Magnetic Properties of Late Holocene Dead Sea Sediments as a Monitor of Regional Hydroclimate

Diagenetic processes in anoxic sedimentary environments influence sediment magnetic properties mainly through dissolution of detrital magnetite and precipitation of authigenic greigite. Recently exposed late Holocene Dead Sea sediments provide an opportunity to study the processes governing greigite formation and preservation, and their relation to different hydrological settings. Magnetic data and pore‐fluid compositions were obtained from three Holocene sections along a N‐S transect on the western Dead Sea shore: Og, Ein‐Feshkha (EF), and Ein‐Gedi. The northern sections are closer to the major freshwater source to the Dead Sea‐the Jordan River. Detrital titanomagnetite is present at all sections, but greigite is the dominant magnetic phase at Og and EF. Bulk rock magnetic data vary between and within the sections by over 3 orders of magnitude, where higher values indicate higher greigite concentrations. At the three sites, pore fluids have similar or lower salinity than the modern and Holocene Dead Sea brine, with variable and dissolved iron (Fe2+) and sulfate (SO42−). Magnetic property changes are reflected by iron and/or sulfate microbial reduction that controlled sedimentary greigite formation. We propose that the N‐S greigite decrease suggests that anoxic microbial activity was controlled by labile organic matter and/or reactive iron brought by, or formed as a result of, freshwater influx from the Jordan River. Hence, greigite concentration changes depended on past freshwater input to the hypersaline lake and proximity to the freshwater source. The apparent relationship between hydrological conditions and magnetic properties provides a new method to trace past hydrological changes in the Dead Sea.

the winter of 1979 due to decreasing lake level and increasing salinity (Stiller & Chung, 1984). The modern lake is regressing rapidly (by > 1 m/year), mainly due to human activity; the current lake surface is at 435 m bmsl. The calcium chloride brine that filled the Holocene Dead Sea (and previous Quaternary lakes) is poor in bicarbonate and sulfate and, thus, deposition of mineral phases such as aragonite and gypsum requires addition of sulfate and bicarbonate to the lake. In the modern Dead Sea, these ions are supplied by freshwaters and springs that discharge to the Dead Sea (Torfstein et al., 2008). Thus, both primary (aragonite and gypsum) and detrital sediments that comprise Dead Sea lacustrine formations contain important information on hydroclimate conditions in the lake watershed and global climate that dictate the synoptic conditions that control rain and dust transportation (e.g., Stein, 2014 and references there).

Sulfate and Iron Reduction in the Dead Sea
The high salinity and high concentration of Mg 2+ > 2 M and Ca 2+ > 0.5 M makes the Dead Sea a harsh environment for microorganisms (Oren, 2001). Nevertheless, sulfate-reducing bacteria (SRB) activity has been detected in the Dead Sea water column, springs, and sediments from sulfur isotopic analyses of sulfide and sulfate (Bishop et al., 2013;Gavrieli et al., 2001;Häusler et al., 2014;Ionescu et al., 2012;Neev & Emery, 1967;Nissenbaum & Kaplan, 1976). Thomas and Ariztegui (2019) analyzed fluid inclusions from within halite and suggested that both archaea and bacteria are involved in the sulfur cycle in the Dead Sea. Overall, the high salinity and poor organic matter content limit MSR activity in the Dead Sea (Häusler et al., 2014;Ionescu et al., 2012;Thomas & Ariztegui, 2019;Thomas et al., 2016).
Little is known about iron reduction in the Dead Sea. Nishri and Stiller (1984) showed that iron reduction occurs in the water column and in sediment pore fluids. They studied the distribution and dissolution of iron and showed that 4,000 tonnes of allochthonous iron enters the Dead Sea via the Jordan River. Sedimentary iron sulfide formation -mackinawite (Fe 2+ S 2− ), greigite (Fe 2+ Fe 3+ 2 S 2− 4 ), and pyrite (Fe 2+ S − 2 ) -depends on the availability of iron and sulfides. When iron reduction outcompetes and overcomes sulfate reduction, dissolved sulfide is removed rapidly to form and preserve greigite (e.g., Kao et al., 2004).

