Sedimentary Trace Elements as Proxies to Depositional Changes Induced by a Holocene Fresh-Brackish Water Transition

Aquatic Geochemistry6: 325–345, 2000.© 2000Kluwer Academic Publishers. Printed in the Netherlands.

325

Sedimentary Trace Elements as Proxies toDepositional Changes Induced by a HoloceneFresh-Brackish Water Transition

JOHN STERNBECK1, GUSTAV SOHLENIUS2 and ROLF O. HALLBERG3

1Swedish Environmental Research Institute (IVL), Box 21060, S-100 31 Stockholm, Sweden;2Swedish University of Agriculture Sciences, Department of Soil Sciences, Box 7014, S-750 07Uppsala, Sweden;3Department of Geology and Geochemistry, Stockholm University, S-106 91Stockholm, Sweden

(Received: 8 July 1999; accepted: 17 February 2000)

Abstract. A halocline developed in the Gotland Deep, Baltic Sea, at c. 8.014C ky BP, as the result ofa transition from fresh to brackish water. The sediment-water interface changed from oxic to predom-inantly anoxic, depositional conditions were periodically euxinic and pyrite formation was extensive.This environmental change led to pyritization of the upper part of earlier deposited sediments. Thisstudy discusses how the distribution of trace elements (As, Ba, Cd, Cu, Co, Mo, Mn, Ni, Pb, U,Zn and V) were affected by the changing redox conditions, productivity and salinity. The reducingconditions led to pyritization of Cu, Co, Ni, Cd, Mo, Mn and As. Lead and Zn concentrations arevery low in pyrite, in agreement with their coordination to sulfide being tetrahedral. Certain elementsare enriched in those sediments deposited under euxinic conditions. This enrichment was causedby scavenging of elements dissolved in the water column and is restricted to elements that have acomparably long residence time in the Baltic Sea. Molybdenum appears to be the most unambigiousproxy for euxinic conditions, whereas enrichment of U also requires brackish water in the productivezone. In the brackish environment, enrichment of Ba and V are linked to the cycling of organiccarbon. Manganese and As are the only elements that have been significantly remobilised due to thedownward moving pyritization front.

Key words: Baltic Sea, euxinic, Holocene, pyrite, trace elements, redox conditions, sediments

1. Introduction

Due to the global eustatic sealevel rise during the latest deglaciation, marine waterentered several silled basins. The environment shifted relatively rapidly from lacus-trine to brackish or marine, and the vertical water circulation became restricted dueto development of haloclines. This resulted in a change in the depositional condi-tions, from oxic to anoxic or even euxinic, that has been documented from, e.g., theBlack Sea and the Kau Bay (Middelburg et al., 1991) and the Baltic Sea (Boesenand Postma, 1988; Sohlenius et al., 1996). These changes may have affected thecycling of trace elements, because elements such as Cu, Zn, Mo, U and V have

326 JOHN STERNBECK ET AL.

frequently shown to be enriched in sediments deposited under anoxic conditions(e.g., Calvert and Pedersen, 1993).

There are several different mechanisms by which trace elements may be en-riched in anoxic sediments. Certain trace elements (e.g., Cu, Pb and Zn) areefficiently scavenged by sulfide, either forming discrete minerals (Luther et al.,1980; Skei et al., 1996) or by coprecipitating in iron sulfide minerals. Huerta-Diazand Morse (1992) showed that the affinity for pyrite differs widely for differ-ent trace elements. In addition to redox conditions, other parameters can also beresponsible for enrichment of trace elements in sediments, e.g., high content oforganic material (U, Cu, Cr, Mo, V and Ba; Anderson et al., 1989a,b; Shaw et al.,1990; Breit and Wanty, 1991; Thomson et al., 1995). High primary productivityis generally a prerequisite for the development of reducing conditions and anoxicsediments usually display high organic carbon content. Consequently, the cause of,e.g., U, Mo, Cu and V enrichments in such sediments has been debated (Andersonet al., 1989a,b; Breit and Wanty, 1991; Piper and Isaacs, 1996). The absence of en-richment of Cd, Cu, Ni, Zn, V, Mo and U was, however, suggested as evidence foroxic bottom water conditions during sediment accumulation (Calvert and Pedersen,1993).

A basic prerequisite for enrichment of a trace element in sediments, not alwaysexplicitly stated, is that the element in concern must be available in excess. Thisrequirement may not be fulfilled in regions with very restricted water exchangeand high particle load. This is the case in the present Baltic Sea where the overallresidence time of water is about 25 years whereas that of many trace elementsis only a few years (Brügmann, 1986; Swedish EPA, 1991). In principle, a shiftfrom oxic to anoxic depositional conditions should then not affect trace elementconcentrations in sediments except during a short transitional period.

When the depositional environment has changed abruptly, non-steady state dia-genesis may erase primary diagenetic fingerprints several hundred or thousandyears after sediment deposition. For example, pronounced redistribution of traceelements from an early Holocene sapropel as a result of a downward movingoxidation front has been demonstrated in sediments from the Mediterranean Sea(Thomson et al., 1995). The opposite process, i.e. sulfidization of sediments origin-ally deposited in an oxic environment, has been documented from the Baltic Sea(Boesen and Postma, 1988; Sternbeck and Sohlenius, 1997) the Kau Bay (Mid-delburg, 1991) and the Black Sea (Leventhal, 1995). However, the effect of thisprocess on the redistribution of trace elements seems not to have been addressed.

This paper discusses how certain trace elements were affected by a well-documented Holocene change from oxic to periodically euxinic conditions (anoxicwater column) in the Gotland Deep of the Baltic Sea (Figure 1). Of special interestis the effect of a sulfide front moving downward from reducing, brackish sedimentsinto sediments that were deposited in an oxic, freshwater environment (Sohlenius etal., 1996; Sternbeck and Sohlenius, 1997). The main questions are: (1) Are certainelements reliable indicators of the bottom water redox conditions or the primary

SEDIMENTARY TRACE ELEMENTS AS PROXIES 327

Figure 1. The Baltic Sea with the sampling site in the Gotland Deep marked +. For the last800014C years there has been a water exchange with the North Sea through the Danish Straits.

productivity of the depositional environment? (2) Are certain elements more pronethan other to stratigraphical redistribution by a downward moving sulfide front? (3)Will those elements forming insoluble sulfide minerals be enriched in euxinic sed-iments? (4) Does the trace element content in pyrite depend on whether the pyriteformed synsedimentary and possibly in the water column, or postsedimentary?

