Millennial to orbital-scale variations of drought intensity in the
Eastern Mediterranean
Mona Stockhecke a, b, c, *
, Axel Timmermann d, **
, Rolf Kipfer e, f, g
, Gerald H. Haug b
,
Ola Kwiecien e, h
, Tobias Friedrich d
, Laurie Menviel i, j
, Thomas Litt k
, Nadine Pickarski k
,
Flavio S. Anselmetti c, l, m
a
Large Lakes Observatory, University of Minnesota Duluth, 10 University Drive 206 RLB, Duluth, MN, 55812-2496, USA
b
Geological Institute, Swiss Federal Institute of Technology (ETH), Sonneggstrasse 5, 8092, Zürich, Switzerland
c
Eawag, Swiss Federal Institute of Aquatic Science and Technology, Department of Surface Waters Research and Management, Überlandstrasse 133, 8600,
Dübendorf, Switzerland
d
IPRC, and Department of Oceanography, SOEST, 1680 East-West Road, University of Hawaii, Honolulu, 96822, HI, USA
e
Eawag, Swiss Federal Institute of Aquatic Science and Technology, Department of Water Resources and Drinking Water, Überlandstrasse 133, 8600,
Dübendorf, Switzerland
f
Institute of Biogeochemistry and Pollutant Dynamics, Swiss Federal Institute of Technology (ETH), Universit€atstrasse 16, 8092, Zürich, Switzerland
g
Institute of Geochemistry and Petrology, Swiss Federal Institute of Technology (ETH), Clausiusstrasse 25, 8092, Zürich, Switzerland
h
Ruhr-University Bochum, Universit€atsstrasse 150, 44801, Bochum, Germany
i
CCRC, University of New South Wales, Matthews Building Level 4, Sydney, NSW, 2052, Australia
j
ARC Centre of Excellence for Climate System Science, Australia
k
Steinmann Institute of Geology, Mineralogy and Paleontology, Bonn University, Nussallee 8, 53115, Bonn, Germany
l
Institute of Geological Sciences, University of Bern, Baltzerstrasse 1þ3, 3012, Bern, Switzerland
m
Oeschger Centre for Climate Change Research (OCCR), University of Bern, Switzerland
a r t i c l e i n f o
Article history:
Received 13 May 2015
Received in revised form
1 December 2015
Accepted 18 December 2015
Available online 2 January 2016
Keywords:
Dansgaard-Oeschger variability
Mediterranean droughts
Milankovitch cycles
Lake Van
ICDP PALEOVAN
a b s t r a c t
Millennial to orbital-scale rainfall changes in the Mediterranean region and corresponding variations in
vegetation patterns were the result of large-scale atmospheric reorganizations. In spite of recent efforts
to reconstruct this variability using a range of proxy archives, the underlying physical mechanisms have
remained elusive. Through the analysis of a new high-resolution sedimentary section from Lake Van
(Turkey) along with climate modeling experiments, we identify massive droughts in the Eastern Mediterranean
for the past four glacial cycles, which have a pervasive link with known intervals of enhanced
North Atlantic glacial iceberg calving, weaker Atlantic Meridional Overturning Circulation and
Dansgaard-Oeschger cold conditions. On orbital timescales, the topographic effect of large Northern
Hemisphere ice sheets and periods with minimum insolation seasonality further exacerbated drought
intensities by suppressing both summer and winter precipitation.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
Rapid climate shifts between North Atlantic cold (stadial) and
warm (interstadial) states occurred frequently during past glacial
periods (Bond et al., 1993; Capron et al., 2010; Deplazes et al., 2013;
Grützner and Higgins, 2010; Menviel et al., 2014). This continuum
of variability, referred to as Dansgaard-Oeschger (DO) variability,
was closely related to the presence or absence of ice-rafted debris
(IRD) layers in the North Atlantic (Hemming, 2004; Hodell et al.,
2008; McManus et al., 1999) or in the Nordic Seas (Bond et al.,
1993; van Kreveld et al., 2000). IRD layers mark periods of
enhanced iceberg calving and freshwater discharge from the major
glacial ice sheets. Various high-resolution paleo-proxy datasets
document that the ratio between the length of DO interstadials and
stadials was strongly modulated by the ice-sheet sizes (Schulz,
2002) and the precession cycle (Timmermann et al., 2010).
One key region that exhibits particularly strong hydroclimate
sensitivity to both DO and orbital-scale variability is the Eastern
* Corresponding author. Large Lakes Observatory, University of Minnesota
Duluth, 10 University Drive 206 RLB, Duluth, MN, 55812-2496, USA.
** Corresponding author.
E-mail addresses: mstockhe@d.umn.edu (M. Stockhecke), axel@hawaii.edu
(A. Timmermann).
Contents lists available at ScienceDirect
Quaternary Science Reviews
journal homepage: www.elsevier.com/locate/quascirev
http://dx.doi.org/10.1016/j.quascirev.2015.12.016
0277-3791/© 2015 Elsevier Ltd. All rights reserved.
Quaternary Science Reviews 133 (2016) 77e95
Mediterranean region (Bartov et al., 2003; Daniau et al., 2007;
Fleitmann et al., 2009; Harrison and Go~ni, 2010; Penaud et al.,
2011; Tzedakis et al., 2006; Van Meerbeeck et al., 2011; among
others). Under present-day conditions, precipitation peaks in
Eastern Turkey in April and in October/November, with a hot and
dry summer season separating the two seasonal precipitation
maxima (Fig. 1A, B). The semi-annual cycle is a characteristic
feature of high-elevation regions. South-easterly wind dominates
in spring likely transporting moisture from the Caspian or Arabian
Sea into the area, while south-westerly winds dominate during fall
and moisture from the Mediterranean Sea precipitates (Wind
vector data from Climate Prediction Centre USA, http://iridl.ldeo.
columbia.edu).
Based on the available meteorological data from Van
(1949e2013), the total annual precipitation amounts to 389 mm and
monthly mean air temperatures varies between 22.1 C and À3.6 C
with an annual mean of 9.1 C. Summer aridity results from largescale
subtropical subsidence caused by the interaction of the midlatitude
Westerlies and the Asian Monsoon (Rodwell and Hoskins,
1996). Anomalously dry (wet) years are typically related to anomalous
high (low) sea level pressure over the North Atlantic (from 45
latitude northward) and the Eastern Mediterranean. These sea level
pressure anomalies promote northerly (southerly) winds over
Turkey (Eshel and Farrell, 2000). This situation shows a direct linkage
between North Atlantic climate variability and hydroclimate
shifts in the Eastern Mediterranean (Fig. 1C).
To further elucidate the drivers of Eastern Mediterranean
hydroclimate variability on orbital and millennial timescales and its
linkage to seasonality changes, we combine new high-resolution
paleo-proxy data obtained from a sediment drill core from the
closed saline Lake Van (Turkey; Fig. 1D) with a suite of transient
earth system model experiments that cover the past 4 glacial
cycles.
2. The Lake Van sedimentary record
The lacustrine record of Lake Van (Fig. 1D) was recovered in
summer 2010 within the framework of the International Continental
Scientific Drilling Program (ICDP) Paleovan project (Litt et al.,
2012). Here we present a multi-proxy analysis of high-resolution
data of the uppermost 150 mcblf (meter composite below lake
floor) of the primary drill site ‘Ahlat Ridge’ (AR, ICDP Site 5034-2,
38.667N, 42.669E, ~360 m water depth, Fig. 1D), which captures
variations in lake level, oxygenation, runoff, sediment transport or
shoreline distance and vegetation.
2.1. Present-day geography and limnology
Endorheic-basin lakes are particularly well suited for highresolution
paleoenvironmental reconstructions as the volume, the
level and the surface area of such lakes are very sensitive to climate
variability (evaporation, precipitation, and river inflow), which also
Fig. 1. Modern climatology of the Mediterranean region and map of Lake Van. (A) Temperature climatology from VAN TURKEY (WMO station code 17170) using monthly mean data
from 1949 to 2013 (from climexp.knmi.nl); (B) Precipitation climatology from VAN TURKEY (WMO station code 17170) using monthly mean data for 1949e2013 (from climexp.knmi.nl);
(C) Correlation between area averaged precipitation minus evaporation anomalies (33e39E, 38e42N) and geopotential height anomalies at 850 hPa using data
from ERA interim reanalysis (Dee et al., 2011). (D) Bathymetric map of Lake Van (1648 m a.s.l.) with the ICDP PALEOVAN drill sites in the Northern Basin (NB, 5034-1, 38.705N,
42.567E) and at Ahlat Ridge (AR, 5034-2, 38.667N, 42.669E), showing major lake basins, inflows and cities. The bathymetry of the lake is based on Seyir Hidrografi ve Os¸ inografi
Dairesi, (1990). Two volcanoes (▵), Nemrut and Süphan, are adjacent to the lake. A third extinct volcano, the _Incekaya hyaloclastite cone, is partly covered by the lake today (Sumita
and Schmincke, 2013). The threshold (TH) at 1737 m a.s.l. prevents water from flowing out to the west. The Bitlis massif rises up to 3500 m a.s.l. Süphan and Mt. Hassanbes¸ ir in the
Kavus¸ s¸ ahap Mountains rise above 3500 m a.s.l. Additionally, major inflowing rivers and cities (*) are shown.
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9578
strongly affects sedimentation processes. Lake Van is situated on a
high plateau in eastern Anatolia, Turkey (Fig. 1). The lake is the
fourth largest terminal lake and the largest soda lake on Earth
(current morphometry: volume: 607 km3
, depth: 451 m depth,
surface area: 3574 km2
). Present lake-level is at an altitude of
1648 m above sea level (a.s.l.). The lake lies within a tectonic
depression with a maximum extension of 130 km ENE-WSW (Fig.1)
and has a relatively small catchment area of about 3.55 times the
present size of the lake (12.522 km2
; Kadioglu et al., 1997). No
outlet, high carbonate concentrations, active regional volcanism of
Nemrut and Süphan volcanoes (2948 m a.s.l. and 4058 m a.s.l.,
respectively) as well as subaquatic hydrothermal exhalations are
responsible for the high alkalinity (pH 9.7, salinity 21.4‰) of the
water (Kaden et al., 2010). The water column undergoes seasonal
stratification with epilimnic temperatures reaching up to 25 C.
Annual water mixing during the cold season affects the uppermost
70 m only and thus deep water temperatures are at ~3.3 C
(Stockhecke et al., 2012). A positive net freshwater balance generates
a halocline that separates the ‘fresh’ surface water from the
saline deep water and thus reduces deep water exchange (Kaden
et al., 2010; Peeters et al., 2000). As mixing is reduced, degrading
organic matter continuously consumes oxygen and the oxiceanoxic
boundary (OAB) rises in the water column (Stockhecke et al., 2012).
The increasing anoxia leads to an enhanced total export and
deposition of organic carbon (TOC). As a consequence of an
enhanced fresh-water input, the lake level of Lake Van rose by 2 m
from 1988 to 1995, deepwater exchange with the surface ceased
and thus the mass of the anoxic deep water body grew (Kaden et al.,
2010; Stockhecke et al., 2012).
2.2. Lithostratigraphy of the drill cores
The stratigraphic backbone including the detailed lithostratigraphy
and the creation of composite section excluding event layers
can be found in Stockhecke et al. (2014b). It was shown that the
recurrence of similar lithological successions follows the glacial/
interglacial cycles since Marine Isotope Stage (MIS) 11. These lithological
successions are reflected by variations in total organic
carbon (TOC) and calcium carbonate content (CaCO3). TOC-rich
varved (annually-laminated) clayey silts correspond to rising lake
levels (in association with a large anoxic deep water body) deposited
during the interglacials, TOC-poor banded and mottled clayey
silts (in association to an oxygenated hypolimnion) are deposited as
consequence of lake-level lowering during the glacials (Stockhecke
et al., 2014b). This pattern was confirmed by geological evidence of
high and low lake levels (terraces and clinoforms, respectively;
Cukur et al., 2014). In order to explore the high frequency
(centennial to millennial-scale) variability captured by the lithology
we distinguish between varved clayey silt, intercalations of varved
clayey silt and frequent turbidites (description and identification of
these graded beds see Stockhecke et al., 2014b) and non-varved
clayey silt. The intervals of varved sediment furthermore imply
pronounced seasonality similar to today (Stockhecke et al., 2012).
The intervals of accumulated turbidites are highlighted, however all
event deposits were eliminated from the timeseries by the use of
the event-corrected depth scale (mcblf-nE, ‘meter composite below
lake floor no event’; Stockhecke et al., 2014b). Varved (e.g. varved
and varved þ turbidites) and non-varved periods are used for statistical
comparison with seasonal precipitation pattern from model
simulations and the precession index.
2.3. Sediment-color data and scanning X-ray fluorescence (XRF)
The surface of the split sediment cores was photographed with a
high-resolution line-scan camera installed in Avaatech XRF core
scanner at MARUM, University of Bremen. The sediment total
reflectance (in a spherical L*A*B* color space with B* denotes a
blue-yellow chromaticity) was measured continuously in transects
from the high-resolution images with a cross-core length of 2 cm
and down-core resolution of 0.03 mm (Fig. S1A). The center of the
sediment surface was used and the down-core transects were only
shifted to one of the sites to avoid sediment disturbances. The B*
reflectance was chosen as it holds highest resolution, highest correlation
to the measured TOC contents (correlation coefficient
r ¼ 0.6, n ¼ 663) and mimics the NGRIP isotopic stratotype
(Andersen et al., 2004). This new high-resolution proxy record is
presented as normalized B* reflectance data (Fig. S1A).
The XRF scanner was used to measure the bulk intensities of
major elements (e. g., Si, K, Ca) on split core surfaces (Jansen et al.,
1998). XRF data were collected every 2 cm down-core over a 1 cm2
area with a cross-core slit size of 12 mm using generator settings of
10 kV, a current of 0.2 mA and a sampling time of 10 s. The XRF data
was described in Kwiecien et al. (2014), confirming the lithological
interpretation of rising lake levels and increasing rainfall during the
last four terminations. In this study we combined statistically
normalized elemental records of K, Ca, Si on the refined age model
(Fig. S1CeE) with the other proxy data B*, TOC and CaCO3.
2.4. Geochemical and palynological analysis of discrete samples
Discrete samples were taken at a spacing of 2.5 cm over the AR
composite record (a total of 1800 samples, Stockhecke et al., 2014b).
The freeze-dried and ground sediment samples were analyzed for
total carbon (TC; HEKAtech Euro Elemental Analyzer), total inorganic
carbon (TIC; Coulometric Inc., 5011 CO2-Coulometer) and
total organic carbon (TOC ¼ TC À TIC). Duplicates of 112 samples
yielded standard errors of ±11% for TN, ±3% for TC and ±5% for TIC.
TIC weight % of total sediment (wt %) was converted to carbonate
wt % by multiplying a stoichiometric factor (8.33) under the
assumption that all inorganic carbon is bound as calcium carbonate
(CaCO3). All wt % data are abbreviated to %. Comparison of the
CaCO3 content with the XRF Ca intensity (Fig. S1C) confirm that the
lower-resolution quantitative measurements are replicable by the
higher-resolution relative changes in XRF Ca intensity as well as by
the inverse changes of Si and K reflecting the siliciclastic minerals.
Pollen assemblages of 320 samples were analyzed at 20 cm and
1 m (Litt et al., 2014) sampling resolution for intervals 0e48 and
48e111.5 mblf-nE, respectively. Pollen preparation is described in
detail in Litt et al. (2014). For pollen identification the reference
collection of the Steinmann Institute (Palaeontology, Bonn) as well
as descriptions of the Mediterranean palynoflora were used
(Chester and Rains, 2001; Reille, 1990, 1992, 1995, 1998; Van Zeist
and Woldring, 1978). The minimum number of terrestrial pollen
grains counted in each sample is at least 500. Here, the percentages
of deciduous Quercus pollen are reported.
2.5. Age model
The age model of the complete sedimentary sequence of Lake
Van recovered within the ICDP Paleovan project was presented by
Stockhecke et al., 2014a. Various acknowledged dating techniques,
such as: absolute and radiometric dating (varve chronology,
radiocarbon dating, and argoneargon single-crystal dating) and
relative dating (tephrostratigraphy, magnetostratigraphy, 10
Be
measurements and climatostratigraphic alignment) techniques
were used to construct a robust and precise chronology of the
600,000 year-old Lake Van record (Stockhecke et al., 2014a, Fig. 2).
We took advantage of that chronology (gray lines in Fig. 2, Fig. S2)
and used it as an initial condition for further refinements by adding
three modifications: (i) a slightly refined correlation between Lake
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 79
Van B* (sediment ) and the d18
O NGRIP record for the time interval
(from 0 to 90 kyr) using the GICC05 (Andersen et al., 2004;
Rasmussen et al., 2006; Svensson et al., 2008) and GICC05modeltext
(Wolff et al., 2010) timescales, (ii) integration of the recently
published new AICC2012 timescale (Antarctic Ice Core Chronology
2012; Veres et al., 2013) for the time interval from 90 to 125 ka to
benefit from the improved dating of Dansgaard-Oeschger (DO)
events 23e25 and (iii) correlation of the millennial-scale variability
depicted by the high-resolution B* record to the synthetic
Greenland record on the Speleo age scale (Barker et al., 2011; Cheng
et al., 2009) for the interval beyond 125 ka (Fig. S2).
Our rationale for using different timescales is that the Greenland
ice core timescale (for the last 60 ka based on absolute ages by layer
counting) and the radiometrically dated speleothem-based timescale
are independent of the marine LR04 oxygen isotopic stack
(Lisiecki and Raymo, 2005) and orbitally-tuned Antarctica ice-core
EPICA Dome C (EDC, Jouzel et al., 2007) timescales, which enables
us to tie the Lake Van record to a global millennial-scale resolved
event stratigraphy. The age model details over the full length of the
PALEOVAN record including further references are provided in
Table S1. Comparison with other timescales shows differences of
less than 1e2 kyr (Table S1).
The refined age model consists of 91 age control points
(Table S1), which constitute one tie point set to mark the sedimentewater
interface, five ages from varve chronology and three
calibrated 14C ages, 78 tuning-based age control points, and four tie
points to set the two gaps and bottom of the record by extrapolation
of linear sedimentation rates. The age control points were
interpolated using a spline interpolation to avoid abrupt gradient
change at each data point, as it would be the case using a linear
interpolation.
The comparison of the refined and the original age model reveals
that the largest age difference (<10 kyr) occurs during MIS6
(Fig. 2, Fig. S2). This is not surprising as this period was previously
Fig. 2. Chronology of the Lake Van record. (A) Age-depth plot with refined age control points (red), the original chronology (Stockhecke et al., 2014a) and further age constraints
from absolute dating and relative ages from other time markers confirming the age model; (B) Sedimentation rates (SR). Gray shaded area represent the interglacials following the
Marine Isotope Stage (MIS) boundaries (Lisiecki and Raymo, 2005). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this
article.)
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9580
linearly interpolated over 65.5 ka without setting an age control.
Additional to the errors, each given in the references of respective
timescale (Table S1), we acknowledge that the tuning itself may
contain biases that reach up to ~200 years. However, new independent
verification of the age model is obtained on orbital time
scales through model data correspondence (see below), as well as
other relative (magnetostratigraphy, tephrostratigraphy) and absolute
(varves counting) and radiometric (argoneargon singlecrystal
dating) dating methods (Stockhecke et al., 2014a, Fig. 2).
Given, these independent constraints, we are confident that the
established age model is adequate to study millennial-scale variability
and its relation to other paleo-climate records for at least the
past 360 ka.
2.6. Statistical analysis of timeseries
The normalized timeseries of the proxy records (Fig. S1) were
resampled at 10-year resolution using linear interpolation for
further analysis of the joint variability of these records. To extract
the common dominant features of multiproxy/temporal variability
we conducted an Empirical Orthogonal Function (EOF) analysis of
the following variables: B*, Si, K, TOC and Ca. The individual EOF
contributions to the overall multiproxy hydroclimate variability can
be obtained by multiplying the time-varying principal components
(Fig. S3AeC) with their respective EOF pattern values (Fig. S3DeF).
For additional analysis the leading principal component (PC1)
was filtered to extract the i) millennial-scale, ii) precessional and iii)
lower frequency orbital components. High and low-pass filter with
cutoff periods of i) <7 kyr (PC1high), ii) >7 kyr (PC17kyr-low), and iii)
>35 kyr (PC135kyr-low), were applied.
The high-frequency millennial-scale PC1 (PC1high) signal was
further subjected to a running variance calculation utilizing a
window length of 3 kyr. A bivariate Gaussian kernel density estimator
was applied to the combined reconstructed global ice volume
timeseries (in meters sea level equivalent, msle; Waelbroeck
et al., 2002) and the DO variance (running variance of PC1high) to
statistically study their relationship.
For statistical pattern analysis and multiple regressions between
reconstruction and model data we also applied two band pass filters
to PC1 focusing on the obliquity/eccentricity and precession
scales using cut-offs at 35e200 and 10e28 kyr, respectively.
3. Climate model experiments
To explore the climate response to changing boundary conditions
on millennial and orbital timescales and to compare the Lake
Van climate reconstructions with numerical simulations, two
transient paleo-climate model experiments were conducted with
the LOVECLIM earth system model (Goosse et al., 2010).
3.1. LOVECLIM earth system model
LOVECLIM is a coupled atmosphere-ocean-sea-ice-vegetation
model. The atmospheric component ECBilt (Opsteegh et al., 1998)
of the coupled model LOVECLIM is a spectral three-level model,
based on quasi-geostrophic equations extended by estimates of
ageostrophic terms. The model is run in T21 horizontal resolution
(~5.6 x 5.6) and contains a full hydrological cycle which is closed
over land by a bucket model for soil moisture and a runoff scheme.
The ocean-sea ice component CLIO (Goosse and Fichefet, 1999) is
based on a free-surface Ocean General Circulation Model with
3

