoort et al., 1996]. Pore water investigations have revealed that it is generally the lower of the two Mn-enriched zones that is forming actively, and that the maximum of this lower Mn-enriched zone marks the present limit of suboxic conditions in the sediments [Van Santvoort et al., 1996].
 The genesis of the upper Mn-enriched zone (from here on upper Mn zone) that often contains higher Mn contents than the lower Mn-enriched zone (from here on lower Mn zone; but usually still <1 wt % Mn) is less certain [Pruysers et al., 1993; Thomson et al., 1995, 1999; Van Santvoort et al., 1996]. Formation of the upper Mn zone usually appears to coincide with the end of sapropel formation, and it seems to be related in some manner to the resumption of ventilation and the return of high O2 levels in bottom waters at the end of S1 sapropel times [Thomson et al., 1995, 1999; De Lange et al., 1999]. The upper Mn zone therefore formed before and is older than the corresponding lower Mn zone that is located a few cm deeper in the sediments. Cita and coworkers [Cita et al., 1989; Camerlenghi et al., 1992; De Capitani and Cita, 1996; Hieke et al., 1996] also reported a prominent dark layer with variable Mn contents up to 22.8 wt % Mn above S1 in sediments of the diapiric crestal area on the Mediterranean Ridge. This layer has been termed the “Marker Bed”, and its formation ascribed to an expulsion of hydrothermal fluid. There is a danger in considering any Mn feature as a marker bed, in the sense of a chronostratigraphic horizon that can be traced over a wide area, because of the early diagenetic mobility of Mn described above. Thomson et al.  noted that the Marker Bed is similar to the upper Mn zone in published photographs, both in appearance and in position relative to the S1 sapropel, and they proposed that the Marker Bed and this Mn zone are equivalent. The unusually large Mn contents measured in some Marker Bed locality cores do however require further explanation.
 1. Regarding the source and formation mechanism, large Mn enrichments might be diagenetic, i.e. formed within the sediments by early diagenesis. Van Santvoort et al.  demonstrated that the lower Mn zone above the residual S1 unit is forming actively by this mechanism as outlined above. Very high Mn contents can be produced by this mechanism in surficial sediments if the surficial oxic zone is a few centimeters or less deep and Mn is continuously added to this thin oxic layer or zone [e.g., Lynn and Bonatti, 1965; Shimmield and Price, 1986; Kadko et al., 1987]. In order to have produced the upper Mn zone in eastern Mediterranean sediments, this mechanism implies that finite bottom water O2 contents were present during sapropel deposition. This would be required to maintain an oxic surficial layer in the sediments and prevent the escape of most of the Mn2+ from the sediments to bottom waters.
 2. Large Mn enrichments might also be hydrogenetic, i.e., precipitated within the water column [e.g., Force and Cannon, 1988; Frakes and Bolton, 1992; Calvert and Pedersen, 1996]. If bottom water oxygen levels fall to sufficiently low but not necessarily zero levels, then significant amounts of Mn2+ from pore waters will diffuse out of the sediments and escape to bottom waters [Balzer, 1982; Kristensen et al., 2003]. This process develops high Mn2+ concentrations in suboxic or anoxic water columns, as in the isolated basins of the Black Sea, the Cariaco Trench, silled fjordic basins [Calvert and Pedersen, 1996], and deep anoxic eastern Mediterranean brine basins [De Lange et al., 1990]. A sharp solubility gradient of Mn exists along the oxic/anoxic interfaces in such water columns. Dissolved manganese contents in anoxic waters might be ∼1000 times greater than in oxic surface waters where Mn concentrations are typically <10 ppb Mn because Mn2+ in solution is favored by the low Eh of anoxic waters [Hem, 1972; Force and Cannon, 1988]. If the stability of such an anoxic water column breaks down, then increased dissolved O2 levels introduced with ventilation will precipitate dissolved Mn2+ as MnOx that rains back down through the water column to the underlying sediments. Such a process occurs regularly in winter in the deep basins of the Baltic Sea [Huckriede and Meischner, 1996; Sternbeck and Sohlenius, 1997]. In order to produce the major upper Mn zone in eastern Mediterranean sediments a two-step process must be involved. First, low bottom water O2 contents must have developed during sapropel deposition in order to allow Mn2+ to escape from sediments into the bottom waters, and second, an increase in ventilation at the end of sapropel formation must have introduced sufficient O2 to reprecipitate this Mn2+. If sapropel formation ended with a single ventilation event, then this latter mechanism might be expected to form the highest enrichment in the upper Mn zone at sites with the deepest water. This is not the case as will be seen below.
 3. A further possibility is the hydrothermal formation of large Mn enrichments, i.e. produced through an emission of hydrothermal fluid. Manganese can be enriched in a range of forms in the vicinity of hydrothermal vents [Roy, 1992; Mills and Elderfield, 1995]. At active vents on the mid-ocean ridge system, Fe2+ tends to precipitate from black smoker plumes in close proximity to vents, whereas Mn2+ precipitates at a variety of near- and far-field sites. Manganese is sufficiently fractionated from Fe2+ that it can still be detected in plumes thousands of kilometers distant from the source vents [Klinkhammer and Hudson, 1986]. It therefore seems likely that the authigenic flux of Mn precipitated in the deep ocean contains a hydrothermal component [Roy, 1992]. To produce the upper Mn zone, this mechanism would imply a sufficiently high water column O2 content to precipitate the emitted Mn, but any relationship in time to sapropel deposition would be fortuitous.
