torfstein et al., 2009 (gca, amora chronology)

8/6/2019 Torfstein Et Al., 2009 (GCA, Amora Chronology)

1/28

U-series and oxygen isotope chronology of the mid-PleistoceneLake Amora (Dead Sea basin)

Adi Torfstein a,b,*,1, Alexandra Haase-Schramm c,d, Nicolas Waldmann e,Yehoshua Kolodnya, Mordechai Stein b

a Institute of Earth Sciences, The Hebrew University of Jerusalem, Jerusalem 91904, Israelb Geological Survey of Israel, 30 Malkhe Israel Street, Jerusalem 95501, Israel

c Max-Planck Institut fur Chemie, Postfach 3060, 55020 Mainz, Germanyd Leibniz-Institut fur Meereswissenschaften, IFM-GEOMAR, Wischhofstrasse 1-3, D-24148 Kiel, Germany

e Department of Earth Science, University of Bergen, Allegt 41, 5007 Bergen, Norway

Received 23 May 2008; accepted in revised form 11 February 2009; available online 20 February 2009

Abstract

This study establishes for the first time the chronology and limnological history of Lake Amora (Dead Sea basin, Israel),whose deposits (the Amora Formation) comprise one of the longest exposed lacustrine records of the Pleistocene time. TheAmora Formation consists of sequences of laminated primary aragonite and silty-detritus, Ca-sulfate minerals, halite andclastic units. This sedimentary sequence was uplifted and tilted by the rising Sedom salt diapir, exposing $320 m of sedimentson the eastern flanks of Mt. Sedom (the Arubotaim Cave (AC) section).

The chronology of the AC section is based on U-disequilibrium dating (230Th234U and 234U238U ages) combined with floatingd

18O stratigraphy andpaleomagnetic constraints.Thedetermination of the230Th234U agesrequiredsignificantcorrectionsto account

for detrital Th and U. These corrections were performed on individual samples and on suites of samples from several stratigraphichorizons.The most reliable corrected ages were used to construct an age-elevation model that wasfurther tunedto theoxygen isotoperecord of east Mediterranean foraminifers (based on the long-term similarity between the sea and lake oxygen isotope archives).

The combined U-series-d18O age-elevation model indicates that the (exposed) Amora sequence was deposited between$740and 70 ka, covering seven glacialinterglacial cycles (Marine Isotope Stages (MIS) 18 to 5).

Taking the last glacial Lake Lisan and the Holocene Dead Sea lacustrine systems as analogs of the depositionallimnologicalenvironment of Lake Amora, the latter oscillated between wet (glacial) and more arid (interglacial) conditions, represented bysequences of primary evaporites (aragonite and gypsum that require enhanced supply of freshwater to the lakes) and clastic sed-iments, respectively. The lake evolved from a stage of rapid shifts between high and low-stand conditions during$740 to 550 kato a sabkha-like environment that existed (at the AC site) between 550 and 420 ka. This stage was terminated by a dry spell rep-resented by massive halite deposition at 420 ka (MIS12-11). During MIS10-6 the lake fluctuated between lower and higherstands reaching its highest stand conditions at the late glacial MIS6, after which a significant lake level decline correspondsto the transition to the last interglacial (MIS5) low-stand lake, represented by the uppermost part of the Formation.

d18O values in the primary aragonite range between 6.0 and 1.3&, shifting cyclically between glacial and interglacial inter-

vals. The lowestd18O values are observed during interglacial stages and may reflect short and intense humid episodes that inter-mittently interrupted the overall arid conditions. These humid episodes, expressed also by enhanced deposition of travertinesand speleothems, seem to characterize the Negev Desert, and in contrast to the overall dominance of the AtlanticMediterra-nean system of rain patterns in the Dead Sea basin, some humid episodes during interglacials may be traced to southern sources. 2009 Elsevier Ltd. All rights reserved.

0016-7037/$ - see front matter 2009 Elsevier Ltd. All rights reserved.

doi:10.1016/j.gca.2009.02.010

* Corresponding author. Fax: +1 845 652 0439.E-mail address: [email protected] (A. Torfstein).

1 Present address: Lamont-Doherty Earth Observatory, Columbia University, USA.

www.elsevier.com/locate/gca

Available online at www.sciencedirect.com

Geochimica et Cosmochimica Acta 73 (2009) 26032630
mailto:[email protected]:[email protected]


2/28

1. INTRODUCTION

The Quaternary climate is characterized by cyclic shiftsbetween different combinations of cold, warm, humid anddry conditions at various sites and time scales. Thesechanges are recorded in many geological archives such asice cores, deep-sea sediments, coral reefs, speleothems and

lacustrine sedimentary sequences.The paleo-climatic exploration of these archives, re-

quires the establishment of reliable and precise chronolo-gies, which beyond the radiocarbon time scale ($50 ka)are usually obtained by U-series methods (e.g., Broecker,1963; Edwards et al., 1987; Ludwig et al., 1992; Winogradet al., 1992; Stein et al., 1993; Kaufman et al., 1998) orluminescence-based techniques (e.g., Stokes, 1999; Lianand Roberts, 2006). The best targets for U-series datingare probably corals (Edwards et al., 2003) and speleothems(e.g., Wang et al., 2001; Bar-Matthews et al., 2003; Rich-ards and Dorale, 2003) although even there, contaminationby detrital material and diagentic effects should be consid-

ered (Thompson et al., 2003; Villemant and Feuillet,2003; Lazar et al., 2004). In the case of lacustrine carbon-ates, UTh dating is hampered by the need to correct forboth detrital additions of U and Th and for the presenceof initial (hydrogenous) Th (Kaufman, 1971, 1993; Linet al., 1996; Israelson et al., 1997; Kaufman et al., 1992;Haase-Schramm et al., 2004; Stein and Goldstein, 2006).Contrary to the 230Th234U and 231Pa235U methods, whichare only applicable for sediments younger than $500 and200 ka, respectively, the 234U238U disequilibrium methodcan be used to extend the datable range up to 1 Ma (Cher-dyntsev, 1971; Ludwig et al., 1992; Richards and Dorale,2003), although it depends on a good knowledge of the ini-tial 234U/238U activity ratio. Additional chronological con-

straints can be obtained from the reconstruction of changesin the stable isotopes, chemical ratios or paleontologicalmarine records, and correlating them to the well-knownchanges in Earths orbital parameters. In the case of conti-nental records (e.g., Winograd et al., 1992; Tzedakis et al.,2003) such a correlation is not as straightforward becausethe sealand climate connection is not always understoodor is not strong. Studying the nature of this connection isa pre-requisite for the correlation of continental and marinerecords (e.g., Kolodny et al., 2005).

In this study we set out to establish a chronologicalframework for the mid to late Pleistocene Amora Forma-tion (Fm), deposited in the lacustrine environment of theDead Sea basin (DSB) (Figs. 1 and 2). Together with itssuccessors (Lake Lisan and the Dead Sea), this sequenceof sediments comprises one of the longest records of itskind.

Most investigations of lacustrine records have focusedon late Pleistocene and Holocene time scales (i.e.,


3/28

chronology is presently available for the past $130 ka, witha few ages reaching $300 ka. Beyond that time, the DSBlacustrine bodies have not been directly dated. Based onstratigraphic considerations, Zak (1967) suggested thatthe Sedom lagoon existed between the late Pliocene andearly Pleistocene, and that the age of the Amora Fm. is be-tween $1 Ma and $100 ka.

2. BACKGROUND

2.1. The Pleistocene Lake Amora

The history of water bodies in the DSB began with theingression of the Mediterranean Sea during Pliocene times(the Sedom Lagoon; Zak, 1967). The interaction of theevaporated seawater filling the Sedom lagoon with the car-bonate wall-rocks of the Dead Sea rift valley together withsulfate reduction led to the production of a Ca-chloridebrine, which played a major role in the subsequent evolu-tion of the lacustrine bodies (Starinsky, 1974). After the dis-

connection of the Sedom Lagoon from the open sea, severallakes developed in the tectonic depressions along the DSBJordan Valley (i.e., the mid to late Pleistocene Lake Amora,the last glacial Lake Lisan ($7014 ka), and the HoloceneDead Sea; Figs. 1 and 2). The size, levels and compositionof these lakes have changed through time, reflecting the cli-matichydrologic history of the region, which fluctuated

from arid to semi-arid conditions (Zak, 1967; Neev andEmery, 1967; Begin et al., 1974; Stein et al., 1997; Stein,2001; Bartov et al., 2003).

Readers should be aware that in several previous pub-lications (e.g., Zak, 1967; Horowitz, 1987, 2001; Steinitzand Bartov, 1991) the name Amora Fm. has been attrib-uted to different sedimentary sections, which do not nec-essarily correlate to each other; in this paper, followingZak (1967), the Amora Fm. refers to the lacustrine se-quence overlying the Sedom Fm. and underlying the Li-san Formation. According to the new chronologypresented here, the Samra Fm. defined at the PerazimValley (PZ2; Waldmann et al., 2007) is time-equivalent

Sedom Fm.

Amora Fm.

Amora Fm.

West EastPZ2

Arubotaim Cave

C

Mt.Sedom

PZ2

Massada

En-Gedi

Zeelim

Sedom-Deep-1

0 10 km

B

Arubotaim

Cave

N

A

0 50 km

29

30

31

33

29

30

31

32

35 34

35 33 32

Med

iterra

nean

Sea

Sinai

Maximum extent

of Lake Lisan

Sea ofGalilee

TheDeadSea

Gulf

ofEla

t

Gulfof

Suez28

Negev

AravaV

alle

y

B

Th

eD

ead

Sea

Fig. 2. Location map. (A) The Dead Sea basin area and maximum extent of the last glacial Lake Lisan. The study site is located in the

southern part of the Dead Sea basin, (B) The main sequence is exposed near the Arubotaim Cave on the north-eastern flanks of MountSedom. Another section is exposed in the Perazim Valley (PZ2 section), west of Mount Sedom, (C) A schematic cross-section of the Sedomdiapir (modified after Zak, 1967). Note the vertical to sub-vertical tilting of the layers. The gray layer in the Amora Fm. marks the AFM-IIIsalt unit.

Lake Amora mid-Pleistocene chronology 2605


4/28

to uppermost Member of the Amora Fm. (i.e., AFM-V,see discussion below).

Although the time of the birth of the early lacustrinebodies in the Dead Sea basin has not yet been established,boreholes indicate that the full subsurface thickness oflacustrine sediments could reach several kilometers (Horo-witz, 1987, 2001; Salhov et al., 1994). There is evidence

for early Pleistocene fresh water-bodies in the northern Jor-dan valley (i.e., Erk-el-Ahmer and Ubedia Fms.; Braunet al., 1991; Raab et al., 1997 and Gesher Benot Yaaqov;Rosenfeld et al., 2004), but their stratigraphic relation tothe lakes in the southern DSB is not clear.

