Paleoceanography

Florisphaera profunda and the origin and diagenesis of carbonate phases in eastern Mediterranean sapropel units

Authors


Abstract

[1] High relative concentrations of the lower photic zone nannofossil Florisphaera profunda have been reported in all recent eastern Mediterranean sapropels. In the most recent sapropel (S1), high bulk sediment Sr/Ca ratios occur along with high F. profunda contents toward the base of the unit, exemplified here in four cores from 1.5–3.5 km water depth. Co-occurring biogenic carbonates contain insufficient Sr to account for these high Sr concentrations, and X-ray diffraction and selective leaching show that the high Sr/Ca ratios are due to aragonite, the CaCO3 polymorph that is rarely preserved in deep marine sediments, with ∼1 wt % Sr. The possible sources of this aragonite include (1) precipitation with surface ocean production, (2) detrital input from shallow-water sediments by high continental runoff, or (3) postdepositional diagenetic formation driven by increased pore water alkalinity resulting from sulphate reduction. The third formation mechanism for the aragonite is favored, in which case the similarity in the positions of the aragonite and F. profunda abundance maxima in sapropels is probably related to Corg accumulation and resulting sulphide diagenesis that produces high pore water alkalinity. There is clear micropaleontological evidence that dissolution of the less soluble biogenic low-Mg CaCO3 is occurring, or has occurred, during early diagenesis in these sediments despite the coexistence of the more soluble high-Mg calcite and aragonite polymorphs. Similar Sr/Ca maxima are also found associated with older sapropels, always located close both to local minima in surface ocean δ18O that signal maximum monsoon-driven runoff and to maxima in diagenetic sediment sulphide contents. High freshwater flows from monsoons are believed to drive eastern Mediterranean sapropel formation through water column stabilization that favors F. profunda production and later through development of deep water column dysoxia/anoxia because of reduced ventilation. The relative abundances of F. profunda are high between 5 and 11 14C kyr B.P. with a maximum at ∼9 14C kyr B.P. The surface ocean production changes marked by F. profunda therefore begin earlier and finish later than the formation of the S1 sapropel, which only develops between 6 and 10 14C kyr B.P.

1. Introduction

[2] At present, the surface waters of the eastern Mediterranean are oligotrophic and the entire water column is well oxygenated [Roether and Well, 2001] with the consequence that Corg-poor (0.1–0.2 wt %) sediments are being deposited throughout the deep basin [e.g., van Santvoort et al., 1996, 2002]. In the past, Corg-rich sediments (i.e., sapropels with Corg > ∼2 wt %) have formed regularly in the basin at those times when the amplitude of the seasonal cycle of insolation in the Northern Hemisphere was at a maximum [e.g., Calvert, 1983; Wehausen and Brumsack, 2000; Calvert and Fontugne, 2001]. These insolation maxima are determined by minima of the precession component of Earth's orbital cycle and occur every 21 kyr. The trigger mechanism for the basin's repetitive transitions into sapropel formation is believed to be amplification of the African summer monsoon by seasonal radiation forcing [e.g., Rossignol-Strick et al., 1982; Hilgen, 1991; Lourens et al., 1996]. Each episode of sapropel formation has a duration of only a few thousand years, during which time it is believed that monsoon runoff leads first to water column stabilization and probably to enhanced production. Deepwater dysoxia or anoxia then develops in the temporarily poorly ventilated deep waters to allow enhanced Corg preservation and sulphide formation and hence sapropel formation in the sediments [Rohling, 1994]. The repetitive occurrence of sapropels is thus an important example of solar orbital forcing on climatic oscillations and of the amplification of climatic signals in a restricted ocean basin. While sapropel formation is understood at this level of detail, however, the precise mechanisms by which the basin makes the transitions from deposition of Corg-poor to Corg-enriched sediments and back are less well understood. An improved mechanistic understanding is relevant to the debate on the relative importance of preservation versus production in formation of Corg-rich marine sediments, including black shales. It seems likely that such an improved understanding of the eastern Mediterranean basin's behavior will come from multidisciplinary investigations that provide data for more sophisticated paleoenvironmental reconstructions and modeling.

[3] Castradori [1993] first pointed out that, in comparison with their enclosing Corg-poor sediments, the last few eastern Mediterranean sapropels contain unusually high numbers of platelets from Florisphaera profunda, an incertae sedis species [Jordan and Kleijne, 1994] with a marked lower photic zone preference [Okada and Honjo, 1973]. Given that most other calcareous nannofossil species flourish higher in the water column, Castradori [1993] argued that production in sapropel times was likely to have been characterized by primary production occurring unusually deep in the photic zone, likely in a deep chlorophyll maximum (DCM) layer as previously proposed by Rohling and Gieskes [1989]. Subsequent detailed investigations of sapropels in which diatoms had unusually been preserved have further supported the DCM hypothesis [Kemp et al., 1999; Corselli et al., 2002]. Laminations in these same diatomaceous sapropel units also provide evidence for an annual depositional pulse of material to the seafloor at the end of seasonal DCM stratification in autumn [Kemp et al., 2000]. High relative abundances of F. profunda are thus interpreted as a persistent carbonate marker of production and water column structure change during times of sapropel formation.

[4] Sapropels, including S1, generally have lower total carbonate contents than their enclosing nannofossil marls. If the ratio of the fluxes of total biogenic production (i.e., opal plus carbonate) to detrital materials does not change materially during sapropel deposition, and if siliceous rather than carbonate production dominates during sapropel episodes, then these lower carbonate contents may be explicable by postdepositional opaline silica dissolution [van Os et al., 1994; Schenau et al., 1999; Kemp et al., 1999, 2000; Corselli et al., 2002]. An unusual feature of eastern Mediterranean sediments is that they contain the more soluble CaCO3 polymorphs high-Mg calcite (>4 mol % MgCO3) and aragonite as well as the less soluble low-Mg calcite (<4 mol % MgCO3). Low-Mg calcite (subsequently referred to simply as “calcite”) is the biogenic polymorph that is generally formed by surface dwelling organisms and is usually the only CaCO3 form found preserved in deepwater sediments. The deep ocean is under-saturated with respect to high-Mg calcite and aragonite, so these polymorphs generally dissolve before or immediately after deposition in deepwater open-ocean sediments [Friedman, 1965; Berner and Honjo, 1981; Fabry and Deuser, 1991]. High-Mg calcite and aragonite are therefore usually only found in sediments under waters where the carbonate system is over-saturated, i.e., at shallow water depth and/or at low latitudes [Kinsman, 1969; Wilkinson and Given, 1986], and the high-Mg calcites found in the eastern Mediterranean marls are believed to be diagenetic [Milliman and Müller, 1973; Aksu et al., 1995; Calvert and Fontugne, 2001].

[5] In this paper, geochemical and micropaleontological analyses are reported that show that high Sr/Ca ratios in the most recent sapropel (S1) appear to be coherent with high F. profunda abundances. This Sr enrichment cannot be introduced into the sediment with biogenic calcites, but it is instead shown to be due to the presence of aragonite. In these sediments, aragonite is also mostly confined to sapropels, while high-Mg calcite is mostly restricted to the intercalated marls. These associations are described and mechanisms proposed for the formation of the carbonate phases in the context of sapropel development.

