The middle Holocene climatic records from Arabia: Reassessing
lacustrine environments, shift of ITCZ in Arabian Sea, and impacts of the
southwest Indian and African monsoons
Yehouda Enzel
a,
⁎
, Yochanan Kushnir
b
, Jay Quade
c
a
The Fredy and Nadine Herrmann Institute of Earth Sciences, The Hebrew University of Jerusalem, Givat Ram, Jerusalem 91904, Israel
b
Lamont–Doherty Earth Observatory, Columbia University, 61 Route 9W, PO Box 1000, Palisades, NY 10964-8000, USA
c
Department of Geosciences, University of Arizona, 1040 E. 4th Street, Tucson, AZ 85721, USA
abstractarticle info
Article history:
Received 14 July 2014
Received in revised form 3 March 2015
Accepted 11 March 2015
Available online 21 March 2015
Keywords:
Arabia
Indian monsoon
African monsoon
Holocene paleoclimate
lake
ITCZ
Somali jet
A dramatic increase in regional summer rainfall amount has been proposed for the Arabian Peninsula during the
middle Holocene (ca. 9-5 ka BP) based on lacustrine sediments, inferred lake levels, speleothems, and pollen. This
rainfall increase is considered primarily the result of an intensified Indian summer monsoon as part of the
insolati on-driven, northward shift of the bore al summer posit ion of the Inter-Tr opical Converge nce Zone
(ITCZ) to over the deserts of North Africa, Arabia, and northwest India.
We examine the basis for the proposed drastic climate change in Arabia and the shifts in the summer monsoon
rains, by reviewing paleohydrologic lacustrine records from Arabia. We evaluate and reinterpret individual lake-
basin status regarding their lacustrine-like deposits, physiography, shorelines, fauna and flora, and conclude that
these basins were not occupied by lakes, but by shallow marsh environments.
Rainfall increase required to support such restricted wetlands is much smaller than needed to form and maintain
highly evaporating lakes and we suggest that rainfall changes occurred primarily at the elevated edges of south-
western, southern, and southeastern Arabian Peninsula. These relatively small changes in rainfall amounts and
local are also supported by pollen and speleothems from the region. The changes do not require a northward
shift of the Northern Hemisphere summer ITCZ and intensification of the Indian monsoon rainfall. We propose
that (a) latitudinal and slight inland expansion of the North African summer monsoon ra ins across the Red
Sea, and (b) uplifted moist air of this monsoon to southwestern Arabia highlands, rather than rains associated
with intensification of Indian summer monsoon, as proposed before, increased rains in that region; the se
African monsoon rains produced the modest paleo-wetlands in downstream hyperarid basins. Furthermore,
we postulate that as in present-day, the ITCZ in the Indian Ocean remained at or near t he equator all year
round, and the Indian summer monsoon, through dynamically induced air subsidence, can reduce rather than en-
hance summer rainfall in the Levant and neighboring deserts, including Arabia. Our summary suggests a widen-
ing to the north of the latitu dinal range of the rainfall associated with the North Africa n summer monsoon
moisture crossing the Red Sea to the east. We discuss other mechanisms that could have potentially contributed
to the formation and maintaining of the modest paleo-wetlands.
© 2015 Elsevier B.V. All rights reserved.
Contents
1. Introduction............................................................... 70
1.1. Structureofthisreassessment ................................................... 70
2. GeographicalandMeteorologicalconstraintsonArabianrainfall ....................................... 70
2.1. Annualandseasonalrainfalldistribution .............................................. 70
2.2. SynopticclimatologyofArabiarainfall................................................ 70
2.2.1. SummerrainfallinArabia ................................................. 71
2.2.2. Thesouthwestmonsoon:SomaliJetovertheArabianSea................................... 71
3. Lakeversusmarsh/wetland/dischargeenvironmentsinaridlands ...................................... 71
Global and Planetary Change 129 (2015) 69–91
⁎ Corresponding author. Tel.: +972 2 6584210; fax: +972 2 5662581.