 3

horizontal resolution which is coupled to a thermodynamicdynamic
sea ice model. The terrestrial vegetation model “VEgetation
COntinuous DEscription” (VECODE; Brovkin et al., 1997) consists
of two plant functional types and non-vegetated deserts
zones. Each grid cell assumes a partial coverage by these three land
cover types depending on the annual mean temperature and
rainfall amount and variability. Vegetation changes feed back to the
atmosphere only through corresponding changes in albedo. The
effect of evapotranspiration on the atmosphere is not explicitly
simulated.
The first LOVECLIM experiment is a global model hindcast from
50 to 11 a (details of the 50e30 ka and 18e11 ka simulations were
presented in Menviel et al., 2014 and Menviel et al., 2011 respectively).
The second experiment was described in Timmermann et al.
(2014) and covers the climate history of the past 408 kyr.
3.2. 50e11 ka global hindcast simulation
Initial conditions for the transient 50e11 ka hindcast simulation
were obtained by conducting an equilibrium spin-up simulation
using an atmospheric CO2 content of 207.5 ppmv, orbital forcing for
the time 50 ka B.P. and an estimate of the 50 ka B.P. ice sheet orography
and albedo which were obtained from a 130 ka off-line ice
sheet model simulation (Abe-Ouchi et al., 2007). In the subsequent
un-accelerated transient run, greenhouse gases, orbital and ice sheet
forcing were updated continuously following the methodology of
Timm and Timmermann, 2007 and Timm et al., 2008. Note that our
coupled model version used here does not include an interactive ice
sheet. Therefore, freshwater release into the ocean as a result of ice
sheet calving and ablation are not explicitly captured. To mimic the
time-evolution of these terms and their effect on the ocean circulation,
we apply an anomalous North Atlantic freshwater forcing to
the North Atlantic region 55