Magnetic Properties of Dead Sea Sediments
Dead Sea sediments contain two dominant magnetic minerals: titanomagnetite and greigite (Ebert et al., 2018;Frank et al., 2007aFrank et al., , 2007bRon et al., 2006). Titanomagnetite is transported to the lake by fluvial and alluvial systems (e.g., Jordan River), while greigite is a diagenetic product of MSR activity. Frank et al. (2007aFrank et al. ( , 2007b investigated four Holocene sediment cores collected along the western Dead Sea shoreline and found large depth variations of magnetic susceptibility, isothermal remanent magnetization (IRM), anhysteretic remanent magnetization (ARM), and the ratio between the low-field IRM and susceptibility (χ). They interpreted these variations as reflecting magnetic mineralogy changes, where greigite-rich sediments have much higher χ and IRM/χ ratio values. However, the controlling mechanism of such variations has remained unclear.
In this study, we aim to reveal the mechanism that controls greigite formation in Dead Sea sediments. Additionally, we assess the effects of greigite on sediment magnetic properties and its environmental implications. We investigate three Holocene sedimentary outcrops, which are exposed freshly in new gullies that were entrenched in the lake floor during the (man-made) retreat of the modern lake ( Figure 1). From each outcrop, we assemble a detailed composite depth profile of magnetic parameters and pore-fluid geochemistry, where intervals with sharp magnetic parameter changes were further characterized using electron microscopy and first-order reversal curve (FORC; Pike et al., 1999) measurements.
ed detritus] described by Haliva-Cohen et al. [2012]) whose major minerals are quartz and calcite along with sequences of laminated aragonite and silty detritus (the aad facies, described by Machlus et al. [2000]). Although the outer surfaces of gully walls appear altered and oxidized, digging less than a few centimeters into the walls reveals dark sediments that contain anoxic hypersaline pore fluid. Images of representative outcrop sediments are presented in Figure 2. The sites selected for this study are from north to south: Nahal Og (Og) (31.7400°E; 35.49029°N), Ein-Feshkha Nature Reserve (EF) (31.70839°E; 35.45524°N), and Ein-Gedi Spa (EG) (31.41917°E; 35.38486°N). The first two sites are located close to freshwater springs (e.g., the EF spring system, which is currently the main freshwater supply to the Dead Sea after damming of the Jordan River in 1964). EG Spa is located close to saline springs that discharge calcium chloride brine into the lake. The studied sedimentary sequences span a similar time interval of ∼2,800-∼1,200 years before present (BP). The chronology of each site is based on new radiocarbon ages (Table S1 and Figure S1) on terrestrial organic debris that are integrated with the existing data (EF [Kagan et al., 2011], EG [Migowski et al., 2006]). The Og section was deposited between ∼2,400 and ∼1,600 years BP; the EF section was deposited between ∼2,800 and ∼1,200 years BP; and the EG section was deposited between ∼3,000 and ∼1,350 years BP. At each outcrop, we dug a vertical cross section and collected sediments in nonmagnetic plastic boxes (23 mm × 23 mm × 19 mm; Figure 2c) for magnetic analyses at ∼25-mm sampling intervals. The thickness of the intervals sampled at the investigated sections is 3.15, 4.6, and 2.6 m for Og, EF, and EG, respectively.