E-mail addresses: yehouda.enzel@mail.huji.ac.il (Y. Enzel), kushnir@ldeo.columbia.edu (Y. Kushnir), quadej@email.arizona.edu (J. Quade).
http://dx.doi.org/10.1016/j.gloplacha.2015.03.004
0921-8181/© 2015 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Global and Planetary Change
journal homepage: www.elsevier.com/locate/gloplacha
4. Lacustrine-likedepositsfromArabia ....................................................73
4.1. Generalconsiderations.......................................................73
4.2. Shorelines ............................................................80
4.3. Marshdeposits ..........................................................81
4.4. Asymmetryindeposition......................................................82
4.5. Faunal and floralremains......................................................82
4.6. Lake-budgetcalculations......................................................82
4.7. Summarizinglacustrine-likeindicators................................................83
5. LimitedvegetationchangesinArabia....................................................83
6. RainfallshiftsbasedonSpeleothems....................................................83
7. MiddleHoloceneregionalhydroclimatologyanditsimplications .......................................84
7.1. Hydroclimaticpattern .......................................................84
7.2. Constraintsonearlytomid-Holoceneclimate ............................................85
7.3. Southwestmonsoonwindsandupwellingoff-OmanassistkeepingArabiadry..............................86
7.3.1. Upwellingoff-southernOman ...............................................86
7.4. AshiftinITCZ? ..........................................................86
7.5. Implicationstopre-HoloceneinterglacialclimatesofArabia ......................................87
7.6. Implicationforhuman-lacustrineassociationinArabia ........................................87
8. Conclusions ...............................................................87
Acknowledgments...............................................................88
AppendixA. Supplementarydata.......................................................88
References ..................................................................88
1. Introduction
Temporal patterns of Holocene lake-level status from Arabia to India
(e.g., Street-Perrott and Harrison, 1985; Street-Perro tt et al. , 1989;
Street-Perrott and Perrott, 1993; Kohfeld and Harrison, 2000; Braconnot
et al., 2004) have led investigators to advocate large hydroclimatic
changes during the early to middle Holocene in Arabia (e.g., Roberts
and Wright, 1993). In a recent influential exampl e, Wanner et al.
(2008) mapped a 12°–15° northward shift in the middle Holocene mon-
soon rains over the Arabia-Near East-West Asia sector, based on paleo-
climatic proxy-records including the lake status reconstructions for 6 ka.
More conservative, but still substantial, estimates of northward dis-
placement of the middle Holocene summer monsoon rainfall belt over
Arabia and West Asia have also been proposed (e.g., Fleitmann et al.,
2007;Fig.11inJunginger et al., 20 14). Existing sum maries of
Holocene paleoclimates of the Arabian Sea and its African, Arabian,
and Indian margins (e.g., Overpeck et al., 1996; Gasse, 2000;
Hoelzmann et al., 2004; Prasad and Enzel, 2006; Sta ubwasser and
Weiss, 2006; Fleitmann et al., 2007; Lézine et al., 2007, 2010;
Wanner et al ., 2008; Conroy and Overpeck, 2012) point primarily to
the Indian summer monsoon (ISM) and to a lesser extent, to the
African summer monsoons as explaining the amounts, seasonality and
spatial distribution of past rainfall in Arabia.
The proposed northward expansion of summer rains across the
Arabian Peninsula has been widely interpreted to represent a north-
ward shift of the Inter-Tropical Convergence Zone (ITCZ) that has
been proposed as a major driver of Holocene changes in the hydrology
and vegetation in that area. This was postulated because of the proposed
increased depth, area, or simply the presence of Arabian paleolakes.
These lakes were taken as part of the evidence of high seasonal rainfall,
which is neede d to overcome the high rates of regional evaporation
in the hyperarid Ar abian Pen insula (currently is N 2200 and as high
as ~3000 mm yr
−1
, with pan or potential evaporation 3000–
4000 mm yr
−1
,seeElNesr et al., 2010; Sorman and Ab dulrazzak,
1995). As several of these lake basins have no contributing watersheds,
this implies a problematic assertion: the annual rainfall must have been
at the magnitude of the above evaporation rates, unless these basins
were primarily fed by regional groundwater flow or were only ephem-
eral; if such rainfall values are too high, then only ephemeral, shallow
ponds could have formed.
1.1. Structure of this reassessment
In this paper we describe the present-day rainfall distribution in the
Arabian Peninsula and the geographical and atmospheric mechanisms
that control it. We then review the paleoenvironmental observations
that underlie the proposed higher stands of the Arabian lakes. We use
the information in the original reports and published research to distin-
guish between marshy or shallow water environments and open water
bodies of lakes. This distinction is critical given the large moisture sur-
plus necessary for the formation and maintenance of open lakes com-
pared with a mosaic of wetland environments. Following this review
we look at published pollen and speleothem data to gain a better picture
of the region's paleoclimatology.