 Manganese enrichments formed by diagenetic, hydrogenetic, and hydrothermal mechanisms are expected to initially form some species of MnOx. Once buried and as anoxic conditions develop over time, all types may contribute to surficial diagenetic MnOx enrichments by the reduction mechanism outlined above.
 This work investigates two Mn zones observed above S1 sapropel units in a suite of 11 cores collected at different locations and water depths (650–3400 m) in the eastern Mediterranean basin (Figure 1). Particular emphasis is placed on cores with a major upper Mn zone in order to explore the source and mechanism of their formation.
2. Material and Methods
 Data obtained at the Utrecht and SOC laboratories at different times and by different analytical methods were assembled from cores selected to give a geographical and bathymetric coverage of the eastern Mediterranean basin (Figure 1 and Table 1). Those from box cores BP10, BP15, BP18, SL125, ABC26, SL9, and SL114 were analyzed by Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES) at 0.5 cm resolution after a three-step digestion to ensure total dissolution: (1) digestion of 125 mg of freeze-dried sediment in a mixture of 2.5 mL 3:2 concentrated HClO4 and HNO3 with 2.5 mL concentrated HF at 90°C; (2) evaporation of the solution to near dryness at 160°C; and (3) dissolution of the residual in 25 mL of 1 M HCl at 90°C. Ten samples from 9.5 to 14.5 cm in core BP18 were not completely digested in this manner; for these samples the procedure was applied to ∼60 mg of sediment with an extra digestion step that involved evaporating the sample to near dryness with 5 mL concentrated HCl after step 2. Analyses of samples at 1 cm resolution from box core MC562 were performed by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) after a three-step total digestion described by Reitz et al. [2004a]. Samples from box core T87-29B and hydraulic piston core 971C from Ocean Drilling Program (ODP) Leg 160 were analyzed by ICP-AES after a 1:5 fusion with LiBO2 [Totland et al., 1992] at 0.5 and 1 cm resolution, respectively. Further analyses of Mo and Li in selected samples from these two cores were performed in the manner described above for core BP18. For the different procedures, sample duplicates, international standards, and in-house standards were processed to monitor the precision (by duplicates) and accuracy (by standards) of the analyses.
Table 1. Positions, Water Depths, and Other Details of the Cores Investigated
Water Depth, m
Maximum Mn Content in Upper Mn Zone, wt%
R/V Pelagia/2001 (Biopass)
R/V Marion Dufresne/1990
R/V Meteor/2001 (M51-3)
R/V Pelagia/2001 (Biopass)
R/V Professor Logachev/1999 (Smilable)
R/V Pelagia/2001 (Biopass)
R/V JOIDES Resolution/1996 (ODP Leg 160)
R/V Professor Logachev/1999 (Smilable)
R/V Professor Logachev/1999 (Smilable)
 Organic carbon content was determined according to the method described by van Santvoort et al. . International and in-house standards and duplicates were processed to monitor precision and accuracy.
 A sequential extraction with different solutions (Table 2) was applied to five 125 mg samples of box core BP18; one background sample from the sediment above the upper Mn peak (5.5–6 cm), three from the upper Mn peak (10.5–11, 12–12.5, and 13–13.5 cm), and one from the lower Mn peak (17–17.5 cm). All samples and two standards (international standard MAG-1 and an in-house manganese nodule standard) were processed in duplicate to monitor the precision of the sequential extraction. Recovery with respect to the standard total element concentrations of Mn, Ba, Fe, and Li was 99%, 94%, 99%, and 107%, respectively.
 Pore water analyses of BP10 were done by ICP-MS after onboard sampling of the box core at in situ temperature (∼13°C) in a nitrogen-filled glove box and centrifugation to separate pore waters. Sample duplicates were analyzed to monitor over all precision and analyses were done in triplicate; reproducibility was better than 10% for Mn concentrations <1.5 μmol L−1 and better than 5% for concentrations >1.5 μmol L−1.
 Determination of molybdenum isotope fractionation was performed at the University of Bern by double-spike inductively coupled plasma-multicollector mass spectrometry by methods described by Siebert et al. .
 Bulk element data (e.g., Fe, Mn, and Ba) are normalized to Al to take account of fluctuations in aluminosilicate content, on the assumptions that the element/Al ratios in detrital material are relatively constant, and that increases in these ratios above detrital levels indicate diagenetic enhancements of the redox-sensitive elements [e.g., Van Os et al., 1994; Van der Weijden, 2002]. For higher concentrations of Mn in the large upper Mn zone, a major fraction of the Ba appears to be associated with MnOx rather than present as biogenic barite. As barite is an important indicator for sapropel formation, a correction for this MnOx-associated Ba is needed. The Ba associated with the maximum upper MnOx ((Ba/Mn)mum ∼0.0031) was determined by sequential extraction; this ratio has been used to correct the Ba/Al ratio in cores with a high Mn concentration in the upper Mn zone (>3 wt % Mn; section 3.2), using
where Batot, Mntot, and Altot are the element concentrations in bulk sediments determined from total digestion. Although Baexcess values could have been calculated [Reitz et al., 2004b], we preferred to use Ba/Al for the correction because the cores are from a wide variety of locations showing detrital background Ba/Al values between 0.0021 (SL9) and 0.0059 (MC562 on the Libyan shelf).
