Gypsum as a monitor of the paleo-limnological–hydrological conditions in Lake Lisan and the Dead Sea
Introduction
Modern deposition of evaporites commonly takes place in shallow-water saline lakes such as Lake Magadi (Eugster, 1970) and Lake Chad (Eugster and Maglione, 1979) in Africa, Lake Eyre basin in Australia (Magee et al., 1995), Great Salt Lake in North America (Spencer et al., 1985), and many others. However, the majority of thick evaporite sequences known from geological records precipitated from seawater-derived solutions such as the Mediterranean Messinian (Hsu et al., 1977, Krijgsman et al., 1999), the Permian basins of Texas and the Zechstein in Europe (cf. Anderson et al., 1972, Glennie and Buller, 1983, Warren, 1989; and references therein). These thick sequences however, could not have precipitated directly from seawater which are undersaturated with respect to the dominant evaporitic minerals (i.e., gypsum, halite). Thus, massive evaporite deposition is generally explained by the occurrence of repeated cycles of desiccation and re-filling or alternatively, a continuous inflow and evaporation configuration (cf. Hsu et al., 1977, Krijgsman et al., 1999, Warren, 1989). The hypersaline nature of the precipitating solutes, the varying degrees of drawdown required for the precipitation of thick evaporate sequences and the depth of water (i.e., deep- vs. shallow-water) in such ancient water bodies are difficult to estimate and have been widely debated (cf. Warren, 1989, Kendall, 1992; and references therein).
The Dead Sea, which is a unique modern analogue to deep-water evaporite systems, is a hypersaline, Ca-chloride lake that despite extremely high calcium concentrations (∼17 g/l) does not support massive gypsum precipitation. This mainly reflects the low sulfate concentration (<0.4 g/l). On the other hand, sediments deposited from Lake Lisan, the late Pleistocene precursor of the Dead Sea, contain thick sequences of gypsum, indicating enhanced supply of sulfate to the lake (Begin et al., 1974, Stein et al., 1997, Torfstein et al., 2005).
In this study we set out to understand the hydrological–limnological conditions that enabled the deposition of primary gypsum in the hypersaline Ca-chloride lacustrine water bodies in the late Quaternary Dead Sea basin. The water bodies represent terminal amplifier lakes whose geochemical and physical characteristics reflect the hydrological-climatic conditions that existed in their large drainage area and therefore provide important information regarding the climatic history of the Levant during the Quaternary (cf. Stein, 2001, Enzel et al., 2003). We focus on the hydrological conditions during the intervals of massive gypsum deposition, which occurred during relatively arid stages and subsequent lake level declines (Stein et al., 1997). We investigate a major cycle of gypsum deposition spanning the transition from Marine Isotopic Stage 2 (MIS2) to MIS1 and use sedimentological and geochemical (δ34S) data to characterize the depositional environment. Exploring the deposition mechanisms of gypsum in the Dead Sea basin water bodies and explaining their connection to the recurring fluctuations between long-term high-stands to long- and short-term low-stands provides strong constraints on the depth of the water and the regional hydrological-climatic regime. In addition, the differences in the limnological–hydrological setting between the last glacial and the Holocene water bodies are discussed.
The series of water bodies that existed in the tectonic depressions along the Dead Sea Rift (DSR) during the Quaternary evolved from the ancient (Pliocene?) Sedom Lagoon which derived its chemical constituents from ingressing seawater and interaction with the carbonate wall-rock. Subsequently, DSR brines obtained their unique Ca-chloride composition (Zak, 1967, Starinsky, 1974, Stein et al., 2000, Stein et al., 2002). After the disconnection of the Sedom lagoon from the open sea, the limnological and geochemical history of the now closed and terminal water bodies was controlled by the interaction between the brine and the freshwater input from the drainage area of the Dead Sea basin (Katz et al., 1977, Stein, 2001, Gavrieli and Stein, 2006).