The Amora Fm. exposure at the Arubotaim Cave (AC)site located on the eastern flanks of Mt. Sedom (Fig. 2) is$320 m thick, consisting of alternating aragonitedetrituslaminae (the aad facies; Marco, 1996), Ca-sulfate minerals,halite and clastics. We divide the Amora Fm. into five mem-bers: Amora Formation Member (AFM) I, II, III, IVand V,corresponding to Zaks (1967) division into Amc, Ama, As,Am and Ams Members. The exposure of AFM-V (the

youngest Member) at the AC site is limited and was de-scribed by Zak (1967) mainly along the northern flanks ofMt. Sedom, which were not studied here. Overall, the lithol-ogy of the sedimentary section is similar to that of the over-lying Lisan and Zeelim Formations, as described by Neevand Emery (1967), Zak (1967), Begin et al. (1974) and manyothers. Exceptions to this generalization are the appear-ances of Ca-sulfate as anhydrite rather than gypsum, espe-cially in the lower part of the AC section, which probablyreflects the burial of the sediments and the consequentdehydration of gypsum.

Segments of the Amora Fm. were uplifted and tilted bythe rising Sedom diapir (Fig. 2C; Zak, 1967). At present,the layers exposed at the AC section, our main study site,

are in sub-vertical position (Fig. 2C) and the thick colum-nar section is exposed along the dry creeks. As mentioned,Zak (1967) estimated the age of the sequence to be between1 Ma and $100 ka. Kaufman (1971) reported four UThages, three of which range between 380 and 190 ka, but in-volve large uncertainties and no stratigraphic coherence.The fourth sample is in secular equilibrium.

2.2. U-series dating of lake sediments

Ages of primary aragonite material can be calculatedaccording to the following equations (after Broecker, 1963):

230

Th238U

1 ek230t

234

U238U

1

k230

k230 k234

1 ek230k234t

1

and

234U238U

initial

234U238U

1

ek234t 1 2

All ratios presented here and in the following discussion inround brackets represent activity ratios. The application ofEqs. (1) and (2) is based on the assumptions that no initial230This initially present in the sample and that all of the U is

derived from the water, rather than from detrital material.These assumptions hold well for corals and speleothems,which typically contain negligible amounts of detrital Thbut rarely hold for other carbonates (e.g., lake sediments)and thus require correction procedures. Many studies inves-tigated ways to determine the appropriate sample process-ing methods and corrections in order to obtain accurate

U-series ages of dirty continental carbonates. These stud-ies discussed the relative advantages of whole rock dissolu-tion versus leaching, and have used UTh isochrontechniques to correct for detrital contributions to the Uand Th (Kaufman, 1971, 1993; Ku and Liang, 1984; Bisc-hoff and Fitzpatrick, 1991; Luo and Ku, 1991; Haase-Sch-ramm et al., 2004; Stein and Goldstein, 2006 and referencestherein).

2.2.1. 230Th234U dating

The problems detailed above are dealt with in this studyby calculating 230Th234U ages of Amora aragonites in twoways:

(1) Single sample correction method, following themethodology of Haase-Schramm et al. (2004) that was ap-plied for the dating of the last glacial Lisan Fm. and assum-ing the Detrital End Member (DEM) component isU/Thatomic 1.2 (10%), U 4.26 ppm and

238U, 234Uand 230Th in secular equilibrium. Haase-Schramm et al.(2004) estimated a rather small correction (i.e., hundredsof years) dictated by the presence of hydrogenous 230Thin Lisan Fm. samples. Assuming similar geochemical prop-erties for the Amora aragonite precipitating solution theseminor age-shifts are neglected for Amora samples, whichare several hundred ka old.

(2) The isochron method, which uses U and Th isotopicdata of a set of coeval samples, including pure detritus.

The ratios used in Eq. (1) are determined by using the230Th234U isochron method, which corrects for thedetrital contributions to the 230Th, 234U and 238U in a sam-ple suite (Fig. 3). The (230Th/238U) ratio is retrieved fromthe y-intercept (where 232Th = 0) of a (232Th/238U) vs.(230Th/238U) Osmond-type diagram. The primary(234U/238U) value is given by the y-intercept in a(232Th/238U) vs. (234U/238U) diagram. The isochron dia-grams have an advantage over the single sample correctionswhen there is a large variation in the mixing proportions ofcoeval detritus and carbonate samples (Kaufman, 1993;Ludwig and Titterington, 1994). Such a set of samples de-fines a large range of either (232Th/238U) ratios (or(238U/232Th) ratios, depending whether Osmond-type orRosholt-type isochrones are used) and the compositionof the DEM does not have to be explicitly determined (orassumed).

2.2.2. 234U238U dating

The Lisan Fm. aragonites display uniform initial(234U/238U) ratios of 1.503 0.037 (2r; Haase-Schrammet al., 2004). Provided pre-Lisan lakes were characterizedby similar activity ratios, Eq. (2) can be used for dating pri-mary aragonite samples based on the measured (234U/238U)ratios alone (Fig. 4). This would provide the possibility toextend the dating range to $1 Ma. In the DSB lacustrine

2606 A. Torfstein et al./ Geochimica et Cosmochimica Acta 73 (2009) 26032630


5/28

bodies, a lower boundary for initial (234U/238U) ratios isidentified in the present hypersaline Dead Sea brine (1.44;Haase-Schramm et al., 2004), while higher values of up to1.61.8 were identified by Waldmann et al. (2007) in pri-mary aragonites deposited during the culmination ofMIS5 ($7090 ka). These authors related the higher(234U/238U) ratios to temporal increases in freshwater inputto the lake.

In the case of Lake Amora, it is assumed that the overalllimnological conditions were similar to those of the last gla-cial Lake Lisan or the Holocene Dead Sea (regarding theAmora glacial and interglacial intervals, respectively). Thisassumption is based on the similar characteristics of AmoraAC section and younger sequences: alternating aragonite

and silty detritus laminae, and gypsum units of Lisan-typeglacial lake, or laminated detrital calcites of Dead Sea-typeinterglacial lake, are the main components of the AC sec-tion; d18O values of the Amora Fm. range between 6&and 5& to 1& and 0&, corresponding to the last glacial Li-san values and to slightly lighter values than those of theHolocene Dead Sea, respectively; In both Formations pri-mary and secondary sulfur deposits display heavy ($1530& and light (


6/28

speleothems has been interpreted to reflect the dominanceof the source effect on the long-term behavior of d18O inthe lake water (e.g., the last glacial samples are heavier by23& compared to the Holocene ones in all three recordspresented in Fig. 5; see also Frumkin et al. (1999), Bar-Mat-thews et al. (2003), Kolodny et al. (2005) and general reviewby Enzel et al. (2008)).

Lake Lisan rose dramatically during the last glacial per-iod to a maximum elevation of 160 m bmsl (more than250 m higher than the present Dead Sea; Bartov et al.,2003). Thus, the commonly applied hydrological interpreta-tion of heavier d18O compositions as reflecting a more salinewater body with stronger evaporation during a dry period isnot the case here. On the contrary; the significant amountsof rain required to raise Lake Lisans water level reveal

intricate controls on the d18

O of the lake, which are dis-cussed by Kolodny et al. (2005). The bottom line is thatwet and dry periods in the DSB lakes are associated withhigher and lower d18O, respectively.

The dominance of the source effect on the d18O in the Ju-dean Mountain speleothems and theDead Seabasin primaryaragonites opens the possibility of using the d18O values inthese records as a regional chronometer (in the sense thatSPECMAP and the ice cores d18O records are globally used).Moreover, the d18O chronometer can be used to extend thechronology of the lacustrine sediments beyond the limits ofthe U-series method, and/or to assess the validity of an age-elevation model constructed from U-series dating.

1.15

1.2

1.25

1.3

1.35

1.4

1.45

1.5

1.55

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4

(230 Th/238U)

(234U/238U)

100

ka

200

ka

20

ka

40

ka

60

ka

80

ka

Fig. 4. Isochron diagram showing the activity ratios of (230Th/238U) vs. (234U/238U) of aragonites from the Dead Sea Group. Legend: GreentrianglesThe Lisan Formation at the PZ1 site (Haase-Schramm et al., 2004); gray squaresPZ2 section (Waldmann et al., 2007; this study);full diamonds (Isochron data) and empty diamonds (single sample data, for clarity error bars are not displayed)The Amora Formation atAC site. The thick dashed curve represents the decay curve of an initial ( 234U/238U) ratio of 1.5. All plotted data falls in the vicinity of thiscurve indicating a similar initial composition. (For interpretation of the references to color in this figure legend, the reader is referred to theweb version of this paper.)

Fig. 5. A 70,000-year correlation between the d18O values ofsynchronously deposited: (A) East Mediterranean foraminifers(Fontugne and Calvert, 1992), (B) Judean Mountain speleothems(Bar-Matthews et al., 2003), (C) Lisan and Zeelim primaryaragonites (Kolodny et al., 2005). Note the similar shift betweenthe last glacial and the Holocene values in graphs AC. These shiftscorrespond to the transition from the high stand Lake Lisan to thelow stand Holocene Dead Sea (D) (Bartov et al., 2003; Migowskiet al., 2006). Modified after Kolodny et al. (2005).

2608 A. Torfstein et al ./G eochimica et Cosmochimica Acta 73 (2009) 26032630


7/28

3. METHODS

3.1. Field work and sample preparation

Field samples approximately 20 20 20 cm3 in sizewere collected from the Amora Fm. exposures at AC andPZ2 sites, wrapped in plastic coating to avoid disintegration

and transported to the laboratory.Approximately 5 g of selected laminae (aragonite, detri-

tus, gypsum and anhydrite) were scraped separately andcollected. Following grinding and homogenization, themineralogical composition of selected samples (Table 1)was identified by XRD.

3.2. UTh dating

3.2.1. Chemistry

Most samples were processed at the Geological Sur-vey of Israel (GSI). Typically, 100 mg of aragonite pow-der were dissolved in 7 N HNO3. Some aragonite and

detrital samples were totally dissolved in an HNO3HFmixture. One sample (AC1-101-6-Da) was dissolved ina weak acid (1 N acetic acid). The solutions were spikedwith a mixed 236U229Th spike and thereafter uraniumand thorium were separated using anion exchange col-umns filled with 2 ml of Bio-Rad AG 1 8 200400mesh resin (see details in Vaks et al., 2006). Additionalsamples were processed at the Leibniz Institute of Mar-ine Sciences at the University of Kiel (IFM-GEOMAR).There, acid-dissolved powder was spiked with a 229Thand a 233U236U double spike mixture. U and Th wereco-precipitated with Fe(OH)2 in alkaline solution (pH8) using NH4OH. They were subsequently separatedusing anion exchange columns filled with Bio-Rad AG1 8 100200 mesh resin (see details in Haase-Schrammet al., 2003).

3.2.2. Mass spectrometry

U and Th isotope ratios were measured using Nu Plas-ma and AXIOM MC-ICPMS instruments at the GSI andat IFM-GEOMAR, respectively.