Magnetic Measurements
The magnetic measurement routine used in this study was as follows: ARM acquisition with a 0.1 mT direct current bias field and 100 mT AF field along the z-axis of the sample; AF demagnetization of the ARM in 11 steps from 5 to 110 mT; and acquisition of an IRM in a 1,500-mT field along the sample z-axis. In addition, all samples were weighed and the mass-normalized susceptibility was measured at 200 A/m at 976 Hz. ARM demagnetization was done using a 2G Enterprises superconducting rock magnetometer (SRM) 750 with in-line two-axis AF demagnetizer coils or using a 2G Enterprises RAPID SRM system with in-line two-axis coil demagnetizer and z-axis ARM coil; the ARM was acquired and measured using the RAPID system; IRM was imparted using an ASC pulse magnetizer and EBERT ET AL.  was measured using an AGICO JR-6A dual speed spinner magnetometer at the Geological Survey of Israel (GSI), Jerusalem. Low-field magnetic susceptibility was measured using an AGICO MFK-1 Kappabridge system. Magnetic analyses were made at the Paleomagnetic Laboratory at the Institute of Earth Sciences, Hebrew University of Jerusalem. FORCs for selected samples were measured at the Black Mountain Paleomagnetic Laboratory, Australian National University (ANU), using a Princeton Measurements Corporation MicroMag TM vibrating sample magnetometer. FORC data were analyzed using xFORC .

Electron Microscopy
Magnetic mineral extracts were isolated from selected intervals representing different magnetic behaviors from all three studied sections. Magnetic minerals were extracted from a mixture of alcohol and the entire sediment content of a sampling box, using a handheld rare-earth magnet within a plastic probe (Nowaczyk, 2011). The extracts were embedded in epoxy and were polished with a series of abrasives down to 1 μm. Mineral identification and composition were performed using two electron probe microanalysis (EPMA) instruments: (1) a JEOL JXA-8530F Plus at ANU and (2) a JEOL 8230 at the Institute of Earth Sciences, Hebrew University of Jerusalem. The EPMA instruments are equipped with an energy-dispersive X-ray spectrometer (EDS) and beam conditions were set to 15 keV for EDS analyses. Data were processed with a PRZ correction procedure, and all phases were analyzed using hematite and pyrite standards. Additional images were taken using a Quanta 200 environmental scanning electron microscope at the Harvey M. Krueger Family Center for Nanoscience and Nanotechnology, Hebrew University of Jerusalem.

Pore-Fluid Compositions
Sediment samples for pore-fluid chemical analyses were taken from each of the three sections in two phases. During the first sampling at EF, 25 samples were taken from a 2.5-m exposed section at 10-cm intervals. Around 50 cm of the exposed surface of the section was removed to ensure the removal of oxidized sediment. Using end-cut 20-ml syringes, sediment was added to 50-ml plastic falcon tubes flushed with Ar gas to prevent oxidation. During the second sampling, four sediment samples were taken from each outcrop at EF, Og, and EG. Here, relatively large quantities of undisturbed sediment were taken in nylon bags, which were analyzed for pore fluids at the GSI and Ben-Gurion University of the Negev. Pore fluid was extracted using both a centrifuge method (for SO 4 2− and Fe 2+ analyses) and a Carver© hydraulic press method (for major ions, salinity, and solution density). For the centrifuge method, pore fluids were extracted from sediments in 50-ml falcon tubes purged with Ar gas placed in a centrifuge at 9,000 rpm for 15 min. Pore-fluid solutions were subsequently purged again to remove any dissolved H 2 S and were then filtered using a 0.45-μm syringe filter. Solution density was measured using an Anton Paar DMA 35 density meter at the GSI. Major cation concentrations (Na + , K + , Ca 2+ , Mg 2+ , Sr 2+ ) were analyzed using inductively coupled plasma atomic emission spectroscopy at the GSI. Bromide (Br − ) and lithium (Li + ) chloride concentrations were analyzed by inductively coupled plasma mass spectrometry (ICP-MS) at the GSI. Combined chloride (Cl − ) and bromide (Br − ) were measured by titration. Br − values were measured using ICP-MS, and Cl − concentrations could then be derived from titration values. Analyses are within a charge-balance error of <2%. SO 4 2− concentrations were measured using a Metrohm© compact ion chromatography flex at BGU. Fe 2+ was measured on centrifuged samples by immediately adding pore fluid to 15-ml falcon tubes treated with ascorbic acid and ferrozine. Samples were then diluted with double-distilled water and Fe 2+ was measured via spectrophotometry. TDS in pore-fluid solutions were calculated by summing cation and anion concentrations.

