The Dead Sea (DS), located in the southern Levant, is a closed-basin lake whose size varies as a function of water availability. Reconstructing DS lake-level variations through time provides a quantitative measure of the natural hydroclimate variability and can inform on the local hydroclimate response to changes in global climate. Here, we constructed an updated lake-level history of the Holocene DS by:
The compilation contains 296 radiocarbon ages described in 19 papers and MSc and PhD theses (Fig. 4, Fig. S5 and Table S6). Of these, samples designated as outliers by the original authors (n = 9), dated by bulk organics (n = 9) or older than 12 kyr cal BP (n = 11) and snail samples with a very large reservoirs effect (n = 2) were not used in our lake-level reconstruction. The data derived from the uplifted caves at Mt. Sedom (Frumkin et al., 2001) are not used for the lake-level reconstruction (n = 25), but were previously applied to calculate the uplift rate of Mt. Sedom diapir and constrain the uplift of the late Holocene shoreline at Arubotaim Cave (Frumkin et al., 2001) (Fig. S5 and supplementary text). The final lake-level was reconstructed based on the remaining 238 ages. Of these, 45 are dated beach ridges or samples from near-shore environments, six are from alluvial units, 166 are from lacustrine sediments, and 21 samples are from gypsum structures (Fig. 4). The compilation of all existing lake-level data show that throughout the Holocene, the DS lake-level fluctuated ~68 m (from 440 to 372 m bsl). In Fig. 4, we present the data using the 1σ error of the calibrated ages. The relatively small errors, force, in some places, the lake-level to vary rapidly. Using a 2σ error reduces some of these rapid fluctuations, we present both records for comparison in the supplementary material (Fig. S5).
The compiled lake-level chronology, derived from the 238 radiocarbon ages from lake-margin exposures, southern basin cores, and from shallow cores from the western margin of the DS, provides a detailed history of the hydrological mass balance of the DS watershed and is the most direct and updated hydrological history of the Levant throughout the Holocene (Fig. 4). Additional important information regarding the DS lake status are derived from the deep core drilled into the depocenter of the northern basin and its lithological and geochemical proxies retrieved by the Dead Sea Deep Drilling Project (DSDDP) (Goldstein et al., 2020; Kiro et al., 2020; Kitagawa et al., 2017; Neugebauer et al., 2014; Torfstein et al., 2015). Below, we combine these two sets of records, the levels from the margins and inferences from the core, and provide a hydrological and limnological history of the Holocene DS and discuss its implications.
The Levant, a locus of some of the most dramatic cultural changes in human history (e.g. the onset of sedentism, agriculture and urbanization, Bar-Yosef and Belfer-Cohen, 1989; Greenberg, 2019), is located at the northern boundary of the global desert belt, and is characterized by high spatial variability of storms and annual rainfall totals (e.g., Enzel et al., 2008) making it a climatically fragile region (Ellenblum, personal communication). In such a region, small changes in water availability could incur substantial ecological and agricultural changes across the landscape and in turn, cultural responses (Peleg et al., 2012). The DS provides a quantitative hydroclimate record of the Levant and thus can be used to assess the potential role of climate in the region’s cultural evolution. As we are not archaeologists, we will not try to interpret the material culture in light of the hydroclimate reconstruction; however, we would like to evaluate whether there are possible relations between regional water availability and population dynamics.
We reconstructed the DS lake-level changes throughout the Holocene using:
Two palynological diagrams are available for this period (Fig. 1.3a,b)––Birkat Ram and the Sea of Galilee (the latter begins in ca. 3150 BCE, corresponding to the later phase of the period). The Mediterranean arboreal pollen curve, including olive trees, appears at its highest percentages, indicating that the Early Bronze I was the most humid phase of the Bronze and Iron Ages. Relatively wet climate conditions are similarly evident from the Sorek Cave isotopic record (Bar-Matthews and Ayalon 2011: Fig. 6) as well as from the Jezreel Valley geoarchaeological data (Rosen 2006: 468–469; Adams et al. 2014). Wet climate conditions in the Early Bronze I may have facilitated the initial phase of the wave of settlement in the Negev Highlands (for the chronology, see Avner and Carmi 2001; Sebbane et al. 1993). The picture is similar in the semi-arid regions of Transjordan, which feature more Early Bronze I than Early Bronze II–III sites (Philip 2008: 189; Bradbury et al. 2014: 211–214).
The high olive pollen frequencies that characterize the Early Bronze I were followed by a dramatic decline in olive pollen percentages during the transition to the Early Bronze II, as is evident in the Birkat Ram and especially the Sea of Galilee pollen records (from a peak of more than 50% to a low of only 5% of the total pollen sum; Fig. 1.3a–b). Decreased olive pollen frequencies around 2900 BCE were also documented in the Lake Ḥula record (van Zeist, Baruch and Bottema 2009: Fig. 5).