Finally, we propose explanations for the paleo-precipitation patterns
emerging from the above lacustrine, speleothems, and pollen records,
based on the modern atmospheric and ocean circulation in the northern
Indian Ocean and its margins.
2. Geographical and Meteorological constraints on Arabian rainfall
2.1. Annual and seasonal rainfall distribution
The Arabian Peninsula is part of the Sahara-Arabian-Thar subtropical
desert belt. As such its climate is dry, with most parts of the peninsula
receiving ≤ 100 mm yr
−1
of rainfall (Fig. 1). The topography strongly
affects the spatia l pattern of the mean annual rainfall (Fig. 1) with
only the elevated areas in the southwestern Peninsula experiencing
≥~500 mm annually (e.g., Subyani and Al-Dakheel, 2009) with highest
points in Yemen receiving up to 800 mm yr
−1
.
The seasonal distribution of rainfall (Fig. 2) indicates that most of the
inner peninsula and the coastal strip along the Persian/Arabian Gulf and
the Gulf of Oman coasts receive most of their rainfall in winter and
spring. In contrast, the mountain ranges of southwestern Saudi Arabia
and Yemen are dominated by summer rains (see also Almazroui,
2011; Almazroui et al., 2012, 2013).
2.2. Synoptic climatology of Arabia rainfall
We present several atmospheric circulation patterns that govern the
winter and summer rainfall in different parts of the Arabian Peninsula.
70 Y. Enzel et al. / Global and Planetary Change 129 (2015) 69–91
Their timing, magnitude and moisture sources are different indicating
that a slight re-organization of these sy stems might create the basis
for the change in rainfall patterns depicted by the paleoenvironmental
records.
We have identified four rain-causing mechanisms that can affect
the Arabian Peninsula durin g late fall, winter and early spring.
a) Mediterranean cyclones that migrate into the Levant are dominant
rainfall mechanisms in this region between October and May (Alpert
et al., 2004; Almazroui et al., 2012). They bring little rain to Arabia, al-
though they can become the dominant source of rainfall in the
peninsula's east-northeast, relative to other seasons. b) The semi-
permanent low-pressure trough over th e Red Sea, when combined
with an eastward moving upper level trough can create local convective
rainfall over the southern Levant, Sinai, and western and northern Ara-
bian Peninsula and cause local flooding. c) The winter–spring tropical
plumes (TPs, also can be considered as across-Sahara atmospheric riv-
ers), which originate over the eastern tropical Atlantic or tropical
Africa and move eastward across northern North Africa (Alpert et al.,
2004; de Vries et al., 2013; Tubi and Dayan, 2014); occasionally, when
combined with the subtropical jet positioned at a relatively southern
latitude, these plumes can bring widespread, one to three days long, tor-
rential rainfall and flooding into Arabia, Sinai, and southern Jordan and
Israel (e.g., Ziv, 2001; Kahana et al., 2002; Barth and Steinkohl, 2004;
Rubin et al., 2007; Tubi and Dayan, 2014); and d) northeast winter mon-
soon winds over the Gulf of Oman could have bring rain to the south-
eastern edges of Arabia.
2.2.1. Summer rainfall in Arabia
The broader regional pattern of monthly precipitation during sum-
mer (June–September, Fig. 3) indicates that southern Arabia falls be-
tween two regional monsoon systems. In the west, the African
summer monsoon extends eastward along the sub-Saharan Sahel to
the prominent local maximum over the Ethiopian Highlands and across
the Red Sea into the mo untains of southwestern Saudi Arabia and
Yemen and westernm ost Oman. Rainfall intensity in this part of the
peninsula varies in phase with African summer monsoon intensity
(Fig. 4). A band of rainfall extends farther east in Yemen into southern
Oman, along the southern coastal topography (Fig. 1B).