A standard reference material for U isotope ratio mea-surements, NBL 112a, was introduced with the samples tocheck instrument performance. The 234U/238U(atomic) ratioobtained for NBL 112a was 5.278 105 3 107 (2r;GSI) and 5.287 105 1.8 107 (IFM-GEOMAR), ingood agreement with that of Cheng et al. (2000;

5.286 105

2.5 107

. The isotopic mass discrimina-tion of the MC-ICP-MS was corrected for by the use of238U/235U as an internal standard. The high precision of238U measurement on MC-ICP-MS allows assuming thatthe (232Th/238U)(230Th/238U) and (232Th/238U)(234U/238U) couplets are free of error correlations (Ludwig,2003). This is a conservative assumption because the exis-tence of error correlations would dictate the true propa-gated age uncertainty to be smaller than reported. Decayconstants used in this study are: k230 9.1577 10

6;k232 4.9475 10

11, k234 2.8263 106 and k238

1.55125 1010 (Le Roux et al., 1963; Jaffrey et al., 1971;Cheng et al., 2000). All related age calculations were per-

formed using a set of MATLABTM scripts available upon re-quest from the corresponding author.

3.3. Stable isotope analyses

Oxygen isotope ratios were measured in aragonitesamples using an elemental analyzer Flash EA 1112 ser-

ies connected online via a ConFlo-III interface to aThermo Finnigan Deltaplus stable-isotope ratio monitor-ing mass spectrometer at the Institute of Earth Sciencesof the Hebrew University of Jerusalem. Oxygen isotopecompositions are expressed as per mil (& deviationsfrom V-PDB standard using the conventional delta nota-tion with a standard deviation better than 0.2&(n 865). The measurements were directly calibratedagainst an in-house standard, which was calibrated tointernationally accepted values by analyzing standardsNBS-19 (d18O 2:2&).

4. RESULTS AND DISCUSSION

4.1. U-series chronology

4.1.1. 230Th234U ages

The isotopic composition of U and Th was analyzed in91 samples. In eight more samples only U isotopic compo-sitions were measured (Table 2). Only samples that met thefollowing arbitrarily-determined requirements wereconsidered reliable for single sample age calculation: (1)(230Th/232Th) > 10, (2) (238U/232Th) >10, (3) [Th] 1). Average single sample ages are reported witha 2r standard error age uncertainty (Table 4). Isochronages were determined by calculating error-weighed linearregressions (York, 1969) and y-intercepts in Osmond-typediagrams (Fig. 3). Isochron age uncertainties represent a95% confidence error. Calculated ages range between

$490 and 155, and 502 and 151 ka for the single sampleand isochron calculations, respectively (Fig. 8).

4.1.2. 234U238U ages

Single sample 230Th234U age calculations lead to anaverage initial (234U/238U) ratio of 1.553 0.264 (2r,or 1.504 0.145 when the relatively elevated initial(234U/238U) ratios of samples located below 185 m (fromthe base of the section) are neglected (Table 3).

Both values are similar, within error, to the correspond-ing value calculated for the Lisan aragonites (1.503 0.037;Haase-Schramm et al., 2004). During other stages in theevolution of the DSB lacustrine bodies, when initial condi-



8/28

Table 1Sample list and XRD results. Traces of minerals are preceded by +. Abbreviations: A=aragonite, G=gypsum, Q=quartz, H=halite,C=calcite, D=dolomite, An=anhydrite, Kf=K-feldspar, B=basanite, Ph=phylosilicates, Pl=plagioclase, n.a.=not available. Elevation refersto height above the Sedom-Amora contact.

Sample name Elevation (m) Field description XRD

AC1-101-6 296.3 aragonite laminae AAC1-101-5 296.3 aragonite laminae A

AC1-101b-11 296.3 detritus laminae C, Q, DAC1-99 295.8 anhydrite layer n.a.AC1-98 295.5 anhydrite layer n.a.AC1-97 295.3 anhydrite layer n.a.AC1-94-5 293.2 aragonite laminae A (+ ?)AC1-94-4 293.2 aragonite laminae AAC1-95-5 293 aragonite laminae AAC1-92-4 292.1 gypsum n.a.AC1-91-1 291.9 aragonite laminae AAC1-88 289.9 gypsum n.a.AC1-87-5 289.2 aragonite laminae AAC1-86 288.2 gypsum layer n.a.AC1-84-l-m 286.6 detritus laminae C,Q,D (+Ph, Pl)AC1-81-5 282.5 aragonite laminae AAC1-80-l-m 281.9 detritus laminae G,C,Q (D, Ph)

AC1-70-LA-1 281.9 detritus laminae G (+Q, C)AC1-70 280.4 gypsum layer n.a.AC1-73-5 273.8 aragonite laminae A (+C, D)AC1-74-2 271.7 aragonite laminae AAC1-77-2 269.4 aragonite laminae AAC1-66-11 268.5 aragonite laminae A (+Calc/Gyp?)AC1-66-1 268.5 aragonite laminae AAC1-64 267.4 native sulfur n.a.AC1-62-1 265.8 gypsum n.a.AC1-61-3 264.6 aragonite laminae AAC1-59-5 263.8 aragonite laminae A (+C)AC1-57-l-m 262.6 detritus laminae A,C,Q,D, (+Ph, H)AC1-56 262.2 native sulfur concretion n.a.AC1-54-5 239 aragonite laminae AAC1-54-4 239 aragonite laminae AAC1-53-6 238.4 aragonite laminae AAC1-50-6 234.2 aragonite laminae AAC1-49-6 233.3 aragonite laminae AAC1-48-1 233 aragonite laminae AAC1-46-5 229.6 aragonite laminae AAC1-46-4 229.6 aragonite laminae AAC1-46A 229.6 detritus laminae G, C, Q, AAC1-42-1 222.4 aragonite laminae AAC1-41-6 220.7 aragonite laminae A (+C)AC1-41-5 220.7 aragonite laminae A (+C)AC1-41-3 220.7 aragonite laminae A (+C)AC1-41-1 220.7 aragonite laminae A (+C)AC1-45-11 219.4 disseminated gypsum (from bulk aad) n.a.AC1-45-6 219.4 aragonite laminae AAC1-44-5 219.1 aragonite laminae A (+G, H, C)AC1-44-2 219.1 aragonite laminae A, CAC1-39-5 216.3 aragonite laminae A (+B/An/H?)AC1-39-3 216.3 aragonite laminae A (+C)AC1-38 216.1 gyp lamina within aad n.a.AC1-37-6 214.4 gyp lamina within aad n.a.AC1-37-3 214.4 aragonite laminae A (+C,G)AC1-37-1 214.4 aragonite laminae A (+C)AC1-36 212 anhydrite layer n.a.AC1-33-4 203.4 aragonite laminae A (+H?,Q,C)AC1-33-2 203.4 detritus laminae D?, C, Q,(+A)AC1-30-2 192.6 araaonite laminae AAC1-29-5 183.4 aragonite laminae A (+C)



9/28

tions are known and 230Th234U ages were calculated, a lar-ger range in the initial (234U/238U) ratios was identified(e.g., 1.44 for the modern Dead Sea brine and 1.8 in fresh-water entering the lake, see Section 2.2.2). Thus, we calcu-lated 234U238U ages assuming an initial (234U/238U) ratioof 1.5, with a total range of 1.441.80 (Fig. 8).

Samples from the bottom part of the Amora Fm.(


10/28


11/28


12/28

234U238U ages cannot be taken at face value but rather,they provide only a rough age assessment.

4.1.3. U-series age-elevation regression

The 230Th234U isochron ages and the 234U238U ages ofsamples below 150 m in the section were used to calculate alinear age-elevation regression curve for the AC Amora sec-

tion. The most reliable isochron ages were selected accord-ing to the following criteria: (1) The difference between thecalculated isochron and average single sample (corrected)ages is less than 5%; (2) isochron age uncertainty is less than10% and (3) at least two primary aragonite samples areused for the age calculation (in addition to the DEM value).The samples defining these isochrones are considered to beanalytically and methodologically reliable and are high-lighted in bold in Table 4. The filtered data and the calcu-lated regression curve are presented in Fig. 9. Accordingto these results, the exposed AC section was deposited be-tween $740 and 1 are displayed. The gray square at thebottom left of the lower diagram represents the detrital end-member (DEM) composition. Symbols are equivalent or larger than error barsexcept when displayed otherwise.

http://-/?-http://-/?-


13/28

our interpretation, glacial values: $56& and light(interglacial values: +1& to 1&) values. The mineral-ogical composition of select samples with low d18O valueswas verified (by XRD) to be pure aragonite (Table 1 andEA 1), thus indicating that the light oxygen isotopic compo-sition represents the primary precipitate and is not the re-sult of diagenesis (e.g., re-crystallization to calcite).

The natural noise ofd18O in the section was estimated

by analyzing consecutive sets of aragonite laminae. Thestandard deviation (1r ofd18O values in each of the samplesets was in all cases less than $1&, with an average of0.43&. This value is well below the range of the d18O inthe sediments ($6&to 1.3&; EA 1). Thus, the averagecomposition of each set of samples can be assumed to repre-

sent the smoothed out steady-state composition of the lake.

4.3. Paleomagnetic constraints

Weinberger (1992) reported four paleomagnetic analysesat the AC section (at $20, 110, 160 and 320 m; Fig. 9). Allsamples possess a normal magnetic pole implying that the

section was deposited after the last major magnetic reversalevent, the MatuyamaBrunhes boundary (MBB), dated at780 ka (Cande and Kent, 1995; Tauxe et al., 1996).

The possibility that the observed normal paleomagneticsignal represents an older normal stage (i.e., the Jaramilloat 9901070 ka; Cande and Kent (1995)) is ruled out basedon the U isotope ratios. The average initial (234U/238U) va-lue is 1.553 0.264 (2r, which is slightly higher than that

of Dead Sea and Lisan aragonites (Haase-Schramm et al.,2004). If the samples from the lower part of the section(i.e., AC2-7B-3,4; Table 3) were deposited during the Jara-millo normal magnetic stage, they would have to have hadextremely high (234U/238U) initial values in the range of 2.52.8 (Eq. (2)), which is unlikely.

4.4. Combined U-series and oxygen isotope age-elevation

model

In the following discussion we describe the combinationbetween the filtered U-series ages (Section 4.1.3) and thewiggle-matched oxygen isotope data at each part of the

(238U/232Th)

(230Th/232Th)

(238U/232Th)

N

1 2 3 4 5 6 7 8 9 1000

5

10

B

A

B

Fig. 7. Detrital end-member (DEM) composition. (A) Rosholt-type diagram of all measured samples in the Amora AC section. Dataextrapolation provides an estimate of the DEM composition at the crossing of the equiline (black line). (B) Histogram of232Th-rich samples.The y-axis pertains to the number of samples (N). Values peak between 3 and 4, at approximately the same DEM composition estimated byHaase-Schramm et al. (2004) for the Lisan Fm. (i.e., (238U/232Th) %3.7). Hence, the latter value is used for age calculations in this study. Notethe possibility for the existence of a multiple DEM composition indicated by a second peak of (238U/232Th) values between 7 and 8.

http://-/?-http://-/?-http://-/?-http://-/?-


14/28

section. The result is a best-fit age-elevation model for theAmora Formation.