Magnetic Data
Samples from the northernmost Og section yielded the highest χ values (around 10 −6 m 3 /kg), whereas those from the southern EG section have the lowest values (10 −7 -10 −8 m 3 /kg). Some parts of the EF section have high χ (10 −6 m 3 /kg) values, similar to Og, whereas others have intermediate values (10 −7 m 3 /kg) that are more similar to the higher values of the EG section ( Figure 3). Variations in all three parameters (χ, ARM, and IRM) are similar for each section, which indicates that χ changes reflect ferrimagnetic mineralogy changes and are not due to paramagnetic contributions. Therefore, the magnetic parameters are affected by the concentration and grain size of two ferrimagnetic minerals that co-occur in the sediment: titanomagnetite and greigite (Frank et al., 2007b;Ron et al., 2006) (see Section 3.2). However, titanomagnetite and greigite have many similar magnetic properties (Roberts, 1995;Roberts et al., 2011), which makes it important to distinguish between the two minerals. It has been suggested that high IRM/χ values are an indicator of greigite (Roberts, 1995;Sagnotti & Winkler, 1999;Snowball, 1991;Snowball & Thompson, 1990). Sediments from Og and part of the EF section are characterized by higher IRM/χ values than the EG section (

Electron Microscopy
Titanomagnetite and greigite are the only magnetic minerals identified in backscattered electron images. EDS analyses of the iron sulfide minerals indicate the presence of three dominant minerals: pyrite (FeS 2 ), greigite (Fe 3 S 4 ), and mackinawite (Fe x+1 S, x = 0-0.11). In general, darker mud is richer in iron sulfide aggregates compared to the gray mud, and its dominant ferrimagnetic mineral is greigite. The Og and EF sediments are richer in iron sulfide aggregates (Figures 4 and 5 Figure 3b). Sediments from the EG section have lower iron sulfide contents compared to the other two outcrops, and the dominant magnetic mineral is titanomagnetite. Iron sulfides in the EG section are illustrated in Figure 6, including pyrite from a sample with weak magnetization (Figure 6a), and aggregates of greigite and titanomagnetite from samples with low and moderate magnetizations, respectively (Figures 6b and 6c). sedimentation rates affect the thicknesses of detrital and aragonitic parts of laminae, which could contribute to magnetic mineral concentration changes via variable aragonite dilution. Bias due to aragonite layers can at most affect the mass-normalized magnetization by a factor of ∼2. Magnetic mineral concentration changes of orders of magnitude are observed between Og/EF and EG (Figure 3; upper three panels), so laminae thickness is not a critical driver of magnetic mineral concentration changes.

FORC Diagrams
FORC diagrams for samples imaged by electron microscopy are shown in the lower panels of Figures 4-6. The shapes of the FORC distributions are consistent with electron microscopy observations and magnetic data. Samples with higher magnetizations from the Og and EF sections, which are rich in greigite aggregates, are characterized by closed concentric contours with wide vertical spreading that are indicative of interacting greigite particles (Roberts et al., , 2011(Roberts et al., , 2014. The contours are centered about B c values that range between ∼25 mT ( Figure 4c) and ∼40 mT (Figure 5e), which is lower than for previously reported FORC diagrams for interacting single-domain greigite (e.g., Duan et al., 2017;Kelder et al., 2018;Roberts et al., 2006Roberts et al., , 2011Sagnotti et al., 2010). The peaks of the FORC distributions are located below the B i = 0 line, as previously observed in greigite-bearing sediments . Samples from the EG section, which have lower iron sulfide contents, have two components (Figures 6d-6f): (1) vortex state to multidomain behavior (e.g., Pike et al., 2001;Roberts et al., 2017), associated with large titanomagnetite, as indicated by a divergent distribution along the B i axis with low B c values; and (2)