2. The Study Site

At present, the Baltic Sea is a semi-enclosed stratified brackish water basin (Fig-ure 1) with salinity decreasing northwards. At the Gotland Deep, salinity is∼7and∼12h in the surface and bottom water, respectively. Water exchange with theocean occurs through the shallow Danish Straits (at present 10-20 m deep). Thewater in the deepest parts of the basin is only occasionally exchanged and euxinicconditions can therefore establish and prevail for several years. Since the latestdeglaciation the Baltic Sea has experienced both fresh- and brackish water stages(Winterhalter et al., 1981). During the Ancylus Lake stage (9.6–8.014C ky BP)the Baltic Sea was isolated from the Sea and freshwater conditions prevailed. Atthe transition from the Ancylus Lake to the Litorina Sea (A-L transition), 8.014Cky BP (Hyvärinen et al., 1988), marine water entered the Baltic Sea through theDanish Straits and since then brackish water conditions have prevailed.


Figure 2. Lithology and stratigraphical units of the investigated sediments. Salinity, redoxconditions and14C dates are from Sohlenius et al. (1996) and pyrite formation from Stern-beck and Sohlenius (1997). HCl-soluble Fe (�) and6Fe ( ) are given for reference to traceelement data.

The trace element data presented in this paper derive from analyses of asediment core from the Gotland Deep. This sediment core has previously been in-vestigated regarding stratigraphy, redox conditions, salinity and authigenic mineralformation (Sohlenius et al., 1996; Sternbeck and Sohlenius, 1997).The lowermostpart of unit I was deposited in the brackish Yoldia Sea, unit II and III in the fresh-water Ancylus Lake and finally unit IV in the brackish Litorina Sea (Figure 2).Salinity increased progressively during deposition of unit IVa, and surface waterswere not fully brackish until unit IVb.

In these papers it is concluded that the sediment surface in the Gotland Deepwas oxic before Litorina time, and that ferric oxides are probably preserved inthese sediments. After the A-L transition (unit III/IV) a halocline developed. Re-stricted bottom water mixing led to a rising redoxcline and euxinic conditionsestablished during the stagnant periods. Although inflows of dense marine waterfrequently perturbed this situation by oxygenating the bottom water, the sedimentsfrom the Litorina Sea are mainly laminated indicating an anoxic sediment surface.The occurrence of benthic foraminiferas at certain levels shows, however, that thesediment-water interface was oxic during short periods (Sohlenius et al., 1998).


During the preceding freshwater stages (units I-III) FeS and FeS2 formed atroughly equal, low, net rates. In the Litorina Sea pyrite formation was extensive,limited by iron and took place at the uppermost sediment or in the water column.This allowed H2S to diffuse downward into the uppermost freshwater sediments,where iron oxides were pyritized. This pyritization has, up till now, affected theupper∼0.5 m of the freshwater sediments which constitutes unit III. Throughoutthe Litorina sediments, a Ca-rich rhodochrosite (MnCO3) occurs as thin whitelaminae. These carbonate minerals were interpreted to have formed in connec-tion with major marine inflows, when the bottom water became oxygenated andMn oxides accumulated at the sediment surface (Huckriede and Meischner, 1996;Sternbeck and Sohlenius, 1997). The presence of MnCO3 is thus, in this area, astrong indicator of alternating euxinic-oxic conditions.

3. Methods

Details on sampling, core processing, and analyses of FeS—S, FeS2—S, and or-ganic carbon are described in Sternbeck and Sohlenius (1997). Trace elements wereleached by a slight modification of a method specially developed to study tracemetals in pyrite (Huerta-Diaz and Morse, 1990). Freeze-dried sediment (∼0.5 g)was leached in 25 ml 1M HCl for 24 hours at room temperature and one hour at 90◦C. By leaching freshly prepared Mn oxides it was found that the higher temper-ature was necessary for their dissolution. The residual from the HCl-leaching wastreated with conc. H2SO4 which quantitatively removes organic matter (Huerta-Diaz and Morse, 1990). Organic matter may otherwise have contributed withmetals to the pyrite fraction. Finally, pyrite-bound trace elements were extractedfrom the residue with conc. HNO3. However, the HF step to remove aluminosilic-ates was omitted. Although HCl and H2SO4 should have leached the most reactivealuminosilicate fractions, it can not be excluded that aluminosilicate-bound traceelements may have been extracted with the HNO3 used to extract pyrite. Results ofKheboian and Bauer (1987) indicate, however, that dilute HNO3 does not attack theillite lattice. Furthermore, the low levels of HNO3-metals in the pyrite-poor units Iand II suggest that this artifact should be of minor importance.

It is possible that certain elements form discrete trace metal sulfides ratherthan substituting for Fe in pyrite. The behavior of possible discrete trace metalsulfides (e.g., MoS2, CuS) during acid leaching is poorly known. However, Cooperand Morse (1998) show that CdS, PbS and ZnS are easily dissolved by 1M HCl,whereas CuS and NiS seem to dissolve slowly. However, except for the sporadicoccurrence of MnS in the Landsort Deep (Suess, 1979) and the Gotland Deep(Böttcher and Huckriede, 1997) we have found no evidence in the literature ofdiscrete trace metal sulfides in the Baltic Sea.

Barium was measured in the HCl fraction, and As, Cd, Cu, Co, Mo, Mn, Ni,Pb, U, Zn and V were measured in both fractions. The elements were analysedwith either flame-AAS, GF-AAS, ICP-AES or ICP-MS. All measurements were


performed in duplicate or triplicate with deviations typically less than 5%. CoHCl

and CuHCl were determined both by GF-AAS and ICP-MS and the discrepanciesare less than 10%. Elements will be denoted MeHCl and Mepyr and are presented ona dry weight basis. Although MeHNO3 would seem more appropriate than Mepyr, itwould misleadingly indicate that the organic matter fraction is included. The sumof the two leachates will be referred to as6Me. HCl-reactive Fe minus FeS-Fe(calculated from AVS-S) will be referred to as HCl-reactive iron (FeR).