The East Mediterranean (EM) is considered to be themain source of water to the DSB drainage area. Accord-ingly, the Amora d18O record is tuned to the EM d18O re-cord compiled by Kroon et al. (1998). However, thisrecord displays some significant off-phase d18O structures

with respect to the globally stacked d18O benthic record(Lisiecki and Raymo, 2005), suggesting that the chronologyof the EM record requires tuning itself. To overcome thisproblem we used Lisiecki and Raymos (2005) chronologyfor the final age determination of the AC record (Fig. 11).While this is probably the best solution for most of the sec-tion, we note that some d18O minima peaks at the base of

Table 3Single sample detritus-corrected ages and 234U238U ages. aages were calculated from measured (uncorrected) data and were added 37 kawhich is the average age difference between detritus-corrected and uncorrected 234U238U ages.

Sample name Elevation (m) 230Th238U

234U238U

Age (ka) 234U

238U

initial

234U238U age (ka)

AC1-101-6-Da 296.3 1.199 1.250 259.9 (+14.5,12.7) 1.521 245AC1-100-4 296 1.203 1.230 283.2 (+12.9,11.5) 1.513 274

AC1-100-7 296 1.182 1.221 274.4 (+11.1,10.0) 1.480 289AC1-94-4 293.2 1.216 1.249 275.1 (+11.4,10.3) 1.542 247AC1-94-5 293.2 1.213 1.250 270.8 (+18.5,15.8) 1.538 245AC1-90-9 290.8 1.262 1.261 305.4 (+21.6,18.0) 1.619 230AC1-90-7 290.8 1.250 1.255 300.9 (+14.0,12.4) 1.597 238AC1-83-8 286.2 1.174 1.245 245.0 (+7.2,6.7) 1.490 252AC1-83-10 286.2 1.157 1.242 236.3 (+4.8,4.5) 1.472 257AC1-81-3 total 282.5 1.193 1.247 258.3 (9.1,8.3) 1.512 250AC1-81-8 total 282.5 1.203 1.243 268.9 (+10.3,9.4) 1.520 255AC1-73-3 273.8 1.167 1.238 246.2 (+6.2,5.9) 1.478 262AC1-66-2 268.5 1.154 1.251 228.3 (+287,69.7) 1.479 243AC1-66-8 268.5 1.147 1.247 226.8 (+212,59.9) 1.468 250AC1-66-8 total 268.5 1.160 1.277 215.9 (+5.8,5.5) 1.510 209AC1-61-3 264.6 1.148 1.238 233.5 (+5.2,4.9) 1.460 263AC1-61-19 264.6 1.141 1.231 234.5 (+4.5,4.3) 1.448 273

AC1-55-1 261.8 1.219 1.250 275.8 (+286,78.1) 1.545 245AC1-55-7 261.8 1.213 1.264 257.5 (+289,74.7) 1.547 226AC1-53-5 238.4 1.293 1.350 245.0 (+29.6,22.9) 1.699 126AC1-53-10 238.4 1.211 1.294 233.6 (+23.6,19.0) 1.569 188AC1-50 234.2 1.153 1.281 210.1 (+7.0,6.5) 1.508 204AC1-49 233.3 1.117 1.263 201.9 (+7.9,7.3) 1.465 228AC1-46-5 229.6 0.965 1.229 155.1 (+3.2,3.1) 1.354 277AC1-43 222.8 1.118 1.282 193.5 (+13.4,11.7) 1.487 202AC1-42-5 total 222.4 1.240 1.296 251.4 (+23.5,19.1) 1.601 186AC1-42-9 total 222.4 1.019 1.285 156.0 (+8.1,7.4) 1.443 199AC1-41-8 220.7 1.078 1.244 192.8 (+5.8,5.5) 1.421 254AC1-41-4 220.7 1.018 1.231 173.3 (+4.6,4.4) 1.378 273AC1-41-3 220.7 1.252 1.240 324.2 (+159,62.8) 1.601 259AC1-45-6 219.4 1.085 1.209 215.1 (+12.1,10.7) 1.383 309AC1-45-7 219.4 1.091 1.204 221.1 (+13.3,11.6) 1.382 317

AC1-39 leach 216.3 1.223 1.205 347.8 (+109,54.1) 1.549 315AC1-37 214.4 1.171 1.213 272.1 (+11.9,10.6) 1.460 301AC1-33-1 203.4 1.234 1.231 317.4 (+19.1,16.3) 1.566 274AC1-32 195.1 1.197 1.255 253.9 (+16.1,13.9) 1.522 239AC1-24 187.9 1.171 1.277 221.7 (+17.2,14.6) 1.519 209AC1-29-1 183.4 1.295 1.303 286.2 (+148,61.9) 1.682 177AC1-28 180.5 1.312 1.262 374.6 (+266,62.0) 1.756 228AC1-25 175.3 1.339 1.246 492.4 (+362,237) 1.991 250AC1-25-3 175.3 1.303 1.263 357.0 (+262,68.6) 1.721 227AC3-50C-5 150.4 1.254 1.197 459.7 (+56.2,38.1) 1.722 330AC3-50B-6 150.2 1.247 1.193 450.0 (+120,59.9) 1.690 336AC3-50B-8 150.2 1.268 1.194 507.5 (+433,183) 1.813 336AC3-50C-4 150.4 1.254 1.187 543.3 (+453,72.9) 1.867 349AC3-45C-10 119.7 426 [a]AC3-36-B-7 79 612 [a]

AC3-28K-8 53.1 476 [a]AC3-28-C2 47.7 531 [a]AC2-15A-3 31.2 591 [a]AC2-7B-3 10.3 736 [a]AC2-7B-4 10.3 652 [a]



15/28

the section (e.g., at $9, 20 and 54 m; see next section) mightrepresent local, East Mediterranean, events, not reflected bythe global record, and therefore their correlation to the glo-bal record may be erroneous (by several thousands ofyears). The ages of these and perhaps other events maybe reconsidered in the future, when the EM record is inde-pendently dated.

4.4.1. Base of section

As explained above, paleomagnetic considerations implythat the base of the Amora (AC) section is younger than780 ka. This observation is supported by the extrapolated(U-series) age-elevation regression curve (Fig. 9). A finalrefinement of the age at the base of the AC section is de-rived from the correlation of three d18O minima at $9, 20

Table 4Isochron and average single sample ages. Ages of sample sets in bold font are considered analytically and methodologically reliable. See textfor details.

Sample name Averageelevation (m)

SS age(ka)

Isochron data234U238U

230Th238U

Age(ka)

234U238U

initial

AC1-101-6-Da

AC1-100-4 296.1 272.5 13.6 1.203 0.006 1.187 0.007 300.6 (+13.1,11.7) 1.476 0.000AC1-100-7

AC1-94-4 293.2 272.9 4.3 1.249 0.006 1.216 0.008 274.4 (+10.8,9.9) 1.541 0.003AC1-94-5

AC1 90-9 290.8 303.2 4.5 1.249 0.005 1.250 0.008 309.8 (+13.4,12.0) 1.597 0.001AC1 90-7

AC1 83-8AC1 83-10 284.4 252.1 14.4 1.242 0.003 1.092 0.006 200.6 (+3.5,3.3) 1.426 0.004AC1-81-3 total dissAC1-81-8 total dissAC1 73-3 273.8 246.2 6.1AC1 66-2

AC1 66-8 268.5 223.6 7.8 1.277 0.007 1.160 0.007 215.8 (+5.8,5.5 1.510 0.005AC1 66-8 total diss

AC1-61-3 264.6 234.0 0.9 1.232 0.004 1.144 0.004 235.8 (+4.3,4.1) 1.451 0.002

ACl-61-19AC1 55-1 261.8 266.7 18.3 1.256 0.081 1.216 0.034 267.5 (+180.6,60.4) 1.545 0.014AC1 55-7AC1-53-5 238.4 239.3 11.4 0.969 0.020 0.721 0.034 150.6 (+21.1,16.8) 0.953 0.018AC1-53-10AC1-50 233.8 206.0 8.2 1.272 0.008 1.150 0.008 213.7 (+7.0,6.5) 1.497 0.003AC1-49

AC1-46-5 229.6 155.1 2.9AC1-43 222.5 222.5 57.9 1.306 0.017 1.154 0.018 198.0 (+13.1,11.5) 1.536 0.004AC1-42-5 total dissAC1-42-9 total diss 222.4 155.4 7.9AC1-41-8 220.7 183.0 19.5 1.240 0.006 1.042 0.007 179.4 (+4.3,4.1) 1.399 0.003AC1-41-4

AC1-45-6 219.4 218.1 6.0 1.206 0.012 1.087 0.012 217.9 (+11.9,10.6) 1.382 0.007AC1-45-7

AC1-39 leach 216.3 347.8 64.0AC1-37

AC1-37-1-M 214.4 272.1 10.4 1.201 0.003 1.162 0.003 276.8 (+4.6,4.4) 1.440 0.001AC1-37-6

AC1-33-1 203.4 317.4 16.0AC1-32 195.1 253.9 13.4AC1-24 187.9 221.7 14.3AC1-29-1AC1-29-5AC1-29-2 183.4 286.2 77.1 1.257 0.007 1.173 0.024 235.5 (+17.0,14.8) 1.501 0.004AC1-29-3AC1-29-4AC1-28 180.5 374.6 77.0 1.249 0.003 1.295 0.004 375.8 (+13.8,12.3) 1.720 0.008AC1-25 175.3 372.5 159.8 1.222 0.002 1.265 0.007 388.9 (+20.9,17.8) 1.665 0.005AC1-25-3

AC3-50C-5AC3-50B-6 150.3 490.1 43.4 1.189 0.003 1.252 0.004 502.5 (+51.5,36.1) 1.783 0.010AC3-50B-8

AC3-50C-4



16/28

and 54 m to excursions in the EM record at 732 ka(MIS18.3), 696 ka (MIS17.1) and 612 ka (MIS15.5), respec-tively (Fig. 11). Thus, the lowermost temporal d18O maximaat the base of the AC section could be placed at $746 ka(MIS18.4). This age is consistent with that inferred by U-series, placing the base of the section at 740 (66 ka).

4.4.2. Amora Formation Member (AFM) I and II

Between the elevation of $54 m and the transition toAFM-III at 154 m, d18O values are positive (45&, Li-san-like) and display limited variations. A striking litho-logical change at $96 m defines the transition betweenAFM-I and II (Figs. 1 & 11). AFM-I is characterized byclastic units, marls, alternating aragonitedetritus laminae(aad) and anhydrite, whereas AFM-II is comprised of con-secutive cycles of marls, anhydrite and sporadic aad ($96154 m). The latter sequence is characteristic of sabkha-likeand salina environments indicating a moderately low level

lake.The age model for the part of the section between 96 and

154 m is based on an extrapolation from under-and over-ly-ing sequences (see below). However, the occurrence of amajor clastic unit at 9096 m, with a determined age of$550 ka, may represent a major hiatus in the lacustrinesedimentation.