Pore-Fluid Chemistry
The distribution of major and some minor ions (Na + , K + , Ca 2+ , Sr 2+ , Li + , Cl − , Br − ) in pore fluids for 35 samples from all sections varies slightly along the depth profiles, except for a few outliers ( Figure S2). TDS profiles (Figure 7a), which represent salinity, have mostly uniform values with an average of 334 ± 45 g/L for all sections. TDS values are similar or lower than the range for Holocene pore fluids in the deep International Continental Scientific Drilling Program (ICDP) Dead Sea core (Levy et al., 2017(Levy et al., , 2018. Mg 2+ (Figure 7b) has an average value of 1.75 ± 0.33 mmol/L for all sections. Mg 2+ is regarded as a conservative ion because pore fluids are undersaturated with respect to carnallite (KMgCl 3 ·6[H 2 O]), the dominant evaporite that can mineralize Mg 2+ in evaporated Dead Sea brines, and Mg/Br is mostly uniform (Levy et al., 2017). Given the low permeability typical of these sediments and the high dynamic viscosity of hypersaline solutions, which prevent advection and diffusion, the pore fluid may be a remnant of brine that has remained in situ since sediment deposition, similar to pore fluids from the deep ICDP Dead Sea core (e.g., Levy et al., 2017). The pore-fluid compositions are indicative of insignificant hydrological modification since recent sediment exposure (e.g., post-exposure evaporation or mixing with fresh groundwater), except for one sample from EG (25 cm), which has geochemical values similar to the Ein-Qedem (EQ) brine (marked in Figure 7 and Figure S2) , which suggests that the lower 20 cm of EG was washed by EQ brine. The main pore-fluid chemistry difference is in the concentration of dissolved Fe 2+ and SO 4 2− . EF has the highest Fe 2+ concentration and EG has the lowest; however, EG has the highest SO 4 2− concentration. Comparisons

Greigite Formation in the Dead Sea
We seek here to understand the processes that affect the magnetic properties of late Holocene Dead Sea sediments. Magnetic analyses and electron microscope observations indicate that the magnetic properties are controlled by the sedimentary greigite content. There is a clear N-S greigite concentration trend in the Holocene Dead Sea sediments (Figure 3). A major difference between these sites is their proximity to the main freshwater source to the hypersaline Dead Sea, the Jordan River, which raises the question of whether there is a connection between greigite formation and proximity to the primary freshwater source.
Several iron sulfides were identified in the studied sediments, which are produced during pyritization: mackinawite (Fe 2+ S 2− ), greigite (Fe 2+ Fe 3+ 2 S 2− 4 ), and pyrite (Fe 2+ S − 2 ). These minerals form as products of two reactants that form during early diagenetic anoxic microbial processes: (1) reactive iron (Fe 2+ )-which is a product of microbial iron (Fe 3+ ) reduction and (2) sulfide (S 2− )-which is a product of MSR. Bishop et al. (2013) and Thomas et al. (2016) suggested that the transformation from monosulfide to pyrite in Dead Sea sediments requires addition of S°. For microbial iron and sulfate reduction to occur, an adequate supply of labile organic matter, iron (Fe 3+ ), and sulfate are required. In the modern Dead Sea, there is a large sulfate supply via freshwater runoff, mostly from the Jordan River (Figure 1), but also via spring discharge (Torfstein et al., 2008). The Jordan River is also an important iron source to the Dead Sea (Nishri & Stiller, 1984), which can be assumed to have remained active throughout the Holocene because pore fluids from the EF and Og sites have relatively large sulfate and dissolved iron (Fe 2+ ) concentrations (Figures 8 and 9). This suggests that sulfate and iron were supplied to the Dead Sea throughout the studied time interval.
Dissolved iron (Fe 2+ ), which is a product of iron reduction, indicates that this process occurred at the studied sites (Figure 7d). The pore fluids are rich in dissolved iron, which suggests that iron reduction outcompetes sulfate reduction and that iron sulfides will form promptly. Availability of dissolved Fe 2+ favors greigite formation (Kao et al., 2004;Picard et al., 2018). The Og and EF sites (∼5 km apart) are located near the Jordan River inlet to the Dead Sea, and are, therefore, expected to have higher reactive Fe 2+ concentrations that promote higher greigite contents (Figure 8). However, the outcrops have variable dissolved Fe 2+ concentrations and the 0-170 cm and 320-460 cm intervals at the EF section have lower χ values than the 170-320 cm interval (Figure 8b). Dissolved Fe 2+ variability at each site and compared to other sites may have variable causes; for example, variable redox conditions, different reactive Fe 2+ supplies at different sites, or site-specific iron sinks (e.g,. pyrite mineralization).
EBERT ET AL. ), and (d) iron (Fe 2+ ), respectively. Green, red, and blue symbols denote data for the Og, EF, and EG sections, respectively. Diamonds and squares for EF represent data collected in the first and second samplings, respectively (see text). The orange line represents the geochemical composition of the EQ brine as given by . EF, Ein-Feshkha; EG, Ein-Gedi; EQ, Ein-Qedem; Og, Nahal Og; TDS, total dissolved solids.

