ArticlePDF Available

Quaternary sediment sources and loess transport pathways in the Black Sea - Caspian Sea region identified by detrital zircon U-Pb geochronology


Abstract and Figures

Constraining the controls on the distribution of sediment at a continental scale is a critical step in understanding long-term landscape and climate evolution. In particular, understanding of the role of rivers in wider sediment routing and impacts on aeolian loess formation on a continental scale remains limited. Extensive Quaternary loess deposits are present on the East European Plain and in the Black Sea - Caspian Sea region and are associated with major rivers draining numerous surrounding cratonic and orogenic hinterland areas. Coupled with this, complex changes in local and global sea level have affected the extent and drainage of the Caspian Sea and the Black Sea, and Quaternary glaciations have impinged on the northern margin of the East European Plain. This suggests that sediment routing and loess formation may show complex patterns and controls. Here, we apply UPb dating of detrital zircons from fluvial, marine and aeolian (dominantly loess) sedimentary records on the East European Plain and in the Black Sea - Caspian Sea region. This shows a strong control of large rivers on the distribution of sediments at a continental scale in the region, through long-distance transport of several 1000 km, sourced from continental and mountain glacier areas prior to marine or atmospheric reworking and transportation. Strong spatial variability in zircon UPb data from loess deposits on the East European Plain reveals multiple diverse sources to the different individual loess sections, whereas no significant temporal variability in loess source is detected during the Late Pleistocene of the Lower Volga loess in South Russia. While the sediment supply from glacial areas via rivers plays an important role for the provenance of East European Plain loess deposits, our data indicate that the stark spatial diversity in loess provenance on the East European Plain is often driven by the input of multiple local sources. Similar to the loess, marine sediments from different basins of the Black Sea and the Caspian Sea also show significant spatial variability. This variability is controlled by the bathymetry of the seas, leading to sedimentary intermixing by sea currents within, but not between different separated sea basins. A direct comparison of marine and aeolian sediments at the same depositional site suggests that although loess and marine sediments are both dominantly sourced from river sediments containing far travelled sedimentary material, local sources play a more important role in many loess deposits.
Content may be subject to copyright.
Global and Planetary Change 209 (2022) 103736
Available online 12 January 2022
0921-8181/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (
Quaternary sediment sources and loess transport pathways in the Black Sea
- Caspian Sea region identied by detrital zircon U-Pb geochronology
Chiara K¨
, Thomas Stevens
, Martin Lindner
, Yunus Baykal
, Amin Ghafarpour
Farhad Khormali
, Natalia Taratunina
, Redzhep Kurbanov
Uppsala University, Dept. of Earth Science, 75236 Villav¨
agen 16, Sweden
Department of Chemistry and Physics of Materials, University of Salzburg, Salzburg, Austria
Department of Soil Science, Gorgan University of Agricultural Sciences and Natural Resources, Gorgan, Iran
MSU, Lomonosov Moscow State University, Faculty of Geography, M.V., Leninskie Gori, 1, Moscow 119992, Russia
Department of Stone Age Archeology, Institute of Archeology and Ethnography SB RAS, 17, Ac. Lavrentieva ave., Novosibirsk, 630090, Russia
RAS, Laboratory of Evolutionary Geography, Institute of Geography, Staromonetny, 29, Moscow 119017, Russia
Editor name: Zhengtang Guo
Detrital zircon U
Pb dating
Sediment routing
East European Plain
Caspian Sea
Constraining the controls on the distribution of sediment at a continental scale is a critical step in understanding
long-term landscape and climate evolution. In particular, understanding of the role of rivers in wider sediment
routing and impacts on aeolian loess formation on a continental scale remains limited. Extensive Quaternary
loess deposits are present on the East European Plain and in the Black Sea - Caspian Sea region and are associated
with major rivers draining numerous surrounding cratonic and orogenic hinterland areas. Coupled with this,
complex changes in local and global sea level have affected the extent and drainage of the Caspian Sea and the
Black Sea, and Quaternary glaciations have impinged on the northern margin of the East European Plain. This
suggests that sediment routing and loess formation may show complex patterns and controls. Here, we apply
Pb dating of detrital zircons from uvial, marine and aeolian (dominantly loess) sedimentary records on the
East European Plain and in the Black Sea - Caspian Sea region. This shows a strong control of large rivers on the
distribution of sediments at a continental scale in the region, through long-distance transport of several 1000 km,
sourced from continental and mountain glacier areas prior to marine or atmospheric reworking and trans-
portation. Strong spatial variability in zircon U
Pb data from loess deposits on the East European Plain reveals
multiple diverse sources to the different individual loess sections, whereas no signicant temporal variability in
loess source is detected during the Late Pleistocene of the Lower Volga loess in South Russia. While the sediment
supply from glacial areas via rivers plays an important role for the provenance of East European Plain loess
deposits, our data indicate that the stark spatial diversity in loess provenance on the East European Plain is often
driven by the input of multiple local sources. Similar to the loess, marine sediments from different basins of the
Black Sea and the Caspian Sea also show signicant spatial variability. This variability is controlled by the ba-
thymetry of the seas, leading to sedimentary intermixing by sea currents within, but not between different
separated sea basins. A direct comparison of marine and aeolian sediments at the same depositional site suggests