4. Results

The distribution and formation of rhodochrosite, FeS and FeS2 are discussed inSternbeck and Sohlenius (1997) and briefly in this introduction. The pyrite andFeHCl contents are shown in Figure 2. All trace element contents are presented vs.sediment depth in Figures 3 and 4. On the average,6Me concentrations of Co, Cu,Fe, Mo, Ni, and V are similar to results from total digestions of sediments fromthe same region (G. Sohlenius, in preparation). This indicates that the HCl-reactiveplus the pyrite fractions constitute a significant part of the sedimentary contentof these metals. For Cd and Zn,6Me concentrations are slightly lower than totaldigestions.

Of the divalent transition metals only Cd shows a large increase in contentupwards in the core. This increase occurs both in the HCl and pyrite fractions.Lead decreases slightly upwards, Zn and Ni show no clear trend with depth, andCu and Co show a slight increase in unit IV. All those elements occurring as oxy-anions under oxic sea water conditions (As, Mo, U, V), as well as Ba, exhibit largeincreases in unit IV. The As content is actually higher in both unit III and IV thanbelow. Arsenic and Mo are about equally partitioned between the HCl and pyritefractions, whereas U and V are mainly recovered in the HCl fraction. The HClfraction of Co, Fe, Ni, Pb, V and Zn decrease upwards from unit I to III. Both Vand BaHCl covary with organic C in unit IVb (Figure 4) but not in the other units.

The role of pyrite in trace element incorporation is illustrated by the degree oftrace metal pyritization (DTMP; Huerta-Diaz and Morse, 1992):

DTMP = MePYR/(MePYR + MeHCl). (1)

This parameter is comparable to DOP for Fe (Raiswell et al., 1988; Canfieldet al., 1992). It is apparent that pyrite is an important sink especially for Cu, withDTMP-Cu values as high as 0.92 in unit IV (Figure 5). Also Co, Ni, Cd, As andMo are to a significant extent found in the pyrite fraction, but Zn and Pb are not(Figure 5). Uranium and V show negligible concentrations in pyrite. The pyrites inunits III and IV are rich in Mn (average Mn/Fe ratio is 7.5%), but6Mn in unit IVis dominated by the high rhodochrosite concentration.


Figure 3. The concentrations of Ni, Cu, Co, Cd, Pb, Zn, U, Mo, V, Mn and As vs. sedimentdepth. The HCl-soluble fraction is represented as (#) and the sum of the HCl-soluble fractionand the pyrite fraction,6Me, as ( ). The Pyrite fraction is the difference between the twocurves. The stratigraphical units are given in Roman numerals.


Figure 4. Organic carbon and the HCl-soluble fraction of Ba and V. The stratigraphical unitsare given in Roman numerals. To allow easy comparison between Ba, V and organic C,organic C is only marked with a symbol in those samples where Ba and V were measured.The complete organic C profile is represented by the dashed line.

5. Discussion

Before discussing the different elements in detail, it is instructive to outline thegeneral physical and biogeochemical factors that may have affected the cyclingof trace elements in the Baltic Sea during Holocene. The allochthonous mineralassemblage was studied in a number of Baltic Sea stations by Gingele and Leipe(1997). In the Gotland Deep, this assemblage showed very little variation through-out postglacial time and should thus not affect the trace metal profiles. Studies inthe present Gotland Deep have clearly shown that depositional conditions exert amajor influence on sedimentary trace metal concentrations, which thus vary widelyalong a depth-transect (Manheim, 1961).

Primary productivity was probably higher in the Litorina Sea (unit IV) than inthe preceding postglacial stages (Sohlenius et al., 1996). The flux to the sedimentsof elements that associate with biota may thus have increased. Such a mechanismwas invoked by Piper (1994) to explain enrichments of several trace elements inmarine sediments. Trace elements in organic matter have not been analysed in thisstudy, but the fraction adsorbed to organic matter is recovered in the HCl fraction.Furthermore, elements reaching the sediments with organic matter may have beenretained in the sediments following mineralisation.

The fact that salinity was considerably higher in the Litorina Sea than in theAncylus Lake may have affected the concentration of trace elements in at least two


Figure 5. Degree of trace metal pyritization vs. degree of pyritization for A. Co and Ni; B. Cuand Zn; C. Pb and Cd; D. As and Mo.

ways: (1) Salinity gradients emerged at the river outlets, leading to flocculationor desorption of certain trace elements. This question seems not to have been in-vestigated thoroughly in the Baltic Sea. However, there is some evidence from thesouthern Baltic Sea that Cu, Ni and Zn are largely unaffected by estuarine mixingwhereas Fe flocculates and settles (Brügmann, 1986). This observation agrees withresults from other estuaries such as the Mississippi River Delta, where Cu, Ni, Znand also Mo were conservative (Shiller and Boyle, 1991); (2) The concentrationof a few trace elements are higher in sea water than in most freshwaters. Of thetrace elements in this study this concerns Ba, U, Mo and possibly V. Whether theeffect of increasing salinity influenced the sediment content of these elements isdiscussed later on.

5.1. Ba AND V

Barium has shown to be a good indicator of paleoproductivity in a number ofmarine sediments (e.g., Dymond et al., 1992; Thomson et al., 1995) although thequantitative significance is questioned (McManus et al., 1998). The mechanismfor Ba uptake is generally considered to be the precipitation of barite (BaSO4) onbiogenic surfaces (e.g., Bishop, 1988) and, therefore, this process may not occur in


waters of low salinity where barite is strongly undersaturated. Hence, the lack ofa Ba-organic C correlation below unit IVb (Figure 4) can be explained by the lowsalinity of the productive zone (the surface waters) in the early Litorina Sea (unitIVa), as well as of the entire water column in the earlier stages (Sohlenius et al.,1996).

The strong correlation between Ba and organic C in unit IVb (Figure 4) may ap-pear slightly surprising since the surface sediments were strongly sulfate-reducingat time of deposition and it could be expected that barite should dissolve in thesediments as a result of porewater sulfate depletion (Dymond et al., 1992). Sulfatereduction rates were obviously not rapid enough to deplete sulfate levels in theporewaters (Boesen and Postma, 1988). Barite has also shown to be preserved inother sulfide-rich sediments, (e.g., Thomson et al., 1995) and the present studyseems to be another example contradicting the reasonable arguments of Dymondet al. (1992). It appears thus that Ba may be a good tracer for paleoproductivity inthe Baltic Sea in sediments younger than unit IVa. This may help to differentiatethe effects of primary production and anoxia on the organic C profile in sediments.