2. Methods

[6] The positions and water depths of the investigated cores in the eastern Mediterranean basin are specified in Table 1 and displayed in Figure 1. For calcareous nannofossil analyses, a small amount of sediment was moistened with distilled water, smeared homogeneously on a slide, and permanently mounted with optical adhesive. No centrifugation or ultrasound cleaning was applied in order to retain the original fine fraction composition. Analyses were performed on 192 samples (cores SL114, SL60, BC19, and LC25) by a cross-polarized light microscope (LM) at 1250 times magnification. The occurrences of the lower photic zone species F. profunda, Gladiolithusflabellatus, and Algirosphaerarobusta were counted separately from upper/middle photic zone species. Their abundance was recorded against 100 coccoliths of all other species following the concepts of Matsuoka and Okada [1989] and Castradori [1993], so the reported data are index values. Both F. profunda var. profunda and F. profunda var. elongata, present in the modern Mediterranean Sea [Cros, 2002], were observed in these sediments, but the plates were not distinguished at the variety level during LM counting. Following Kleijne [1992], all specimens of the genus Algirosphaera were assigned to A. robusta.

Figure 1.

Map of the eastern Mediterranean Sea with the positions of the cores discussed in the text. Contours are at 1000 m.

Table 1. Positions and Water Depths of the Cores Investigated
CoreCorer TypeWater Depth,a mLatitude, °NLongitude, °E
  • a

    Units are uncorrected.

BC19box275033°47.85′28°36.50′
LC25large diameter piston306032°36.01′27°23.25′
SL60piston152235°39.69′26°34.99′
SL114box339035°17.24′21°24.52′
T87-19Bbox348336°48.9′19°10.7′
MD84641piston137533°02′32°38′

[7] The inorganic geochemical data available were from cores that had been analyzed in different laboratories at different times, either by X-ray fluorescence (XRF) analysis of fusion beads or compressed powder pellets for major and minor elements, respectively [Mercone et al., 2000], or by inductively coupled plasma atomic emission spectroscopy (ICP-AES) analysis of dissolved (with HF/HClO4 acids) or fused (LiBO2) samples for both minor and major elements [van Santvoort et al., 1996]. Organic carbon and CaCO3 were determined coulometrically from released CO2 [Mercone et al., 2000], with inorganic C (biogenic calcium carbonate) measured as the CO2 evolved by the addition of 10% (v/v) H3PO4 and total C measured as the CO2 generated by total sample combustion at 900°C. Selected samples including some with high Sr contents were investigated by X-ray diffraction (XRD) to determine the proportions of the different CaCO3 polymorphs and by mild leaching with 1 molar acetic acid to separate carbonate (dissolved in the leachate) from detrital phases (left as the leaching residue). Analysis for a full suite of elements was then performed on the leachates by ICP-AES, and the residues were fused with lithium metaborate, dissolved, and similarly analyzed by ICP-AES. In both cases the analyses were calibrated with matrix-matched standards, and the residue analyses were confirmed by including the international standard reference material MAG-1 (marine mud). Four or five AMS radiocarbon analyses from planktonic foraminifera >150 μm were obtained in and around each sapropel unit in order to convert depth in core to radiocarbon time [Mercone et al., 2000].

[8] Selected samples from cores LC25 and T87-19B with high Sr/Ca values were also examined by scanning electron microscopy (SEM) at Kiel and Utrecht in an attempt to discriminate between biological and inorganic/diagenetic precipitates. Filter preparation for SEM study was based on a modification of the method of Andruleit [1996]: ∼0.05 g of wet sediment was diluted with ethanol, ultrasonified for 5 s, and split in 10 aliquots with a rotary splitter. The suspension (0.005 g of sediment) was filtered on to a polycarbonate membrane filter (0.4 μm pore size; 50 mm diameter) and dried for 2 hours at 50°C. A segment from each filter was mounted on a SEM stub and coated with Au/Pd. Selected stubs that had been examined by SEM were also reanalyzed by ion microprobe at Utrecht in order to determine the Sr/Ca ratios in the different carbonate particle morphologies that had been identified by SEM.

3. Results and Discussion

3.1. Sr/Ca, S, and Florisphaera Profunda Profiles Compared

[9] Postdepositional oxidation has affected the upper reaches of many sapropel S1 units such that their present visual thickness is typically several centimeters thinner than their inferred original thickness [Higgs et al., 1994; Thomson et al., 1995; van Santvoort et al., 1996] (Figure 2). The limits of the original sapropel S1 units are therefore defined from the whole sediment Ba/Al profiles on the premises that (1) the biogenic Ba that accompanies deposited Corg during sapropel formation increases Ba/Al over a detrital value of 0.0030–0.0035 and (2) that Ba/Al is a more persistent indicator than Corg because it does not undergo the postdepositional oxidation and remineralization that results in Corg loss from the sapropel [Thomson et al., 1999; Rutten and de Lange, 2002].

Figure 2.

Downcore profiles of Ba/Al (wt ratio), total Corg and total S (wt %), Sr/Ca (wt ratio), and Florisphaera profunda index (number of specimens per 100 other coccoliths counted) for cores LC25, and BC19, SL114, and SL60. The oxidized and unoxidized parts of the S1 sapropel are indicated by light and dark shading, a thin turbidite in core LC25 and an ash layer in core BC19 are marked by dark shading. The F. profunda profile for core BC19 was obtained from a different subcore that was 2.5 cm shorter to the base of the sapropel than the subcore used for geochemical analysis.

[10] As will be seen later, the Sr/Ca mass ratio of deep-sea carbonate marls is generally around 0.004–0.005, but higher values are measured consistently in S1 units. There appear to be similarities in the positions and possibly the shapes of the profiles of whole sediment Sr/Ca ratio, whole sediment S content, and F. profunda relative abundance profiles through these sapropel units (Figure 2). This is demonstrated with profiles from two cores with the largest F. profunda index and Sr/Ca ratio excursions encountered, BC19 and LC25 with maximum values >100 for the F. profunda index and 0.009–0.01 for Sr/Ca, and two further cores, SL60 and SL114 with peak F. profunda index values of ∼60 and Sr/Ca mass ratios of 0.006–0.007. The magnitudes of the F. profunda index and Sr/Ca values thus increase together, but they do not appear to be directly proportional to Corg contents. Negri and Giunta [2001] and A. Rutten and G. J. de Lange (Sequential extraction of aragonite/calcite and dolomite from eastern Mediterranean sediments, submitted to Marine Geology, 2004, hereinafter referred to as Rutten and de Lange, submitted manuscript, 2004) have previously reported similar F. profunda and Sr/Ca profile shapes through sapropel S1, respectively.

[11] Data from different S1 units can be compared directly if a few radiocarbon ages from in and around each sapropel are available [Mercone et al., 2000]. This comparison is achieved by regressing the accelerator mass spectrometric (AMS) radiocarbon ages of planktonic foraminifera on their depth in core in order to construct an independent timescale for each core. This procedure avoids selection of a reservoir age correction for the Mediterranean area that is necessary to convert radiocarbon convention ages to calibrated time [Reimer and McCormac, 2002; Keenan, 2002] but has the drawback that marine radiocarbon convention ages do not have an exact linear relation with true time [Stuiver and Braziunas, 1993]. When the four core profiles of Figure 2 are placed on such a radiocarbon timescale (Figure 3), a consistent picture emerges. The enhanced F. profunda and Sr/Ca values reach their maxima toward the base of high Ba/Al (and residual Corg) values, and high F. profunda and Sr/Ca values occur consistently over a longer time than the S1 unit defined by the Ba/Al ratio. (Note that, as indicated in Figure 2, the Corg and S profiles in the upper parts of the S1 units in Figure 3 have suffered loss from postdepositional oxidation. This has affected sediments laid down after ∼7.5 14C kyr B.P. in cores LC25, BC19, and SL114.)