We10

W, 50

Ne65

N (Fig. S4B).
Negative forcing anomalies can be interpreted as periods of reduced
precipitation, calving and runoff into the anomalously saline Arctic
Ocean and into Nordic Seas and the absence of Bering Strait
Throughflow during the glacial. Positive freshwater anomalies
represent times of negative net mass balance of the Northern
Hemisphere ice sheet, associated for instance with massive calving
events, subsurface melting at the ice-sheet/ice-shelf/ocean interface
or surface ablation. The freshwater forcing (FWF) time series is obtained
through an iterative procedure (Menviel et al., 2014), that is
designed such that the simulated temperature anomalies in the
eastern subtropical North Atlantic best match a target alkenonebased
sea surface temperature (SST) reconstruction from the Iberian
margin core MD01-2443 (Fig. S4D; Martrat et al., 2007). The simulated
Iberian margin SST (15

We8

W, 37

Ne43

N) and Atlantic
Meridional Ocean Circulation (AMOC) strength (maximum of
meridional streamfunction in the North Atlantic) are depicted in
Fig. S4D, E.
The model index regions for the paleodata/model comparison
are chosen as 41

Ee44

E, 37

Ne40

N for Turkey rainfall,
15

We8

W, 37

Ne43

N for Iberian margin SST and 38.5

Ne44.2

N,
19.7

Ee25.2

E for the simulated forest fraction changes in Greece. It
should be noted that for model/data comparison the model data
were projected onto the GICC05 timescale (see Menviel et al., 2014
for details).
3.3. Transient simulation over the past 408 ka
To study the time-evolving aspects of orbital and greenhouse
gas-driven climate changes in the Mediterranean region over the
past 408 ka, we also conducted and analyzed a transient LOVECLIM
model experiment (Timmermann et al., 2014), which uses timedependent
boundary conditions with an acceleration factor of 5
for orbital parameters (Berger, 1978), CO2 concentrations, northern
hemispheric ice-sheet orography (Ganopolski and Calov, 2011) and
albedo over ice sheets, i.e. one coupled model year corresponds to
five orbital calendar years. No additional freshwater forcing was
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 81
used to mimic millennial-scale DO variability. The model simulation
is compared in detail with the Lake Van hydroclimate data.
Furthermore, the simulated precipitation and evaporation data are
used to force a simplified lake model for different lake configurations
in Turkey and the Levant, mimicking roughly the conditions
for Lake Van and Lake Lisan.
3.4. Regression analysis between reconstructed PC1 and model
simulations
The normalized hydroclimate PC1 was multiplied with À1
([PC1*¼(À1) (PC1À<PC1>)/(<(PC1À<PC1>)2
>)0.5
], where <..>
represents the long-term mean value) to determine the typical
atmospheric patterns corresponding to the reconstructed drought
conditions in the Eastern Mediterranean on millennial, eccentricity
and precessional timescales. Positive values of PC1* correspond to
drought conditions. For millennial timescales, we just focus on the
period 50e30 ka and use an unfiltered reconstructed PC1*. For the
eccentricity (precession) timescales, we used the 35e200 kyr bandpass
filter to PC1*, respective 10e28 kyr for precession scales. PC1*
for the respective timescales is then regressed onto the LOVECLIM
simulated anomalies of 800 hPa geopotential height anomalies and
relative rainfall changes from the 408 ka transient experiment (for
eccentricity and precession timescales; Timmermann et al., 2014)
and the DO hindcast simulation (for millennial timescales; Menviel
et al., 2014).
4. Lake-level simulations
We develop a simplified lake level model for two different regions
(Turkey and Levant) that is driven by local precipitation (Pm),
and evaporation (Em) output from the 408 ka transient LOVECLIM
simulation. The model index regions for the lake level models are
chosen as 41

Ee44

E, 37

Ne40

N for simulation of Lake Van levels
and 30

Ne33

N for simulation of Lake Lisan levels and we focus on
the last 85 ka and orbital timescales only. For the Lake Lisan model,
the LOVECLIM model timeseries for precipitation (Pm) and evaporation
(Em) were bias corrected with
E Ã ðtÞ ¼ ½EmðtÞ À Emðt ¼ 0 kaÞ
Ã
þ Eobs (1)
PðtÞà ¼ b½PmðtÞ À Pmðt ¼ 0 kaފ þ Pobs (2)
such as to match approximately the modern mean Dead Sea
evaporation rates Eobs ¼ 110 cm/year and the mean precipitation
over the catchment area of Pobs ¼ 10 cm/yr, similar to the values
used in ref. Rohling (2013). Taking into account the effective ratio
between catchment area and lake surface area (g), we further
assume that the long-term mean freshwater balance is zero.
〈g$P Ã À E Ã 〉¼ 0 (3)
where 〈g$P Ã À E Ã 〉 denotes the mean from 0 to À85 ka. This
balance Equation (3) together with Equations (1) and (2) allow us to
determine the precipitation scaling factor b in Equation (2). To
Fig. 3. Schematic of lake dynamics during the transitions from Dansgaard-Oeschger events 11 to 10 (Stockhecke, 2013). (A) NGRIP d18
O from 43.5 to 46.5 kyr GICC05 (Andersen
et al., 2004; Wolff et al., 2010); (B) High-resolution proxy records (B*, TOC, CaCO3, XRF Ca, XRF K and XRF Si intensities) and deciduous Quercus pollen percentages on composite
depth (21.6e22.75 mcblf); (C) High-resolution line scan image of corresponding sediment surface; (D) Reconstructed lake level variations and changing thickness of the anoxic
deep water body.
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9582
obtain estimates for the evolution of the anomalous Lake Lisan level
we integrate the freshwater balance in time L(t) ¼ !t¼85ka dt [g
P*(t) e E*(t)]. Representing the fact that a considerable fraction of
rainfall in the catchment area of Lake Lisan, which was about 40
times larger (Rohling, 2013) than the current size of the Dead sea, is
lost through evaporation, we choose an idealized, yet representative,
value of g ¼ 9.8. A similar approach is pursued for Lake Van,
using g ¼ 3.55 (see Section 2.1) and Pobs ¼ 37 cm (Fig. 1) and
Eobs ¼ 110 cm/yr, similar to the values used in Lemcke (1996). The
idealized lake model experiments discussed here should be
considered possible scenarios, which are based on reasonable assumptions,
not as an exact reproduction of the real lake level trajectory
over the last glacial cycle.
5. Results and discussion
5.1. Hydroclimate timeseries from Lake Van
Extending back to 600 ka (of the sedimentStockhecke et al.,
2014b), the partly-varved sedimentary section of Lake Van allows
for an unprecedented view into the millennial to orbital-scale
modulation of subtropical hydroclimate changes and its relation to
North Atlantic climate change and external forcing. Rising conditions
of Lake Van water levels, associated for instance with
increased precipitation (or reduced evaporation), typically lead to
both a reduction of surface salinity and deep lake mixing, which
resulted in an expansion of the volume of the anoxic deep water
body (Kaden et al., 2010). Such conditions caused the enhancement
of net export, deposition and preservation of organic carbon
(Stockhecke et al., 2014b). These hydroclimate changes are well
documented by shifts in sediment colorof the sediment(B*;
Fig. 3B). Furthermore, pluvials can lead to an increase in the
terrestrial sediment transport distance between the coring site
and the lake shore-line as well as to an increase in calcium ion
supply. Both of these factors resulted in a reduction of the siliciclastic
minerals and in an increase of autochthonous carbonate
and subsequent relative shifts in the composition of the sediments.
Thus, complementary information on hydroclimate
changes in the Lake Van region can be derived from other sedimentary
data such as total organic carbon (TOC), calcium carbonate
(CaCO3, Ca), silica (Si) from quartz, feldspar, clay minerals
and potassium (K) from latter two (Fig. S1C to E).
Fig. 4. Paleoreconstruction and model comparison for Dansgaard-Oeschger events from 0 to 80 kyr. (A) NGRIP d18
O (Svensson et al., 2008); (B) Simulated maximum of the North
Atlantic meridional streamfunction [1 Sv ¼ 106
m3
/s] representing the Atlantic Meridional Overturning Circulation (AMOC) in LOVECLIM (Menviel et al., 2014); (C) Lake Van B*
reflectance color record; (D) Lake Van PC1 hydroclimate record; (E) deciduous Quercus pollen percentage (on logarithmic ordinate) from Lake Van; (F) Arboreal pollen from Tenaghi
Philippon, Greece (Muller et al., 2011) synchronized to the NGRIP; (G) Simulated LOVECLIM precipitation changes over Turkey. Black diamonds depict the age control points. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 83
The high-resolution B* sediment color record (Figs. 3B and 4C) is
correlated with Greenland oxygen-isotope variations (Andersen
et al., 2004, Figs. 3A and 4A) and North Atlantic temperature
anomalies (Fig. 5B) of DO variability during MIS 5-2. It captures
millennial-scale variability during glacial periods for the past
360 kyr.
Applying an Empirical Orthogonal Function (EOF) analysis to the
timeseries of B*, Si, K, TOC and Ca (all sensitive to hydroclimate
variations in the Lake Van region) helps us to extract the common
features of multiproxy/temporal hydroclimate variability (see Section
2.6).
The explained variances for each EOF mode attain values of 56%
(EOF1), 28% (EOF2) and 9% (EOF3). Positive values of the leading
principal component (PC1) are associated with high B* (anoxic
conditions, intensified stratification), anomalously low Si and K
(increased shore-line distance, increased runoff and forested
vegetation in watershed), positive TOC contents (anoxic conditions,
intensified stratification and productivity) and positive CaCO3
anomalies (increased supply via runoff of Ca2þ
ion into lake). PC1
(Figs. 4D and 5D) captures both orbital-scale variability and
millennial-scale flickering of hydroclimate conditions in the
Eastern Mediterranean associated with DO variability.
5.2. Millennial-scale climate variability
5.2.1. Comparison of Lake Van hydroclimate reconstruction with DO
hindcast simulation
To explore the underlying dynamics of DO-related Eastern
Mediterranean hydroclimate variability from 50 to 11 ka (Fig. 4), we
compare the reconstructed changes in B* and PC1 with rainfall
changes simulated by the transient DO LOVECLIM hindcast experiment
(Menviel et al., 2014) described in Section 3.2. In response to
the applied anomalous North Atlantic freshwater forcing (Fig. S4A),
the model simulates a weaker Atlantic Meridional Overturning
Circulation (AMOC) during DO and Heinrich stadials (Fig. 4B),
cooling of the North Atlantic in accordance with Iberian margin SST
reconstructions (Martrat et al., 2007; Fig. S4C, D) and extended
droughts in the Eastern Mediterranean (Fig. 4G), in agreement with
the Lake Van B* and PC1 records and deciduous Quercus pollen
percentages (Fig. 4C to E). This close qualitative correspondence
between AMOC changes and Eastern Mediterranean droughts is
further supported by arboreal pollen percentages from Tenaghi
Philippon, Greece (Tzedakis et al., 2006, Fig. 4F), which show
massive retreat of temperate forest during stadial periods and an
increase to about 40e60% during interstadials.
Fig. 5. North Atlantic and Mediterranean climate variability from 0 to 360 ka. (A) Percentage of ice-rafted debris (IRD on Speleo timescale, in gray) from core North Atlantic core
ODP980 (McManus et al., 1999) on logarithmic ordinate and reconstructed sea level data (on its original timescale, in yellow; Waelbroeck et al., 2002); (B) d18
O of planktic
foraminifera from Iberian Margin core (on Speleo timescale; Hodell et al., 2013); (C) Lake Van B* reflectance color record; (D) Lake Van PC1 hydroclimate record; (E) deciduous
Quercus pollen percentage from Lake Van (on logarithmic ordinate) from Lake Van; (F) Simulated annual mean precipitation anomalies (dark blue; averaged over 41e44