The VHCl profile is very similar to that of BaHCl in unit IVb, and thus also toorganic carbon (Figure 4). Previous studies have shown that the accumulation of Vin marine sediments is linked to the flux of organic carbon (Shaw et al., 1990; Breitand Wanty, 1991; Shiller and Boyle, 1991). It was also suggested for the BalticSea that V in the water column is controlled by primary productivity (Boström etal., 1981; Prange and Kremling, 1985). The lack of a VHCl-organic C correlation inunits I-III may be explained by a combination of the lower organic matter contentand the presence of ferric oxides contributing with V to this leaching fraction (Tre-fry and Metz, 1989; Balistrieri et al., 1994). A better VHCl-organic C correlationcould have been expected in unit IVa. It is, however, mainly marine organisms thathave been reported to accumulate V (Rehder, 1992) and salinity of the productivezone was very low during deposition of unit IVa.

5.2. THE DIVALENT ELEMENTS: Cd, Co, Cu, Mn, Ni, Pb AND Zn

The DTMP values for Co and Ni correlate to DOP (slope 0.91 and 0.78 respect-ively, Figure 5), indicating that Co and Ni behave similarly to Fe during earlydiagenesis, and that the supply of Co and Ni to the sediments are associated withthe supply of Fe. It is commonly observed that Co and Ni substitute for Fe inpyrite (Morse et al., 1987) and a close to 1 : 1 relationship between DTMP-Co/Niand DOP was also found in a large number of coastal sediments in the Gulf ofMexico (Huerta-Diaz and Morse, 1992). We suggest that Co(II) and Ni(II) substi-tute in pyrite better than most other divalent metals investigated since they bothhave octahedral coordination (like Fe in pyrite) with sulfide ligands (Cotton andWilkinson, 1988) and since their ionic radii are very similar to that of Fe(II).

Several studies have shown enrichments of Cu in anoxically deposited BalticSea sediments (Hallberg, 1974; Sohlenius and Westman, 1998) and, in this study,


the6Cu concentration is slightly higher in the laminated sediments of unit IVthan below. However, the speciation of Cu is more strongly affected by the euxinicconditions with Cupyr forming at the expense of CuHCl (Figure 3). The Cupyr in unitIII is also formed at the expense of CuHCl, but to a lesser extent. This pyritization ofCu is most likely to have occurred synchronously with the postdepositional pyriteformation. Copper generally shows a higher degree of pyritization than Fe and theDTMP-DOP correlation is not as good as in the case of Co and Ni (Figure 5). Itappears that Cu is not as closely related to Fe as Co and Ni are. The coordinationnumber of Cu with sulfide is generally three or four (e.g., Pattrick et al., 1997) andit is thus not obvious that Cu actually occurs in pyrite. Cooper and Morse (1998)recently showed that CuS is only sparingly soluble in HCl and, if present, shouldthen be recovered in the pyrite fraction. The true nature of the Cu-S phase in thesesediments remains to be proved with other analytical techniques.

The concentration of6Cd increases gradually from unit I to the unit III/IVtransition and occurs at high, varying levels in unit IV. The Cd molar% in pyriteis significantly higher in the synsedimentary pyrite (4.5± 2.5× 10−4; unit IV)relative to the postdepositionally formed pyrite (1.2± 0.2× 10−4; unit III). Thefairly high DTMP values for Cd (Figure 5) are in contrast to, e.g., Huerta-Diazand Morse (1992). Cadmium favors a tetrahedral coordination but also forms 6-coordinate octahedral coordination (Greenwood and Earnshaw, 1984) and can thusbe expected to have a higher affinity to pyrite than Zn. CdPyr did not form at theexpense of CdHCl, in contrast to Co, Ni, Cu and As which all were transformedfrom the HCl-reactive phase to pyrite. Obviously CdHCl is not readily availablefor pyritization, and these observations suggest an external Cd source, unrelated toallochthonous particles. Cadmium dissolved in the water column may either havebeen scavenged efficiently under euxinic conditions or been assimilated by biotain the productive zone. Actually,6Cd correlates with organic carbon throughoutthe core (Figure 6). The Cd-organic matter association is well-known from theoceans, and Cd is the only metal for which a strong negative coupling betweenconcentration in surface waters and primary productivity has been demonstratedin the Baltic Sea (Schneider and Pohl, 1996). Overall the geochemical cycling ofCd appears to be more strongly related to C and S than to Fe, which agrees withthe low of affinity of Cd for natural inorganic particles such as ferric oxides (e.g.,O’day et al., 1998).

Lead shows no pronounced variations throughout the core, which contrasts witha number of studies showing that dissolved Pb is efficiently scavenged below theredoxcline in euxinic environments (e.g., Lewis and Landing, 1992). Because Pbis very particle-reactive, its residence time in aquatic environments is shorter thanfor most other divalent metals, and has been estimated to a few months in thepresent Baltic Sea (Brügmann, 1986; Swedish EPA, 1991). This indicates thatPb is strongly retained in the Baltic Sea and that the pool of Pb in the watercolumn is fairly low. Therefore, a shift from oxic to euxinic conditions may leadto a more rapid turnover of water column Pb, but the flux to sediments cannot


Figure 6. Correlation between6Cd and organic C (all samples).

increase (except for very temporary) since Pb is completely retained even underoxic conditions. Zinc does not either show variations that appear related to theredox conditions, which may be explained in the same way as for Pb. NeitherPb nor Zn is incorporated into pyrite to a significant extent, in accordance witha previous study (Huerta-Diaz and Morse, 1992). This may be explained by thefact that Pb and Zn have tetrahedral coordination with sulfide ligands (Cotton andWilkinson, 1988) and thus do not fit into the pyrite lattice. Possibly, these metalsoccur as PbS and ZnS which should be recovered in the HCl fraction (Cooper andMorse, 1998).