Figure 3.

Ba/Al (wt ratio), Corg (wt %), S (wt %), Sr/Ca (wt ratio), and F. profunda index versus a 14C convention years timescale for cores LC25, BC19, SL114, and SL60. The timescales were obtained by regression of C14 ages in and around sapropel S1 on depth in each core. In this case the shaded area between 6 and 10 14C kyr B.P. covers the period of enhanced Ba/Al values, but sapropel with high Corg and Ba/Al values is present for a lesser time (∼6.5–9.5 14C kyr B.P.), while F. profunda values are high between 5 and 11 C kyr B.P. Note that postdepositional oxidation has affected the Corg and S contents of sediments laid down after ∼7.5 14C kyr B.P. in cores LC25, BC19, and SL114. The scale on the far left is the Stuiver and Braziunas [1993] conversion of marine 14C convention ages to calibrated time with a constant fixed surface ocean reservoir age of 400 years.

3.2. Geochemical Evidence From Older Sapropel Units

[12] The cyclicity of sapropel formation in the eastern Mediterranean suggests that similarities might be found between S1 and older sapropels. Core MD84641, located approximately midway between the Nile delta and Cyprus, has been used to demonstrate the distinct decrease in salinity experienced by the surface waters of the eastern Mediterranean basin during formation of the last several sapropels [Kallel et al., 1997, 2000] and that marked compositional contrasts exist between sapropels and the enclosing marl sediments [Calvert and Fontugne, 2001]. In this core, Sr/Ca values in a background range of 0.004–0.005 are observed between sapropels, coincident with those sections where high Mg-calcite levels are present (Figure 4). Higher Sr/Ca values (of a magnitude similar to or greater than those measured in S1) are found at each of the sapropels S1, S3, S4, S5, S6, S7, S8, and S9, and aragonite is detected in the XRD scans of these samples (not shown). Within the resolution of sampling, maximum Sr/Ca values (always >0.006) occur at every sapropel either within or at the base of the high Ba/Al values and close to the maximum diagenetic S (sulphide) values developed in each sapropel. Zones with high-Mg calcite content thus alternate with zones of high Corg, Ba/Al, Sr/Ca, and S values (i.e., the sapropels; Figure 4). Low but finite values of high-Mg calcite values are present in sapropel S1, and these values are higher than those in any of the older sapropels.

Figure 4.

Downcore profiles of Ba/Al (wt ratios), total S (wt %), Sr/Ca (wt ratio), high-Mg calcite (relative XRD measure, counts per 0.1 s), and δO18 from foraminifer G. ruber for core MD848641 [Kallel et al., 1997, 2000; Calvert and Fontugne, 2001]. Sapropels are identified on the basis of high Ba/Al ratios and indicated by light shading.

[13] Most of the Sr/Ca peak values in core MD84641 also occur close to depths where the planktonic foraminifer Globigerinoides ruber records low (<0.0‰) surface ocean δ18O values, as discussed by Kallel et al. [1997, 2000]. The single exception is sapropel S6, formed in glacial times, where the lowest δ18O value is +0.5‰. In this basin, such low surface ocean δ18O values are due to the runoff from heavily fractionated monsoon rains that drive the environmental changes that cause the episodes of sapropel formation [Kallel et al., 1997; Rohling, 1999; Bar-Matthews et al., 2000; Gasse, 2002; Bard et al., 2002]. Since the original report of Castradori [1993], it has also been shown that all the sapropels present in core MD84641 and much older sapropels also display increased F. profunda abundances [Castradori, 1998; Negri et al., 1999; Negri and Giunta, 2001; Corselli et al., 2002].

3.3. Source of the Additional Sr

[14] Distinctly different mechanisms can be invoked to link high Sr/Ca values to F. profunda abundance or S content but cause and effect uncertainties arise because maxima in signals of preserved production (F. profunda, Ba/Al, and Corg), induced diagenesis (S as sulphide from sulphate reduction by Corg), and Sr/Ca all occur close together in sapropels (Figure 3). A priori, the obvious possibilities are that the extra Sr relative to Ca is contributed by F. profunda or another component with a flux closely related to that of F. profunda or is diagenetic but related to surface ocean production as traced by F. profunda.

[15] The first requirement is to identify the host phase for the additional Sr observed in sapropels. One possibility is preservation of celestite (SrSO4) from acantharia in the sediments, but this seems unlikely in view of the high solubility of celestite [Reardon and Armstrong, 1987; Bernstein et al., 1992, 1998]. No acantharia were identified during micropaleontological analyses, and celestite was not identified in XRD analysis. Marine barite has a variable Sr content, and SrSO4 molar fractions as high as 0.26 have been reported [Dehairs et al., 1980; Bertram and Cowen, 1997; Rushdi et al., 2000]. There is, however, insufficient barite for Sr in barite to dominate the Sr/Ca ratio in these sediments with a CaCO3 content of 40–50 wt %, and the shapes of the Sr/Ca and Ba/Al profiles are quite different (Figure 3).

[16] In shallow-water marine carbonates, increased Sr relative to Ca often implies that a proportion of the total CaCO3 is present as aragonite rather than calcite [Kinsman, 1969], and aragonite was detected in those levels of core MD 84641 where the highest Sr/Ca values were detected. The nature of the Sr association was therefore investigated in the core with the highest Sr/Ca values (LC25) by mild acid leaching of the whole sediment to yield carbonate and detrital aluminosilicate fractions (Figure 5). Excluding samples from the thin Nile turbidite that interrupts S1 in this core [Reeder et al., 1998], the acetic acid leach residue representing detrital aluminosilicate material has a Sr content in the range 34–95 μg g−1, or 77–122 μg g−1 on a carbonate-free basis. Rutten and de Lange (submitted manuscript, 2004) have reported even lower detrital aluminosilicate Sr contents (21–47 μg g−1) after multistep leaching of recent eastern Mediterranean sediments including sapropel S1. Detrital aluminosilicate Sr thus makes a minor contribution to the total Sr content of these sediments (1000–1800 μg g−1), and most of the Sr present appears to be associated with carbonate.

Figure 5.

Leach experiment and X-ray diffraction data profiles for core LC25. (left) Total data (XRF), leached and detrital data (ICP-AES) for Sr (μg g−1). (middle) Total data (XRF), leached and detrital data (ICP-AES) for Ca (wt %). (right) Coulometric data for total CaCO3 and XRD data for aragonite, high-Mg calcite, and calcite. The oxidized and unoxidized parts of the S1 sapropel are indicated by light and dark shading, and a thin turbidite in sapropel S1 is marked by darker shading.