E, 37e40

N)
from a transient earth system model simulation conducted with LOVECLIM and 500-yr running mean (light blue; Timmermann et al., 2014). Black diamonds depict the age control
points. Shaded area represents the gap (gray), the glacials (blue) and the terminations (yellow) following the Marine Isotope Stage (MIS) boundaries (Lisiecki and Raymo, 2005).
Interglacial substages follow the nomenclature of Jouzel et al., 2007. Dansgaard-Oeschger events of the last glacial (A1-25, see Table S1 for references), of MIS 6 (B16-1) and of MIS 8
(C4eC1) are enumerated continuously. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9584
5.2.2. Modulation of DO variability by ice volume
DO variability in the high-resolution Lake Van hydroclimate
records can be traced back far beyond the last glacial period. The DO
variance timeseries from PC1high shows several episodes of DO
bursting within glacial periods (Fig. 6C), such as from 140 to 160 ka,
184e205 ka, 220e237 ka, 250e263 ka and 307e321 ka in alignment
with periods of high IRD in the North Atlantic (Fig. 6A) representing
unstable ice-sheets. The comparison between the
variance changes of the high-pass PC1 and of a synthetic record of
millennial-scale bipolar seesaw activity (Barker et al., 2011; Fig. S2)
reveals a very good agreement until 220 ka, but insignificant correlations
prior to that time. We conclude that at least until MIS7, the
Lake Van hydroclimate record accurately tracks the timing of icesheet
induced changes in the strength of the AMOC and the continuum
of DO and Heinrich variability in the North Atlantic. More
high-resolution and well-dated records of DO variability from MIS8
and beyond are needed to further establish the exact timing of the
DO bursting periods and their individual DO events.
Quantification of the relationship between reconstructed global
ice volume (in meters sea level equivalent, msle; Waelbroeck et al.,
2002) and DO variance (normalized high-frequency PC1) using a
bivariate Gaussian kernel density estimator documents that neither
interglacials (Fig. 6D, region A), nor deep glacial periods (Fig. 6D,
region C) support high levels of millennial-scale variability. Within
the window corresponding to ~20e100 m sea-level drops (Fig. 6D,
region B), higher millennial-scale DO variance resulted from intermittent
ice-sheet calving and associated AMOC fluctuations. If icesheets
were small, they took a long time to reach the ocean and
build up instabilities that were large enough to generate freshwater
and iceberg surges, which affected the AMOC and hydroclimate in
the remote Eastern Mediterranean. Thus, during MIS5, interstadials
were very long relative to the stadial periods. For very large icesheets,
negative mass balance terms were much larger and background
calving and ablation caused lower North Atlantic sea surface
salinities, which lead to a longterm moderate weakening of the
AMOC and enhanced drought in the Eastern Mediterranean. In this
scenario, MIS2 can be viewed as a near-continuous DO stadial, which
was further amplified by additional freshwater discharge during
Heinrich event 2 around 24 ka and additional orbital-scale effects.
5.3. Orbital-scale climate variability and seasonality
The Lake Van PC1 hydroclimate record and deciduous Quercus
percentages (Fig. 5D, E) exhibit pronounced variability on orbital
timescales i.e., on precession, obliquity and eccentricity timescales
(Fig. 7). To further elucidate the underlying mechanism for the
orbital modulation of drought intensities in the Eastern Mediterranean
we compare the Lake Van record with the transient LOVECLIM
earth system model experiment that covers the past 408 ka
(Timmermann et al., 2014) and simulates the global climate
Fig. 6. Orbital-scale modulation of Dansgaard-Oeschger variance. (A) Si/Sr XRF data (black line) from North Atlantic site U1308 (Hodell et al., 2008) on logarithmic ordinates, as a
proxy for silicate-rich ice-rafted debris and sea level reconstruction (Waelbroeck et al., 2002; yellow); (B) High-frequency contribution of Lake Van hydroclimate PC1 (PC1high); (C)
Normalized running standard variance of PC1high. (D) Normalized millennial variance of PC1high compared with the global ice volume (in meters sea level equivalent, msle) in a
two-dimensional histogram. Region A reflects the interglacial periods of low ice volume and no calving events, region B reflects the periods of intermediate ice volume and
intermittent calving events, region C marks the periods of high global ice volume during the glacials associated with continuous calving. (For interpretation of the references to color
in this figure legend, the reader is referred to the web version of this article.)
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 85
response to orbital, ice sheet and greenhouse gas forcing (see Section
3.3). The simulated mean annual rainfall changes over Turkey agree
qualitatively well with orbital-scale variability of PC1 (Fig. 5D, F).
To better illustrate the effect of orbital forcings on seasonality
changes in precipitation, we calculated the percentage of monthly
rainfall for each year relative to the annual mean for this year
(Fig. 8E). Under various climate conditions the LOVECLIM model
simulates a semi-annual cycle in rainfall with peak rainfall in May
and October (Fig. 8F), in good agreement with present-day observations
(Fig. 1). Compared to the observations, however, the
simulated late fall/early winter peak is somewhat underestimated.
The Hovmoller diagram in Fig. 8D documents orbitalscale
shifts in the seasonality of the simulated rainfall relative to
the annual mean for each year. The largest orbital-scale changes in
seasonal rainfall occur in boreal summer, when drought conditions
alternate on precessional and eccentricity timescales with
wet conditions (15% of the annual precipitation falling in June).
These variations are also reflected in Fig. 7, which shows a near
quadrupling of simulated summer precipitation rates between
extreme orbital/ice-volume conditions. Furthermore, we used
sections of the Lake Van core, where laminations are present and
respectively absent and calculated a composite seasonal cycle for
these times of rainfall using the LOVECLIM climate model data and
a composite normalized frequency distribution of the precession
index (Fig. 8F). We can clearly see that periods of lamination in the
Lake Van record corresponds an enhanced seasonal cycle of
simulated rainfall and a stronger precession index (boreal summer
in perihelion).
Under current conditions a hot and dry summer season separates
the two seasonal precipitation maxima in April and in
October/November (Fig. 1A, B). According to the transient model
simulation (Fig. 8D, E) and previous studies (Kutzbach et al., 2014),
periods with maximum seasonality in boreal insolation (positive
precession index) caused an increase of both spring to summer
rainfall (Fig. 8E) and storm track-driven winter precipitation
(Fig. 7C). This process in turn affected the amplitude of precipitation
seasonality, the mean annual rainfall and runoff into Lake Van,
as documented by the correspondence between periods of varved
Fig. 7. Lake Van hydroclimate orbital variability compared with model precipitation, external forcing and sea level changes. (A) 7 kyr low-pass filtered PC1 (PC17kyr-low); (B)
Simulated summer (JJA) precipitation (on model timescale); (C) Simulated winter (DJF) precipitation (on model timescale); (D) Precession index; (E) 35 kyr low-pass filtered PC1
(PC135kyr-low) and (F) sea level (Waelbroeck et al., 2002). Summer or winter precipitation changes could have caused changes in the precession cycle which affected the PC1. Gray
shaded area represent the interglacials following the Marine Isotope Stage (MIS) boundaries (Lisiecki and Raymo, 2005).
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9586
sediment (Fig. 8B), enhanced seasonal forcing and increased
simulated rainfall seasonality (Fig. 8F).
5.4. Large-scale hydroclimate patterns
Multiple regressions between the Lake Van hydroclimate PC1
and the LOVECLIM model simulation were conducted to determine
the typical atmospheric patterns corresponding to the
reconstructed drought conditions in the Eastern Mediterranean
on millennial, eccentricity and precessional timescales. This
analysis searches for the large-scale simulated circulation patterns
that correlate closely with reconstructed hydroclimate
changes in Lake Van. The results are shown in Fig. 9 and have been
schematized for Fig. 10. We find that on all timescales, droughts in
the Eastern Mediterranean are a large-scale phenomenon,
associated with stronger mean annual anticyclonic circulation and
increased atmospheric subsidence (not shown here) in the
Eastern Mediterranean and typically cold surface conditions in the
North Atlantic. A reconstruction of the associated atmospheric
circulation pattern reveals that precessional-scale hydroclimate
anomalies in Lake Van PC1 were part of a large-scale hydroclimatic
pattern extending across the Eastern Mediterranean,
North Africa and the Middle East (Figs. 9C and 10D depicting
drought conditions).
A significant portion of eccentricity-scale (100e120 kyr)
variability can also be seen in PC1 and the simulated rainfall
changes (Fig. 8). Interglacials in Turkey were relative wet (positive
PC1 loading) and glacials were relatively dry (negative PC1
values), in accordance with lake-level reconstructions (Cukur
et al., 2014; Kuzucuoglu et al., 2010; Stockhecke et al., 2014b)
Fig. 8. Ice sheet and orbital effects on hydroclimate in Turkey from 0 to 360 ka. (A) Low-pass filtered PC1 (PC17kyr-low); (B) Periods of lamination in Lake Van (red: varved, pink:
varved and turbidites, white: non-varved); (C) Reconstructed global sea level changes (on its original timescale; Waelbroeck et al., 2002); (D) Hovmoller diagram of monthly
insolation changes at Lake Van latitude; (E) Hovmoller diagram of simulated monthly precipitation changes (averaged over 41e44