The freshwater influx and the limnological configuration of the lakes (stratified vs. homogeneous water column) reflect the climatic–hydrologic history of the region, which fluctuated during the Quaternary between arid to semiarid conditions (Begin et al., 1974, Bartov et al., 2002, Bartov et al., 2003, Prasad et al., 2004, Haase-Schramm et al., 2004) and correlate to global climatic trends, as recorded by oxygen isotopic ratios and other proxies in Greenland ice and deep-sea cores.
During wet stages, lake level rose and water column stratification developed. Long-term stratification required a positive water balance maintained by freshwater influx. Overturn and mixing of the water column occurred when the salinity of the upper water body (the mixolimnion) increased again, which occurred upon a shift to a negative water balance. Changes in the water column configuration and consequent geochemical processes are reflected by the chemical and isotopic composition of the sediments (i.e., Sr/Ca and 87Sr/86Sr ratios (Katz et al., 1977, Katz and Kolodny, 1989, Stein et al., 1997) and δ34S values (Torfstein et al., 2005)).
The Lisan Formation (Fm.), deposited by the Late Pleistocene Lake Lisan (∼70 to 14 ka BP; Kaufman, 1971, Haase-Schramm et al., 2004), is exposed along the Dead Sea basin, from the northern Arava valley in the south to the Sea of Galilee in the north (Fig. 1).
The lake was sensitive to hydrological changes in its large drainage area, and its level, which responded to wet and arid episodes within the last glacial interval, fluctuated between ∼160 to 370 m below mean sea level (mbsl) (Fig. 2; Bartov et al., 2003). During high stands the lake precipitated sequences of thin (∼1 mm) aragonite laminae alternating with silt-sized detrital laminae (this facies is termed alternating- aragonite-detritus (“aad”) facies; Machlus et al., 2000). While the aragonite precipitated chemically from the lake, the detrital laminae, composed mainly of calcite, quartz, dolomite and clay, are the erosion products from the Dead Sea rift shoulders as well as eolian input from more distant terrains (Neev and Emery, 1967, Begin et al., 1974, Haliva et al., 2003). During low stands, the lake deposited gypsum as the main evaporitic mineral phase.
The main study site in this work is the M1 sedimentary section located on the foothills of the Massada archeological site (Fig. 1). Here, the base of the formation is located at 374 mbsl, overlying the exposed top of the last interglacial Samra Fm. (Waldmann et al., 2007), and is approximately 30 m thick. Additional samples were collected from Perazim Valley, Mt. Yizrach, Nahal Mor, Nahal Mishmar, Bet Ha’arava and Deir Shaman (Fig. 1).
The Massada M1 section consists of three stratigraphic Members (Fig. 2). The Lower Member is 5.6 m thick and consists mainly of aad sequences, which are capped by a sequence containing three gypsum layers. The latter can be correlated over large distances in the Dead Sea basin (Waldmann et al., 2007). The Middle Member is 11.5 m thick and is comprised of abundant clastic beds (sands and silts), which alternate with aad packages and some gypsum layers. The Upper Member is 13.5 m thick and consists of a 9 m thick aad sequence which is capped by a ∼2 m thick sequence of gypsum layers alternating with aad bundles. These are overlain by an additional unit of aad and gypsum. The transitions from the Lower to Middle and from the Middle to Upper Members coincide with the main Marine Isotopic Stages (MIS) boundaries: 4–3 and 3–2, respectively (Stein, 2001, Bartov et al., 2003, Haase-Schramm et al., 2004).
The Lisan Fm. is overlain by sediments of the Ze’elim Fm. deposited from the Holocene Dead Sea, whose typical levels were ∼400 ± 30 mbsl (Bookman (Ken-Tor) et al., 2004, Migowski et al., 2006). The Ze’elim Fm. consists of layered calcitic marls, alternating laminae of aragonite–silty detritus couplets or aragonite–gypsum–silty detritus triplets, and several sand layers (Migowski et al., 2004, Migowski et al., 2006). No massive gypsum units were deposited by the Holocene Dead Sea and the overall amount of aragonite is small relative to the Lisan Fm.