4.4.3. AFM-IIIthe salt unit

The age of the salt unit is constrained by UTh dating toapproximately 400 ka (corresponding to MIS11; Fig. 11).This unit represents the only exposed occurrence of massivehalite in the Quaternary DSB lake deposits. While massive

halite is relatively abundant in the central and southern ba-sins of the Dead Sea (Neev and Emery, 1995) its precipita-tion or preservation on the shallow margins of the basin arerare; in fact, only one other massive salt layer was identifiedalong the lake margins: at the base of the Holocene ZeelimFm., dated to $1110 ka (Yechieli et al., 1993; Migowskiet al., 2004; Fig. 2).

Many reports described a similarity between the geolog-ical records of MIS11 and the Holocene and discussed theirpossible relation to similar orbital forcing conditions duringboth periods (McManus et al., 2003; Berger and Loutre,2003; EPICA community members, 2004 and referencestherein). This lends support to the identification and datingof the Amora salt at MIS11. Furthermore, the deposition ofsalt during the PleistoceneHolocene transition ($11 to10 ka) indicates that the Amora salt (AFM-III) probablyprecipitated upon the transition from MIS12 to MIS11, at$420 ka.

4.4.4. AFM-IV and V

Overlying the AFM-III sequence, ages are determinedby the UTh smoothed-out age curve and are fine-tunedusing the d18O record. Stages 9, 8 and 7 are well definedby the d18O record (Fig. 11C), which corresponds within er-ror to UTh ages. In addition, sub-stage MIS7.4 (223 ka) isidentified at 222 m, where an excursion toward heavier iso-topic compositions is observed.

The MIS6-sequence seems to be excessively thick, war-ranting the possibility of serious structural disturbances:according to UTh age calculations the chronostratigraphicorder is disrupted above 290 m (Fig. 9). However, no major

Elevation(m)

Age (ka)

A B

IV

II

I

V

III

Isochron age

Single sample > 150 m

Single sample < 150 m

Single sample age

234U-238U ages:

Isochron

230Th-234U ages:

Fig. 8. Isochron, single sample and 234U238U ages of the Amora section. (A) 230Th234U isochron and single sample ages. Error barsrepresent the 2r uncertainty and 95% confidence error for single sample and isochron ages, respectively. (B) 234U238U ages for all measuredsamples assuming initial activity ratios were 1.5. Samples below 150 m represent back-calculated ages from measured (uncorrected) datawith an added 37 ka (see text for details). Error bars represent an age range of45 and +166 ka, corresponding to the possible range of initialactivity ratios (1.44 and 1.8, respectively). Note the similar ages and overall trend in both diagrams. The Amora Fm. members (AFM) aregiven on the right.

2618 A. Torfstein et al ./G eochimica et Cosmochimica Acta 73 (2009) 26032630


17/28

sedimentary or structural disturbances (e.g., faults or folds)were detected in this part of the section; no repetitions ofdistinct lithological sequences were observed (such as the$20-m-thick horizon of native sulfur concretions at$240260 m); the oxygen isotope data appears undisturbed(namely, does not display the typically-light d18O values of

the underlying MIS7 sequence). Thus, while it appears thatthe overall stratigraphic order remains intact, UTh ages ofaragonites above 290 m seem to deviate from the generalage-elevation regression curve, suggesting disturbance ofthe UTh isotope system. Nevertheless, at this stage ofthe study we cannot rule out thickening of the MIS-6 se-quence due to local structural disturbances.

At $300 m in the AC section a lithological transitionfrom lacustrine to clastic sediments represents the MIS6-MIS5 transition at $130 ka. This transition correspondsto the timing of a large lake level drop recognized by Bartovet al. (2007) and Waldmann et al. (2007). Additional evi-dence for a lake retreat during the early part of the last

interglacial period was provided by Davis et al. (2007),who reported two OSL ages of 127 ( 27) and 125( 10) ka for abandoned erosion terraces south of the DeadSea (the Zin valley).

4.4.5. Sedimentation rates and age uncertainties

The integrated U-series -d18

O tuned age-elevation modelyields an average sedimentation rate of 0.49 m/ka (Table 5& Fig. 12), only slightly lower than that of Lake Lisan atthe Perazim Valley ($0.7 m/ka; Haase-Schramm et al.,2004). The sedimentation rates in the Holocene Dead Seawere generally more variable, in the range of $0.31.3 m/kain the marginal Zeelim site (Ken-Tor et al., 2001) or$1.7 m/ka at the lacustrine environment off the En-Gedishore (Migowski et al., 2006).

The relatively low sedimentation rate at the AC sitecould reflect the occurrence of hiatuses throughout the sec-tion, most likely associated with clastic units. It could alsorepresent an artifact of sediment compaction, which might

Fig. 9. Arubotaim Cave (AC) section U-series age compilation. The graph displays reliable isochron ages (see Table 4) for the upper part ofthe section (above 150 m) and back-calculated 234U238U ages for the lower part of the section (


18/28

have had a larger impact on the lower part of the section.Although the effects of these issues on the age-elevationmodel remain unresolved at this stage, they are probablythe main contributors to the linear regression age uncertain-ties (66 ka; see Section 4.1.3). Our d18O tuning allows forsignificant narrowing of uncertainties at specific controlpoints (Fig. 11D) where the uncertainty represents how wellthe Amora d18O record could be matched to the regionaland global records. Overall, the d18O-tuned age-elevationmodel is in agreement with the U-series-based regressioncurve, and provides a more realistic representation of thevariability of the sedimentation regime.

4.4.6. The chronology of the pre-Lisan sediments at the

Perazim Valley

Lacustrine sediments deposited synchronously to theAmora AC deposits are exposed in the Perazim Valley(PZ2 section), some 5 km west of the AC site on the westernside of Mt. Sedom (Fig. 2).

A compilation of UTh ages obtained on Amora arago-nites at the PZ2 section and one TL age are listed in Table6. This data, together with complementary d18O composi-tions of aragonites from the PZ2 section (EA 2) is presentedin Fig. 13. We calculated an error-weighed linear age-eleva-tion regression using only samples highlighted in bold inTable 6. We excluded samples with uncertainties extendingto equilibrium, as well as two samples significantly deviat-ing from the overall stratigraphic order (i.e., PZ2-SAM-3

and PRZ-7). The resulting age-elevation regression curve(Y 71 T 0:211 0:021, where Y= elevation (m) andT= age (ka)) suggests that the Amora part of the sedimen-tary column was deposited at the PZ2 site between $340and 80 ka, throughout MIS10 to 5. This range correspondsto the upper part of the AC section (i.e., AFM-IV and V),and contains the section described by Waldmann et al.(2007) as the last interglacial Samra Formation.

The AC section is more than twice as thick compared tosynchronous deposits at PZ2 (Fig. 13). This difference re-flects the occurrence of significant erosional unconformitiesat PZ2. Its marginal location (Fig. 2) is reflected by thelithology of the sediments that contain, relative to the morecentral AC site, less primary evaporites (aragonite and gyp-sum) and more marls and clastic components. In addition,

the PZ2 site was susceptible to significant hiatuses, whichare reflected by the occurrence of several major clastic units.These correlate mainly to interglacial stages (MIS9, 7, 5 andpart of the glacial MIS8) indicating lake low-stand condi-tions. By their nature, the sedimentary deposits in deeperbasinal sites reflect long-term changes in the depositionalenvironment, i.e., the lithology tends to reflect major cli-matic transitions (glacial-interglacial) while the lithologyof marginal-shallow sites tends to be a better monitor ofshort-term limnological shifts (i.e., lake level drops orfloods). Thus, the limnologicalhydrological history of thewater body can be obtained from the integration of thetwo types of sections.

Elevation(m) E

levation(m)

18O () PDB

18O standard deviation ()

Fig. 10. Oxygen isotopic compositions of Amora Fm. aragonites. The right graph displays all measured data (empty triangles, analyticaluncertainty is


19/28

4.5. The climaticlimnologic history of Lake Amora

4.5.1. Interpretation of runoff/evaporation ratio

The chronology of the lacustrine sequence of the AmoraFm. extends the previously studied chronologies of the Zee-

lim,Lisan andSamraFormations by more than 500 ka (from$130 to beyond 700 ka) and provides a long-term record ofthe limnological history of the DSB during this period.

In general, the DSB lakes shifted between two climatic-limnologic regimes (Begin et al., 1974; Stein et al., 1997;

Fig. 11. The chronology of the Amora Fm. based on full data compilation. (A) Global stacked d18O benthic record (Lisiecki and Raymo,2005) and list of MISs (including several substages); (B) East Mediterranean d18O record (Kroon et al., 1998); (C) Amora Fm. d18O record;(D) 230Th234U and 234U238U ages of aragonites in the Amora Fm. (see Fig. 9). The black curve represents the linear regression with a 66-kauncertainty (dashed curves). The thick blue curve represents the d18O-tuned age-elevation correlation. Error margins for control points (seeTable 5) represent the uncertainty in matching between the Amora d18O record and the regional and global records. A general description ofthe lithological composites of the section is given on the right side of graph D (legend follows Fig. 1 except for: yellowclastic, green

gypsum or anhydrite, yellow circlesnative sulfur concretions). The y-axis of (A) and (B) pertains to age (ka) while the y-axis of (C) and (D)pertains to elevation (meters above base of section). (For interpretation of the references to color in this figure legend, the reader is referred tothe web version of this paper.)

Table 5Control points and summary of the Amora AC section age-elevation model. The average sedimentation rate of the full section is 0.49 m/ka.

Elevation (m) Member Age (ka) MIS Sedimetation rate (m/ka) Considerations

320 AFM-V 90 5 0.60 Extrapolated age296 AFM-IV 130 Top 6 0.95 Significant lithological transition239 190 76 transition 0.52 d18O matching and U-series222 223 7.4 0.60 d18O matching and U-series210 243 87 transition 0.37 d18O matching and U-series180 324 9 0.22 d18O matching and U-series

159 AFM-III 420 1211 transition 0.41 Resemblance to Holocene record and U-series120 AFM-II 514 13 0.67 d18O matching and U-series54.5 AFM-I 612 15.5 0.41 d18O matching and U-series20 696 17.1 0.31 d18O matching and U-series9 732 18.3 0.64 d18O matching and U-series0 746 18.4 d18O matching and U-series



20/28

Torfstein et al., 2005): (1) a relatively wet regime with highrunoff/evaporation ratios, represented by the high standlate Pleistocene Lake Lisan deposits, which are dominatedby primary minerals (aragonite laminae and gypsum layers)and (2) a relatively arid regime with low runoff/evaporationratios, represented by the low stand Holocene Dead Sea(the Zeelim Fm.). The Holocene deposits are largely com-prised of detrital calcites, marls and sands, with a limitedpresence of aragonite and practically no primary gypsum.The lithology of the last interglacial deposits (at both ACand PZ2 sites) resembles that of the Holocene ZeelimFormation.

Thus, the fraction of the primary minerals (i.e., arago-nite, gypsum, anhydrite) in the lake deposits can be usedas a proxy for the runoff/evaporation ratio (Fig. 14D),which in itself is indicative of hydrological conditions andlake level fluctuations.