E, 37e40

N) over Turkey relative to the annual
mean of any given year. Gray shaded area represent the interglacials following the Marine Isotope Stage (MIS) boundaries (Lisiecki and Raymo, 2005); (F) Normalized frequency
distribution of simulated rainfall in Lake Van region for varved intervals (red, left) and for non-varved intervals (blue, right) according to the lithostratigraphy of the record
(Stockhecke et al., 2014b) with inlets of normalized frequency distribution for precession index (PI). (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 87
Fig. 9. Relationship between reconstructed droughts and simulated climate data on millennial, eccentricity and precession timescales. (A) Model/paleodata regressions between
simulated 800 hPa geopotential height changes (contour, m2
/s2
), relative precipitation changes (%) and negative normalized Lake Van PC1 (Fig. 1) from 50 to 30 ka, capturing the key
pattern of DO stadials; (B) same as (A), but using reconstructed negative normalized band-pass filtered (35e200 kyr) PC1 from Lake Van (Fig. 7), capturing the ice sheet effects; (C)
same as (A), but using reconstructed negative normalized band-pass filtered (10e28 kyr) PC1 from Lake Van, corresponding to negative precession index values and minimum
insolation seasonality (Fig. 7).
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9588
and the deciduous Quercus pollen percentages (Litt et al., 2014)
from Lake Van, which show advances and retreats of the oak
steppe-forest, respectively (Fig. 5E). This variability results from
the ice-sheet orographic forcing (Fig. 7E, F), the associated
anomalous atmospheric subsidence (Timm et al., 2010) and shifts
in atmospheric pressure patterns (Figs. 9B and 10C). The icesheet
effects from both the Laurentide and the Eurasian ice
sheets caused massive drying over the Mediterranean and
Northern Africa (Fig. 10C).
5.5. Glacial droughts and lake levels in the Levant region
The evidence presented here (Figs. 4, 5 and 8) for reduced glacial
rainfall in eastern Turkey is consistent with the larger-scale pattern
of dry glacial and humid interglacial conditions reconstructed from
lakes, speleothems and peat bogs in the Mediterranean and
northern Levant (Lago Grande di Monticchio (Allen et al., 2000);
Lake Ohrid (Vogel et al., 2010); Ioannina (Tzedakis, 1994); Tenaghi
Philippon (Fletcher et al., 2013; Tzedakis et al., 2006); Sofular Cave
(Fleitmann et al., 2009); Karaca Cave (Rowe et al., 2012); Lake
Urmia (Stevens et al., 2012); Lake Yammoûneh (Gasse et al., 2015
and references therein; Fig. 11)). Furthermore our data lend support
to previous interpretations of speleothem data from the central
Levant region (Bar-Matthews et al., 2003). Speleothems from
the southern Negev deserts (Vaks et al., 2010) and dated lake deposits
from the Nafud desert (Rosenberg et al., 2013) indicate
interglacial pluvial periods synchronous with sapropel layers in the
Mediterranean (Zhao et al., 2011). These findings have been linked
to an intensification of the Atlantic-Mediterranean cyclone activity
and an enhancement of monsoon-like rain.
However, the evidence for large-scale glacial drying in the
Eastern Mediterranean and Levant region, appears to be at odds
with the evidence for high (low) glacial (interglacial) lake stands of
lacustrine bodies which evolved within Dead Sea basin (DSB;
central Levant, 400 m b.s.l.), such as the mid-Pleistocene Lake
Amora, the late Pleistocene Lake Samra, the last glacial Lake Lisan
Fig. 10. Schematic pattern of sea-level pressure, wind (cyan/orange contours), precipitation (green/brown shading) anomalies and ocean currents (red-blue arrow) for Eastern
Mediterranean droughts occurring for present day conditions (A), on millennial (B; DO stadials), eccentricity (C; glacial periods) and precessional timescales (D, negative precession
index). The schematic is based on the analysis in Figs. 1 (for A) and 9 (for B to D). Yellow triangle marks Lake Van. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of this article.)
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 89
(~70e14 kyr), and the Holocene Dead Sea (Torfstein et al., 2015; and
references therein) as well as with the increased growth of speleothems
under glacial conditions (Frumkin et al., 2011). The
anomalous lake stands and speleothem growth periods have been
interpreted in terms of increased (decreased) glacial (interglacial)
rainfall (Enzel et al., 2008). While lake levels variations of the DSB
lakes contrasts regional reconstructions on orbital timescales, are
they in line with other regional reconstructions on millennial
timescales, e.g. rise during interstadials (including MIS5.5, 5.3, 5.1)
and fall during stadials (including MIS5.4 and 5.2). In particular,
Fig. 11. Comparison of different paleo-archives over 260 ka from the Mediterranean and the Levant. (A) d18
O from the Sofular cave (Badertscher et al., 2011); (B) Woody taxa from
the Lago Grande di Monticchio (Allen et al., 1999); (C) d18
O from the Karaca cave (Rowe et al., 2012) and d18
O from the Corchia cave (Drysdale et al., 2005); (D) Lake Van B* (this
study); (E) Hydroclimate PC1 (this study); (F) d18
O and Arboreal pollen (AP) from the Lake Yammoûneh (Develle et al., 2011; Gasse et al., 2015); (G) d18
O from the Soreq and Peqiin
caves (Bar-Matthews et al., 2003); (H) d18
O from the Eastern Mediterranean Sea (Almogi-Labin et al., 2009; Kroon et al., 1998); (I) lake levels of the Dead Sea Basin (DSB) lacustrine
bodies (Waldmann et al., 2009; Waldmann et al., 2010). All data are shown on it original timescales. Gray shaded area represent the interglacials following the Marine Isotope Stage
(MIS) boundaries (Lisiecki and Raymo, 2005).
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9590
Heinrich stadials are reflected by Gypsum units in the Lisan formation,
which were formed during lake-level decreases and water
column mixing (Stein et al., 1997; Torfstein et al., 2015, 2009,
Neugebauer et al., 2014). Numerous studies have tried to explain
or reconcile this conundrum (Enzel et al., 2008; Frumkin et al.,
2011; Kutzbach et al., 2014; Rohling, 2013; Stein et al., 2010;
Torfstein et al., 2013; Tzedakis, 2007), which is apparent on
orbital, but not on millennial timescales related to DansgaardOeschger
and Heinrich variability (Bartov et al., 2003).1
To reconcile the interpretations of Lake Van hydroclimate and
Lake Lisan level data in terms of the large-scale hydroclimate patterns
identified in Figs. 9 and 10 we present a simple numerical
model that is driven by the simulated LOVECLIM precipitation and
evaporation changes in these regions. We suggest, that the relative
glacial changes in precipitation may have been smaller than those
for evaporation in DSB, thus leaving a delicate, but positive freshwater
balance for Lake Lisan, even though precipitation was also
reduced in glacial times. We would like to stress that according to
our model simulation precipitation was reduced in both localities,
but clearly not completely absent. We thus suggest that a longterm
reduction in precipitation does not rule out the possibility that Lake
Lisan experienced occasional “freshening” and supply of bicarbonate
and sulfide via runoff or subsurface saline aquifers (Lazar
et al., 2014) or via freshwater (Neugebauer et al., 2014, Stein
et al., 1997; Torfstein et al., 2015, 2009). This would help explain
the aragonite precipitation and gypsum deposition (Stein et al.,
1997). However, the possibility of a positive water balance is
illustrated by using precipitation and evaporation data (Pm, Em) for
the DSB region from our 408 ka climate model simulation as forcings
for our highly idealized lake level model. Note, that this model
Fig. 12. Lake level experiment for Eastern Turkey (A) and the Levant (B) from 0 to 80 kyr. From top to bottom: (A) Timeseries of precipitation anomalies over Turkey (P anom.), lowpass
PC1 and speleothem d18
O record from Karaca (eastern Turkey; Rowe et al., 2012), evaporation (E), temperature anomaly (T anom.), precipitationeevaporation (PeE) and timeintegrated
lake level (LVan); (B) Timeseries of precipitation anomalies over Levant (P anom.), speleothem d18
O record from Soreq and Peqiin cave in Israel (400 m respective 650 m
a.s.l.; Bar-Matthews et al., 2003), evaporation (E), temperature anomaly (C, T anom.), precipitationeevaporation (PeE) and time-integrated lake level (LDSB). The running means
(31-point moving window) are overlain and the second ordinates reflect the relative change of the anomalies in P and E at each site.
1
It should be noted that the interpretation of speleothem growth for only wet
periods is quite controversial, because as a result of changing infiltration coefficients,
less precipitation is required for speleothem growth in cold climates
Bonacci, O., 2001, Analysis of the maximum discharge of karst springs. Hydrogeology
Journal, 9(4): 328e338. Furthermore, changes in relative humidity may have
also contributed to increased speleothem growth during glacials Frumkin, A., BarYosef,
O. and Schwarcz, H.P., 2011, Possible paleohydrologic and paleoclimatic effects
on hominin migration and occupation of the Levantine Middle Paleolithic.
Journal of Human Evolution, 60(4): 437e451, even without invoking changes in
annual mean precipitation.
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 91
is not designed to generate the exact lake level evolution of Lake
Lisan, but rather to illustrate in a very idealized thought experiment
the fact that even for reduced glacial precipitation, lake level can be
considerably higher (lower) than for present-day conditions, if the
glacial evaporation was reduced by an even larger (smaller)
amount.
The lake level evolution of the two locations are shown in Fig.12.
The simulated and idealized Lisan Lake level evolution (LDSB(t))
exhibits a steady increase from low levels around 50e60 kyr into
the Last Glacial Maximum. Peak lake levels are attained around 16
ka followed by a steady Holocene decline of about 250 m. The P-E
timeseries also documents that a tiny imbalance of about several
centimeters per year (several percent of the mean evaporation
rates) can accumulate on orbital timescales into considerable lake
level anomalies of several hundred meters. Contrasting the glacial
evolution of Lake Lisan, the idealized lake level simulation for Lake