Manganese sulfide is generally considered to be of low significance in sediment-ary Mn cycling, although sporadic occurrence of MnS is reported from the BalticDeeps (Suess, 1979; Böttcher and Huckriede, 1997). However, because Mn(II) hasoctahedral coordination to sulfide and an ionic radius similar to Fe(II), it could beexpected to substitute readily for Fe in pyrite. The pyrites in units III and IV sys-tematically show significantly higher Mn/Fe ratios (3–16 molar%; average 7.5%)than most previous studies (see Huerta-Diaz and Morse, 1992). SEM-EDS analysisconfirm these high ratios in individual pyrite crystals and do not show any discreteMnS phases. However, due to the abundance of rhodochrosite (MnCO3) in unitIV (Sternbeck and Sohlenius, 1997), pyrite is not a quantitatively important sinkfor Mn in this unit. The results indicate that coprecipitation of Mn in pyrite maybe the main authigenic Mn phase in those sulfidic sediments where rhodochrositedoes not form. Because reduction of Mn oxides (production of Mn(II)) occurs at ahigher redox potential than do reduction of ferric oxides or sulfate, pyritization of


Mn may be less important in sediments where the different electron acceptors arewell separated vertically. Mn(II) will then diffuse out of the sediment rather thancoprecipitate with the pyrite forming at greater sediment depth.

5.3. THE OXYANIONIC ELEMENTS: Mo, U AND As

Both6Mo and6U are highly enriched in major parts of unit IV relative to theunderlying units (Figure 3). It has previously been shown that U and Mo are en-riched in anoxic sediments from the present Baltic Sea (Manheim, 1961; Hallberg,1974; Sohlenius and Westman, 1998) and elsewhere (e.g., Barnes and Cochran,1993; Calvert and Pedersen, 1993; Piper and Isaacs, 1996). However, Mo and Uare also strongly correlated to salinity in the present Baltic Sea waters (Prange andKremling, 1985), which indicates that the water column concentrations of theseelements increased progressively following the A–L transition. The initial increasewas restricted to the bottom waters, because salinity gradually increased upwardsfrom the deepest parts. Thus, there is a possibility that the enrichment of U andMo in unit IV solely is due to increased adsorption as a result of the higher watercolumn concentrations of U and Mo during the more saline conditions. In moredetail, it is clear that the elevations of Mo and U start above the depth (unit III/IVtransition) where the salinity of the bottom water started to increase (Figure 7).

Furthermore, the burial rate of Mo in unit IV (average 0.3µmol dm−2 y−1) isvery similar to the settling rate of particulate Mo in the hypolimnion of a seasonallyanoxic freshwater lake (Magyar et al., 1993). This indicates that the availability ofMo in lacustrine environments, such as the Ancylus Lake (units II-III), should nothave been limiting for enrichment in the sediments if the conditions had been eu-xinic. The lower part of unit I was also deposited in a brackish water environment.Only one sample from this section was analysed for trace elements, but showed noenrichment of either U or Mo. We hold these facts as evidence of that sedimentaryenrichment of these elements is not primarily controlled by salinity. The details ofthe U and Mo profiles will be discussed in the following.

Molybdenum is strongly pyritized in unit IV but the DTMP values show no rela-tion to the DOP values (Figure 5). Several pathways leading to the incorporation ofMo in anoxic sediments have been suggested: e.g., (1) diffusion over the sediment-water interface and precipitation within the sediments (Emerson and Huested,1991); (2) ligand exchange with HS− in a euxinic water column to form thiomolyb-dates which are adsorbed on settling organic matter or organic Fe particles (Helz etal., 1996). In the present Gotland Deep large amounts of Mo were coprecipitatedwith Mn oxides, forming as a result of a major oxic marine water inflow (Brügmannet al., 1998). This is probably one major mechanism for Mo transport to thesesediments. Following reduction of the Mn oxides at the sediment surface (to formMnCO3; Sternbeck and Sohlenius, 1997), the molybdate ion may be convertedto the more particle reactive thiomolybdate ion (MoS2−

4 ) which is retained in the


Figure 7. HCl-soluble Mo, U and Mn at the A-L transition. The entire core depth is notdisplayed.

sediments (Helz et al., 1996). This hypothesis explains why MoHCl increases at thesame depth as rhodochrosite starts to appear in the sediments (Figure 7).

It is well-known that U is enriched in sediments deposited under reducingconditions (Klinkhammer and Palmer, 1991; Barnes and Cochran, 1993). The con-centration of6U (essentially UHCl) starts to increase close to the unit IVa/IVbtransition (385 cm) whereas MoHCl and MnHCl increase at 405 cm, i.e., whenconditions became euxinic (Figure 7). Previous studies from anoxic basins haveshown two dominating processes that supply U to surface sediments: (1) uptakeby plankton in the photic zone and subsequent burial and preservation in the sedi-ments; and (2) diffusion over the sediment-water interface and precipitation withinthe sediments (Anderson et al., 1989a,b). Biogenic particles in Baltic Sea waterswere recently shown to be enriched in U (Andersson et al., 1998). Furthermore,U concentrations in Baltic Sea water are strongly positively correlated to salinity(Prange and Kremling, 1985) and the surface water did not become fully brackishuntil the unit IVb (Sohlenius et al., 1996). Thus, enrichment of U in the sediments


Table I. Estimated average residence times (years) fortrace metals in the Baltic Sea. Residence times canbe calculated in several ways and the values must beconsidered as approximate.

As Cd Cu Pb Zn

Brügmann, 1986 – 5.2 3.6 0.3 1.9

Swedish EPA, 1991 17 6 5 0.4 2

was dependent on the facts that (1) the surface waters were brackish; (2) biogenicmatter scavenged U in the photic zone; and (3) the biogenic matter was depositedin a strongly reducing environment. Obviously, reducing bottom waters were notsufficient on their own to lead to the accumulation of authigenic U.

Arsenic is enriched in unit IV suggesting scavenging of dissolved As undereuxinic conditions, or diffusion over the sediment-water interface. However,6Asis also enriched in the postdepositionally formed unit III which suggests that Asmust have been supplied to unit III postdepositionally. This will be discussed innext section. In these units, As is dominated by the pyrite fraction, with DTMPvalues up to 0.75 (Figures 3 and 5). DTMP-As values are not closely related toDOP although they form two clusters with low and high values (Figure 5). Thisindicates that the supply of As to the sediments is unrelated to the supply of Febut both elements are precipitated as pyrite in the sediments. In contrast to theseobservations, pyrite-As and pyrite-Fe were strongly related in sediments originallydeposited in an oxic environment (Belzile, 1988). This would suggest that supplyof As and Fe are related under oxic conditions due to the well-known affinity ofAs for ferric oxides (e.g., Widerlund and Ingri, 1995), and that the supply of theseelements are decoupled under euxinic conditions.