4.5.2. d18O excursions

Kolodny et al. (2005) interpreted shifts in the d18O val-ues of aragonite packages from the last glacial Lisan andthe Holocene Zeelim Fms. as reflecting corresponding

shifts in the d18

O values of the source (i.e., the EasternMediterranean surface water). In the Amora AC section,the glacial intervals display relatively heavy d18O valuesof$46&, similar to those of the last glacial Lake Lisan.However, some interglacial intervals of the AC section dis-play substantial 18O-depletion compared to the HoloceneDead Sea Zeelim Formation. For example, between$210240 m above the base of the AC section (MIS7),some d18O values reach $0&, which is approximately2& lower than Holocene Dead Sea aragonites (typicallyaround $23&; Katz et al., 1977; Kolodny et al., 2005).The extremely low d18O values find no equivalence in theEM record where the MIS7 is only slightly (


21/28

The specific mechanism driving the above mentionedshort-term wet phases during overall dry interglacials inthe DSB region is not known. Enhanced deposition of trav-ertine at the Arava valley during MIS7 and 5 (Livnat andKronfeld, 1985) as well as enhanced formation of cavedeposits in the Negev Desert during MIS9, 7 and 5 (Vaks,2007) might indicate a southern (relative to the DSB) source

of humidity (further support for a southern humiditysource during these stages comes from an increase in theformation of cave deposits in Arabia (e.g., Fleitmannet al., 2003) and the short-term formation of lakes in the Sa-hara (e.g., McKenzie, 1993; Szabo et al., 1995)). In this con-text, we point out that EM sapropel events, which aregenerally associated with enhanced summer monsoon pre-cipitation over subtropical Africa and subsequent heavyNile floods in the EM, usually corresponds to interglacialtimes (Rossignol-Strick, 1983; Rossignol-Strick and Pa-terne, 1999; Fig. 14). Other possible mechanisms that couldexplain the occurrence of short humid episodes duringMIS7 (and other interglacials) are Red Sea Troughs (e.g.,

Kahana et al., 2002; Greenbaum et al., 2006) or TropicalPlumes which pertain to the transport of moisture fromthe Tropics (i.e., West Africa) to extra-tropical latitudes(e.g., Ziv, 2001; Rubin et al., 2007). Such episodes havebeen recorded during the 20th century and have occasion-ally triggered extremely heavy rain over short periods(hours and days) in the Negev Desert.

Alternatively, Vaks (2007) suggested that interglacialcave deposits in the Negev formed due to a southwardand inland shift of Mediterranean systems. While thismechanism could explain our observations it contradictsthe overall limnologicclimatic regime that is interpretedto have existed during MIS7; the bulk of the sequencedeposited at the AC site during this stage consists of detrital

marls, which are considered to represent a relatively low-stand water body (Waldmann et al., 2007). One would ex-pect that periods of enhanced humidity originating in theMediterranean would most probably be characterized byan overall lake level rise.

Significant changes in temperature compared to present-day conditions could be responsible for deviations from thesource effect -dominated signal. Indeed, during intergla-cial MIS7, it is quite possible that the temperature was war-mer than today. However, the light-d18O excursions wouldbe equivalent to a $10C temperature increase (assuming1& represents a temperature shift of 46C), which is con-sidered improbable (especially in view of the short (annual?)perturbation recorded by the sediments).

In summary, several possible mechanisms controllingthe increased input of freshwater to Lake Amora andresulting in light-d18O excursions are considered, but this is-sue is far from resolved and will require furtherinvestigation.

4.5.3. The limnological history of Lake Amora

The history of climate changes in the Levant during thelate Quaternary is well documented and indicates a closelyconnected climatic response between North Atlantic eventsand the hydrologic conditions that prevailed in the easternMediterranean (e.g., Bar-Matthews et al., 2003; BartovT

able6

PZ2sectionagesummary.Sources:1-R

ecalculatedfromWaldmannetal.(2007);2-thisstudy;3-IsochronagerecalculatedfromKaufmanetal.(1992);4-Agesrecalcula

tedfromWaldmann

(2002);ErrorsretrieveddirectlyfromW

aldmann(2002);5-SheinkmanandShlukov

(2001).

a-Errorsretrievedfromuncorrected

data.Sampleinboldfontwereusedforcalculationofregression

curvepresentedinFig.13(seetextfor

details).

Sample

name

Elevation

(m)

Source

234U

238U

230Th

238U

U

(ppm)

Th(ppm)

Uncorrected

age(ka)

Corrected

age(ka)

234U

238U

initial

PZ2-34

56.4

4

1.459n.a.

0.675n.a.

3.0

3

0.182

65.3

(3.0)

59.9

(3.0)a

1.459

PZ-2-33

56

1

1.4240.003

0.8130.004

1.3

4

0.046

87.6

(+0.9,0.9)

84.7

(+1.2,

1

.2)

1.424

PZ-2-7

33

1

1.2900.002

1.0670.006

1.2

4

0.045

170.4

(+2.6,2.6)

167.1

(+3.7,

3.6)

1.291

PRZ2-40

31.6

4

1.300n.a.

1.160n.a.

3.2

0

0.072

203.6

(

62.0)

201.5

(

62.0)a

1.300

PZ2-SAM

3

29

2

1.2830.002

1.3590.003

2.6

6

0.410

418.3

(+15.1,13.9)

337.5

(+145,

49)

1.284

PRZ-8

25

3

1.2160.021

1.2510.121

n.a.

n.a.

/

376.2

(+1,1

87)

1.216

PRZ-7

18

3

1.3460.016

0.9630.061

n.a.

n.a.

/

126.8

(+18.8,

15.8)

1.346

PZ2-41

7

2

1.231n.a.

1.267n.a.

2.4

7

0.112

368.6

(+13.0,11.7)

364.9

(+24.8,

20.3)

1.231

PZ241-d

7

1

0.9530.002

1.0140.013

1.4

1

2.627

/

PZ2-26Top

3.6

2

1.2170.002

1.3220.004

4.7

5

0.293

338.2

(+43,27)

318.4

(+1,49.1)

1.217

PZ-2-28

1.8

2

1.2280.003

1.2400.005

1.8

1

0.069

329.6

(+13.0,11.6)

326.3

(+18.2,

15.6)

1.228

PZ2-47

50

5

82.0

(

16)



22/28

et al., 2003). For its own part, the DSB lacustrine record re-flects the integrated interplay between incoming freshwater,the Ca-chloride subsurface brine and processes such asevaporation, anoxia and brinerock interaction.

In light of the strong similarity between the characteris-tics of previously studied sequences (Lisan and Zeelim) and

the current one (i.e., Amora AC section), we assume thatthe strong global climatic modulation attributed to thehydrologicallimnological regime during late Quaternary(e.g., Bartov et al., 2003; Haase-Schramm et al., 2004)was also dominant during the Amora period, and it is there-fore reflected by the lithological and chemical compositionof the Amora sediments.

In the following discussion we present a summary of themain episodes in the lake evolution, based on the chrono-logical framework established here (Fig. 14):

(1) 740 to 550 ka (MIS18-14), AFM-I: rapid transitionsbetween a deep-lake environment during which primaryaragonite and gypsum precipitated, and a low-stand lake

that deposited fine to coarse clastic units. The sharp nega-tive excursions ofd18O observed at the base of the section(612, 696 and 732 ka) support the occurrence of a flood re-gime that was dominant during this period. Thus, the lakewas unstable and was characterized by rapid shifts betweenhigh and low stands. Nevertheless, the occurrence of arago-

nite sequences and intermittent Ca-sulfate layers indicatesthat the typical DSB limnologicalhydrological configura-tion identified during the last glacial and during the Holo-cene (e.g., stratified water column and interaction betweenthe Pliocene Ca-chloride brines and incoming freshwater;Stein et al., 1997) already existed during mid Pleistocenetimes ($700 ka).

The sharp lithologic transition at $96 m (approximately550 ka) implies a major change in the path of sediment sup-ply to the lake. Considering the very long time scale of theeffect of tectonic processes, it is hard to imagine that thisshift was tectonically controlled (i.e., decrease in the reliefof surrounding land surface). On the other hand, it might

Fig. 13. The chronology of the PZ2 section and its correlation to the AC section. Ages in the PZ2 section (full diamonds) pertain to data givenin Table 6, and d18O data (empty diamonds) is derived from EA 2. d18O data of AC section (average values represented by red diamonds,connected by blue curve) is from EA 1. MIS boundaries in the PZ2 section are determined on the basis of absolute dating, complementaryd

18O wiggle matching and lithological transitions, and are correlated (red dashed curves) to corresponding sequences in the AC section (notethe different vertical scales). The black curve in the PZ2 section describes an error-weighed linear regression age curve and its error margins(dashed curves; 19 ka). A lithological boundary at $29 m in the PZ2 section might represent the timing of the MIS7-6 transition; d18O valuesat $30 m might indicate that the MIS6-7 transition is located a few meters higher. The PZ2 columnar section is modified after Waldmann(2002). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this paper.)

http://-/?-http://-/?-http://-/?-http://-/?-


23/28

be explained by a cessation in the activity of a major paleo-river nourishing clastic material into the southern part ofthe DSB (e.g., the paleo-Arava river?). According to thisscenario, the river died away over a relatively short period,

perhaps due to a shift to drier climatic conditions or alter-natively, its flow route shifted so that it stopped nourishingthe AC site with clastic material.

(2) 550 to 420 ka (MIS14-12), AFM-II: recurring cyclicdeposition of marls, gypsum and aragonite at the AC site.During this period the lake was generally low and the ACsite was submerged beneath a shallow, sabkha-like watercolumn.

(3) 420 ( 10) ka (MIS11), AFM-III: the transitioninto the dry and warm MIS11 interglacial is marked bythe precipitation of a massive halite unit, reflecting a signif-icant rise in lake salinity. Assuming the MIS11 conditionsresembled those that existed during the Holocene, the level

of Lake Amora could have dropped during the precipita-tion of the AFM-III salt to below 400 m bmsl.

(4) 400 to 240 ka (MIS11/10 to MIS8), the lower part ofAFM-IV: the lake deposited intermittent sequences of pri-

mary aragonite and marls with sporadic deposition of gyp-sum or clastics. The relatively low sedimentation rate andoccasional appearance of sand layers imply that unidenti-fied hiatus(es?) might exist throughout this part of the sec-tion, indicating significant lake level drops.

(5) 240 to 190 ka (MIS7): the lake deposited a thick se-quence of massive and laminar marls with some primaryaragonite laminae as well as a few gypsum and coarse sandlayers, indicating an arid climate. At the same time, thelight d18O excursions are interpreted as reflecting recurringfreshwater flooding of the lake (see Section 4.5.2).