[17] XRD analysis revealed that aragonite is present throughout the sampled section of core LC25 in variable amounts, increasing from ∼5 wt % of the total sediment in the marls above the sapropel to ∼15 wt % in the sapropel itself, and it is those samples with the highest aragonite contents that show the highest Sr contents and Sr/Ca values (Figure 5). High-Mg calcite has an abundance pattern opposite to that of aragonite, decreasing from ∼15% of the total sediment in the overlying marls to <5% within the sapropel. From the core MD84641 and LC25 XRD data (Figures 4 and 5), this alternation of high-Mg calcite preferentially in the marls and aragonite preferentially in the sapropels therefore appears to be general in eastern Mediterranean sediments. Between them, aragonite and high-Mg calcite contribute 20–50% of the total CaCO3 in core LC25. This paper focuses on Sr and aragonite rather than on Mg and high-Mg calcite because the relatively high Mg content of detrital aluminosilicates means that the Mg content is divided more evenly between the aluminosilicate and carbonate phases than is the Sr content. The XRD and leachate data from core LC25 samples can be combined to derive a relationship of leachate Sr/Ca with carbonate mineralogy (Figure 6), which indicates a pure aragonite end-member with Sr/Ca wt ratio = 0.027 (10,700 μg Sr g−1 CaCO3) and a zero aragonite end-member with Sr/Ca = 0.0033 (1340 μg Sr g−1 CaCO3). This contrast implies that the Sr/Ca ratio will be double that of the zero aragonite end-member when 15% of the total CaCO3 is present as aragonite.

Figure 6.

Leach fraction Sr/Ca data (ICP-AES) for selected samples of core LC25 versus aragonite fraction of total CaCO3 (sum of XRD aragonite, high-Mg calcite, and calcite), shown as filled circles. The orthogonal best fit line is solid and shown along with zero and unity aragonite end-member values. Similar data from samples of the Augias turbidite in core T87-19B are also displayed as open squares and with a dashed fit line.

[18] The observed range of Sr/Ca variations in Figure 6 is much larger than those encountered in investigations of Sr/Ca in separated and cleaned foraminiferal and coccolith calcite [e.g., Stoll and Schrag, 1999; Shen et al., 2001]. Stoll and Schrag [1999] measured Sr/Ca molar ratios in foraminiferal and coccolith calcites of ∼1.3 × 10−3 and ∼2.2 × 10−3 (Sr/Ca wt ratios = 0.0028 and 0.0048; 1100 and 1900 μg Sr g−1 CaCO3), respectively, in a sediment with a bulk carbonate Sr/Ca ratio of 1.8 × 10−3 (Sr/Ca wt ratio = 0.0039; 1600 μg Sr g−1 CaCO3). Subsequent work has confirmed that the Sr/Ca range for various coccoliths either separated from sediment or grown in culture is in the range 2–6 × 10−3 mol [Rickaby et al., 2002; Stoll and Zivieri, 2002], and specifically the Sr/Ca ratio of a F. profunda fraction separated from all other coccoliths by differential size settling corresponds to ∼1900 μg Sr g−1 CaCO3 [Stoll and Ziveri, 2002]. F. profunda and all other coccolith or foraminiferal calcites cannot be the source of the observed increases in aragonite and Sr content.

[19] Biogenic aragonite often, but not invariably, contains much more Sr than biogenic calcite. As examples, the Sr/Ca molar ratio in the aragonite of the scleractinian hermatypic coral Porites lobata is ∼9.2 × 10−3 (Sr/Ca wt ratio = 0.021; 8100 μg Sr g−1 CaCO3; Beck et al. [1992]), whereas in the calcareous green alga Halimeda, it is ∼10.9 × 10−3 (Sr/Ca wt ratio = 0.024; 9500 μg Sr g−1 CaCO3; Delaney et al. [1996]). Recrystallization in corals and Halimeda has been shown to produce early diagenetic aragonite with somewhat higher Sr/Ca values than the original biogenic aragonite at ∼10,000 μg Sr g−1 CaCO3 [Müller et al., 2001; Ribaud-Laurenti et al., 2001], similar to the Sr/Ca in aragonite cements [Kinsman, 1969]. Pteropod aragonite accounts for at least 12% of the total sinking CaCO3 flux in the open ocean, mainly in the large (>1 mm) size fraction [Berner and Honjo, 1981; Fabry and Deuser, 1991]. It is not possible to explain the data of Figures 5 and 6 by means of pteropod aragonite, however, because unlike coral, algal, or precipitated inorganic aragonite, pteropods have a Sr content no greater than that of biogenic calcites at 1000–1500 μg Sr g−1 CaCO3 [Krinsley and Bieri, 1959; Kinsman, 1969].

3.4. Mechanisms for Aragonite in Eastern Mediterranean Sapropels and Paleoceanographic Implications

[20] Three hypotheses to account for the presence of high-Sr aragonite in eastern Mediterranean sapropels are presented and examined using the data above and constraints imposed on the likely scenarios by other paleoceanographic information.

3.4.1. Aragonite Production Accompanies Florisphaera Profunda Production in Surface Waters

[21] This hypothesis posits that aragonite precipitation is directly linked to monsoonally driven surface ocean production. This is analogous to one set of mechanisms that has been invoked to explain whitings, the drifting patches of very fine CaCO3 turbidity (mainly aragonite) observed suspended in the water column over shallow subtropical carbonate banks. Some workers have ascribed whitings on the Bahama Banks to a biological or chemical precipitation of fine-grained aragonite in the water column [e.g., Shinn et al., 1989; Robbins and Blackwelder, 1992; Macintyre and Reid, 1992; Milliman et al., 1993]. Robbins and Blackwelder [1992] have specifically proposed that whitings result from the precipitation of CaCO3 by picoplankton in the water column.

[22] Friedman [1993, 1994] noted the importance of a proper interpretation of whitings and proposed the Dead Sea as the type locality for abiotic chemical precipitation. It has recently been claimed, however, that the early interpretation of the mechanism for aragonite precipitation events in the Dead Sea is incorrect. Rather than summer evaporation and warming of the CaCl2-enriched Dead Sea brine causing aragonite precipitation as initially envisaged, Barkan et al. [2001] suggest instead that aragonite is precipitated after winter flooding when a riverine fresh water layer enriched in HCO3 overlies and gradually mixes with the underlying Ca2+-rich Dead Sea brine. These workers propose that similar conjunctions of two different water bodies, one with high Ca2+ and the other with high HCO3, may also promote CaCO3 precipitation in other aquatic environments. Calcium carbonate formation induced by phytoplankton growth is indeed observed in lakes, although in these cases it is generally calcite that is formed [e.g., Thompson et al., 1997; Dittrich et al., 2004].

[23] Eastern Mediterranean salinities are much lower than those of the Dead Sea, but in so far as the scenario described above is similar to sapropel formation during monsoon episodes, the production hypothesis envisages F. profunda growth and aragonite formation occurring together at the base of the fresher surficial layer, possibly through picoplankton biomineralization [Robbins and Blackwelder, 1992]. This fine material would be removed from the water column along with settling particulate matter [Deuser et al., 1981, 1983]. The problem with this explanation is that no biogenic high-Sr aragonitic form could be identified in the sediments.

3.4.2. Aragonite Is Detrital

[24] A different hypothesis links the high Sr/Ca in sapropels to an introduction of detrital aragonite with the monsoonal floodwaters that initiate and sustain DCM formation and hence favor the production of F. profunda. This hypothesis is analogous to the proposition that bottom sediment resuspension (rather than precipitation in the water column) is the explanation for whiting formation on the Bahama Banks [Morse et al., 1984, 2003; Boss and Neumann, 1993; Broecker et al., 2000, 2001].