5. In addition to the Mediterranean climate system, precipitation in these regions is influenced by the Red Sea Trough
Based on the Sea of Galilee pollen record, the Early Bronze III is also characterized by relatively high arboreal percentages (reaching up to 46.2%), indicating the continuity of relatively wet climate conditions. The minor increase in oak pollen documented in the Birkat Ram and Sea of Galilee records signals a slight expansion of the natural Mediterranean forest/maquis (Fig. 1.3a). The Sorek Cave isotopic record also indicates humid climate conditions in the region, with estimated annual rainfall above 520 mm (the mean annual rainfall today; Bar-Matthews and Ayalon 2004 2011), though a slight gradual decline was documented over the course of this period.6 The Olea pollen values retain their low frequencies. Still, in the Sea of Galilee record slightly higher olive pollen percentages were recorded in the early phase of this period (ca. 2900– 2650 BCE) in comparison to the later phase (ca. 2650–2500 BCE), reaching maximum of olive pollen values of 11.3% versus 5.4% (of the total pollen sum) respectively. This means that neither the decline in the number of sites in the central hill country nor the southward expansion of urban settlement in the lowlands were climate-related (Langgut, Adams and Finkelstein 2016: Table 2). New radiocarbon dates and other lines of evidence from the copper mining districts in the Arabah, the Negev Highlands and Arad also reveal activity in the arid regions during the Early Bronze III, contra previous theories (Ben Yosef et al. 2016; Finkelstein et al. 2018). While the wet climate conditions which are reconstructed for this period may have facilitated activity in the arid lands of the Southern Levant, it seems that other factors were more influential, first and foremost the demand for copper in Egypt (Finkelstein et al. 2018).
6. The differences between the pollen and the isotopic records lay only within the fluctuations of the general trends, since both records point to relatively humid climate conditions. The slight discrepancies may derive from differences in sampling resolution (the Sorek Cave record was sampled at a higher resolution), and/or differences within the dating methods (14C versus Uranium-Thorium). They may also be related to the slower response of the vegetation in comparison to the more sensitive isotopic proxy.
At both Birkat Ram and the Sea of Galilee, this period shows no major change in the distribution of the Mediterranean arboreal vegetation (Figs. 1.3a–b). Therefore, it seems that the crisis in the urban system, which started at the end of the Early Bronze Age (ca. 2500 BCE; Regev et al. 2012) and lasted through the entire Intermediate Bronze Age, was not a result of climate change. Yet, two short events pointing to drier conditions were recorded: at ca. 2350 BCE (based on the Sea of Galilee record) and at the end of the Intermediate Bronze/beginning of the Middle Bronze I (Sea of Galilee and Zeʾelim; Fig. 1.3b,d; Langgut et al. 2015). These dry events were also documented by the declining level of the Dead Sea (Kagan et al. 2015). The latter dry event was also identified in the new pollen record from Tel Dan (Kaniewski et al. 2017). The Intermediate Bronze Age features evidence for strong settlement activity in the Negev Highlands (Cohen 1999; Dunseth et al. 2017), which was related to the copper industry in the Arabah Valley south of the Dead Sea (Ben-Yosef et al. 2016; Finkelstein et al. 2018). The humid climate conditions did not stimulate dry farming in the region but could have supported pastoral nomadic activity (Rosen 2017; Dunseth, Finkelstein and Shahack-Gross 2018). The cessation of activity in the region during the middle of the Intermediate Bronze Age, ~2300–2200 BCE, is probably related to diminishing demand for copper as a result of the decline of the Old Kingdom in Egypt (Finkelstein et al., 2018).
7. Since Olea pollen production has a strong response to cessation of cultivation (a dramatic decrease in pollen production was documented in deserted orchards after several decades of abandonment—Langgut, Lev-Yadun and Finkelstein 2014), the olive pollen that was identified during the Intermediate Bronze Age represents well-maintained orchards.
During the Middle Bronze I, olive tree percentages appear at the same magnitude as in the Intermediate Bronze Age in both records (Sea of Galilee and Zeʾelim), representing olive production likely sufficient only for local consumption From the beginning of the period until about 1800 BCE, Mediterranean tree values remain low. It seems that the dry period that began at the end of the Intermediate Bronze Age lasted about two centuries (ca. 2100–1800 BCE). Based on other paleoclimate records already mentioned, this event may have lasted slightly longer—about 300–400 years, between 2200–1800 BCE (Langgut et al. 2015; Laugomer 2017). Can this dry event be associated with the much discussed 4.2 BP event, suggested by Weiss et al. (1993; Weiss 2012; 2017) as the region-wide “mega-drought” that brought about the collapse of the Akkadian Empire (see critique in, e.g., Wanner et al. 2008; Finné et al. 2011)? The answer is that climatic proxies of the Southern Levant cannot be easily projected onto Mesopotamia, at the very least because the latter is influenced by parameters other than just the Mediterranean climate system (Finkelstein and Langgut 2014 and references therein).