(6) 190 to 130 ka (MIS6), upper part of AFM-IV: the lakedeposited mainly sequences of primary aragonite and gypsum

Fig. 14. Chronology and generalized scheme of climatic fluctuations in the Dead Sea basin during the past $740 ka. (A) Global stacked d18Obenthic record (Lisiecki and Raymo, 2005); (B) East Mediterranean d18O and sapropel records (blue curve and gray rectangles, respectively;Kroon et al., 1998); (C) Age-normalized d18O record of the Dead Sea basin lakes (data for Lisan and Holocene Dead Sea is from Kolodnyet al. (2005)). The orbital eccentricity (gray curve; Berger and Loutre, 1991), is in phase with the Amora d18O record, reflecting the orbital-tuning of this sequence; (D) Relative lake level curve for past $140 ka and estimated runoff/evaporation ratio for Lake Amora. The thickblack curve represents the compiled lake level curve between 70 ka and the present (Bartov et al., 2003) and the dashed black curve representsthe estimated lake level curve between $130 and 70 ka (Waldmann et al., 2007), Both black curves pertain to the upper abscissa. The dashedgray curve pertains to the lower abscissa (runoff/evaporation ratio) and was constructed on the basis of the age-elevation model and thelithological components of the AC section. The thick gray curve represents a 5-point running average for the dashed gray curve. Grayrectangles represent the lake conditions inferred from the PZ2 columnar section and chronology. A generalized description of the temporallimnological conditions is given on the right of (D). (For interpretation of the references to color in this figure legend, the reader is referred tothe web version of this paper.)



24/28

and the sedimentarysectionis overall similar to the last glacialPZ2 and Massada sections (Lisan Fm.). This indicates en-hanced long-term supply of freshwater and subsequent lakelevel rise as well as build up of long-term density-stratification(in contrast, during MIS7 the low-stand lake was typicallymixed). The development of anoxic conditions in the lowerlayer (brine) of the lake are reflected by the well-defined

appearance of native sulfur concretions between $200 to160 ka. These are the product of partial and late oxidationof reduced sulfide concretions in the lake sediments (see Torf-stein, 2008). The high fraction of primary minerals observedin the upper part of AFM-IV, implies that this time periodprobably represents the longest wet stage in the recorded his-tory of Lake Amora; only the sedimentary sequence of MIS2reflects a wetter stage during the past 740 ka.

(7) 130 to 70 ka (MIS5; last interglacial), AFM-V: asharp transition from high-stand to low-stand conditions(MIS6-5) is marked by a lithological change from arago-nite-gypsum facies to a 20-m-thick sequence dominatedby clastic material and totally devoid of aragonites. Inde-

pendent lake level curves constructed by Waldmann et al.(2007) and Bartov et al. (2007) are in agreement with thislimnological transition. While this period, which corre-sponds to AFM-V (and the Samra Fm.), is identified atthe AC site on the basis of extrapolation from underlyingsediments (where it is estimated that the youngest layer is$90 ka old), it is well defined at PZ2 and has been compre-hensively discussed by Waldmann et al. (2007).

Two more sedimentologicclimatic stages are identifiedin this region and although they are not discussed in thisstudy they provide a strong reference for the interpretationof the limnological history of Lake Amora:

(8) 70 to 10 ka (MIS4-2; last glacial), the Lisan Fm.: thelast glacial Lake Lisan was characterized by high levels and

typically deposited thick sequences of primary aragonitesand silty detritus with intermittent gypsum and sand layers(at PZ1 and Massada sites).

(9) 10 ka to present, the Holocene Zeelim Fm.: the DeadSea was characterized by low levels ($400 30 m bmsl)and typically deposited marls and sporadic aragonitelaminae.

In summary, the AmoraLisanDead Sea sequence dis-plays oscillating sedimentological and geochemical patternsover the past $740 ka. A notable shift in the frequency ofchanges occurs at 96 m, where the section shifts fromAFM-I, characterized by repeated lithological transitionsbetween shallow and deep lake facies (i.e., clastic vs. arago-nite and anhydrite-dominated layers, respectively) to AFM-II, characterized by a generally low-stand lake facies. Moreimportantly, this shift corresponds to a shift from a rapidlyfluctuating lake to a more stable one. The age of AFM-I isapproximately 740 to 550 ka, corresponding to the laststage in the transition from the orbital-controlled 41 Kato 100 ka-dominated climate cycles, known as the mid-Pleistocene transition ($920 to 640 ka; Mudelsee andSchulz, 1997). Establishing whether or not the Amora re-cord reflects this event requires a thorough chronological-lithological-geochemical study of a consecutive sequenceof the Amora Fm. reaching back to the early Pleistocene.Such a sequence can only be found buried in the subsurface.

4.5.4. Amora Formation in the subsurface

Lake Amora deposits overlay the thick halite sequencethat precipitated from the (Pliocene?) Sedom Lagoon. Theage of the base of the exposed Amora Fm. at the AC site,$740 ka, refers to the lower-most exposed part of the sedi-mentary section. While the age of the base of the AmoraFm. is not known, the new chronology established here

provides the possibility of correlating between exposedand buried sequences, which reach depths of several kilo-meters (Horowitz, 1987, 2001; Salhov et al., 1994). In theSedom-Deep-1 borehole located several kilometers southof Mt. Sedom (Fig. 1), the SedomAmora transition is lo-cated at a depth of$3.8 km (Salhov et al., 1994). A singleoccurrence of massive (i.e., several meters thick) halitethroughout this section is recorded at a depth of 540 m. Ifthis salt correlates with the AFM-III salt at the AC site(age of$420 ka), the average sedimentation rate at the Se-dom-Deep-1 site during the past 420 ka would be $1.3 m/ka. Assuming the effect of sediment compaction was con-stant at these depths, extrapolation of the above sedimenta-

tion rate provides a first order approximation of the timingof the Sedom-Amora transition at 2.95 Ma, close to previ-ous estimations of the age of the Sedom lagoon (3.44 Ma(Steinitz and Bartov, 1991) and $3.2 Ma (Stein and Agnon,2007)).

The extension of the chronology of the Amora Fm. tothe late Pliocene requires recovery of suitable samples fromdeep drills at the bottom of the Dead Sea and developmentof appropriate dating methods, such as the oxygen isotopestratigraphy presented in this study.

5. CONCLUSIONS

1. The chronology of the Amora Fm. was established bycombining U-series, d18O wiggle matching, stratigraphiccorrelations to dated sequences, extrapolated sedimenta-tion rates and complementary paleomagnetic data. Thiseffort extends the chronology of the Dead Sea basin fill,thus establishing one of the longest records of its kind.Each of the chronological tools discussed in this studycould not independently suffice for the determinationof the chronology of the Amora Fm., and it is only theircombination and iterative age determinations that pro-vided the chronological framework of Lake Amora.

2. The ages of the deposits of Lake Amora at the Arubot-aim Cave (AC) site range from 740 to less than 100 ka,covering seven glacialinterglacial cycles (MIS18-5).These deposits (the Amora Fm.) are exposed on the east-ern and western flanks of Mount Sedom, where theywere uplifted and tilted by the rising Sedom salt diapir.

3. The Amora sequence exposed at the Perazim valley westof Mt. Sedom spans the time interval of 340-70 ka(MIS10-5), correlative to AFM-IV and AFM-V definedat the Arubotaim Cave site. This sequence is overlain bya sequence deposited during MIS4-2 (the Lisan Fm.).

4. The limnological history of Lake Amora displays cyclicshifts between dry-low-stand conditions and wet-high-stand conditions. After a period of frequent level fluctu-ations ($ 740 and 550 ka), the lake declined and the



25/28

Arubotaim Cave site was characterized by Sabkha-likeconditions. This period was terminated with the instiga-tion of MIS11 (at $420 ka), during which a prominentsalt layer was deposited. MIS7 was characterized by alow-stand and sporadic intensive flooding events. MIS6was a high-stand period that was followed by a signifi-cant lake level decline corresponding to the transition

to last interglacial (MIS5).5. The transition in the global climatic records from a 41 ka

to 100 ka cycles (i.e., the mid-Pleistocene transition)might be reflected by a shift from frequent lake standfluctuations pre-550 ka to a more stable lake thereafter.

6. During MIS7 (and to a lesser extent during MIS17, 15and 9), d18O excursions to lighter values are interpretedto reflect short but intense freshwater flooding of thelake, possibly caused by rain storms originating fromsouthern latitudes. This issue requires furtherinvestigation.

7. The new chronology presented here for the Amora For-mation opens up the possibility of correlating between

exposed and buried sequences and extending the avail-able chronology further back, possibly to the late Plio-cene. A first order calculation yields an age ofapproximately 3 Ma for the SedomAmora transition.Thus, an important but yet unstudied geological archivecovering the entire Quaternary is unveiled for futureresearch.

ACKNOWLEDGMENTS

We thank Y. Enzel (HUJI) and A. Ayalon (GSI) for their crit-ical reading of an earlier version of this work, as well as A. Starin-sky (HUJI), I. Gavrieli and M. Bar-Matthews (GSI), and S.Goldstein (LDEO) for fruitful discussions. A. Hofmann (MPI,Mainz) was one of the initiators of this project and provided assis-tance along the way. E. Shelef, B. Tatarsky, N. Jesselson, U. Shaa-nan, U. Kodington and S. Reuveni helped in field and lab work.We also thank E. Barkan for performing the oxygen isotope anal-yses. We benefited from comments by two anonymous reviewers.The study was supported by the German Israel Science FoundationGIF Grant No. G-717.129.8/2001 to Y.K. and A.H., and the Bi-National Science Foundation BSF Grant No. 2006.363 to M.S.

APPENDIX A. SUPPLEMENTARY DATA

Supplementary data associated with this article can befound, in the online version, at doi:10.1016/j.gca.2009.

02.010.

REFERENCES

Bar-Matthews M., Ayalon A., Gilmour M., Matthews A. andHawkesworth C. J. (2003) Sea-land oxygen isotopic relation-ships from planktonic foraminifera and speleothems in theEastern Mediterranean region and their implication for paleo-rainfall during interglacial intervals. Geochim. Cosmochim. Acta67, 31813199.

Bartov Y., Enzel Y., Porat N. and Stein M. (2007) Evolution of thelate Pleistocene - Holocene Dead Sea basin from sequence

stratigraphy of fan deltas and lake-level reconstruction. J. Sed.Res. 77, 680697.

Bartov Y., Goldstein S. L., Stein M. and Enzel Y. (2003)Catastrophic arid episodes in the Eastern Mediterranean linkedwith the North Atlantic Heinrich events. Geology 31, 439442.

Bartov Y., Stein M., Enzel Y., Agnon A. and Reches Z. (2002)Lake levels and sequence stratigraphy of Lake Lisan, the latePleistocene precursor of the Dead Sea. Quat. Res. 57, 921.

Begin Z. B., Ehrlich A. and Nathan Y. (1974) Lake Lisan: thePleistocene precursor of the Dead Sea. Geol. Surv. Isr. Bull. 63,130.

Berger A. and Loutre M.F. (2003) Climate 400,000 years ago, a keyto the future? In: Earths Climate and Orbital Eccentricity (eds.A. W. Droxler, R. Z. Poore and L.H. Burckle). AmericanGeophysical Union, Washington, DC, pp. 1726.