[25] One unequivocally detrital aragonite occurrence in the eastern Mediterranean is the Augias turbidite that floors the Sirte and Ionian abyssal plains [Hieke and Werner, 2000; Cita and Aloisi, 2000]. This turbidite is formed from sediments from shallow (photic zone) waters in the Gulf of Sirte and has an aragonite content of up to 15% ascribed to the green alga Halimeda out of a total carbonate content of 50–60% [Hieke and Werner, 2000]. Although the Augias turbidite is >12 m thick on the Ionian Plain floor [Hieke and Werner, 2000], it is only ∼14 cm thick in core T87-19B [Rasmussen, 1991; Troelstra et al., 1991], which was retrieved 500 m shallower in the northeastern part of the plain (Figure 1). The S1 sapropel is also present in this core at 30–44 cm (Figure 7), and like the cores in Figure 2, the Sr/Ca profile exhibits values of >0.005 between 29 and 45 cm with a maximum value of 0.007 at 43 cm. Chemical and XRD analyses of the Augias turbidite, which has Sr/Ca values up to 0.018, carried out in a similar manner to that applied to the core LC25 samples, yield a pure aragonite end-member with 10800 μg Sr g−1 CaCO3 (Sr/Ca wt ratio = 0.027) and a zero aragonite end-member with 1680 μg Sr g−1 CaCO3 (Sr/Ca = 0.0042; Figure 6). The slight offset between the data from cores T87-19B and LC25 may be due to minor low-Sr pteropod aragonite that is present in core LC25 but absent in the Augias turbidite.

Figure 7.

Downcore profiles of Ba/Al, Si/Al, Zr/Al (all wt ratios), CaCO3 (wt %), and Sr/Ca (wt ratio) for core T87-19B. The oxidized and unoxidized parts of the S1 sapropel are indicated by light and dark shading between 30 and 44 cm, and the Augias turbidite is marked by darker shading from 4.5 to 18.5 cm. The positions of turbidite samples selected for leaching (Figure 6) are indicated by triangles on the far right. (Note that higher Sr/Ca values are associated with sapropel S1 from 29 to 45 cm, with the maximum value at 43 cm, close to the sapropel base.)

[26] The Augias turbidite data (Figure 6) prove that a source of algal high-Sr aragonite of a composition similar to that observed in the deepwater sediments studied exists in the shallow water sediments on the north African margin. Halimeda is abundant on many Mediterranean coasts and many coastal ridges of north Africa contain carbonate sands in which Halimeda is a major component [Burrolet et al., 1979] with resultant high aragonite and Sr contents [Holail, 1993]. The detrital hypothesis presented here envisages that monsoon floodwaters discharging from north Africa into the Mediterranean Sea entrain fine-grained shallow-water aragonite from the coastal photic zone and carry it into the surface waters of the basin, whence it would be removed in the settling flux to the sediments [Deuser et al., 1981, 1983]. In this case the linkage between Sr/Ca and F. profunda must be through the volume of fresh water discharged from the continent, first in the amount of aragonite it can entrain and second in the subsequent degree of stratification it can produce. The Sr/Ca and F. profunda profiles through S1 might therefore be ascribed to a pulse of freshwater that rises to a maximum and then decreases during sapropel times, so that the volume of discharge flux controls both parameters. Halimeda is very productive, with estimates of annual production ranging between 300 and 3000 g CaCO3 m−2 yr−1 [Ballesteros, 1991; Freile et al., 1995], so that in principle it can rapidly rebuild its concentration in shallow photic zone sediments. This appears to be a viable explanation for the observed high-Sr aragonite in sapropels.

3.4.3. Aragonite Is Formed As a Consequence of Sulphate Reduction During Sediment Diagenesis

[27] The third hypothesis is that the high-Sr aragonite in sapropel units is diagenetic, like the high-Mg calcite found in the marls. The characteristic difference between sapropel units and the enclosing marls is their high contents of preserved Corg from primary production and of diagenetic S produced from Corg oxidation by SO42− reduction. Corg remineralization affects sediment pore water alkalinity and hence the dissolution versus precipitation behavior of the sediment solid phase carbonates. Oxic remineralization of Corg produces HCO3, HPO42−, and H+ and overall a net reduction in alkalinity. This leads to a lowering of the saturation state with respect to carbonate minerals in the sediment pore waters and induces dissolution of CaCO3, particularly of aragonite and high-Mg calcite in preference to calcite [Mucci et al., 2000; Green and Aller, 2001]. Such oxic remineralization occurs during formation of the Corg-poor marls laid down between sapropel events [van Santvoort et al., 2002].

[28] Anoxic remineralization of Corg by SO42− produces HS along with HCO3, HPO42−, and H+, so there is a net increase in alkalinity, and the H+ produced is consumed in the formation of FeS minerals if sufficient reactive Fe is available [Mucci et al., 2000]. Sulphate reduction therefore tends to induce CaCO3 precipitation rather than promote CaCO3 dissolution [Gaillard et al., 1989; Mucci et al., 2000; Green and Aller, 2001]. While the oxygenation condition of bottom waters during formation of S1 is not well established [Casford et al., 2003], the presence of sulphides in and below S1 units demonstrates that anoxic conditions must have been established in pore waters at shallow depth very soon after sediment deposition [Passier et al., 1996].

[29] The sulphate reduction reaction may be summarized as

equation image

and the precipitation reaction as

equation image

[30] Reactions (1) and (2) are both usually bacterially mediated [Castanier et al., 1999], and Ehrlich [1996] reports that the precipitation of aragonite and other CaCO3 phases often accompanies the activities of sulphate-reducing bacteria. Each 1 wt % of diagenetic S formed is accompanied by sufficient HCO3 from Corg to form ∼3 wt % CaCO3, so that the 3+ wt % S in the core LC25 and BC19 sapropel samples with the largest Sr/Ca ratios could produce 10% diagenetic CaCO3. This hypothesis leads to the expectation that the Sr/Ca and diagenetic S profiles should be similar, with the additional Sr and S both sourced from bottom waters, and indeed the maximum levels of diagenetic S in sapropels always occur close to peak Sr/Ca ratio values (Figure 3). The Sr data of cores LC25 and SL60 were therefore modeled (Figure 8) assuming three components implied by the systematics of Figures 5 and 6, i.e., (1) 1 mol of aragonite with Sr/Ca = 0.027 is produced along with each mole of diagenetic S, (2) all remaining Ca present as CaCO3 with Sr/Ca = 0.0033, and (3) detrital material with Sr/Al = 0.0137. In order to reproduce a satisfactory fit, this calculation requires that all S produced from SO4 by Corg remineralization is preserved in the sediments, whereas in fact much sulphide S undergoes reoxidation after formation [Boudreau et al., 1992]. Migration of sulfur and carbonate system species in the pore water solution is a further complication that is not taken into account. Although the calculation based on the stoichiometry of equations (1) and (2) therefore underestimates diagenetic Sr in cores LC25 and SL60, there are nevertheless similarities in the shape of the measured and modeled Sr profiles at around the correct level in the sapropel.

Figure 8.