(a) (b) (c) (d)
Similar to dissolved iron, sulfate appears to have been in ample supply (Figure 7c), and correlation with χ (Figure 9) may suggest a relationship between MSR and greigite formation. High dissolved Fe 2+ in pore fluids and abundant greigite imply that iron reduction is relatively high compared to sulfate reduction. However, precipitation/dissolution of calcium sulfate minerals can also result in decreased/increased sulfate concentrations, respectively (Levy et al., 2019). Although there are limitations in determining the factors that control dissolved Fe 2+ and sulfate distribution, there appears to be a similarity between changes in dissolved Fe 2+ concentrations (Figure 8), sulfate (Figure 9), and χ. Together, this evidence suggests a correlation between greigite formation and anaerobic-microbial activity (iron and sulfate reduction).
Assuming a connection between microbial activity and greigite formation, data from the northern sites (EF and OG) suggest greater greigite formation and predominant iron reduction compared to the south (EG). High χ occurs when iron reduction is relatively high compared to sulfate reduction (thus favoring greigite relative to pyrite). This raises the question of whether the distance from the Jordan River played a role in determining the degree of microbial activity and greigite formation. The Jordan River is the major freshwater source to the Dead Sea, so salinity differences between sites might have favored microbial activity in the north rather than the south. However, pore-fluid TDS in Figure 7a, which is representative of salinity, has similar values at all three studied sections (all are hypersaline). Furthermore, salinity and conservative ion concentrations at all sites are comparable to those of pore fluids from Holocene ICDP Dead Sea core sediments (Levy et al., 2017(Levy et al., , 2019, which suggests that the studied sites were probably closer to the hypolimnion (deep water) than to the fresher epilimnion (surface water) when the sediments were deposited.
Microbial activity might have been controlled by labile organic matter availability as shown by Häusler et al. (2014) and Thomas et al. (2016). Organic matter in the terminal Dead Sea sediments arrives from two sources: (1) allochthonous organic matter brought mainly via the Jordan River and (2) autochthonous organic material formed in the Dead Sea water-column by microorganisms, such as algal Dunaliella (Oren, 2010;Oren & Shilo, 1985). The amount of allochthonous organic material may decrease as a function of distance from the Jordan River. Additionally, autochthonous organic material in this hypersaline environment is related to freshwater influx via the Jordan River. Algae blooms under meromictic conditions, when a fresher surface layer forms due to sufficient freshwater input to the lake (Oren, 2010). Algal blooms were observed during the last significant formation of a fresher surface layer during the rainy winters of 1980 and 1992 (Oren, 1995;Oren & Shilo, 1982). Similar conditions to those of the winters of 1980 and 1992 prevailed occasionally in the lake during the late Holocene leading to algal blooms that would have supplied organic matter to the sediment.
To summarize, we suggest that allochthonous and/or autochthonous organic matter and/or iron supply may have been controlling factor(s) on microbial activity and subsequent greigite formation within sediments along the Holocene Dead Sea margin. All three factors are related to freshwater supply from the Jordan River. Therefore, we conclude that the χ of fresh sediment could be a useful site-specific proxy for Jordan River influx to the Paleo-Dead Sea.
EBERT ET AL.