2.1. Model Description
 A Mathematica®-based, one-dimensional, reactive transport model scheme was composed to quantify MnOx formation in the sediments. A similar model setup was described by Hensen and Wallmann . The model enables for the calculation of sedimentation, molecular diffusion, and degradation of organic matter. Rates of Corg degradation (RCorg in wt % yr−1) are given by
where kCorg is the kinetic constant for the degradation of reactive Corg yr−1 and [Corg] is the concentration of reactive organic carbon in wt % that reaches the sediment surface.
 Likewise, the rate of Mn2+ oxidation (RMnOx in wt % yr−1) by oxygen is calculated as
where kMnOx is the kinetic constant for Mn2+ oxidation in cm3 μmol−1 yr−1, and [O2] and [Mn2+] are the corresponding concentrations in μmol cm−3 of pore solution.
 To keep the model simple, the relevant processes of organic carbon degradation were reduced to oxic respiration and manganese-oxide reduction. These processes were implemented by rate laws using Monod-type kinetics:
where RO2, RMn, and RMnOx are the rate of oxidation, manganese reduction, and manganese oxidation in mmol dm−3 yr−1 and wt % yr−1 (for RMnOx), [MnOx] is the species concentration in mmol dm−3, kO2 is the Monod constant for oxic respiration in mmol dm−3, kMn is the Monod constant for manganese-oxide reduction in mmol dm−3, and kIO2 is the inhibition constant for manganese-oxide reduction in mmol dm−3. A summary of all parameter values used is given in Table 3.
Table 3. Summary of The Parameter Values Used in the Model
Dissolved Mn2+ at the sediment surface (post-S1 times)
0.0 mmol dm−3
Dissolved Mn2+ below the Mn peaks (present-day)
0.004 mmol dm−3
MnOx input at the sediment surface
0.06 wt %
3. Results and Discussion
3.1. General Characteristics of Mn-Enriched Horizons Above Sapropel S1
 Interpretations of Corg, Ba/Al, Fe/Al, and Mn/Al profiles through S1 units were reviewed by Thomson et al. , and the data for the new cores discussed in this section have many similarities to the limited number (about a dozen) of eastern Mediterranean cores containing S1 for which detailed inorganic data have been reported. A set of five parameters is presented for core BP10 (Figure 2) to illustrate the geochemical criteria that define the original and residual S1 boundaries. Mn/Al and Fe/Al profiles are also shown for cores BP15, ABC26, SL9, and SL114, where the thickness of the visual S1 unit and the amount of oxidation that occurred since deposition were evaluated in a similar manner to that applied to core BP10 (Figure 2).
 Unless S1 units are from areas of unusually high sediment accumulation rates [Mercone et al., 2000, 2001], they are generally thinner than the unit originally laid down, because of postdepositional oxidation. This oxidation greatly diminished or destroyed the original Corg and sulfide contents in the upper part of the sapropel and thus lightened the characteristic dark color that accompanies high Corg and S contents in sapropels [De Lange et al., 1989, 1994; Higgs et al., 1994; Thomson et al., 1995, 1999; Van Santvoort et al., 1996]. Fortunately, biogenic Ba (as barite) that accompanied settling Corg appears to be preserved in the sapropels, even after the Corg has been extensively oxidized [Thomson et al., 1995]. Thus the positions and shapes of the Ba/Al and Mn/Al profiles allow deduction of the original thickness of S1 [Thomson et al., 1995, 1999; Van Santvoort et al., 1996; Mercone et al., 2000; Rutten and De Lange, 2002] (Figure 2). For all cores shown in Figure 2, except core BP15 from the shallowest water depth, the upper Mn/Al peak occurs exactly at the top of the zone of high Ba/Al values that marks the original S1 unit, as seen in the detailed profiles of core BP10; the lower Mn/Al peak is situated immediately above the zone of high Corg. This is evidence of the postdepositional oxidation of sapropel S1 that occurred between the two Mn peaks over the last 6 kyr. The pore water Mn2+ profile (Figure 2) indicates continuing suboxic/anoxic reduction of MnOx in underlying sediments that fuels formation of the lower Mn zone.
 There is little or no evidence of Fe enrichment associated with the upper Mn peak in these cores. By contrast, a clear enrichment of Fe is associated with the lower (diagenetic) Mn peak in all cores. This indicates that the precipitation of Mn oxide and Fe oxyhydroxide is fueled by Mn2+ and Fe2+ diffusing up from below and by bottom water O2 penetrating into the sediment [De Lange et al., 1989; Van Santvoort et al., 1996; Passier et al., 2001; Rutten and De Lange, 2003]. The lower Mn zone with associated Fe formed after S1 formation due to the penetration of the oxidation front into the sediment. This process continues at present, except at very shallow water sites (∼<500 m). The behavior of Fe in sapropels differs from Mn in that Fe and S form pyrite. High Fe/Al values in the visually observable sapropel, where Mn values are low and Corg and S values are high, indicate the presence of pyrite [Passier et al., 1996]. Passier et al. also reported sulfide formation in excess of the available Fe for sapropel S1. This excess may not only have resulted in the downward sulfidization process and associated consumption of all available Fe2+ during S1 formation, but may also have resulted in a HS− flux into the water column. The latter would have maintained low Fe2+ concentrations in the water column.