Model simulation of the total Sr in cores LC25 (upper) and SL60 (lower) versus depth compared with the measured Sr data. The total model assumes three components on the basis of the systematics of Figures 5 and 6. (1) 1 mol of aragonite with an Sr/Ca content of 0.027 is produced along with each mole of S, (2) all other Ca is present as CaCO3 with Sr/Ca = 0.0033, (3) detrital material has Sr/Al = 0.0137. This calculation requires that all S produced from SO42− by Corg remineralization is preserved in the sediments, whereas some fraction may in fact be reoxidized.

[31] It might be anticipated from the changes in pore water alkalinity that must occur during Corg remineralization that carbonate dissolution should occur between sapropel events under oligotrophic conditions with an oxic water column, while precipitation occurs in anoxic conditions during sapropel events. In fact, high contents of precipitated high-Mg calcite are also found in the marls deposited between sapropel events [Calvert and Fontugne, 2001] (Figure 4). This formation of high-Mg calcite has been ascribed to the high salinity of eastern Mediterranean bottom waters [Milliman and Müller, 1973], but the solubilities of aragonite and calcite both increase with increasing salinity [Mucci, 1983]. If the high-Mg calcite and aragonite are diagenetic, the high Mg and Sr contents developed in them suggests that formation must occur early so that the necessary Mg and Sr can be supplied from bottom waters for incorporation into CaCO3. The CaCO3 polymorph that actually precipitates from a saturated pore water solution is a complex function of temperature and pore water composition [Burton and Walter, 1987; Burton, 1993]. At the seawater Mg/Ca molar ratio of 5:1, aragonite is precipitated at temperatures >6°C, while calcites form at lower temperatures under similar conditions [Morse et al., 1997]. The well-mixed water column in the eastern Mediterranean basin everywhere has a temperature >12.5°C [Klein et al., 1999], so that aragonite might be expected always to be the CaCO3 polymorph favored to precipitate in eastern Mediterranean sediments. The fact that high-Mg calcite is dominant outside sapropels (Figures 4 and 5) suggests that some critical local difference in pore water composition must dictate which CaCO3 polymorph forms, i.e., aragonite in and around sapropels but high-Mg calcite in the enclosing Corg-poor marls. Some pore water species associated with sapropels or sulphide formation (e.g., pH, dissolved organic matter, soluble reactive phosphate, or sulphide) may be involved, and pore water composition can suppress CaCO3 precipitation and allow oversaturation to reach high levels [Burton and Walter, 1987; Patterson and Walter, 1994].

[32] In summary, this diagenetic hypothesis is that Corg remineralized by SO42− produced alkalinity that was precipitated to form aragonitic CaCO3 that incorporated Sr from the pore waters. Although the major difference between this and the production hypothesis is that the aragonite forms in the sediments rather than in the surface ocean, this mechanism uncouples aragonite formation from F. profunda production.

3.5. Florisphaera Profunda Abundance and Dissolution and Recrystallization in Biogenic Carbonates

[33] The increases in the relative abundance of F. profunda observed in sapropels might be the result of a decrease in production of upper photic zone coccolithophores, and/or by an increase in F. profunda production in the lower photic zone, or by increased postdepositional dissolution of upper photic zone coccoliths in the sediments. In eastern Mediterranean sediments, Crudeli et al. [2004] have recognized both dissolution and overgrowth effects on coccoliths of Emiliania huxleyi, the most abundant coccolithophorid. Overgrowth effects were particularly marked in the marls above and below S1, corresponding to the levels where Calvert and Fontugne [2001] report high-Mg calcite (Figure 4). Well-preserved, heavily etched, and heavily overgrown E. huxleyi coccoliths could sometimes be identified in the same sample. It is not clear whether dissolution and precipitation occurred simultaneously in different microenvironments or in episodes at different times after deposition.

[34] Coccolithophore primary production changes are mainly influenced by nutrient availability and light [Brand, 1994], such that production occurs high in the photic zone when the nutricline is shallow and lower when the nutricline is deeper [Molfino and McIntyre, 1990a, 1990b; Castradori, 1993; Beaufort et al., 1997; Henriksson, 2000; de Garidel-Thoron et al., 2001]. Gladiolithus flabellatus and Algirosphaera robusta dominate the modern nannoflora assemblage at >100 m in the Mediterranean Sea along with F. profunda [Knappertsbusch, 1993; Ziveri et al., 2000; Cros, 2002]. Although simultaneous production of all three species is expected [Broerse et al., 2000; Takahashi and Okada, 2000], G. flabellatus and A. robusta are rarely reported from the fossil record. Instead, F. profunda is often the only lower photic zone species preserved in deepwater sediments because it is the most solution-resistant species of the three [Matsuoka and Okada, 1989; Okada and Matsuoka, 1996; Okada and Wells, 1997; Takahashi and Okada, 2000; Okada, 2000]. The patterns of G. flabellatus and A. robusta were therefore compared with those of F. profunda in order to evaluate whether coccolithophore dissolution might affect abundance patterns.

[35] G. flabellatus and A. robusta are at much lower abundance, and their profiles are quite different in shape from those of F. profunda in cores BC19, SL114, and SL60 (Figure 9). In cores BC19 and SL114, both species are variably present both above and below sapropel S1 but are at low abundance or completely absent within the sapropel itself. In contrast, both species are continuously present and relatively abundant through the sapropel in core SL60 where the sediments are more rapidly accumulated (Figure 9). Crudeli and Young [2003] observed by SEM that single lamellar fragments and broken coccoliths of A. robusta were abundant in sapropel S1 sediments, but such small lamellar fragments are not detected by LM, and consequently the number of A. robusta coccoliths is underestimated. Whereas G. flabellatus has been reported to have a depth preference closely similar to that of F. profunda, A. robusta has a somewhat wider depth distribution and has also been reported from the middle photic zone [Jordan and Winter, 2000; Cros, 2002]. While the marked decrease of A. robusta in sapropel S1 in cores SL114 and BC19 might therefore be ascribed to some unfavorable condition occurring higher in the photic zone during sapropel deposition, the absence of G. flabellatus in the presence of F. profunda is more difficult to explain on the basis of the known ecological preferences of the two species. It is therefore suspected that the different trends of G. flabellatus and A. robusta by comparison with F. profunda in cores BC19 and SL114 are related to their dissolution and fragmentation in the water column or sediments rather than to environmental conditions. This selective dissolution hypothesis is supported by the presence of etched and broken E. huxleyi coccoliths in sapropel S1 [Crudeli et al., 2004] and by the dominance of structurally solid holococcoliths over more delicate forms [Crudeli and Young, 2003]. Moderately overgrown coccoliths of E. huxleyi were also found in the S1 sapropel sediments of core SL60, whereas such forms are completely absent in the sapropels of cores SL114 and BC19 [Crudeli et al., 2004]. This may imply that early diagenetic carbonate precipitation has allowed a better preservation of the delicate G. flabellatus and A. robusta forms in core SL60, and their slight increase in abundance within S1 in this core supports the idea of proliferation of both species during sapropel deposition.

Figure 9.

Downcore profiles of Florisphaera profunda, Algirosphaera robusta, and Gladiolithus flabellatus in cores SL114 (upper panels), SL60 (middle panels), and BC19 (lower panels). Light and dark shading indicate the oxidized and unoxidized parts of the S1 sapropels.