By definition, Ca-chloride solutions are low in sulfate (Hardie and Eugster, 1970, Starinsky, 1974, Eugster and Jones, 1979). Indeed, the Ca-chloride water bodies of the DSR are (and were) characterized by relatively low sulfate concentrations (e.g., <400 mg/l in the 20th century Dead Sea, compared to calcium concentration of ∼17,000 mg/l; Neev and Emery, 1967). The modern Dead Sea brine is supersaturated with respect to gypsum (Katz et al., 1981) and the residence time of dissolved sulfate in it is relatively short (100–1500 years; Torfstein et al., 2005). Thus, it has been suggested (Stein et al., 1997) that the sulfate dissolved in the DSR lakes is derived mainly from inflowing freshwater.
Based on the analyses of various freshwater sources in the DSR drainage area, the bulk isotopic composition of dissolved sulfate presently discharging into the Dead Sea is estimated to lie within δ34S ≈ 6–14‰. Since the same water sources were active in the Late Pleistocene too, it can be assumed that this range also represents the isotopic composition of the dissolved sulfate that reached Lake Lisan (Torfstein et al., 2005). However, sulfur occurrences in the Lisan Fm. display a bimodal isotopic composition (<1‰ and >14‰; Fig. 3, Fig. 4; Torfstein et al., 2005), which is inconsistent with direct precipitation of gypsum from the water column. Rather, these compositions imply the occurrence of sulfur isotopic fractionation induced by bacterial sulfate reduction (BSR). This process takes place under anoxic conditions following the simplified reaction:
The anoxic conditions required for BSR could have evolved in the lake’s lower water body (the monimolimnion), when the lake was stratified, or within the bottom sediments. The source and type of the organic matter that was oxidized during the BSR is not known although it has been attributed to algal blooms in the diluted mixolimnion (Oren et al., 2004, Kolodny et al., 2005). BSR is typically accompanied by isotopic fractionation whereby preferential reduction of 32 results in the remaining sulfate anions becoming enriched in 34S (Kaplan and Rittenberg, 1964, Kemp and Thode, 1968, Chambers et al., 1975, Habicht and Canfield, 1996, Canfield, 2001). In the modern Dead Sea and the surrounding saline springs, the isotopic fractionation associated with BSR has been determined to be 25–30‰ (Nissenbaum and Kaplan, 1976, Gavrieli et al., 2001). A similar value was attributed to the Lisan system (Torfstein et al., 2005).
During wet periods, when the lake was stratified and undersaturated with respect to gypsum, dissolved sulfate that entered the lake first accumulated in the mixolimnion. When enough sulfate accumulated in the mixolimnion gypsum saturation was reached and additional sulfate inflow lead to gypsum precipitation. In contrast, intensive BSR in the monimolimnion decreased the sulfate concentration therein, which became under-saturated with respect to gypsum. Any gypsum that precipitated from the mixolimnion and sank through the monimolimnion was thus susceptible to dissolution. Accordingly, inflowing sulfate was transported from the mixolimnion via continuous gypsum precipitation to the monimolimnion where it re-dissolved. Concurrent BSR in the monimolimnion produced residual 34S-enriched sulfate as well as 34S-depleted sulfide which was removed to the sediments. Torfstein et al. (2005) calculated that a steady state value of was reached in the monimolimnion after several thousands years of meromictic conditions.
Upon a decrease in the freshwater flux into the lake, lake level dropped and the increased salinity (and thus density) of the mixolimnion led to overturn and mixing of the water column and to massive precipitation of gypsum. The isotopic composition of sulfur in this gypsum represents that of the mixed water body which in turn depends on the relative volumes, dissolved sulfate concentrations and isotopic compositions of sulfate in the two water layers before the overturn.
The 34S-depleted sulfides were most probably removed from the brine through precipitation of Fe-sulfide phases and were buried in the sediments. The high sedimentation rate of Lake Lisan decoupled these phases from the aqueous system soon after deposition. Thus, during water column mixing and oxidation the reduced sulfides were not susceptible to immediate oxidation and therefore were not recycled back into the brine. Sulfide oxidation took place only after the retreat of the lake and sediment exposure. Because sulfide oxidation involves only minor isotopic fractionation (Fry et al., 1986, Fry et al., 1988, Habicht et al., 1998) the original low δ34S values were preserved in the oxidized sulfate which, given the abundance of calcium in the exposed sediments, promptly precipitated and formed disseminated and thin laminae of gypsum. Evidence for the extensive oxidation of the Lisan Fm. sediments was provided by Ron et al. (2006). Based on the magnetic properties of the sediments, they suggested that only negligible fractions of the original reduced phases were preserved. By comparison, they observed that the freshly exposed Holocene Ze’elim Fm. displays a higher content of ferromagnetic Fe-sulfide reduced phases (i.e., greigite).