3.2. An Upper Mn layer in Oxic Conditions Above Sapropel S1
Section 3.1 demonstrated that S1 units in different cores display some compositional variability and consistent evidence for postdepositional oxidation after their formation. Among the major elements, Mn shows the greatest concentration variations in S1 units. In this section, five cores are examined that have a much larger upper Mn peak than the <1 wt% levels (Figure 2). Core BP18 from the Libyan slope has the highest Mn concentration (24.6 wt %) in the upper Mn peak yet reported for the eastern Mediterranean.
3.2.1. Libyan Continental Slope
 Core MC562 was recovered west of Cape Sirte in 1391 m water depth and has a maximum Mn content of 9.4 wt % in the upper Mn zone. Core BP18 was recovered northwest of Cape Sirte at 1850 m and has a maximum Mn content of 24.6 wt % in the upper Mn zone. XRD analysis of the core BP18 sample with the highest Mn concentration established birnessite as the main Mn phase, and an absence of Mn-carbonate minerals (rhodochrosite/kutnahorite) (C. Vogt, personal communication, 2004). This finding is supported by sequential extractions that reveal that nearly all the Mn in the upper zone is extracted in the poorly crystalline and amorphous oxide/oxyhydroxide fraction (Figure 3).
 The uncorrected Ba/Al profile of core BP18 displays a narrow peak that coincides exactly with the upper Mn peak (Figure 3). For the samples discussed in this section that have Mn concentrations >0.5 wt%, a correction for Ba absorbed to MnOx was made because sequential extraction confirmed that some Ba is associated with the poorly crystalline/amorphous oxide fraction and is therefore unrelated to biogenic barite (see section 2). The Fe/Al profiles of both cores display small Fe enrichments that correspond to the upper Mn enrichment. This is interpreted as a sorption of Fe2+ on MnOx [Koschinsky and Halbach, 1995] because sequential extraction of core BP18 samples released a fraction of the total Fe from the sediments of the upper Mn zone in the absorbed ion fraction, a feature that is not observed in samples from above or below the upper Mn zone. It seems that sorption of Fe2+ onto MnOx can be discerned when Mn concentrations exceed 5 wt % (compare Figures 3 and 5; MC562 with ∼9 wt % Mn and T87-29B with ∼5 wt % Mn). Both Fe/Al profiles show a clear and distinct enrichment of Fe that corresponds with the lower Mn enrichment. However, the fraction of Fe released in the poorly crystalline/amorphous (oxyhydr)oxide fraction by sequential extraction of the BP18 samples increases in the lower Mn zone, whereas there is negligible Fe leached with the absorbed ion fraction (17.5 cm, Figure 3).
Casford et al.  reported the presence of benthic foraminifera throughout S1 for core MC562. They postulated conditions of either continuous dysoxia or intermittent ventilation at ∼2000 m water depth. Conversely, benthic foraminifera are absent between ∼28 cm (∼8.73 kyr; interpolated) and 14 cm (5.5 kyr, direct radiocarbon dating) sediment depth in core BP18, implying extensive anoxia during sapropel formation.
3.2.2. Marker Bed Area
 Box core SL125 with a maximum Mn content of 2.9 wt % in the upper Mn zone was retrieved from 1946 m water depth in the diapiric summit area of the Mediterranean Ridge south/southwest of Crete where high Mn values have been reported [Cita et al., 1989; De Capitani and Cita, 1996]. The profiles in this core are very similar to those discussed above (Figures 2 and 3), with the exception that the upper Mn peak is much broader (Figure 4) than the others discussed in this paper. As in the other examples, this peak corresponds exactly to the top of the S1 unit.
 ODP hydraulic piston core 971C is located in the circular depression or “moat” that surrounds the Napoli mud dome in the Marker Bed area. The S1 unit in this core is unusual in that it is ∼65 cm thick, whereas all other S1 units investigated are 15–20 cm thick (Figures 2, 3, and 5). The ODP shipboard description of 971C notes that the S1 unit is “expanded”, but our interpretation, based on micropaleontology (E. J. Rohling, personal communication, 2004) and inorganic chemistry is that the unusual thickness of this S1 unit is due to a redeposition near the end of sapropel time of ∼50 cm of S1 sapropel sediment, as indicated in Figure 4.
 The point of interest of core 971C S1 for this study is that the upper Mn peak is large and very sharp, with the upper and lower Mn peak maxima containing 14.2 and 0.62 wt % Mn (Figure 4). The upper peak corresponds to the top of the high Ba/Al values, and the separation of the maxima at 31.5 and 36.5 cm indicates 5 cm of postdepositional oxidation, which is similar to the oxidation depth seen in most other S1 units (e.g., BP10 in Figure 2). Although different hydrogenetic, diagenetic, or diapiric (deep-seated) possibilities for Mn supply can be invoked for a highly enriched upper Mn zone at this locality, this Mn zone must have formed very late in S1 times, if it was caused by redeposition (Figure 4). It is most likely that the upper Mn zone represents a thin surficial diagenetic MnOx layer that formed from a pore water flux of Mn2+ that was generated from the redeposited sapropel sediment in this unusually thick S1 unit. This interpretation requires that bottom waters were oxic. The unusual thickness of the S1 unit in comparison with the other cores from the Marker Bed area (ABC26 and SL125, Figures 2 and 4) suggests that a redeposition event is the major factor that produced such a highly enriched upper Mn zone. Its sharpness demonstrates the rapidity of MnOx remobilization and redistribution.