Van (LVan(t)) exhibits declining lake levels from MIS5b to MIS2 and
minimum values around 16 ka and a Holocene increase (Fig. 12A).
Thus, reduced MIS3 and MIS2 precipitation (consistent with
speleothem hydroclimate data from the Levant region and PC1 of
Lake Van; Fig. 11) can lead to either higher or, lower lake levels,
depending on the delicately balanced relative changes in
temperature-driven evaporation. Therefore, decreasing simulated
and reconstructed glacial precipitation in the Eastern Mediterranean
region is not inconsistent with the high glacial lake stands in
DSB lakes. We find that although glacial evaporation and precipitation
were both reduced for two locations compared to presentday,
the model, for reasonable parameter values, can still simulate
opposing glacial trajectories with lower LGM lake level for Lake
Van (Fig. 12A) and higher one for Lake Lisan (Fig. 12B).
6. Summary
Our study presented a 360 kyr-long new color record from Lake
Van showing DO variability with unprecedented decadal-scale
resolution. The millennial-scale variability recorded by the ultrahigh
resolution color variations and changes in pollen percentages,
TOC, CaCO3, Si and K correspond to large-scale hydroclimate
changes. Dansgaard-Oeschger and Heinrich stadials lead to massive
drying of the Eastern Mediterranean and rapid shifts from aridity to
pluvials are recorded for stadial/interstadial transitions. By
combining individual hydroclimate-sensitive paleo-proxies from
Lake Van using EOF analysis, we derived a combined hydroclimate
index (PC1), which correlates well with precipitation changes
simulated by a transient DO hindcast simulation performed with
the LOVECLIM earth system model. From this simulation it is
deduced that North Atlantic stadials, associated with a weaker
AMOC, generate a massive anticyclonic circulation anomaly
extending from northern Africa, the Mediterranean to Northern
Europe and east into Central Asia. This large-scale circulation
feature advects dry and cold continental air masses towards the
Lake Van site, thus explaining the paleo-proxy evidence. The length
of millennial-scale droughts tracks that of DO stadials. The most
severe droughts in the Eastern Mediterranean are found when
stadials and low precession coincide with the presence of large
northern Hemisphere ice sheets, such as during MIS2.
Our statistical analysis of the PC1 record showed evidence for
enhanced DO and millennial-scale activity during the presence of
medium-sized ice-sheets. This supports the notion that conditions
with maximum ice-sheet size can be regarded as extended stadials,
with continuous freshwater background discharging into the North
Atlantic and hence weaker AMOC. The near absence of millennialscale
activity in PC1 during interglacials points to the importance of
ice-sheets in generating DO variability.
Analyzing the orbital-scale variability in the Lake Van
hydroclimate record we found that increasing precession leads to
an enhancement of spring to early summer precipitation, as well as
winter precipitation through the enhancement of the winter storm
track. This situation enhances the mean annual rainfall considerably,
as well as the seasonality between spring precipitation and
late summer drought, as documented by the transient LOVECLIM
model experiments and by the varved sections in the Lake Van
record and their preferred occurrence during high precession
phases. Such conditions were particularly pronounced during
MIS5e, 5c, 5a.
Hydroclimate changes in the Eastern Mediterranean were also
modulated on the 100 kyr ice-sheet/eccentricity timescale. The
atmospheric response to the orographic forcing from the mature
Laurentide and Eurasian ice-sheets caused enhanced subsidence in
the western Mediterranean and an anomalous low pressure in the
Eastern Mediterranean and Levant region, which advected dry air
from northeastern Africa towards the Lake Van region.
The Lake Van hydroclimate reconstructions are a new member
in a long list of records that shows large-scale glacial drying in the
Levant and Eastern Mediterranean. Using a simplified lake level
model, forced by the transient LOVECLIM simulations, we demonstrated
that this finding is not necessarily inconsistent with higher
Lake Lisan lake levels during this period, if for Lake Lisan a reduction
of precipitation occurred in conjunction with an even stronger
reduction in evaporation. However, based on our modeling experiments,
a regional-scale anomaly for Lake Lisan cannot be excluded.
LOVECLIM is a relatively coarse-resolution climate model, which
does not capture the detailed small-scale responses that play a role
for hydroclimate variability in eastern Turkey, Anatolia and the
Levant. Nevertheless, the fact that our model simulates a realistic
semi-annual cycle in rainfall under present-day conditions and the
notion that millennial-scale and orbital-scale droughts are largescale
features, which are both reconstructed and simulated, supports
the robustness of our model-based conclusions.
Methodologically, our study makes significant advances in
combining transient climate model simulations with the Lake Van
hydroclimate reconstructions. Statistical analysis relating to the
Lake Van precipitation changes on millennial and orbital scales was
made by directly regressing changes in PC1 reconstructions with
simulated atmospheric data. The resulting patterns provided key
insights into the atmospheric drivers of subtropical megadroughts.
Furthermore we see from the high correlations between PC1 and
the simulated orbital-scale rainfall changes in LOVECLIM, that the
model simulations could have been used to generate an alternative
age model for the paleo-climate data. Driven by atmospheric CO2
concentrations (on EDC3), ice-sheet orography and orbital-forcings,
the physical model simulates rainfall variability in eastern Turkey,
which could have been used in conjunction with the multiproxy
PC1 data to derive a physically-tuned age model. This idea, which
relies on the dating of the external forcings, the realism of the
model simulations and the interpretation of the paleo-proxy data,
will be pursued in future studies.
Acknowledgments
Michael Sturm deserves special recognition for his numerous
contributions to the scientific exploration of Lake Van over the
course of the last two decades. We thank the PALEOVAN team for
support during collection and sharing of data. The authors
acknowledge funding of the Swiss National Science Foundation
(SNF) 200021_124981 and 200020_143330, the PALEOVAN drilling
campaign by the International Continental Scientific Drilling Program
(ICDP), the Deutsche Forschungsgemeinschaft (DFG) LI 582/
20-1 and the Scientific and Technological Research Council of
Turkey (Tübitak). Thanks go to A. Feray G€oktepe for her cooperation
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9592
and support and the Yüzüncü Yıl Üniversitesi of Van, Turkey for
providing the infrastructure. AT. and T.F. are supported through the
US NSF (grants ANT-1341311, 1010869) and L. M. acknowledges
support from the Australian Research Council (grant DE150100107).
The MIS3 and MIS2 transient LOVECLIM simulations were performed
on a computational cluster owned by the Faculty of Science
of the University of New South Wales. The reported data are
available as Tables 1e6 in the Supplementary Information. The data
for the transient LOVECLIM experiments can also be accessed
through http://apdrc.soest.hawaii.edu/dods/public_data/
Paleoclimate_modeling/LOVECLIM/800k and http://apdrc.soest.
hawaii.edu/dods/public_data/Paleoclimate_modeling/LOVECLIM/
Dansgaard-Oeschger.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.quascirev.2015.12.016.
References
Abe-Ouchi, A., Segawa, T., Saito, F., 2007. Climatic Conditions for modelling the
Northern Hemisphere ice sheets throughout the ice age cycle. Clim. Past 3 (3),
423e438.
Allen, J.R.M., Brandt, U., Brauer, A., Hubberten, H.W., Huntley, B., Keller, J., Kraml, M.,
Mackensen, A., Mingram, J., Negendank, J.F.W., Nowaczyk, N.R., Oberhansli, H.,
Watts, W.A., Wulf, S., Zolitschka, B., 1999. Rapid environmental changes in
southern Europe during the last glacial period. Nature 400 (6746), 740e743.
Allen, J.R.M., Watts, W.A., Huntley, B., 2000. Weichselian palynostratigraphy,
palaeovegetation and palaeoenvironment; the record from Lago Grande di
Monticchio, southern Italy. Quat. Int. 73e74, 91e110.
Almogi-Labin, A., Bar-Matthews, M., Shriki, D., Kolosovsky, E., Paterne, M.,
Schilman, B., Ayalon, A., Aizenshtat, Z., Matthews, A., 2009. Climatic variability
during the last 90 ka of the southern and northern Levantine Basin as evident
from marine records and speleothems. Quat. Sci. Rev. 28 (25e26), 2882e2896.
Andersen, K., Azuma, N., Barnola, J., Bigler, M., Biscaye, P., Caillon, N., Chappellaz, J.,
Clausen, H., Dahl-Jensen, D., Fischer, H., Fluckiger, J., Fritzsche, D., Fujii, Y., GotoAzuma,
K., Gronvold, K., Gundestrup, N., Hansson, M., Huber, C., Hvidberg, C.,
Johnsen, S., Jonsell, U., Jouzel, J., Kipfstuhl, S., Landais, A., Leuenberger, M.,
Lorrain, R., Masson-Delmotte, V., Miller, H., Motoyama, H., Narita, H., Popp, T.,
Rasmussen, S., Raynaud, D., R€othlisberger, R., Ruth, U., Samyn, D., Schwander, J.,
Shoji, H., Siggard-Andersen, M., Steffensen, J., Stocker, T., Sveinbj€ornsdottir, A.,
Svensson, A., Takata, M., Tison, J., Thorsteinsson, T., Watanabe, O., Wilhelms, F.,
White, J., Project, N.G.I.C, 2004. High-resolution record of Northern Hemisphere
climate extending into the last interglacial period. Nature 147e151.
Badertscher, S., Fleitmann, D., Cheng, H., Edwards, R.L., Gokturk, O.M., Zumbuhl, A.,
Leuenberger, M., Tuysuz, O., 2011. Pleistocene water intrusions from the Mediterranean
and Caspian seas into the Black Sea. Nat. Geosci. 4 (4), 236e239.
Bar-Matthews, M., Ayalon, A., Gilmour, M., Matthews, A., Hawkesworth, C.J., 2003.
Seaeland oxygen isotopic relationships from planktonic foraminifera and speleothems
in the Eastern Mediterranean region and their implication for paleorainfall
during interglacial intervals. Geochim. Cosmochim. Acta 67 (17),
3181e3199.
Barker, S., Knorr, G., Edwards, R.L., Parrenin, F., Putnam, A.E., Skinner, L.C., Wolff, E.,
Ziegler, M., 2011. 800,000 Years of abrupt climate variability. Science 334
(6054), 347e351.
Bartov, Y., Goldstein, S., Stein, M., Enzel, Y., 2003. Catastrophic arid episodes in the