[36] It is important to note that overall dissolution of the carbonate nannoflora within sapropel S1 is of moderate intensity in all these cores since some very delicate forms are still present [Crudeli and Young, 2003]. G. flabellatus and A. robusta were quantified independently of F. profunda platelets, and their abundance patterns do not affect the corresponding F. profunda profiles. Nevertheless, the G. flabellatus and A. robusta trends are probably evidence that species-selective carbonate dissolution is or has been active in these sediments, and it is possible that dissolution of other coccoliths has affected F. profunda profiles to some extent in cores SL114 and BC19 at least.

3.6. Aragonite Morphology and Composition by Scanning Electron Microscopy and Ion Probe

[37] It was hoped that a choice between the alternatives for the high-Sr aragonite formation might be assisted by SEM and Ion Probe studies, but recrystallization effects complicate the recognition and differentiation of biogenic and diagenetic aragonite. Aragonite and Mg-calcite dissolution and recrystallization are both widespread during very early diagenesis in shallow water carbonates [Macintyre and Reid, 1995; Reid and Macintyre, 1998; Hover et al., 2001]. Substrate control often appears to favor isomorphic recrystallization, and Hover et al. [2001] concluded that the driving force in recrystallization was a minimization of crystal surface free energy through an increase in crystal size. In the case of eastern Mediterranean sediments, however, the E. huxleyi dissolution observed by Crudeli et al. [2004], and the likely dissolution of other species discussed above is evidence that some dissolution of the less soluble biogenic low-Mg calcite must be occurring, or have occurred, in these sediments, despite the presence of the more soluble high-Mg calcite and aragonite forms (Figures 4 and 6).

[38] In samples from the S1 sapropel in cores LC25 and T87-19B, the dominant forms identified on morphological grounds as aragonite by SEM were aggregates of needles or “rods,” typically 1–5 × ∼0.2 μm in dimension and very small (0.5 μm) equant crystals (Figure 10). Ion microprobe determination of Sr and Ca in particles on stubs that had also been examined by SEM confirmed that all biogenic nannoplankton consistently analyzed with a Sr/Ca mass ratio in the range 0.001–0.003, whereas the rod aggregates had Sr/Ca mass ratios in the range 0.02–0.03.

Figure 10.

SEM images of forms interpreted as aragonite in sapropel S1 and Augias Turbidite sediments, illustrating the variations in size and cementation/compaction of rod aggregates. (a) Aggregate of loosely compacted rods, most of which are >1 μm in length. Micritization appears less pervasive in this example than in those of Figures 1c and 10d, although the rods in the lower center of this aggregate appear more cemented. The surfaces of single rods are less irregular than those in Figures 10c and 10d (S1 sapropel at 83 cm in section 10 in core LC25). (b) Small aggregate of rods in a loose framework (S1 sapropel at 83 cm in section 10 in core LC25). (c) Aggregate of compacted rods. Here most rods are ∼1 μm long, and their surfaces are irregular as if etched. (S1 sapropel at 43 cm in core T87-19B.) (d) Aggregate of compacted rods in which most rods are slightly longer than those in Figure 10c but have the same irregular surface. There are distal views of etched E. huxleyi coccoliths to the upper right and upper left, and a proximal view of a well-preserved E. huxleyi to the bottom left (S1 sapropel at 43 cm in core T87-19B). (e) Well-preserved coccoliths of A. robusta (center) and F. profunda (immediately left of center). There is a subrounded aragonite(?) nanograin (0.5 μm in diameter) three quarters way up the frame on the right hand side (S1 sapropel at 43 cm in core T87-19B). (f) Loose aggregate of very thin rods mostly >1 μm long. Organic matter(?) obscures thin rods at the lower right side of the image (S1 sapropel at 83 cm in section 10 in core LC25). (g) Aggregate of closely spaced, medium-sized rods with a faecal pellet-like morphology (Augias Turbidite at 12 cm in core T87-19B). (h) Aragonite rods surrounding a coccolith of E. huxleyi that shows etching on the elements of the distal shield (Augias Turbidite at 12 cm in core T87-19B). (i) Unidentified particle with extremely high Sr/Ca values (center) with a F. profunda platelet at the upper left (S1 sapropel at 83 cm in section 10 in core LC25). (j) Magnification of a particle with a high Sr/Ca ratio showing inclusion of a small Syracosphaera coccolith that indicates a probable later diagenetic origin for this particle (S1 sapropel at 83 cm in section 10 in core LC25).

[39] Both well-separated and more fused rods were observed in the different aggregates (Figures 10a–10d and 10f). In those aggregates with a loose open framework of randomly oriented rods, the rods commonly had smooth surfaces and rounded terminations (Figures 10a and 10b) and resembled the original aragonitic skeletal needles of the calcareous alga Halimeda reported by Macintyre and Reid [1995, Figures 2b and 2c]. The thicker rods in these aggregates did not, however, exhibit the blunt crystal terminations and well-developed crystal faces typical of needles of living Halimeda [Hover et al., 2001, Figure 6c]. The second type of aggregate, in which the rod framework appears more fused, is more common (Figures 10c and 10d), and in this type the rods have more irregular surfaces, rounded terminations, and some cementation between them. The surfaces of these rods appear better preserved than the etched surface of bioneedles illustrated by Macintyre and Reid [1995, Figure 7b]. Morphologically, these rods resemble etched aragonitic needles from carbonate platforms reported by Melim et al. [2002, Plate II, Figure c]. A third type of aggregate with a loose framework, characterized by well-spaced elongate crystals (typically 1–5 m long × 0.07 m thick) was also observed. This form was commonly surrounded by what appeared to be amorphous organic matter (Figure 10f), and the crystals vaguely resemble the inorganic precipitates reported by Robbins and Blackwelder [1992, Figure 3b] and aragonitic needles from living calcareous algae [Reid and Macintyre, 1998, Figure 1e]. In addition to aggregates, randomly spaced single rods were also common, and subcircular anhedral microcrystals (0.5 m) were occasionally present (Figure 10e) that are similar to the nanograins produced by progressive micritization of original skeletal needles documented by Macintyre and Reid [1995, Figures 3c, 3e, 6b, 6e] and to the crystal form of many calcareous algae [Macintyre and Reid, 1992, Figure 2].

[40] The sapropel S1 aggregates are typically smaller (rarely >20 μm with an average of all types of ∼8 μm) and more irregular in shape than those from the Augias turbidite (Figures 10g and 10h) that were referred to Halimeda sp. by Hieke and Werner [2000]. Some larger aggregates were subcircular to elliptical in shape, 5–10 m in length (Figure 10g), and resembled faecal pellets [e.g., Wassmann et al., 2000]. Although the individual needles comprising the different aggregates appear similar, the degree of compaction and fusion always appears less in the sapropel S1 aggregates than in the Augias turbidite grain illustrated by Hieke and Werner [2000cite-ref>, Figure 2].

[41] The irregular rod surfaces and cements might indicate recrystallization of original aragonite needles (J. Reijmer, personal communication, 2003), and the predominantly needle- or rod-like morphology might suggest a biogenic origin of these aggregate structures. Since etching and progressive micritization have been shown to alter original aragonitic needle morphology dramatically, however [Macintyre and Reid, 1995; Hover et al., 2001], postdepositional alteration of the original morphology is also possible.