5.4. SOME CONSTRAINTS OF USING TRACE ELEMENTS TO DECIPHER

ORIGINAL DEPOSITIONARY CONDITIONS

If the concentration of a trace element is higher in euxinic sediments, it meansthat the supply of the element increased as conditions shifted from oxic to euxinic.The average residence time for water in the present Baltic Sea is about 25 years,whereas the residence times for many trace elements are much shorter (Table I).There is no reason to suspect that the relative order between these residence timesshould have been different earlier. The short residence times for in particular Pb butalso Zn, in relation to that for water, demonstrate that these elements are efficientlytrapped in the Baltic Sea. Therefore, even if trapping efficiency were to increase,there are no additional amounts of Pb or Zn available that could lead to enrichmentin the sediments.


Table II. Comparison between average elementary accumula-tion rates in unit IV and the dissolved amounts in a 100 m watercolumn (based on present-day concentrations in the open BalticSea, because historical concentrations are not available).

Cd Cu Mo Pb Zn

Average acc. rate in 0.0025 0.15 0.25 0.05 0.35

unit IV, mg/dm2/y

Amount in 100 m 0.015 0.4 3.0 0.03 0.7

water, mg/dm2

Since it has been shown that Mo and U are closely related to salinity in the BalticSea (Prange and Kremling, 1985), it can be assumed that the residence times forMo and U approach that of water. The supply of Mo and U to sediments can thusincrease significantly without necessarily affecting the source strength, agreeingwith the very strong enrichment these elements display in unit IV. The residencetimes of Cd, Cu and As are intermediate and consequently these elements doalso show moderate enrichments under euxinic/high productivity conditions. Theseprinciples are illustrated in Table II, where the sedimentary accumulation rates ofselected elements under euxinic conditions are compared to the amounts availablein dissolved forms in the water column. It is apparent that strongly enriched ele-ments are available in large excess and vice versa. The sedimentary accumulationof trace elements is also supported by the flux of allochtonous particles. The tableillustrates that strong enrichment of certain elements in the euxinic sediments canbe explained by scavenging in the water column.

When using trace elements as paleoceanographic proxys it must be ascertainedthat they have not been redistributed after sediment deposition. Changes in, e.g.,sedimentation rate or primary productivity can lead to postdepositional redistri-bution of elements by oxidation fronts moving downward into reduced sediment(Thomson et al., 1993, 1995). However, the reverse process, i.e., sulfidization ofsediments originally deposited under oxic conditions, seems not have been studiedregarding trace elements.6Mn and6As increase above the unit II/III transition, mainly due to increasing

Mnpyr and Aspyr. This transition is diagenetic due to the late pyritization causedby downward diffusing H2S (Boesen and Postma, 1988; Sternbeck and Sohlenius,1997), and do not represent a change in the original depositional environment: Theincreases of6Mn and6As in unit III thus suggest that Mn and As diffused down-wards from unit IV to coprecipitate with pyrite in unit III. When the reactive Fe at acertain level is pyritized, the rate of pyritization decreases and the pyritization frontprogresses downwards. Manganese and As generally display high solubility undersulfidic conditions (Balistrieri et al., 1994; Widerlund and Ingri, 1995) and may,


therefore, diffuse downwards to keep up with the downward moving pyritizationfront. That is, they are mobile in the presence of6H2S but are strongly retainedwhen pyrite precipitate.

Although the degree of pyritization of Cu, Ni and Co increases above the unitII/III transition, 6Cu,6Ni and6Co are not pronouncedly affected. This suggeststhat pyritization occurred at the expense of metal sources available in the sedi-ments. These metals were thus not added to unit III by diffusion. Molybdenumand U mainly derive from seawater. The solubility of these elements are frequentlyfound to be lower under reducing conditions (Emerson and Huested, 1991; Barnesand Cochran, 1993). Thus, their mobility should be lower in reducing sedimentswhich explains why they are not enriched in the originally oxic sediments that havebeen pyritized. The low content of Mo in units I-III agrees with oxic bottom watersduring sediment deposition (Sohlenius et al., 1996). The low U content does notgive this information because U enrichment could also have been limited by thelow salinity and productivity. Neither are Pb nor Zn significantly affected at theunit II/III transition.

A non-laminated sequence occurs in unit IV around 330 cm depth, implyingless reducing or even oxic depositional conditions. Consistent with our proposalof reducing conditions as the cause of enrichment of Mo, U, Cd and Cu, theseelements display minima in this non-laminated section. Similar variations betweenlaminated and non-laminated have been shown previously in the Baltic Sea (Hall-berg, 1974; Sohlenius and Westman, 1998) and elsewhere (e.g., Piper and Isaacs,1996). Vanadium appears not affected by this event, supporting the suggestion thatV in brackish water sediments primarily is controlled by primary productivity.

6. Conclusions

Because the Baltic Sea has a comparably long hydrologic residence time, elementswith short residence times are not enriched in euxinic sediments even though someof them have shown to be depleted in sulfidic water columns. This is the case forthe divalent transition elements Pb, Zn and Ni. In contrast, Mo and U are stronglyenriched in the euxinic sediments. This is explained by the fact that they are al-most conservative in the Baltic Sea but are insoluble under reducing conditions.However, enrichment of U also appears to require brackish/marine water in theproductive zone. Arsenic and Cu are slightly enriched in the euxinic sediments.

Enrichment of Cd appears related primarily to primary productivity rather thanto euxinic conditions. Barium and V closely follow the organic carbon profilein sediments deposited during a period with a brackish productive zone. Theseelements could thus be used to separate the effects of primary productivity andpreservation under anoxic conditions on the organic carbon content.

Pyrite scavenges As, Cd, Co, Cu, Mn, Mo and Ni efficiently during formation,but only minor amounts of Pb, Zn, V, and U are found in pyrite. For the Me2+metals, this pattern can be explained by the coordination number of the Me2+


ions with sulfide ligands, relative to that of Fe(II) in pyrite. The Cd/Fe ratio inpyrite is higher in the synsedimentary pyrite as compared to the pyrite formedpostdepositionally. The average Mn/Fe molar ratio in pyrite is 7.5% which showsthat pyrite can be an important sink for Mn in anoxic sediments, in addition to thewell-known coprecipitation of Mn in carbonates under anoxic conditions.