Berger A. and Loutre M. F. (1991) Insolation values for the climateof the last 10 million years. Quat. Sci. Rev. 10, 297317.

Bischoff J. L. and Fitzpatrick J. A. (1991) U-series dating of impurecarbonates: An isochron technique using total-sample dissolu-tion. Geochim. Cosmochim. Acta 55, 543554.

Braun D., Ron H. and Marco S. (1991) Magnetostratigraphy of thehominid tool bearing Erk el Ahmar formation in the northernDead Sea rift. Isr. J. Earth Sci. 40, 191197.

Broecker W. S. (1963) A preliminary evaluation of uranium seriesinequilibrium as a tool for absolute age measurement of marinecarbonates. J. Geophys. Res. 68, 9, 2817283.

Cande S. C. and Kent D. V. (1995) Revised calibration of thegeomagnetic polarity timescale for the Late Cretaceous andCeneozoic. J. Geoph. Res. 100, 60936095.

Cheng H., Edwards R. L., Hoff J., Gallup C. D., Richards D. A.and Asmerom Y. (2000) The half-lives of uranium-234 andthorium-230. Chem. Geol. 169, 1733.

Cherdyntsev V. V. (1971) Uranium-234. Moskva, Atomizat(translated from Russian by J. Schmorak). J. Israel Prog. Sci.Trans., 234 p.

Davis M., Enzel Y., Matmon A. and Zilberman E. (2007)Morphological processes in the meeting point of an activetectonic lineament, a drainage system and a retreating laketourguide for the Nahal Zin and the Amatzyahu fault. In: IsraelGeological Society Annual Meeting, Neve Zohar (In Hebrew).

Eardley A. J., Shuey R. T., Gvosdetsky V., Nash W. P., DanePicard M., Grey D. C. and Kukla G. J. (1973) Lake cycles inthe Bonneville Basin, Utah. Geo. Soc. Am. Bull. 84, 211216.

Edwards R. L., Gallup C. D. and Cheng H. (2003) Uranium-seriesdating of marine and lacustrine carbonates. Rev. Mineral.Geochem. 52, 361405.

Edwards R. L., Chen J. H. and Wasserburg G. J. (1987) 238U,234U, 230Th, 232Th systematics and the precise measurementsof time over the past 500,000 years. Earth Planet. Sci. Lett. 81,175192.

Enzel Y., Amit R., Dayan U., Crouvi O., Kahana R., Ziv B. andSharon D. (2008) The climatic and physiographic controls of

the eastern Mediterranean over the late Pleistocene climates inthe southern Levant and its neighboring deserts. Glob. Planet.Chan. 60, 165192.

EPICA community members (2004) Eight glacial cycles from anAntarctic ice core. Nature 429, 623628.

Fleitmann D., Burns S., Neff U., Mangini A. and Matter A. (2003)Changing moisture sources over the last 330,000 years inNorthern Oman from fluid inclusion evidence in speleothems.Quat. Res. 60, 223232.

Fontugne M. and Calvert S. E. (1992) Late Pleistocene variabilityof the carbon isotopic composition of organic matter in theEastern Mediterranean: Monitor of changes in carbon sourcesand atmosphere CO2 concentrations. Paleoceanography 7, 120.

http://dx.doi.org/10.1016/j.gca.2009.02.010http://dx.doi.org/10.1016/j.gca.2009.02.010http://dx.doi.org/10.1016/j.gca.2009.02.010http://dx.doi.org/10.1016/j.gca.2009.02.010


26/28

Frumkin A., Ford D. C. and Schwarcz H. P. (1999) Continentaloxygen isotopic record of the last 170,000 years in Jerusalem.

Quat. Res. 51, 317327.

Gat J. R. (1984) The stable isotopic composition of Dead Seawaters. Earth Planet. Sci. Lett. 71, 361376.

Greenbaum N., Ben-Zvi A., Haviv I. and Enzel Y. (2006) Thehydrology and paleohydrology of the Dead Sea tributaries. In:New frontiers in Dead Sea paleoenvironmental research, GSA

Special Paper (eds. Y. Enzel, A. Agnon and M. Stein), vol. 401,pp. 6393.

Haase-Schramm A., Goldstein S. L. and Stein M. (2004) UThdating of Lake Lisan (late Pleistocene Dead Sea) aragonite andimplications for glacial East Mediterranean climate change.

Geochim. Cosmochim. Acta 68, 9851005.

Haase-Schramm A., Bohm F., Eisenhauer A., Dullo W.-C.,Joachimski M. M., Hansen B. and Reitner J. (2003) Sr/Caratios and oxygen isotopes from sclerosponges: Temperaturehistory of the Caribbean mixed layer and thermocline duringthe Little Ice Age. Paleoceanography 18, 1073. doi:10.1029/2002PA000830, 2003.

Horowitz A. (2001) The Jordan Rift Valley. Balkema, Lisse, theNetherlands 730 pp.

Horowitz A. (1987) Palynological evidence for the age and rate ofsedimentation along the Dead Sea Rift, and structural impli-cations. Tectonophysics 141, 107115.

Israelson C., Bjorck S., Hawkesworth C. J. and Possnert G. (1997)Direct UTh dating of organic-and carbonate-rich lake sedi-ments from southern Scandinavia. Earth Planet. Sci. Lett. 153,251263.

Jaffrey A. H., Flynn K. F., Glendenin L. W., Bentley W. C. andEssling A. M. (1971) Precision measurements of half-lives andspecific activities of 235U and 238U. Phys. Rev. 4, 18891906.

Kahana R., Ziv B., Enzel Y. and Dayan U. (2002) Synopticclimatology of major floods in the Negev Desert, Israel. Int. J.Clim. 22, 867882.

Katz A., Kolodny Y. and Nissenbaum A. (1977) The geochemicalevolution of the Pleistocene Lake Lisan-Dead Sea system.

Geochim. Cosmochim. Acta 41, 16091626.

Kaufman A. (1993) An evaluation of several methods for deter-mining 230Th/U ages in impure carbonates. Geochim. Cosmo-chim. Acta 57, 23032317.

Kaufman A. (1971) U-series dating of Dead Sea basin carbonates.Geochim. Cosmochim. Acta 35, 12691281.

Kaufman A., Wasserburg G. J., Porcelli D., Bar-Matthews M.,Ayalon A. and Halicz L. (1998) UTh isotope systematics fromthe Soreq Cave, Israel and climatic correlations. Earth Planet.Sci. Lett. 156, 141155.

Kaufman A., Yechieli Y. and Gardosh M. (1992) Re-evaluation oflake-sediment chronology in the Dead Sea basin, Israel, basedon new 230Th/U dates. Quat. Res. 38, 292304.

Ken-Tor R., Stein M., Enzel Y., Agnon A., Marco S. and

Negendank J. F. W. (2001) Precision of calibrated radiocarbonages of historic earthquakes in the Dead Sea basin. Radiocarbon43, 13711382.

Kolodny Y., Stein M. and Machlus M. (2005) Sea-Rain-Lakerelation in the last Glacial East Mediterranean revealed by 18O -13C in Lake Lisan aragonites. Geochim. Cosmochim. Acta 69,40454060.

Kroon D., Alexander I., Little M., Lournes L. J., Matthewson A.,Robertson A. H. F. and Sakamoto T. (1998) Oxygen isotopeand sapropel stratigraphy in the Eastern Mediterranean duringthe last 3.2 million years. Proc. ODP 160, 181189.

Ku T. L. and Liang Z. C. (1984) The dating of impure carbonateswith decay-series isotopes. Nuclear Instrum. Methods Phys. Res.223, 563571.

Lazar B., Enmar R., Schossberger M., Bar-Matthews M., Halicz L.and Stein M. (2004) Diagenetic effects on the distribution ofuranium in live and Holocene corals from the Gulf of Aqaba.Geochim. Cosmochim. Acta 68, 45834593.

Le Roux L. J. and Glendenin L. E. (1963) Half-life of 232Th. Proc.Nat. Meet. Nuclear Energy. Pretoria, South Africa. pp. 8394.

Lian O. B. and Roberts R. G. (2006) Dating the Quaternary:progress in luminescence dating of sediments. Quat. Sci. Rev.25, 24492468.

Lin J. C., Broecker W. S., Anderson R. F., Hemming S.,Rubenstone J. L. and Bonani G. (1996) New 230Th/U and14C ages from Lake Lahontan carbonates, Nevada, USA, and adiscussion of the origin of initial thorium. Geochim. Cosmo-chim. Acta 60, 28172832.

Lisiecki L. E. and Raymo M. E. (2005) A Pliocene-Pleistocenestack of 57 globally distributed benthic d18O records. Paleoce-anography 20, PA1003.

Livnat A. and Kronfeld J. (1985) Paleoclimatic implicatins of U-series dates for lake sediments and travertines in the Arava RiftValley, Israel. Quat. Res. 24, 164172.

Ludwig K. (2003) Mathematical-statistical treatment of data anderrors for 230Th/U geochronology. Rev. Mineral. Geochem. 52,631656.

Ludwig K. R. and Titterington D. M. (1994) Calculation of230Th/U isochrones, ages, and errors. Geochim. Cosmochim. Acta 58,50315042.

Ludwig K. R., Simmons K. R., Szabo B. J., Winograd I. J.,Landwehr J. M., Riggs A. C. and Hoffman R. J. (1992) Mass-spectrometric 230Th234U238U dating of the Devils Hole calcitevein. Science 258, 284287.

Luo S. and Ku T. L. (1991) U-series isochron dating: A generalizedmethod employing total-sample dissolution. Geochim. Cosmo-chim. Acta 55, 555564.

Machlus M., Enzel Y., Goldstein S. L., Marco S. and Stein M.(2000) Reconstructing low levels of Lake Lisan by correlatingfan-delta and lacustrine deposits. Quat. Int. 73/74, 137144.

Marco S. (1996) Paleomagnetism and Paleoseismology in the LatePleistocene Dead Sea Graben. Ph.D. Thesis, The HebrewUniversity of Jerusalem.

McKenzie J. (1993) Pluvial conditions in the eastern Saharafollowing the penultimate deglaciation: implications forchanges in atmospheric circulation patterns with global warm-ing. Palaeogeog. Palaeoclim. Palaeoec. 103, 95105.

McManus J. F., Oppo D. W., Cullen J. L. and Healey S. (2003)Marine isotope stage 11 (MIS11): analog for Holocene andfuture climate? Geophys. Monogr. 137, 6985.

Migowski C., Stein M., Prasad S., Negendank J. F. W. and AgnonA. (2006) Holocene climate variability and cultural evolution inthe Near East from the Dead Sea sedimentary record. Quat.Res. 66, 421431.

Migowski C., Agnon A., Bookman R., Negendank J. F. W. andStein M. (2004) Recurrence pattern of Holocene earthquakes

along the Dead Sea transform revealed by varve-counting andradiocarbon dating of lacustrine sediments. Earth Planet. Sci.Lett. 222, 301314.

Mudelsee M. and Schulz M. (1997) The Mid-Pleistocene climatetransi

torfstein et al., 2009 (gca, amora chronology)

Documents