Magnetic Susceptibility as a Proxy for Freshwater Input to the Holocene Dead Sea
Late Holocene Dead Sea sediments have χ values that fall within three ranges: high, intermediate, and low. We suggest that χ can be used to indicate different hydrological and environmental conditions, which rely on Jordan River freshwater inputs. Sediments with high χ values reflect higher freshwater inputs to the lake, intermediate χ values imply lower inputs, and lower χ values indicate almost no input. In addition, outcrop locations relative to the Jordan River inlet likely dictate χ changes because the freshwater influence from the Jordan River decreases to the south. Therefore, χ values at sites proximal to the Jordan River inlet (Og and EF) are sensitive to freshwater input changes, while χ values at EG are almost insensitive. Therefore, χ changes at northern sites (Figures 3a and 3b) reflect freshwater input changes from the Jordan River, and the χ profile provides a proxy for freshwater volume entering the lake. We correlate lake-level changes with the EF χ profile because it spans a longer time period.
The relationship between the EF χ record that is interpreted as a monitor of freshwater supply to the Dead Sea and other regional hydroclimate recorders is examined in Figure 10 in relation to Dead Sea lake levels for the past 3.5 Kyr (based on Bookman et al., 2004;Kushnir & Stein, 2019;Migowski et al., 2006). We note that high χ values correspond to lake-level rises above the sill at ∼402 m bmsl that separates the northern (deeper) and southern (shallow) Dead Sea basins (yellow line). During most of the Holocene, the southern Dead Sea basin behaved like a large evaporation space (similar to the modern situation) that buffered lake-level rise. Thus, significant freshwater inputs are required to fill the southern basin to allow it to rise above the sill (Bookman et al., 2004). Higher χ values are recorded during these periods (∼2,100-1,700 and ∼1,500-1,300 years BP; Figure 10b) of enhanced freshwater supply to the lake.
Temporal planktonic foraminiferal δ 18 O variations (Globigerinoides ruber) in an eastern Mediterranean deep-sea core  and in a Soreq Cave speleothem  are shown in Figure 10c. Foraminiferal δ 18 O values represent the composition, salinity, and temperature of eastern Mediterranean seawater, and mainly have a narrow range of values (δ 18 O = 0.2‰ ± 0.2‰), with pronounced negative peaks at 3.2 and 1.3 ka BP. These negative peaks were attributed by  to enhanced Nile River freshwater ingressions into the eastern Mediterranean, which reflect enhanced monsoon rains in the Ethiopian Highland Blue Nile source area. Enhanced Nile River inflows to the Mediterranean are most marked during sapropel events (e.g., sapropels S5 at ∼128-121 ka and S1 at ∼10 −6 ka [Rohling et al., 2002;Rossignol-Strick & Paterne, 1999]  within 14 C dating errors of the marine sediment core (U-Th dating was used for the speleothem). In general, speleothem δ 18 O values reflect a combination of factors, including the composition of source rains near the cave (in this case, Mediterranean seawater) and the rainfall amount. Mediterranean seawater composition dominates speleothem δ 18 O in the late Pleistocene Judean and Galilean caves (e.g., Bar-Matthews et al., 2003;Frumkin et al., 1999;Kolodny et al., 2005;Roberts et al., 2008). Nevertheless, variations between cave and Mediterranean seawater δ 18 O values (Δ 18 O cave-sea ) are also evident in Figure 10c, where black arrows denote times when Δ 18 O cave-sea was more positive (e.g., both sides of the 1.4 ka BP peak in the speleothem record), which indicates lower precipitation in the Judean Mountains. Blue arrows mark times with more negative Δ 18 O cave-sea that indicate higher Judean Mountain precipitation (in the Jordan River watershed; Kushnir & Stein, 2010), which also correlate with high χ values. We note that at 1.4 ka BP, Δ 18 O cave-sea is less negative, which reflects enhanced Nile water contributions to the eastern Mediterranean that "overcome" the "rain amount" effect (see temporal Δ 18 O cave-sea changes in Figure S3). In summary, the two independent regional hydroclimate archives-the Dead Sea lake-level curve and Soreq Cave δ 18 O values-indicate that the rainfall ("amount") pattern in the Judean Mountains and the Dead Sea watershed are both consistent with our conclusion that magnetic susceptibility data from EF sediment monitor freshwater inputs to the lake in high temporal detail. We note, however, that relative changes in the amplitudes of lake level and magnetic susceptibility peaks are different, for example, during the late Byzantine abrupt lake-level rise (∼1.4 ka BP). This possibly reflects the nature of each of these hydroclimate recorders. Dead Sea level is a "regional" hydroclimate recorder that integrates climatic inputs from its large drainage area that extends from the Gulf of Aqaba desert area to Mount Hermon in the Mediterranean climate zone (e.g., Stein, 2014). Magnetic susceptibility data from EF record mostly Jordan River freshwater inputs. The late Byzantine 1.4 ka BP level peak correlates with peak negative eastern Mediterranean δ 18 O values, which reflects enhanced fresh Nile River input to the sea. Thus, the Dead Sea level also responds to freshwater inputs from southern sources, as suggested for previous Dead Sea hydroclimate episodes that correlate with intervals of enhanced Nile River input to the Mediterranean Sea (sapropel events, for example, the last interglacial S5 and the early Holocene S1 [Torfstein et al., 2015;Waldmann et al., 2010]). Nevertheless, magnetic susceptibility provides a sensitive high-resolution archive of Jordan River freshwater inputs and is, thus, a potential recorder of the hydroclimate history of the eastern Mediterranean climate zone and discharge history of Mount Hermon, which is the largest natural water reservoir of the southern Levant.