Neither Co, Cu, Ni, Mo, Pb, U nor Zn are relocated when sediments aresulfidized postdepositionally, although Co, Cu and Ni are redistributed from theHCl-reactive phase to pyrite. Molybdenum, U, and Cdpyr provide most informationof the depositional redox conditions and are reliable redox proxies at oxic-anoxictransitions, such as may occur when an aquatic environment becomes eutrophic.The low content of Mo throughout the freshwater sediments thus confirm thatthe sediment-water interface was oxic during sediment deposition. In contrast, Asand Mn appear to be mobile in presence of porewater sulfide and As and Mn areenriched in the postdepositionally sulfidized sediments.

Acknowledgments

We gratefully acknowledge the help of Lauri Niemistö and Boris Winterhalterwho retrieved the sediment core with the Finnish R/V Aranda. We also thankGeorge Luther for inspiring discussions. The Stockholm Centre for Marine Re-search (SMF) and The Swedish Natural Science Research Council (NFR) financedthis investigation. Comments by the two journal reviewers are appreciated.

References

Anderson, R. F., LeHuray, A. P., Fleisher, M. Q. and Murray, J. W. (1989a) Uranium deposition inSaanich Inlet sediments, Vancouver Island.Geochim. Cosmochim. Acta53, 2205–2213.

Anderson, R. F., Fleisher, M. Q. and LeHuray, A. P. (1989b) Concentration, oxidation state, andparticulate flux of uranium in the Black Sea. Geochim.Geochim. Cosmochim. Acta53, 2215–2224.

Andersson, P. S., Porcelli, D., Wasserburg, G. J. and Ingri, J. (1998) Particle transport of234U–238Uin the Kalix River and in the Baltic Sea.Geochim. Cosmochim. Acta62, 385–392.

Balistrieri, L. S., Murray, J. W. and Paul, B. (1994) The geochemical cycling of trace elements in abiogenic meromictic lake.Geochim. Cosmochim. Acta58, 3993–4008.

Barnes, C. E. and Cochran, J. K. (1993) Uranium geochemistry in estuarine sediments: Controls onremoval and release processes.Geochim. Cosmochim. Acta57, 555–569.

Belzile, N. (1988) The fate of arsenic in sediments of the Laurentian Through.Geochim. Cosmochim.Acta52, 2293–2302.

Bishop, J. K. B. (1988) The barite-opal-organic carbon association in oceanic particulate matter.Nature332, 341–343.

Boesen, C. and Postma, D. (1988) Pyrite formation in anoxic environments of the Baltic.Am. J. Sci.288, 575–603.

Boström, K., Burman, J.-O., Pontér, C. and Ingri, J. (1981) Selective removal of trace elements fromthe Baltic by suspended matter.Mar. Chem. 10, 335–354.

Böttcher, M. E. and Huckriede, H. (1997) First occurrence and stable isotope composition ofauthigenicγ -MnS in the central Gotland Deep (Baltic Sea).Mar. Geol. 137, 201–205.


Breit, G.N. and Wanty, R. B. (1991) Vanadium accumulation in carbonaceous rocks: A review ofgeochemical controls during deposition and diagenesis.Chem. Geol. 91, 83–97.

Brügmann, L. (1986) The influence of costal zone processes on mass balances for trace metals in theBaltic Sea.Rapp. P.-v. Réun. Cons. Int. Explor. Mer. 186, 329–342.

Brügmann, L., Hallberg, R., Larsson, C. and Löffler, A. (1998) Trace metal speciation in sea andpore water of the Gotland Deep, Baltic Sea, 1994.Appl. Geochem. 13, 359–368.

Calvert, S.E. and Pedersen, T. F. (1993) Geochemistry of recent oxic and anoxic marine sediments:Implications for the geological record.Mar. Geol. 113, 67–88.

Canfield, D. E., Raiswell, R. and Bottrell, S. (1992) The reactivity of sedimentary iron mineralstoward sulfide.Am. J. Sci. 292, 659–683.

Cooper, D. C. and Morse, J. W. (1998) Extractability of metal sulfide minerals in acidic solutions.Environ. Sci. Technol. 32, 1076–1078.

Cotton, F. A. and Wilkinson, G. (1988)Advanced Inorganic Chemistry. Wiley-Interscience.Dymond, J., Suess, E. and Lyle, M. (1992) Barium in deep-sea sediment: a geochemical proxy for

paleoproductivity.Paleoceanogr. 7, 163–181.Emerson, S. R. and Huested, S. S. (1991) Ocean anoxia and the concentrations of molybdenum and

vanadium in seawater.Mar. Chem. 34, 177–196.Gingele, F. X. and Leipe, T. (1997) Clay mineral assemblages in the western Baltic Sea: Recent

distribution and relation to sedimentary units.Mar. Geol. 140, 97–115.Greenwood, N. N. and Earnshaw, A. (1984)Chemistry of the Elements. Pergamon Press.Hallberg, R. O. (1974) Paleoredox conditions in the Eastern Gotland Basin during the recent

centuries.Havsforskningsinst. Skr.238, 3–16.Helz, G. R., Miller, C. V., Charnock, J. M., Mosselman, J. F. W., Pattrick, R. A. D., Garner, C. D. and

Vaughan, D. J. (1996) Mechanism of molybdenum removal from the sea and its concentration inblack shales: EXAFS evidence. Geochim. Cosmochim. Acta60, 3631–3642.

Huckriede, H. and Meischner, D. (1996) Origin and environment of manganese-rich sediments withinblack-shale basins.Geochim. Cosmochim. Acta60, 1399–1413.

Huerta-Diaz, M. A. and Morse, J. W. (1990) A quantitative method for determination of trace metalconcentrations in sedimentary pyrite.Mar. Chem. 29, 119–144.

Huerta-Diaz, M. A. and Morse, J. W. (1992) Pyritization of trace metals in anoxic marine sediments.Geochim. Cosmochim. Acta56, 2681–2702.

Hyvärinen, H., Donner, J., Kessel, H. and Raukas, A. (1988) The Litorina Sea and Limnaea Sea in thenorthern and central Baltic. InProblems of the Baltic Sea History(eds. Donner, J. and Raukas,A.). Annales Acad. Scient. Fenn. 148A, 25–35.

Kheboian, C. and Bauer, C. F. (1987) Accuracy of selective extraction procedures for metal speciationin model aquatic sediments.Anal. Chem. 59, 1417–1423.

Klinkhammer, G. P. and Palmer, M. R. (1991) Uranium in the oceans: Where it goes and why.Geochim. Cosmochim. Acta55, 1799–1806.