The peninsula's interior is hyperarid, receiving b 80 mm yr
−1
. The
subsidence aloft induced by the heavy convection associated with the
ISM (Rodwell and Hoskins, 1996), which is strongest over the East Med-
iterranean, extends over much of the Arabian Peninsula and possibly
also brings northerly winds to southern Arabia and even to northern
Arabian Sea region (Figs. 1 and 5). Moreover, the local high surface albe-
do together with the overlying dry air leads to a net loss of radiative en-
ergy, solar and infrared. The resulting radiative cooling of the
atmosphere must be balanced by the adiabatic warming associated
with large-scale subsidence, which further intensifies the drying of the
desert (Charney, 1975). The moisture transport vectors (Fig. 3)and
the rainfall minimum inland (Fig. 1b) suggest that the narrow zone of
rainfall in coastal southern Arabia is associated with local orographic
uplift (e.g., see left panels in Fig. 5 of Fleitmann et al., 2007). Very little
moisture is transported inland across and over the coastal mountains.
The low-level flow over the interior is mostly from the northwest
(Fig. 2 in Ziv et al., 2004) and is dry and weak. Thus, in the interior of
the Arabian Peninsula summer rainfall drops rapidl y north of it s
Yemeni–Omani southern coast to b 10 mm for the entire summer sea-
son (Almazroui et al., 2012).
2.2.2. The southwest monsoon: Somali Jet over the Arabian Sea
The summertime southwesterly monsoon winds, which supply the
moisture to the mountainous southernmost Arabian Peninsula, form a
low-level jet, which peaks in intensity at altitudes of 1000–180 0 m
above sea level (Findlater, 1969a,b, 1971, 1977; Slingo et al., 2005). Re-
ferred to as the Somali or Findlater jet, these winds cross the equator in
the trade wind belt north of Madagascar. The jet parallels the northeast
African coast, in part constrained by the high East African orography
(Findlater, 1969a; see Fig. 5.31 in Krishnamurti et al., 2013)andforced
by the northward pressure gradient. North of the equator, the low-
level Somali jet is deflected by the Coriolis effect eastward from the
East African highlands and across the Arabian Sea and heading directly
towards India. Over the Arabian Sea the winds evaporate large amounts
of moisture and eventually, over India and the Bay of Bengal, it feed the
ISM rains (Findlater, 1971, 1977; Halpern and Woiceshyn, 2001; Slingo
et al., 2005; Turner an d Annamalai, 2012; Krishnamurti et al., 2013)
with rainfall maxima where uplifted (Fig. 3).
3. Lake versus marsh/wetland/discharge environments in arid lands
Quantitative estimates of past changes in rainfall and temperature
on the continents rely heavily on solving hydrological/energy budgets
of past lakes (e.g., Mifflin and Wheat, 1979; Street-Perrott and Harrison,
1985; Benson, 1986; Benson and Paillet, 1989; Enzel, 1992). The main
concept for these reconstructions is that at steady state in closed basin
lakes, direct rainfall and runoff into a lake are balanced by evaporative
losses (e.g., Street-Perrott and Harrison, 1985). The primary source of
inflow for lakes in arid and semiarid basins is surface runoff
(e.g., Mifflin and Wheat, 1979; Street-Perrott and Harrison, 1985;
Enzel, 1992), as groundwater recharge and direct rainfall in these envi-
ronments are usually low (e.g., Yechieli and Wood, 2002). The little re-
charg e of groundwater when it occurs is more ef
ficient in supplying
water to terminal basins in arid regions as it is less susceptible to evap-
oration losses and leads to the formation of fresh to saline wetlands.
Clearly, an accurate reconstruction of a paleolake area is vital, since
evaporative loss from a lake depends primarily on its surfac e area
(e.g., Benson and Paillet, 1989). Surface area reconstruction is straight-
forward where paleolake shorelines are well preserved, but can be
more difficult where shore deposits are old and/or local erosion rates
are high. It is also important to clearly distinguish between lake and
other lacustrine deposits in we t environments characterized by
ground-wat er discharge such as springs, wet meadows , shallow
marshes or discharging saline playas (Mifflin and Wheat, 1979; Rosen,
1991, 1994; Pigati et al., 2014).
In their seminal study of the hydrological basins in the Great Basin–
Mojave deserts of the western United States, Mifflin and Wheat (1979)
outlined for the first time a hydro-physiographic criterion for separating
basins with evidence for paleolake existence from those without. Until
this 1979 study, researchers had not been rigorous in distinguishing
true lake deposits from the other types of deposits of similar appear-
ance. This distinction is of the utmost importance given the very differ-
ent paleohydrologic conditions required to support an open water body,
even if it is a shallow lake, compared with other la custrine environ-
ments. It requires much less water input into a drainage basin to main-
tain a shallow and patchy wetland or discharging playa, fed by
groundwater that is mainly sheltered from direct evaporation than an
open surface of a lake under ~2 m yr
−1
of evaporation or even less.