Eastern Mediterranean linked with the North Atlantic Heinrich events. Geology
439e442.
Berger, A., 1978. Long-term variations of caloric insolation resulting from the earth's
orbital elements. Quat. Res. 9 (2), 139e167.
Bonacci, O., 2001. Analysis of the maximum discharge of karst springs. Hydrogeol. J.
9 (4), 328e338.
Bond, G., Broecker, W., Johnsen, S., McManus, J., Labeyrie, L., Jouzel, J., Bonani, G.,
1993. Correlations between climate records from North-Atlantic sediments and
Greenland Ice. Nature 365 (6442), 143e147.
Brovkin, V., Ganopolski, A., Svirezhev, Y., 1997. A continuous climate-vegetation
classification for use in climate-biosphere studies. Ecol. Model. 101 (2e3),
251e261.
Capron, E., Landais, A., Chappellaz, J., Schilt, A., Buiron, D., Dahl-Jensen, D.,
Johnsen, S.J., Jouzel, J., Lemieux-Dudon, B., Loulergue, L., Leuenberger, M.,
Masson-Delmotte, V., Meyer, H., Oerter, H., Stenni, B., 2010. Millennial and submillennial
scale climatic variations recorded in polar ice cores over the last
glacial period. Clim. Past 6 (3), 345e365.
Cheng, H., Edwards, R.L., Broecker, W.S., Denton, G.H., Kong, X.G., Wang, Y.J.,
Zhang, R., Wang, X.F., 2009. Ice age terminations. Science 326 (5950), 248e252.
Chester, P.I., Rains, J.I., 2001. Pollen and spore keys for Quaternary deposits in the
northern Pindos Mountains, Greece. Grana 40 (6), 299e387.
Cukur, D., Krastel, S., Schmincke, H.U., Sumita, M., Çagatay, N., Meydan, A.F.,
Damcı, E., Stockhecke, M., 2014. Seismic stratigraphy of Lake Van, eastern
Turkey. Quat. Sci. Rev. 104, 63e84.
Daniau, A.L., Sanchez-Goni, M.F., Beaufort, L., Laggoun-Defarge, F., Loutre, M.F.,
Duprat, J., 2007. Dansgaard-Oeschger climatic variability revealed by fire
emissions in southwestern Iberia. Quat. Sci. Rev. 26 (9e10), 1369e1383.
Dee, D.P., Uppala, S.M., Simmons, A.J., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U.,
Balmaseda, M.A., Balsamo, G., Bauer, P., Bechtold, P., Beljaars, A.C.M., van de
Berg, L., Bidlot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer, A.J.,
Haimberger, L., Healy, S.B., Hersbach, H., Holm, E.V., Isaksen, L., Kallberg, P.,
Kohler, M., Matricardi, M., McNally, A.P., Monge-Sanz, B.M., Morcrette, J.J.,
Park, B.K., Peubey, C., de Rosnay, P., Tavolato, C., Thepaut, J.N., Vitart, F., 2011. The
ERA-Interim reanalysis: configuration and performance of the data assimilation
system. Q. J. R. Meteorol. Soc. 137 (656), 553e597.
Deplazes, G., Luckge, A., Peterson, L.C., Timmermann, A., Hamann, Y., Hughen, K.A.,
Rohl, U., Laj, C., Cane, M.A., Sigman, D.M., Haug, G.H., 2013. Links between
tropical rainfall and North Atlantic climate during the last glacial period. Nat.
Geosci. 6 (3), 213e217.
Develle, A.L., Gasse, F., Vidal, L., Williamson, D., Demory, F., Van Campo, E.,
Ghaleb, B., Thouveny, N., 2011. A 250 ka sedimentary record from a small karstic
lake in the northern Levant (Yammoûneh, Lebanon). Paleogeogr. Paleoclimatol.
Paleoecol 305, 10e27.
Drysdale, R.N., Zanchetta, G., Hellstrom, J.C., Fallick, A.E., Zhao, J.X., 2005. Stalagmite
evidence for the onset of the Last Interglacial in southern Europe at 129 þ/- 1
ka. Geophys. Res. Lett. 32 (24).
Enzel, Y., Arnit, R., Dayan, U., Crouvi, O., Kahana, R., Ziv, B., Sharon, D., 2008. The
climatic and physiographic controls of the eastern Mediterranean over the late
Pleistocene climates in the southern Levant and its neighboring deserts. Glob.
Planet. Change 60 (3e4), 165e192.
Eshel, G., Farrell, B.F., 2000. Mechanisms of eastern Mediterranean rainfall variability.
J. Atmos. Sci. 57 (19), 3219e3232.
Fleitmann, D., Cheng, H., Badertscher, S., Edwards, R.L., Mudelsee, M.,
Goektuerk, O.M., Fankhauser, A., Pickering, R., Raible, C.C., Matter, A., Kramers, J.,
Tuysuz, O., 2009. Timing and climatic impact of Greenland interstadials recorded
in stalagmites from northern Turkey. Geophys. Res. Lett. 36.
Fletcher, W.J., Muller, U.C., Koutsodendris, A., Christanis, K., Pross, J., 2013.
A centennial-scale record of vegetation and climate variability from 312 to 240
ka (Marine Isotope Stages 9c-a, 8 and 7e) from Tenaghi Philippon, NE Greece.
Quat. Sci. Rev. 78, 108e125.
Frumkin, A., Bar-Yosef, O., Schwarcz, H.P., 2011. Possible paleohydrologic and paleoclimatic
effects on hominin migration and occupation of the Levantine
Middle Paleolithic. J. Hum. Evol. 60 (4), 437e451.
Gasse, F., Vidal, L., Van Campo, E., Demory, F., Develle, A.-L., Tachikawa, K., Elias, A.,
Bard, E., Garcia, M., Sonzogni, C., Thouveny, N., 2015. Hydroclimatic changes in
northern Levant over the past 400,000 years. Quat. Sci. Rev. 111 (0), 1e8.
Ganopolski, A., Calov, R., 2011. The role of orbital forcing, carbon dioxide and
regolith in 100 kyr glacial cycles. Clim. Past 7 (4), 2391e2411.
Goosse, H., Brovkin, V., Fichefet, T., Haarsma, R., Huybrechts, P., Jongma, J.,
Mouchet, A., Selten, F., Barriat, P.Y., Campin, J.M., Deleersnijder, E.,
Driesschaert, E., Goelzer, H., Janssens, I., Loutre, M.F., Morales Maqueda, M.A.,
Opsteegh, T., Mathieu, P.P., Munhoven, G., Pettersson, E.J., Renssen, H.,
Roche, D.M., Schaeffer, M., Tartinville, B., Timmermann, A., Weber, S.L., 2010.
Description of the Earth system model of intermediate complexity LOVECLIM
version 1.2. Geosci. Model Dev. 3 (2), 603e633.
Goosse, H., Fichefet, T., 1999. Importance of ice-ocean interactions for the global
ocean circulation: a model study. J. Geophys. Res. 104 (C10), 23337e23355.
Grützner, J., Higgins, S.M., 2010. Threshold behavior of millennial scale variability in
deep water hydrography inferred from a 1.1 Ma long record of sediment
provenance at the southern Gardar Drift. Paleoceanography 25 (4), PA4204.
Harrison, S.P., Go~ni, M.F.S., 2010. Global patterns of vegetation response to
millennial-scale variability and rapid climate change during the last glacial
period. Quat. Sci. Rev. 29 (21e22), 2957e2980.
Hemming, S.R., 2004. Heinrich events: massive late pleistocene detritus layers of
the North Atlantic and their global climate imprint. Rev. Geophys. 42 (1),
RG1005.
Hodell, D., Crowhurst, S., Skinner, L., Tzedakis, P.C., Margari, V., Channell, J.E.T.,
Kamenov, G., Maclachlan, S., Rothwell, G., 2013. Response of Iberian Margin
sediments to orbital and suborbital forcing over the past 420 ka. Paleoceanography
28 (1).
Hodell, D.A., Channell, J.E.T., Curtis, J.H., Romero, O.E., R€ohl, U., 2008. Onset of
“Hudson Strait” Heinrich events in the eastern North Atlantic at the end of the
middle Pleistocene transition (~640 ka)? Paleoceanography 23 (4), PA4218.
Jansen, J.H.F., Van der Gaast, S.J., Koster, B., Vaars, A.J., 1998. Short communication
CORTEX, a shipboard XRF-scanner for element analyses in split sediment cores.
Mar. Geol. 145e153.
Jouzel, J., Masson-Delmotte, V., Cattani, O., Dreyfus, G., Falourd, S., Hoffmann, G.,
Minster, B., Nouet, J., Barnola, J.M., Chappellaz, J., Fischer, H., Gallet, J.C.,
Johnsen, S., Leuenberger, M., Loulergue, L., Luethi, D., Oerter, H., Parrenin, F.,
Raisbeck, G., Raynaud, D., Schilt, A., Schwander, J., Selmo, E., Souchez, R.,
Spahni, R., Stauffer, B., Steffensen, J.P., Stenni, B., Stocker, T.F., Tison, J.L.,
Werner, M., Wolff, E.W., 2007. Climate variability over the past 800 000 years.
Science 317, 793.
Kaden, H., Peeters, F., Lorke, A., Kipfer, R., Tomonaga, Y., Karabiyikoglu, M., 2010.
Impact of lake level change on deep-water renewal and oxic conditions in deep
saline Lake Van, Turkey. Water Resour. Res. 46.
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 93
Kadioglu, M., Sen, Z., Batur, E., 1997. The greatest soda-water lake in the world and
how it is influenced by climatic change. Ann. Geophys. Atmos. Hydrospheres
Space Sci. 15 (11), 1489e1497.
Kroon, D., Alexander, I., Little, M., Lourens, L.J., Matthewson, A., Robertson, A.H.F.,
Sakamoto, T., 1998. Oxygen isotope and sapropel stratigraphy in the Eastern
Mediterranean during the last 3.2 million years. In: Proceedings of the ODP
Scientific Results, pp. 181e190.
Kutzbach, J.E., Chen, G., Cheng, H., Edwards, R.L., Liu, Z., 2014. Potential role of
winter rainfall in explaining increased moisture in the Mediterranean and
Middle East during periods of maximum orbitally-forced insolation seasonality.
Clim. Dyn. 42 (3e4), 1079e1095.
Kuzucuoglu, C., Christol, A., Mouralis, D., Dogu, A.F., Akkoprou, E., Fort, M.,
Brunstein, D., Zorer, H., Fontugne, M., Karabiyikoglu, M., Scaillet, S., Reyss, J.L.,
Guillou, H., 2010. Formation of the Upper Pleistocene terraces of Lake Van
(Turkey). J. Quat. Sci. 25 (7), 1124e1137.
Kwiecien, O., Stockhecke, M., Pickarski, Nadine, Litt, T., Sturm, M., Anselmetti, F.,
Haug, G.H., 2014. Dynamics of the last four glacial terminations recorded in
Lake Van, Turkey. Quat. Sci. Rev. 104, 42e52.
Lazar, B., Sivan, O., Yechieli, Y., Levy, E.J., Antler, G., Gavrieli, I., Stein, M., 2014. Longterm
freshening of the Dead Sea brine revealed by porewater ClÀ and d18O in
ICDP Dead Sea deep-drill. Earth Planet. Sci. Lett. 400, 94e101.
Lemcke, G., 1996. Pal€aoklimarekonstruktion Am Van See (Ostanatolien, Türkei)
(PhD thesis). Swiss Federal Institute of Technology Zurich. ETH, p. 182.
Lisiecki, L.E., Raymo, M.E., 2005. A Pliocene-Pleistocene stack of 57 globally
distributed benthic d18O records. Paleoceanography 20, PA1003 (Data archived
at the World Data Center for Paleoclimatology, Boulder, Colorado, USA).
Litt, T., Anselmetti, F.S., Baumgarten, H., Beer, J., Cagatay, N., Cukur, D., Damci, E.,
Glombitza, C., Heumann, G., Kallmeyer, J., Kipfer, R., Krastel, S., Kwiecien, O.,
Meydan, A.F., Orcen, S., Pickarski, N., Randlett, M.-E., Schmincke, H.-U.,
Schubert, C.J., Sturm, M., Sumita, M., Stockhecke, M., Tomonaga, Y., Vigliotti, L.,
Wonik, T., 2012. 500,000 years of environmental history in eastern Anatolia: the
PALEOVAN Drilling Project. Sci. Drill. 18e29.
Litt, T., Pickarski, N., Heumann, G., Stockhecke, M., Tzedakis, P., 2014. A 600,000 year
long continental pollen record from Lake Van, eastern Anatolia (Turkey). Quat.
Sci. Rev. 104, 30e41.
Martrat, B., Grimalt, J.O., Shackleton, N.J., de Abreu, L., Hutterli, M.A., Stocker, T.F.,
2007. Four climate cycles of recurring deep and surface water destabilizations
on the Iberian margin. Science 317 (5837), 502e507.
McManus, J.F., Oppo, D.W., Cullen, J.L., 1999. A 0.5-million-year record of millennialscale
climate variability in the North Atlantic. Science 283 (5404), 971e975.
Menviel, L., Timmermann, A., Friedrich, T., England, M.H., 2014. Hindcasting the