3.2.3. Medina Rise Area
 Core T87-29B was retrieved from a water depth of 1530 m depth on Medina Rise and has a maximum Mn content of 4.7 wt % in the upper Mn zone [Rasmussen, 1991; Troelstra et al., 1991]. The separation of the upper and lower Mn zones related to postdepositional oxidation is similar to that seen in the other cores, but an important difference is that the upper Mn zone is located in a slightly different position. Its maximum is located 2 cm below the top of the zone of high Ba/Al values that define the full S1 sapropel, rather than exactly on its upper boundary (Figure 5). According to a hydrogenetic origin for Mn in the upper Mn zone, this zone would have formed at the seafloor before the end of sapropel time. Alternatively, if the upper layer formed diagenetically, it would represent oxygen penetration into the surface-sediment layer at this shallower water location during the last phase of sapropel formation.
3.3. Molybdenum Sorption on Manganese Oxide (MnOx)
 The large Mn peaks discussed above are accompanied not only by enrichments of adsorbed Ba and Fe (section 3.2), but also by variable enrichments of several other elements. The elements that exhibit the most pronounced and consistent enrichments are Mo and Li. All the cores with a large Mn peak in the upper Mn zone display an exact coincidence of Mn/Al, Mo/Al, and Li/Al maxima, with the Mo and Li enrichments clearly dependent on the magnitude of the Mn peak (Figures 3, 4, and 5).
 In contrast, an estimated 47–85% of the total Mo removal from the ocean occurs under oxic conditions by pelagic sediments [Bertine and Turekian, 1973; Morford and Emerson, 1999]. Shimmield and Price  summarized evidence for the scavenging and uptake of Mo by oxic Mn-containing sediments, Mn-containing particulate material, and ferromanganese nodules. It was demonstrated that the Mn:Mo mass ratio in Mn-enriched oxic sediments, nodules, and crusts is consistently ∼500:1 up to 25 wt % Mn (500 μg g−1 Mo [Takematsu et al., 1985; Shimmield and Price, 1986; Calvert and Pedersen, 1993]). Unlike Mo enrichments with sulfide, burial and preservation of these Mo enrichments in the sedimentary record is not expected because of reductive dissolution of MnOx during burial. Shimmield and Price  suggest that Mo adsorbed by MnOx is preferentially released and lost to solution when the Mn[IV] and Mn[III] in MnOx are reduced to Mn[II]. This contrasts with the behavior of Mn itself, because residual MnOx can adsorb Mn2+ after reductive release to solution.
 Most of the Mn in both zones above the residual S1 is present as MnOx (section 3.2.1) and is therefore also expected to adsorb Mo. In the oxic sediments above the residual sapropel, where Mn contents exceed 1500 μg g−1 (0.15 wt %), Mo correlates linearly with Mn (average r2 and n of the 5 cores discussed in this section is 0.97 and 8, respectively). The average enrichments of ∼600 (MC562 419:1; BP18 696:1; SL125 769:1; ODP971C 617:1; T87-29B 620:1) is slightly higher than the 500:1 ratio reported by Shimmield and Price . The average Mn:Mo ratio of 600:1 appears to be the ratio for Mo adsorbed from eastern Mediterranean bottom waters on to MnOx.
 Molybdenum isotope fractionation was determined for two samples from the upper Mn zone of BP18. The sample with the highest Mn concentration at 12–12.5 cm has a Mo isotope composition of δ98/95MoMOMO = −2.6 ‰ (±0.1) and the sample immediately below has a value of −2.4 ‰ (±0.1). These isotopic compositions are distinctly offset from the values found by Siebert et al.  for ocean waters (δ98/95MoMOMO = 0 ‰) and reducing sediments (δ98/95MoMOMO = −1.8 to −0.5 ‰), but lie between the estimated mean crustal composition (δ98/95MoMOMO = −2.3 to −2 ‰) and Fe-Mn crusts and pelagic sediments (δ98/95MoMOMO = −2.7 to −2.9 ‰). The preferential sorption of lighter Mo isotopes to the upper Mn peak MnOx therefore appears consistent with sorption from seawater under oxic conditions (T. Nägler, personal communication, 2004), although more measurements are required for a fuller interpretation.