Following studies using Mifflin and Wheat's (1979) approach, exam-
ined lacustrine-like sediments in lowland basins lacking shorelines,
mainly in North and South America the distinc tion was possible. Re-
searchers were able to define deposits of spring heads, wet meadows,
shallow marshes and even shallow local pools, where these could be
easily be mistaken for, and in some cases were actual ly reported as,
lake deposits (Quade, 1986; Quade and Pratt, 1989; Quade et al., 1995,
1998, 2003; Rech et al., 2002, 2003; Pigati et al., 2009, 2011, 2014).
Other examples of basins first defined as paleolakes and later shown
as shallow wetlands are from northwest India (e.g., Enzel et al., 1999)
and southern Jordan (Catlett et al., 2013, Catlett, 2014; Mischke et al.,
2016). Table 1 summarizes the geomorphic, sedimentologic, and floral
and faunal criteria that dis tinguish lake from wetland deposits (see
also Pigati et al., 2014). This table guides us in re-interpreting the pub-
lished information on the deposits in the basins of Arabia.
71Y. Enzel et al. / Global and Planetary Change 129 (2015) 69–91
A
B
72 Y. Enzel et al. / Global and Planetary Change 129 (2015) 69–91
4. Lacustrine-like deposits from Arabia
4.1. General considerations
Based on the criteria shown in Table 1 we re-appraised the pub-
lished sedimentary, floral and faunal characteristics and coexistence of
these indicators in Holocene lowland deposits in Arabia (Table 2;
Fig. 6). We did not consider any pre- Holocene propos ed lacustrine-
like deposits such as Mudawwara in southernmost Jordan (Petit-Maire
et al., 2010), which we think could also have benefited from additional
scrutiny regarding its actual depositional environment, weak chrono-
logical controls, and identification and mapping of shoreline features
as suggested by Catlett et al. (2013) and Rech et al. (2014). Recently,
the Jubbah basin and other an-Nafud basins were proposed to
Fig. 1. A. A schematic and simplistic representation of the main boreal summer, low-level atmospheric and oceanic features controlling the climate in the Arabian Sea and its surrounding
regions in the northern Indian Ocean. It is presented for listing the main features to an unfamiliar reader. Note the very low latitude [Equator to ~5°N] of the Northern Hemisphere weak
summer Inter-Tropical Convergence Zone (ITCZ). B. Mean annual precipitation across the Arabian Peninsula and neighboring areas (contours in mm — note that isohyets contour resolu-
tion is changing with the amount from 40 mm to 100 mm at 200 mm yr
−1
) superimposed on the topography (colors in m). Source: University of East Anglia — Climate Research Unit
compilation and gridding of land station reports version TS3.1. Note (1) that the scale of the coastal rainfall band may appear broader than real because of the relatively low spatial res-
olution of the gridded precipitation data, (2) the strong imprint of topography on the precipitation pattern, (3) the extension of East Africa rainfall maxima across the Red Sea to the Yemen
highlands and further to the east along the south Yemen and south Oman mountains, and (4) The relatively steep gradient in rainfall from the Red Sea highlands in Yemen and southwest
Saudi Arabia coast into the heart of Arabia to the east.
0 102030405060708090100
precipitation [% of annual rainfall]
Fig. 2. The proportion of rainfallby seasons (compare with Fig. 1): winter (December–March, top left panel), spring (April–May, top right), summer (June–September, bottom left), and fall
(October–November, bottom right). Unites are percent of total annual rainfall. Source: University of East Anglia — Climate Research Unit TS3.1 dataset. Note that most of the rainfall in
Arabia is during the winter season with total annual very small, except in the southwest and to lesser extent in southern Arabia where there is summer rain maximum, which is an ex-
tension of the East African monsoon. The Horn of Africa does not receive summer rain but experiences spring rainfall maximum. Note: (1) The impact of the Mediterranean storms on
northeastern Arabia and the amplification by the Persian/Arabian Gulf. (2) The Gulf of Oman imprint on rainfall in the nearby areas of coastal northern Oman and the Oman Mountains.
73Y. Enzel et al. / Global and Planetary Change 129 (2015) 69–91