Conclusions
We present magnetic property and pore-fluid chemistry results for late Holocene Dead Sea lacustrine sediments. Sediments were sampled from three sections exposed on the retreating shores of the modern Dead Sea along a north-south transect: Nahal Og and EF Nature Reserve on the northwestern side of the lake and EG Spa to the south. Magnetic property variations (e.g., magnetic susceptibility, χ) of more than 3 orders of magnitude mainly reflect variable sedimentary greigite contents. Greigite precipitation is controlled by iron and sulfate microbial reduction within the sediment, which varied along the N-S transect as a function EBERT ET AL. The yellow line marks the sill level (402 m bmsl), which is a natural barrier that separates the northern, deeper lake from the shallow southern part. Gray shading marks the association between lake levels above the sill and higher χ.  Figure  S3) are related to the amount effect in the vicinity of the cave in the Judean Mountains. At 1.4 ka, the sea-cave difference reflects enhanced inflow of low-δ 18 O Nile River water to the eastern Mediterranean. Correlation between lake level, χ, and δ 18 O values is evident, where high χ corresponds to high lake levels that reflect mainly increased Jordan River freshwater inputs. See text for discussion of the figure and its implications. bmsl, below mean sea level. BP, before present.
of distance from the Jordan River, and by the availability of labile organic matter and/or iron influx. Thus, magnetic property variations indicate freshwater influx changes to the lake. χ is extremely sensitive to freshwater availability in this depositional environment and provides a new tool to reconstruct decadal regional hydroclimate regime fluctuations. Relatively easily measured χ can, thus, be used as site-specific tracers of Dead Sea catchment hydrological conditions. The dependence of χ on the local hydrological regime provides a cautionary note for using χ for stratigraphic correlation, which should be assessed carefully when magnetic mineral assemblages are dominated by authigenic minerals such as greigite.