Section snippets
Sampling
Samples, approximately 30 × 30 × 30 cm3 in size, were collected from several sections along the DSR (Fig. 1). The samples were wrapped in adhesive plastic foils to prevent disintegration of the soft sediments. Gypsum samples from Holocene (Dead Sea) sediments were picked by C. Migowski from a sedimentary core drilled at the En-Gedi shore (a detailed description of the core is given in Migowski et al., 2006).
The gypsum content in two primary gypsum layers in the M1 section was estimated by extracting
Sulfur occurrences in the Lisan Formation
Sulfur in the Lisan Fm. occurs in several forms (Table 1, Table 2, Table 3 and Fig. 3, Fig. 4, Fig. 5):
Limits on sulfate concentration in the lakes
In the Ca-chloride Lake Lisan brine, high Ca2+ concentration would not be significantly modified by mineral precipitation or dissolution; rather, it would display a relatively conservative behavior whereby Ca2+ concentration reflects the degree of dilution of the brine. In contrast, dissolved sulfate concentration in the lake would be dictated by the rate of sulfate inflow, gypsum solubility and Ca2+ concentrations. Thus, the lower the salinity and Ca2+ concentration, the higher the
Conclusions
- (1)
Sulfur occurs in the late Pleistocene Lisan and Holocene Ze’elim Formations in a variety of forms, each characterized by specific sedimentologic properties and isotopic compositions. The main sulfur reservoir in the sediments, primary gypsum, is 34S-enriched (14–28‰) compared to the other sulfur reservoirs and compared to the source of sulfur to the lake (∼10‰). The isotopic shifts are attributed to bacterially-mediated sulfate reduction and concurrent isotopic fractionation, that took place in
Acknowledgments
We thank A. Amrani and Z. Aizenshtat for advice and support in laboratory work and E. Shalev for valuable discussions. We also thank Blair Jones and two anonymous reviewers for helpful and constructive remarks. The study was supported by BSF 2000271 (to I.G. and M.S.) and GIF 717.129.8/2001 (to M.S. and Y.K.).
References (69)
- et al.
Lake levels and sequence stratigraphy of Lake Lisan, the late Pleistocene precursor of the Dead Sea
Quat. Res.
(2002) - et al.
Southward migration of rain tracks during the last glacial, revealed by salinity gradient in Lake Lisan (Dead Sea rift)
Quat. Sci. Rev.
(2004) Isotope fractionation by natural populations of sulfate-reducing bacteria
Geochim. Cosmochim. Acta
(2001)- et al.
Late Holocene climates of the Near East deduced from Dead Sea level variations and modern regional winter rainfall
Quat. Res.
(2003) - et al.
Brines and evaporites of the Lake Chad basin, Africa
Geochim. Cosmochim. Acta
(1979) - et al.
The sulfur system in anoxic subsurface brines and its implication in brine evolutionary pathways: the Ca-chloride brines in the Dead Sea area
Earth Planet. Sci. Lett.
(2001) - et al.
The Permian Weissliegend of NW Europe: the partial deformation of eolian sands caused by the Zechstein transgression
Sed. Geol.
(1983) - et al.
U–Th dating of Lake Lisan (late Pleistocene Dead Sea) aragonite and implications for glacial East Mediterranean climate change
Geochim. Cosmochim. Acta
(2004) - et al.
Sulfur isotope fractionation during bacterial reduction and disproportionation of thiosulfate and sulfide
Geochim. Cosmochim. Acta
(1998) - et al.
Hypersaline brine diagenesis and evolution in the Dead-Sea Lake Lisan system (Israel)
Geochim. Cosmochim. Acta
(1989)