3.4. Lithium Sorption on Manganese Oxide (MnOx)
 Alkali elements like lithium are unaffected by redox reactions [James and Palmer, 2000] and most of the uptake of Li into particulate material occurs in the water column before burial in sediments [Zhang et al., 1998]. In order to establish the authigenic fraction in the large upper Mn zone, Liexcess was calculated as
because of a relatively large Li-clay contribution with a constant Li/Al ratio. Litot is the bulk Li concentration and (Li/Al)det is the Li concentration associated with the clay minerals, determined by sequential extraction as 0.0007. The average Mn:Liexcess ratio is ∼750:1 (MC562 652:1; BP18 898:1; SL125 724:1; T87-29B 797:1) within the upper Mn peak. A positive correlation (except core T87-29B) of Liexcess and Mn contents (the average r2 and n of the first three cores is 0.92 and 9 and that of the last core is 0.24 and 3, respectively) for samples formed under oxic conditions has not been previously reported to our knowledge. Some but by no means all hydrothermal samples rich in Mn oxides also show distinct enrichments of Li [Hodkinson et al., 1994; Glasby et al., 1997; Koschinsky and Hein, 2003; Dekov et al., 2003]. In fresh precipitates from seawater, Sato et al.  found that concentrations of alkali and alkaline earth elements were higher in Mn oxides than in Fe(III) oxides. The ionic radii of Li+ and Mo6+ are the same (0.68 Å), and are believed to substitute for Mn4+ with an ionic radii of 0.62 Å [Koschinsky and Hein, 2003]. On the basis of the sequential extractions, the largest Li fraction was released in step 4 that dissolves poorly crystalline/amorphous oxides (Figure 3), but a significant fraction is also released in step 3 that dissolves carbonates. According to Koschinsky and Hein , who found similar Li behavior in a leaching investigation, Li may be present either as a loosely bound fraction or as a discrete Li phase. In core BP18, the main Liexcess fraction is associated with the poorly crystalline/amorphous-oxide phase and a minor fraction is found as the easily exchangeable phase/absorbed ions. Earlier investigations [Gieskes, 1983; Zhang et al., 1998; James and Palmer, 2000] showed that Li is a readily exchangeable ion because seawater Li deposited along with marine sediments is readily displaced from the solid to solution by less hydrated cations like NH4+. Thus Li may be adsorbed directly from seawater or from Li released from underlying sediments.
3.5. An Upper Mn Layer Now Located in Anoxic Conditions: S1 Sapropel in Core MD90-917
Mercone et al. [2000, 2001] reported that a single layer of high Mn content (3.3 wt %) exists in the S1 sapropel of core MD90-917 recovered from a depth of 1010 m in the Otranto Strait, Adriatic Sea (Figure 6). At this locality the sediments accumulate so rapidly that the S1 Corg content is low, postdepositional oxidation of the original sapropel is insignificant, and no lower Mn zone exists. As a result, the zones of Corg enrichment and higher Ba/Al ratio coincide in this S1 [Mercone et al., 2001]. Like the upper Mn zone in cores T87-29B (Figure 5) and BP15 (Figure 2) that were also retrieved from shallow water depths, the Mn zone (max. 3.3 wt % Mn) in core MD90-917 is within the S1 unit. Thus it is located in sediments laid down before the end of S1 times. However, this S1 unit is now located so deep in the sediments that the pore waters at this level must now be anoxic.
 X-ray diffraction of material from this Mn zone revealed a 2.94 Angstrom d-spacing peak characteristic for kutnahorite, an authigenic Mn-Ca carbonate mineral [Mercone et al., 2001] (Figure 6). The diffusion of Mn2+ in carbonate oozes and marls in anoxic environment is often limited by the uptake of Mn2+ onto carbonate surfaces [Boyle, 1983; Thomson et al., 1986; Middelburg et al., 1987]. A rough estimate of the composition of the authigenic Mn-Ca carbonate in core MD90-917 was obtained by plotting the Ca and CO32− contents against Mn content for samples from the Mn zone with Mn contents >0.5 wt %. The molar balance of additional Mn plus Ca versus additional CO32− corresponds to (Mn0.77Ca0.23)CO3. This is similar to the Mn molar fraction of 0.7–0.8 found in the authigenic Mn-Ca carbonates that form episodically as laminations in the sediments of the Baltic Sea deeps. The critical requirement for the formation of these Baltic Sea Mn layers is the development of very high (>0.2 mmol kg−1) pore water Mn2+ concentrations from MnOx reduction in the presence of alkalinity [Kulik et al., 2000; Neumann et al., 2002]. The Mn zone in core MD90-917 shows that a conversion of MnOx to kutnahorite also occurs in eastern Mediterranean sediments when MnOx is reduced. Given the repetitive nature of sapropel formation, such an authigenic Mn-Ca carbonate zone can also be expected to have formed in the sedimentary record in association with older sapropels, although their long-term persistence is not assured [Van Os et al., 1991, 1994; De Lange et al., 1994; Heiser et al., 2001].
 If adsorbed Mo is lost preferentially when MnOx is reduced, as suggested by Shimmield and Price , Mo would not be expected to incorporate into the kutnahorite. A clear Mo enrichment does coincide with the Mn enrichment, although the Mn/Mo mass ratio at and above this Mn zone averages ∼ 1700:1. This ratio is considerably larger than the Mn/Mo ratios observed in MnOx in the cores discussed above where oxic conditions still exist (Figure 6). Loss of Mo from MnOx on reduction therefore appears to have been extensive but not quantitative.
3.6. A Conceptual Model for Formation of the Upper Mn Layer Above Sapropel S1
 Some generalizations can be drawn from consideration of S1 sapropel units with an upper Mn zone with a maximum Mn content of >3 wt %:
 1. Such large Mn concentrations are found at other eastern Mediterranean localities besides the crestal diapiric ridge (the Marker Bed area), e.g., on the Libyan slope, Medina Rise, and the Straits of Otranto (Figure 1).
 2. Apart from core ODP971C (2152 m), all S1 units with unusually large upper peak Mn contents occur in sediments retrieved from water depths between 1100 and 2000 m (Figure 7). Not all cores from this water depth range exhibit a strongly Mn-enriched zone however, and Mn contents are highly variable in the two areas from which multiple cores are available, namely the Marker Bed area [Cita et al., 1989; De Capitani and Cita, 1996; this work] and the Libyan slope (this work).