Lapp, B. and Balzer, W. (1993) Early diagenesis of trace metals used as an indicator of pastproductivity changes in coastal sediments.Geochim. Cosmochim. Acta57, 4639–4652.

Leventhal, J. S. (1995) Carbon-sulfur plots to show diagenetic and epigenetic sulfidation in sedi-ments.Geochim. Cosmochim. Acta59, 1207–1211. Lewis, B. L. and Landing, W. M. (1992) Theinvestigation of dissolved and suspended-particulate trace metal fractionation in the Black Sea.Mar. Chem. 40, 105–141.

Luther, G.W., Meyerson, A. L., Krajewskim J. J. and Hires, R. (1980) Metal sulfides in estuarinesediments.J. Sed. Petrol. 50, 1117–1120.

Magyar, B., Moor, H. C. and Sigg, L. (1993) Vertical distribution and transport of molybdenum in alake with a seasonally anoxic hypolimnion.Limnol. Oceanogr. 38, 521–531.

Manheim, F. T. (1961) A geochemical profile in the Baltic Sea.Geochim. Cosmochim. Acta25,52–70.


McManus, J., Berelson, W. M., Klinkhammer, G. P., Johnson, K. S., Coale, K. H., Anderson, R. F.,Kumar, N., Burdige, D. J., Hammond, D. E., Brumsack, H. J., McCorkle, D. C. and Rushdi, A.(1998) Geochemistry of barium in marine sediments: implications for its use as a paleoproxy.Geochim. Cosmochim. Acta62, 3453–3473.

Middelburg, J. J. (1991) Organic carbon, sulphur, and iron in recent semi-euxinic sediments of KauBay, Indonesia.Geochim. Cosmochim. Acta55, 815–828.

Middelburg, J. J, Calvert, S. E. and Karlin, R. (1991) Organic-rich transitional facies in silled basins:Response to sea-level change.Geology19, 679–682.

Pattrick, R. A. D., Mosselmans, J. F. W., Charnock, J. M., England, K. E. R., Helz, G. R., Garner, C.D. and Vaughan, D. J. (1997) The structure of amorphous copper sulfide precipitates: An X-rayabsorption study.Geochim. Cosmochim. Acta61, 2023–2036.

Piper, D. Z. (1994) Seawater as the source of minor elements in black shales, phosphorites and othersedimentary rocks.Chem. Geol. 114, 95–114.

Piper, D. Z. and Isaacs, C. M. (1996) Instability of bottom-water redox conditions during accumula-tion of Quaternary sediment in the Japan Sea.Paleoceanogr. 11, 171–190.

Prange, A. and Kremling, K. (1985) Distribution of dissolved molybdenum, uranium and vanadiumin Baltic Sea waters.Mar. Chem. 16, 259–274.

Raiswell, R., Buckley, F., Berner, R. A. and Anderson, T. F. (1988) Degree of pyritization of iron asa paleoenvironmental indicator of bottom-water oxygenation.J. Sed. Petrol. 58, 812–819.

Rehder, D. (1992) Structure and function of vanadium compounds in living organisms.Biometals5,3–12.

Schneider, B. and Pohl, C. (1996) Time series for dissolved cadmium at a coastal station in theWestern Baltic sea.J. Mar. Sys. 9, 159–170.

Shaw, T. J., Gieskes, J. M. and Jahnke, R. A. (1990) Early diagenesis in differing depositionalenvironments: The response of transition metals in pore water.Geochim. Cosmochim. Acta54,1233–1246.

Shiller, A. M. and Boyle, E. A. (1991) Trace elements in the Mississippi River Delta outflow region:Behavior at high discharge.Geochim. Cosmochim. Acta55, 3241–3251.

Skei, J. M., Loring, D. H. and Rantala, R. T. T. (1996) Trace metals in suspended particulate matterand in sediment trap material from a permanently anoxic fjord – Framvaren, South Norway.Aq.Geochem. 2, 131–147.

Sohlenius, G. and Westman, P. (1998) Salinity and redox alternations in the northwestern Balticproper during the late Holocene.Boreas27, 101–110.

Sohlenius, G., Sternbeck, J., Andrén, E. and Westman, P. (1996) Holocene history of the Baltic Seaas recorded in a sediment core from the Gotland Deep.Mar. Geol. 134, 183–201.

Sohlenius, G., Wastegård, S. and Sternbeck, J. (1998) Evidence of benthic colonisation duringformation of laminated sediments in the Gotland Deep, Baltic Sea.Geologiska FöreningensFörhandlingar120, 293–296.

Sternbeck, J. and Sohlenius, G. (1997) Authigenic sulfide and carbonate mineral formation inHolocene sediments of the Baltic Sea.Chem. Geol. 135, 55–73

Suess, E. (1979) Mineral phases formed in anoxic sediments by microbial decomposition of organicmatter.Geochim. Cosmochim. Acta43, 339–352.

Swedish EPA (1991) Metaller i Svenska Havsområden. Swedish Environmental Protection Agency,report 3696 (in Swedish).

Thomson, J., Higgs, N. C., Croudace, I. W., Colley, S. and Hydes, D. J. (1993) Redox zonation ofelements at an oxic/post-oxic boundary in deep-sea sediments.Geochim. Cosmochim. Acta57,579–595.

Thomson, J., Higgs, N. C., Wilson, T. R. S., Croudace, I. W, De Lange, G. J. and Van Santwoort, P.J. M. (1995) Redistribution and geochemical behaviour of redox-sensitive elements around S1,the most recent eastern Mediterranean sapropel.Geochim. Cosmochim. Acta59, 3487–3501.


Trefry, J. H. and Metz, S. (1989) Role of hydrothermal precipitates in the geochemical cycling ofvanadium.Nature342, 531–533.

Widerlund, A. and Ingri, J. (1995) Early diagenesis of arsenic in sediments of the Kalix River estuary,northern Sweden.Chem. Geol.125,185–196.

Winterhalter, B., Flodén, T., Ignatius, H., Axberg, S. and Niemistö, L. (1981) Geology of the BalticSea. InThe Baltic Sea(ed. A. Voipio), pp. 1-121. Elsevier Oceanographic Series 30.

Sedimentary Trace Elements as Proxies to Depositional Changes Induced by a Holocene Fresh-Brackish Water Transition

Documents