[42] A few irregular small particles with much higher Sr/Ca mass ratios in the range 0.12–0.15 were also observed in ion microprobe analysis (Figure 10i), and the fact that one of these particles appears to have grown around a Syracosphaera coccolith (Figure 10j) suggests that they are likely to be diagenetic. While it was not possible to identify the nature of these high Sr particles further, their occasional presence may contribute to the aragonite end-member that has a Sr/Ca mass ratio higher than that anticipated for biogenic aragonite (Figure 6). Rare examples of distinctive aragonitic didemnid ascidians with sparse aragonite needle overgrowths were observed (not shown), and these are likely to have been sourced from shallow waters [Varol and Houghton, 1996]. Ion probe determinations of these didemnid ascidians indicated Sr/Ca mass ratios in the range 0.02–0.03, similar to the aragonite rods.

3.7. Preferred Mechanism for Aragonite in Sapropels and Implications for the Evolution of Sapropel Formation Episodes

[43] The difficulties experienced in ascribing genesis to the aragonite morphologies observed by SEM do not allow an unequivocal selection between the three alternatives of aragonite supply discussed above. No evidence was found for any high Sr biogenic forms that would match the production mechanism, however, and although the detrital hypothesis cannot be discounted, the diagenetic explanation appears the most likely. The reasons for this are the observed alternation of aragonite with diagenetic high-Mg calcite (Figures 4 and 5), the indicated high Sr end-member contents that exceed biogenic high-Sr aragonite and resemble those of aragonitic cements (Figure 6), and the co-occurrence of highest Sr/Ca ratios with high S contents because of the inescapable alkalinity increases that accompany sulfate reduction (Figure 3). Diagenetic carbonate formation is so widespread in eastern Mediterranean sediments that hardgrounds, generally composed of high-Mg calcite, are sometimes formed at the sediment surface [Allouc, 1990]. Hardgrounds with a different genesis have also been reported at and just under the sediment water interface at cold seep sites in the eastern Medoterranean basin [Aloisi et al., 2000, 2002]. In cold seep examples, methane oxidation and sulphate reduction are both active at shallow depth in the sediments, so that an increase in pore water alkalinity accompanies methane oxidation by sulfate [Lein et al., 2002]:

equation image

[44] Aloisi et al. [2002] ascribed an observed alternation of aragonite at the surface and high-Mg calcite at greater depth in cold seep crusts to an inhibition of high-Mg calcite precipitation by pore water sulfate, with aragonite formed at higher pore water sulphate levels and high-Mg calcite formed where pore water sulfate was depleted. Such an explanation will not serve for the sediment observations here of preferential high-Mg calcite formation in the marls and aragonite formation in the sapropels. It is clear that the diagenetic precipitation of either high-Mg calcite or aragonite must be in delicate balance in eastern Mediterranean sediments, and the alternations observed in the sediments here and in the cold seep hardgrounds could both be explained by a preferential formation of aragonite along with sulfate reduction, possibly by sulfate reducing bacteria.

[45] If the alkalinity generated with sulphide formation is the source for the formation of diagenetic aragonite in sapropels, then formation of aragonite and the proliferation of F. profunda must be indirectly related. From studies of laminated sapropel intervals, Kemp et al. [2000] have suggested that the autumn pulse of diatom production in the DCM represents an important fraction of total annual production. Although this could lead to the speculation that F. profunda might be a useful indicator for overall (as opposed to just DCM) production during sapropel formation, the differences in the F. profunda and Ba/Al (as an indicator of the original Corg profiles in the S1 units) profiles clearly suggest rather different production patterns through time.

[46] S1 formation is usually equated with the Holocene African Humid Period [e.g., Rossignol-Strick, 1999; Gasse, 2002]. Application of Castradori's [1993] concept to the F. profunda profiles through S1 implies that the DCM component of total upper ocean production and hence monsoon outflow from the continents were at a maximum early in S1 times. F. profunda values begin to increase from low (baseline) values at ∼11 14C kyr B.P., peak around 9 14C kyr B.P., and subsequently decline toward baseline until 5 14C kyr B.P. (Figure 3). In contrast, the increased Corg contents and high Ba/Al ratio values used to indicate sapropel thickness do not increase above their low baseline values until after 10 14C kyr B.P., rise sharply between 10 and 9 14C kyr B.P., maintain consistently high levels between 9 and 6.5 14C kyr B.P., and then fall off between 6.5 and 6 14C kyr B.P. These differences between the F. profunda and Ba/Al profiles suggest that production in the surface ocean must increase well before sapropel formation begins in the sediments. In turn, this suggests that development of anoxia/dysoxia between 11 and 10 14C kyr B.P in a stabilized, poorly ventilated deeper water column was necessary before high Corg contents could begin to be preserved in the sediments to form sapropel. Consistently high values of Corg and Ba/Al were then maintained in the sediments deposited between 9 and 6.5 kyr, whereas the F. profunda profiles suggest that at least the DCM component of total production was continuously decreasing between 9 and 5 14C kyr B.P. The return of Ba/Al to its baseline values at 6 14C kyr B.P. also occurs before F. profunda returns to its baseline at 5 14C kyr B.P. (Figure 3). In reverse of the scenario to initiate sapropel development, this suggests that water column stabilization must have broken down and good ventilation must have resumed to reoxygenate deep water after 6.5 14C kyr. Thus sapropel formation terminated after 6 14C kyr B.P., even though surface ocean production did not return to the low levels of F. profunda that typify nonsapropel times until 5 14C kyr B.P.

4. Conclusions

[47] Unlike most other deepwater marine sediments, eastern Mediterranean basin sediments contain both the more soluble CaCO3 forms aragonite and high-Mg calcite, predominantly in sapropels and in the intercalated marls, respectively. The repeated finding of maxima in the F. profunda index and Sr/Ca ratio profiles around the same depths in the most recent sapropel was unexpected because F. profunda, other biogenic calcite, and pteropod aragonite cannot be the source of the observed high-Sr aragonite. Although the aragonite could have been formed or emplaced by other mechanisms, the alkalinity generated by sulphate reduction appears to be the likely principal source for the formation and recrystallization of aragonite in and around sapropels.

[48] Similar high Sr/Ca ratios and inferred aragonite formation also occur at, or immediately below, the bases of the last few eastern Mediterranean sapropels, suggesting that aragonite generally accompanies sapropel formation in this basin. Regardless of the mechanism of formation of high-Sr aragonite, the F. profunda maxima and δ18O monsoon rainfall minima always occur very early in the episodes when high Corg contents are preserved as sapropels in the sediments along with highest diagenetic sulfide contents. In the case of sapropel S1 the development of the production pulse in the upper water column that is signaled by F. profunda clearly begins earlier and finishes later than the resultant formation of sapropel in the sediments.

Acknowledgments

[49] This work was partly supported by the EU Marine Science and Technology Framework V project Sapropels and Paleoceanography (MAS3-CT97-0137) and the Natural Sciences and Engineering Research Council of Canada (SEC). We are grateful to Trevor Clayton of SOC for XRD analyses, to Simon Troelstra of The Free University, Amsterdam, for the core T87-19B samples, to Guy Rothwell of the British Ocean Sediment Core Research Facility at SOC for the core LC25 samples, to Ute Schuldt and the Ocean Gateways Research Unit at Kiel University for assistance with and support of the SEM analyses, and to John Reimer of IFM-GEOMAR for discussions on the aragonite aggregates. Constructive referee comments from T. Lyons and H. Stoll on the original manuscript greatly improved this version.

Ancillary