Based on the three pollen records available for the Middle Bronze II‒III (Birkat Ram, Sea of Galilee and Zeʾelim; Fig. 1.3), wetter climate conditions, which are recognized by the increasing percentages in Mediterranean trees around 1800 BCE, prevailed throughout this period (Langgut et al. 2015), in comparison to the previous period. During this timespan the Zeʾelim sediments accumulated in a lacustrine environment, representing an increase in Dead Sea levels and more humid conditions (as opposed to the beginning of Middle Bronze I, when sediments accumulated in a shore environment; Langgut et al. 2014; Kagan et al. 2015). Indeed, according to the reconstruction of the Dead Sea levels, it was in the Middle Bronze II–III, that the lake reached its highest level during the last four millennia—up to 370 m bmsl (Migowski et al. 2006; Kushnir and Stein 2010).
According to the northern pollen diagrams (Birkat Ram and Sea of Galilee; Figs. 1.3a, b) during the beginning of the period, the Mediterranean arboreal pollen percentages remain relatively high, representing continuity of a well-developed Mediterranean forest/ maquis. The ʿEin Feshkha record begins in the middle of the Late Bronze Age; it exhibits high values for arboreal pollen, which decline toward the end of the period, reflecting drier climate conditions (Fig. 1.3c). No pollen data for the Late Bronze Age is available from the Zeʾelim record (Fig. 1.3d) due to sedimentary erosion and unfavorable conditions for pollen preservation in sandy sediments (Langgut et al. 2014; 2015).
All four pollen records for the Iron I show an increase in oak, total Mediterranean trees and olive pollen percentages. In the Birkat Ram record, where only two samples fall within the Iron I, a minor increase of Mediterranean tree pollen is visible in the transition from the previous period (Fig. 1.3a). In the Sea of Galilee record, a significant Olea pollen peak is notable; a similar peak appears in the ʿEin Feshkha and Zeʾelim pollen diagrams, together with a rise in oak pollen (Fig. 1.3b,d). As a result of the increase in moisture following the severe dryness at the end of the Late Bronze Age, both the Mediterranean forest/maquis and olive orchards expanded (Langgut et al. 2015). This is also evident in the En-Gedi pollen record starting from ca. 1000 BCE (Litt et al. 2012).
8. Evacuations at Megiddo have revealed olive presses dating to the Late Bronze III and Iron I (Frankel 2006) Yoqneʿam’s “Oil Maker’s House” dates to the late Iron I (Zarzecki-Peleg 2005). In addition, Megiddo’s Iron I strata have produced an exceptionally high percentage of olive charcoal remains (Benzaquen, Finkelstein and Langgut 2019). Olives were also cultivated in the Beth-Shean Valley. The largest percentages of olive charcoal and olive pits were documented at Tel Reḥov from the Iron IIA (Liphschitz 2020). At Tel Beth-Shean the Iron IB features high frequencies of olive charcoal remains (Baruch 2006; Liphschitz 2020: Table 52.4). At Tel Kinrot an olive-oil press was found in Area U, where olive cultivation played a significant role in the economy. In Tel Rekhesh, an extraordinary number of five Iron I oil presses were discovered (Onozuka 2012). It seems, then, that the areas around both Tel Kinrot and Tel Rekhesh included significant territory devoted to olive orchards. The proximity of these two sites to the Sea of Galilee may account for the exceptional percentage of olive pollen recorded in the sediment core extracted from the lake. This is especially true for Tel Kinrot, located relatively close to the place where the Sea of Galilee sediment core was extracted (Finkelstein and Langgut 2018).
During the Iron IIA (ca. 950–780 BCE), pollen from Mediterranean trees maintained relatively high ratios, reflecting a developed Mediterranean forest/maquis and relatively wet climate conditions, while the Iron IIB (ca. 780–680 BCE) and the Iron IIC (ca. 680–586 BCE) were characterized by a slight reduction in Mediterranean trees as evident in all three northern pollen records (Birkat Ram, Sea of Galilee and ʿEin Feshkha; Fig. 1.3a–c). The slight decline in arboreal percentages may represent moderate climate conditions but may also have been the result of anthropogenic activity such as tree clearing, spread of agriculture and grazing activities (Langgut et al. 2014; 2015). Indeed, a surge in human activity that had started in the Iron I, and increased in the Late Iron IIA, reached its zenith during the Iron IIB–IIC. The picture is less clear in the southernmost record (Zeʾelim; Fig. 1.3d): Mediterranean arboreal pollen appears in relatively low values starting from the end of the Iron I and continuing through the entire Iron II; at the same time, the lithology of this sequence has shown that sediments were deposited in a lake environment, and therefore represent relatively high Dead Sea lake levels (ca. 408 m below msl) (Langgut et al. 2014; 2015).
This review presents the role of climate fluctuations in shaping south Levantine human history ca. 3600–600 BCE (the Bronze and Iron Ages) as evidenced in four palynological archives. Three of the four show similar vegetation fluctuations (Sea of Galilee, ʿEin Feshkha and Zeʾelim), indicating that at least during the Bronze and Iron Ages different regions of the Southern Levant were characterized by similar climate patterns. The Birkat Ram record does not point to any dramatic vegetation and climate fluctuations because of its northern location within an area that receives more than 1,000 mm of annual rainfall, which makes it a less sensitive climate recorder.