 3. In the Libyan slope cores studied here, a benthic foraminiferal fauna is present throughout S1 in core MC562 from 1391 m [Casford et al., 2003] but completely absent from the S1 unit of core BP18 from 1850 m. This indicates that bottom waters at the depth interval 1100–1800 m could not have been continuously anoxic during S1 formation.
 4. The systematics of parallel Mo and Li enrichments with MnOx are conspicuous in all cores containing the distinct upper Mn zone.
 It is commonly considered that solid-phase Mn enrichments in sediments result from the contact of oxic/anoxic interfaces in the water column with bottom topography [e.g., Force and Cannon, 1988; Frakes and Bolton, 1992; Calvert and Pedersen, 1996]. The accumulation of fine particulate MnOx precipitated in the water column also results in variable localized Mn enrichments. Two independent lines of evidence suggest that it is likely that during S1 formation eastern Mediterranean surface sediments in deeper water had a lower redox potential than sediments in shallower water. Murat and Got  found that the Corg of S1 units increased systematically between water depths of 500 and 4000 m and inferred that deeper sediments must have experienced lower mean bottom water O2 levels than shallower sediments. Casford et al.  noted that benthic foraminifera occur throughout S1 sapropel units in several cores retrieved from water depths <2000 m, and inferred that bottom waters must have been dysoxic or at least intermittently oxic.
 The combination of less intense or less frequent bottom water ventilation and higher preserved Corg contents in the deepest sediments during sapropel formation would be expected to lead to anoxia in the overlying bottom waters. Anoxia would consequently allow for the loss of considerable amounts of Mn2+ from the sediments to bottom waters and would promote sulfide formation. The presence of high O2 values higher in the water column, either continuously or intermittently during S1 formation would be expected to precipitate this Mn2+ where and when oxic and anoxic waters met and mixed.
 In the present anoxic water column of the Black Sea, the highest dissolved Mn2+ contents occur in the few hundred meters below the redox interface that is located at ∼200 m [e.g., Lewis and Landing, 1991]. This results from precipitation of MnOx at the oxic/anoxic redox interface in the water column in response to mixing and diffusion, and from the continuous dissolution of precipitated MnOx as it sinks back down into the deeper, anoxic sulfidic water column. Kempe et al.  reported that a fine particle layer (FPL) with a large increase in dissolved Mn2+ below it follows the pycnocline in the Black Sea. This FPL also has higher concentrations of particulate manganese toward the shelf where MnOx deposited in the sediment is exposed to reducing bottom waters. A similar process should be expected for the eastern Mediterranean during S1 times, but the geometry of the oxic/anoxic interface must have been quite different. Some level of ventilation to considerable depth is required to introduce O2 at depth to react with deep anoxic waters containing Mn2+ and form MnOx.
 Differences in oxygenation level with water depth during sapropel S1 formation thus readily provided a mechanism through which both a preferential loss of Mn2+ from the deepest sediments and a preferential deposition of MnOx on higher topography in the basin might have occurred. These differences in the mean oxygenation levels must have occurred throughout the time of S1 formation as indicated by the distribution of benthic foraminifera and the increasing Corg contents [Casford et al., 2003; Murat and Got, 2000]. A single basin-wide ventilation event at the end of S1 times could not have caused the Mn distributions now observed in the sediments, nor produced the major upper Mn zone. However, the record of multiple hydrogenetically formed inputs of MnOx to midwater depth (∼1000–2000 m) sediments would not be preserved as spikes in Mn content in S1 sediments, because they would be dissolved by anoxic pore waters in the sapropel and concentrated into a single surficial diagenetic MnOx layer. This process is exemplified by the upper narrow (∼4 cm) strongly Mn-enriched zone in core ODP971C that sits at the top of the thick, largely redeposited S1 unit. This narrow upper Mn zone is consistent with the redistribution of Mn as MnOx from within that unusually thick (redeposited) sapropel, which implies that bottom waters must have been oxic to some degree at the time of its formation late in S1 times.
 A one-dimensional, reactive transport model was used to evaluate our genetic model for the enrichment of Mn above S1 sediments in the eastern Mediterranean. First, S1 conditions were simulated by assuming that high dissolved Mn2+ and low dissolved O2 concentrations (Table 3) existed at the sediment-water interface for a period of about 4100 years (Figure 8a). The results of this S1 model run were employed as the initial conditions for the post-S1 model runs. The first post-S1 model run tested if the upper high-Mn zone might have formed purely by diagenesis within the sediment. In order to simulate the shift from high Mn2+ concentrations during S1 to present-day conditions in the post-S1 model, an exponential function was used that “slowly” decreased the Mn flux at the lower boundary according to
where [Mn S1] and [Mn PD] is the concentration of Mn2+ during S1 (results of the S1 model run) and present-day, respectively.
 This model run simulates the possibility of MnOx precipitation within the sediment at the transition from S1 to recent conditions, assuming fairly high dissolved Mn concentrations during S1 and a sudden shift to high oxygen and low Mn2+ levels at the sediment surface. The model results after 1000, 3000, and 5500 year simulations (Figures 8b–8d) show that pure sediment diagenesis could not have produced the high Mn values that have been found above the S1 horizon in sediments between 1000 and 2000 m water depth in the eastern Mediterranean.