continuum of Dansgaard-Oeschger variability: mechanisms, patterns and
timing. Clim. Past 10 (1), 63e77.
Menviel, L., Timmermann, A., Timm, O.E., Mouchet, A., 2011. Deconstructing the Last
Glacial termination: the role of millennial and orbital-scale forcings. Quat. Sci.
Rev. 30 (9e10), 1155e1172.
Muller, U.C., Pross, J., Tzedakis, P.C., Gamble, C., Kotthoff, U., Schmiedl, G., Wulf, S.,
Christanis, K., 2011. The role of climate in the spread of modern humans into
Europe. Quat. Sci. Rev. 30 (3e4), 273e279.
Neugebauer, I., Brauer, A., Schwab, M.J., Waldmann, N., Enzel, Y., Kitagawa, H.,
Torfstein, A., Frank, U., Dulski, P., Agnon, A., Ariztegui, D., Ben-Avraham, Z.,
Goldstein, S.L., Stein, M., 2014. Lithology of the long sediment record recovered
by the ICDP Dead Sea Deep Drilling Project (DSDDP). Quat. Sci. Rev. 102,
149e165.
Opsteegh, J.D., Haarsma, R.J., Selten, F.M., Kattenberg, A., 1998. ECBILT: a dynamic
alternative to mixed boundary conditions in ocean models. Tellus A 50 (3),
348e367.
Peeters, F., Kipfer, R., Achermann, D., Hofer, M., Aeschbach-Hertig, W., Beyerle, U.,
Imboden, D.M., Rozanski, K., Frohlich, K., 2000. Analysis of deep-water exchange
in the Caspian Sea based on environmental tracers. Deep Sea Res. Part I
Oceanogr. Res. Pap. 47 (4), 621e654.
Penaud, A., Eynaud, F., Sanchez-Go~ni, M., Malaize, B., Turon, J.L., Rossignol, L., 2011.
Contrasting sea-surface responses between the western Mediterranean Sea and
eastern subtropical latitudes of the North Atlantic during abrupt climatic events
of MIS 3. Mar. Micropaleontol. 80 (1e2), 1e17.
Rasmussen, S.O., Andersen, K.K., Svensson, A.M., Steffensen, J.P., Vinther, B.M.,
Clausen, H.B., Siggaard-Andersen, M.L., Johnsen, S.J., Larsen, L.B., DahlJensen,
D., Bigler, M., R€othlisberger, R., Fischer, H., Goto-Azuma, K.,
Hansson, M.E., Ruth, U., 2006. A new Greenland ice core chronology for the last
glacial termination. J. Geophys. Res. 111 (D6), D06102.
Reille, M., 1990. Lecons de Palynologie et d'Analyse Pollinique. Editions du Centre
Nacional de la Recherche Scientifique (CNRS), Paris.
Reille, M., 1992. Pollen et Spores d'Europe et d'Afrique du Nord. Laboratoire de
Botanique Historique et Palynologie. URA CNRS, Marseille.
Reille, M., 1995. Pollen et Spores d'Europe et d'Afrique du Nord. Suppl. 1. In: Laboratoire
de Botanique Historique et Palynologie. URA CNRS, Marseille.
Reille, M., 1998. Pollen et Spores d'Europe et d'Afrique du Nord. Suppl. 2. Laboratoire
de Botanique Historique et Palynologie. URA CNRS, Marseille.
Rodwell, M.J., Hoskins, B.J., 1996. Monsoons and the dynamics of deserts. Q. J. R.
Meteorol. Soc. 122 (534), 1385e1404.
Rohling, E.J., 2013. Quantitative assessment of glacial fluctuations in the level of
Lake Lisan, Dead Sea rift. Quat. Sci. Rev. 70, 63e72.
Rosenberg, T.M., Preusser, F., Risberg, J., Plikk, A., Kadi, K.A., Matter, A., Fleitmann, D.,
2013. Middle and Late Pleistocene humid periods recorded in palaeolake deposits
of the Nafud desert, Saudi Arabia. Quat. Sci. Rev. 70, 109e123.
Rowe, P.J., Mason, J.E., Andrews, J.E., Marca, A.D., Thomas, L., van Calsteren, P.,
Jex, C.N., Vonhof, H.B., Al-Omari, S., 2012. Speleothem isotopic evidence of
winter rainfall variability in northeast Turkey between 77 and 6 ka. Quat. Sci.
Rev. 45, 60e72.
Schulz, M., 2002. The tempo of climate change during Dansgaard-Oeschger interstadials
and its potential to affect the manifestation of the 1470-year climate
cycle. Geophys. Res. Lett. 29 (1).
Stein, M., Torfstein, A., Gavrieli, I., Yechieli, Y., 2010. Abrupt aridities and salt
deposition in the post-glacial Dead Sea and their North Atlantic connection.
Quat. Sci. Rev. 29 (3e4), 567e575.
Stein, M., Starinsky, A., Katz, A., Goldstein, S.L., Machlus, M., Schramm, A., 1997.
Strontium isotopic, chemical, and sedimentological evidence for the evolution
of Lake Lisan and the Dead Sea. Geochim. Cosmochim. Acta 61, 3975e3992.
Stevens, L.R., Djamali, M., Andrieu-Ponel, V., de Beaulieu, J.L., 2012. Hydroclimatic
variations over the last two glacial/interglacial cycles at Lake Urmia, Iran.
J. Paleolimnol. 47 (4), 645e660.
Stockhecke, M., 2013. Exploring the 600,000 Year Old Sedimentary Record of Lake
Van (Turkey): Lithostratigraphy, Chronology and Hydroclimate Reconstruction.
ETH, Zurich (Dissertation).
Stockhecke, M., Anselmetti, F.S., Meydan, A.F., Odermatt, D., Sturm, M., 2012. The
annual particle cycle in Lake Van (Turkey). Palaeogeogr. Palaeoclimatol. Palaeoecol.
333, 148e159.
Stockhecke, M., Kwiecien, O., Vigliotti, L., Anselmetti, F., Beer, J., Çagatay, N.,
Channell, J.E.T., Kipfer, R., Lachner, J., Litt, T., Pickarski, N., Sturm, M., 2014a.
Chronostratigraphy of the 600,000 year old continental record of Lake Van
(Turkey). Quat. Sci. Rev. 104, 8e17.
Stockhecke, M., Sturm, M., Brunner, I., Schmincke, H.-U., Sumita, M., Cukur, D.,
Kwiecien, O., Anselmetti, F., 2014b. Sedimentary evolution and environmental
history of Lake Van (Turkey) over the past 600 000 years. Sedimentology 61 (6),
1830e1861.
Sumita, M., Schmincke, H.U., 2013. Impact of volcanism on the evolution of Lake
Van II: temporal evolution of explosive volcanism of Nemrut Volcano (eastern
Anatolia) during the past ca. 0.4 Ma. J. Volcanol. Geotherm. Res. 253, 15e34.
Svensson, A., Andersen, K.K., Bigler, M., Clausen, H.B., Dahl-Jensen, D., Davies, S.M.,
Johnsen, S.J., Muscheler, R., Parrenin, F., Rasmussen, S.O., R€othlisberger, R.,
Seierstad, I., Steffensen, J.P., Vinther, B.M., 2008. A 60 000 year Greenland
stratigraphic ice core chronology. Clim. Past 4 (1), 47e57.
Timm, O., K€ohler, P., Timmermann, A., Menviel, L., 2010. Mechanisms for the onset
of the African Humid Period and Sahara greening 14.5-11 ka BP. J. Clim. 23 (10),
2612e2633.
Timm, O., Timmermann, A., 2007. Simulation of the last 21000 years using accelerated
transient boundary conditions. J. Clim. 20 (17), 4377e4401.
Timm, O., Timmermann, A., Abe-Ouchi, A., Saito, F., Segawa, T., 2008. On the definition
of seasons in paleoclimate simulations with orbital forcing. Paleoceanography
23 (2), PA2221.
Timmermann, A., Friedrich, T., Timm, O.E., Chikamoto, M.O., Abe-Ouchi, A.,
Ganopolski, A., 2014. Modeling obliquity and CO2 effects on Southern Hemisphere
Climate during the past 408 ka. J. Clim. 27 (5), 1863e1875.
Timmermann, A., Knies, J., Timm, O.E., Abe-Ouchi, A., Friedrich, T., 2010. Promotion
of glacial ice sheet buildup 60e115 kyr B.P. by precessionally paced Northern
Hemispheric meltwater pulses. Paleoceanography 25 (4), PA4208.
Torfstein, A., Goldstein, S.L., Stein, M., Kushnir, Y., Enzel, Y., Haug, G., 2015. Dead Sea
drawdown and monsoonal impacts in the Levant during the Last Interglacial.
Earth Planet. Sci. Lett. 412, 235e244.
Torfstein, A., Goldstein, S.L., Stein, M., Enzel, Y., 2013. Impacts of abrupt climate
changes in the Levant from Last Glacial Dead Sea levels. Quat. Sci. Rev. 69, 1e7.
Torfstein, A., Haase-Schramm, A., Waldmann, N., Kolodny, Y., Stein, M., 2009. Useries
and oxygen isotope chronology of the mid-Pleistocene Lake Amora (Dead
Sea basin). Geochim. Cosmochim. Acta 73 (9), 2603e2630.
Tzedakis, P.C., 1994. Vegetation change through glacial-interglacial cycles e a long
pollen sequence perspective. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 345
(1314), 403e432.
Tzedakis, P.C., 2007. Seven ambiguities in the Mediterranean palaeoenvironmental
narrative. Quat. Sci. Rev. 26 (17e18), 2042e2066.
Tzedakis, P.C., Hooghiemstra, H., Palike, H., 2006. The last 1.35 million years at
Tenaghi Philippon: revised chronostratigraphy and long-term vegetation
trends. Quat. Sci. Rev. 25 (23e24), 3416e3430.
Vaks, A., Bar-Matthews, M., Matthews, A., Ayalon, A., Frumkin, A., 2010. Middle-Late
Quaternary paleoclimate of northern margins of the Saharan-Arabian Desert:
reconstruction from speleothems of Negev Desert, Israel. Quat. Sci. Rev. 29
(19e20), 2647e2662.
van Kreveld, S., Sarnthein, M., Erlenkeuser, H., Grootes, P., Jung, S., Nadeau, M.J.,
Pflaumann, U., Voelker, A., 2000. Potential links between surging ice sheets,
circulation changes, and the Dansgaard-Oeschger cycles in the Irminger Sea,
60-18 kyr. Paleoceanography 15 (4), 425e442.
Van Meerbeeck, C.J., Renssen, H., Roche, D.M., Wohlfarth, B., Bohncke, S.J.P.,
Bos, J.A.A., Engels, S., Helmens, K.F., Sanchez-Goni, M.F., Svensson, A.,
Vandenberghe, J., 2011. The nature of MIS 3 stadial-interstadial transitions in
Europe: new insights from model-data comparisons. Quat. Sci. Rev. 30 (25e26),
3618e3637.
Van Zeist, W., Woldring, H., 1978. A postglacial pollen diagram from Lake Van in
East Anatolia. Rev. Palaeobot. Palynol. 249e276.
Veres, D., Bazin, L., Landais, A., Kele, H.T.M., Lemieux-Dudon, B., Parrenin, F.,
Martinerie, P., Blayo, E., Blunier, T., Capron, E., Chappellaz, J., Rasmussen, S.O.,
Severi, M., Svensson, A., Vinther, B., Wolff, E.W., 2013. The Antarctic ice core
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e9594
chronology (AICC2012): an optimized multi-parameter and multi-site dating
approach for the last 120 thousand years. Clim. Past 9 (4), 1733e1748.
Vogel, H., Wagner, B., Zanchetta, G., Sulpizio, R., Rosen, P., 2010. A paleoclimate
record with tephrochronological age control for the last glacial-interglacial
cycle from Lake Ohrid, Albania and Macedonia. J. Paleolimnol. 44 (1), 295e310.
Waelbroeck, C., Labeyrie, L., Michel, E., Duplessy, J.C., McManus, J.F., Lambeck, K.,
Balbon, E., Labracherie, M., 2002. Sea-level and deep water temperature
changes derived from benthic foraminifera isotopic records. Quat. Sci. Rev. 21
(1e3), 295e305.
Waldmann, N., Stein, M., Ariztegui, D., Starinsky, A., 2009. Stratigraphy, depositional
environments and level reconstruction of the last interglacial Lake Samra in the
Dead Sea basin. Quat. Res. 72 (1), 1e15.
Waldmann, N., Torfstein, A., Stein, M., 2010. Northward intrusions of low- and midlatitude
storms across the Saharo-Arabian belt during past interglacials. Geology
38 (6), 567e570.
Wolff, E.W., Chappellaz, J., Blunier, T., Rasmussen, S.O., Svensson, A., 2010. Millennial-scale
variability during the last glacial: the ice core record. Quat. Sci. Rev.
29 (21e22), 2828e2838.
Zhao, Y.L., Liu, Z.F., Colin, C., Paterne, M., Siani, G., Cheng, X.R., DuchampAlphonse,
S., Xie, X., 2011. Variations of the Nile suspended discharges during
the last 1.75 Myr. Palaeogeogr. Palaeoclimatol. Palaeoecol. 311 (3e4), 230e241.
M. Stockhecke et al. / Quaternary Science Reviews 133 (2016) 77e95 95