 To simulate an exceptional situation similar to that favored in this paper, where the common interplay of increasing O2 and decreasing Mn2+ in the bottom water is suppressed, a second post-S1 model run was adapted. This model run assumes a “slow” increase of O2 accompanied by a slow decrease of Mn2+ over time at the sediment surface (thus in the bottom water; upper boundary conditions) by implementing the following exponential equations:
where [O2 S1] and [O2] are the corresponding concentration during S1 times (results of the S1 model run) and present-day at the sediment surface (Table 3). Since the recent eastern Mediterranean is a well-ventilated basin, the Mn2+ background concentration has been taken from the uppermost pore water sample (0.25 cm) of core BP18 at 0.00005 mmol dm−3.
 This model run considers slowly decreasing Mn2+ concentrations as lower and upper boundary conditions as well as slowly increasing O2 concentrations within the water column (upper boundary conditions) at the transition from S1 to recent time and consequently initiates enhanced MnOx precipitation. The results after 1000, 3000, and 5500 year simulations produce an extreme MnOx enrichment with the magnitude depending on the exponential factor in the exponential equations (Figures 8e–8g).
 Consequently, these model results confirm that the Mn in the upper zone is mainly diagenetic, even though it initially formed at intermediate water depth sites as a hydrogenetic precipitate. Indeed, the unusual strongly Mn-enriched upper zone occurs at and above the water depth (∼2000 m), at which mud mounds are found in the eastern Mediterranean [Robertson and Kopf, 1998; Kopf et al., 2000]. These mounds emit CH4 and possibly Mn2+, but a Mn2+ contribution from this source would be expected to oxidize to fine particulate MnOx in the water column that would behave similarly to the hydrogenetic MnOx just discussed [Klinkhammer and Hudson, 1986].
 A corollary of the mechanism envisaged is that sediments from the deepest parts of the basin are expected to have encountered the most marked sediment and bottom water anoxia and therefore to have lost the most Mn. All published profiles from cores at >2500 m water depth exhibit upper Mn zone contents of <1 wt % (Figure 7), for example the deep-water core SL114 (Figure 2 and Table 1). On the other hand, higher mean O2 concentrations in bottom waters at shallower water sites (cores BP15, T87-29B, MD90-917; Figure 2, 5, 6, and 7) can account for the occurrence of the upper Mn zone within, rather than overlying the S1 unit. Higher bottom water O2 concentrations at these sites during sapropel formation would have resulted in deeper penetration of oxygen into the sediment. The available Mn in the sediments would then have been diagenetically enriched in this deeper oxic layer, within the sapropel, rather than at the very top of the sapropel.
 A set of cores with large (>3 wt %) Mn contents in the uppermost of two Mn-enriched horizons that are usually observed above sapropel S1 have been examined in the light of other S1 studies. Although mostly equivalent in time to the smaller upper Mn zone, the magnitude of the large Mn zone appears variable within particular areas. The wide variation in Mn content does not appear to be reflected in other geochemical parameters apart from elements associated with the MnOx. Our preferred scenario involves a diagenetic release of Mn2+ from the deepest water (>2000 m) sediments into anoxic bottom waters, a reprecipitation of this Mn2+ as particulate MnOx where it encounters O2 higher in the water column, and a removal of the particulate MnOx onto higher topography underlying oxic waters. The precipitation step occurs when anoxic and oxic waters meet at intermediate water depths (1000–2000 m). This mechanism is consistent with higher mean O2 levels at intermediate depths than at the deepest depths in the basin, which is also indicated by micropaleontological and Corg evidence. This situation must also have occurred regularly during sapropel formation rather than in a single event at the end of the S1 period, because benthic foraminifera and lower Corg contents are found throughout S1 units formed at shallower water depths. The combination of pore water anoxia in sapropel surface sediments and some level of oxygen in the overlying bottom water would maintain the excess Mn introduced by this hydrogenetic mechanism as a large diagenetic zone in sediments at midwater depth (∼1000–2000 m). Model results indicate that this extreme Mn enrichment can form only by slowly decreasing Mn2+ and slowly increasing O2 supplies at the sediment surface. This emphasizes MnOx formation within the water column or at the sediment/water boundary. Molybdenum and lithium are incorporated into this diagenetic MnOx at Mn:Mo and Mn:Li ratios that vary near 600:1 and 750:1, and with a δ98/95MoMOMO ratio of ∼−2.5 ‰. These ratios and isotopic composition support our preferred scenario.
 We are grateful to Sebastian Meier (Natural History Museum, London), Simon Troelstra (Free University, Amsterdam), Martine Paterne (LSCE, Gif-sur-Yvette), and Eelco Rohling (SOC) for the provision of samples (cores MC562, T87-29B, MD90-917, and ODP971C, respectively). Two anonymous reviewers are thanked for their valuable comments. Robin Keir is gratefully acknowledged for constructive comments and thorough editing of the text. Caroline Slomp is thanked for discussions on SAP project data. We thank Darryl Green, Thomas Appel, Helen de Waard, Eric van Vilsteren, and Rinske Knoop for their contribution to laboratory analysis and the crew and the shipboard parties of the cruises named in Table 1 for their contribution to sample collection. Thomas Nägler (University Bern) and Christoph Vogt (University of Bremen) are acknowledged for the molybdenum isotope and XRD measurements, respectively. This work was partly funded by EU-MAST-III project MAS3-CT97-0137, Sapropels and Paleoproductivity (SAP); NWO project Sapropel-related Paleoceanographic Studies in Sediments of the eastern Mediterranean (SAPS). This is NSG contribution 2006 0987).