that although loess and marine sediments are both dominantly sourced from river sediments containing far
travelled sedimentary material, local sources play a more important role in many loess deposits.
1. Introduction
One of the most fundamental topics in Earth Sciences is how the
material for the formation of sediments is produced, transported and
deposited; i.e. the sedimentary cycle. These processes are crucial to
constrain in order to understand the long-term development of
landscapes, the forcing and effects of climate and environmental change,
and a host of other related questions. In particular, knowledge of where
and how material is eroded, transported and deposited, and which
process agents are active in this (e.g. water, ice, wind, etc.), is essential
to link geomorphic work with forcing of other systems, such as climate
(e.g. Bridge and Demicco, 2008). Indeed, constraining lags and
* Corresponding author.
E-mail address: (C. K¨
Contents lists available at ScienceDirect
Global and Planetary Change
journal homepage:
Received 7 May 2021; Received in revised form 2 December 2021; Accepted 7 January 2022
Global and Planetary Change 209 (2022) 103736
pathways in the cycling of sedimentary material is also central to
interpretation of rock archives of erosion on land, in terms of climate or
landscape processes (e.g. Frostick and Jones, 2002).
Rivers play a major role in this sediment cycling, and climate-
landscape interactions can be revealed by analysis of how and where
land is denuded via river systems, and how sediment is stored en route to
sedimentary basins. Thus, rivers may also play a fundamental role in the
distribution of material that later forms aeolian sediments by facilitating
the movement of large volumes of sediment to areas where it can easily
be deated and subsequently deposited as loess or aeolian sands (e.g.
Smalley et al., 2009; H¨
allberg et al., 2020). In particular, aeolian dust
transport in atmospheric suspension (silt and clay sized particles) is
important to understand as a major component of the climate system,
both driving and responding to climate change (e.g. Maher et al., 2010;
Choobari et al., 2014; Harrison et al., 2001). Examination of the wide
scale distribution of dust particles is central to understand how much
material has been transported, for how long, and over what pathways.
This information is used to better simulate the potential impact of at-
mospheric dust on the Earth system (Albani et al., 2015). Loess com-
prises a record of past wind-blown dust, and in particular is dominated
by ‘coarse dust(520
m), which itself has specic climate forcing ef-
fects and has been signicantly underestimated in terms of importance
and atmospheric abundance (Adebiyi and Kok, 2020). Constraining
prior river transport in dust and loess distribution is a key part in un-
derstanding pathways of coarse dust transport, as the extent of this prior
uvial transport has implications for the extent and duration of particle
atmospheric transport.
While this importance is well-known, understanding of wide-scale
and long-term sediment cycling in loess formation is limited by a lack
of knowledge of source and transport of sedimentary material. Loess
deposits are extensively preserved over large areas of the mid latitudes,
more or less continuously over thousands of km in Eurasia (e.g.
c et al., 2015; Lehmkuhl et al., 2020). However, the spatial
changes in sources for these deposits over a wider, continental scale
have seldom been investigated. One of these few studies used bulk
geochemistry data and atmospheric modelling to suggest that last glacial
dust material in Europe only underwent atmospheric transport of a few
hundred km or less (Rousseau et al., 2014). However, the specic role of
rivers in wider distribution of material prior to atmospheric transport
was not examined, even though several studies propose the importance
of river transport in the formation of European loess deposits and river
sediment sources are strongly suggested by reconstructed transport di-
rections (e.g. Smalley et al., 2009; Ujv´
ari et al., 2012; Nawrocki et al.,
2018; Pa´
nczyk et al., 2020; Baykal et al., 2021).
Despite this uncertainty, the generation of atmospheric dust and
ultimately the formation of loess deposits in Europe and western Asia is
strongly associated with the overall cool climate during the Quaternary,
and its glacial and interglacial cyclicity. Glacial grinding by continental
ice sheets and mountain glaciers, as well as cold climate weathering
processes in mountains, led to an increased production of ne-grained
sedimentary material (Muhs et al., 2003; Smalley and Derbyshire,
1990). Generally, the aeolian deation of dust is facilitated in cold and
poorly vegetated semi-arid and arid areas experiencing strong winds,
which are more common in glacial periods (e.g. Pye, 1995). However,
transport of material to suitable depocentres where loess can be pre-
served may require extended uvial transport prior to aeolian deation
and nal deposition. As such, material eventually forming loess can
experience several cycles of erosion, accumulation, and transport before
its nal deposition, and as a result of this can contain sediment from
several proto sources, transported via multiple agents and pathways (e.
g. Licht et al., 2016).
To test the inuence of rivers on sediment supply and subsequent
loess formation, to determine the processes and patterns in wide scale
sediment cycling, and to facilitate better constraint of the possible ef-
fects of dust on climate, environment and landscape, it is important to
pinpoint loess provenance in detail using techniques that can reveal
multiple proto- and secondary sources, thereby allowing analysis of
stepwise dust transport from sources to sink. Furthermore, constraining
the provenance of loess distributed over a wide area where sources likely
vary geographically, allows analysis of the general role of rivers in wide
scale sediment distribution, facilitating insight into recycling and source
to sink sediment movement. In addition, the detailed analysis of how
loess provenance varies at a continental scale, may yield insight into the
controls on wide-scale generation and distribution of aeolian dust during
the Quaternary, which is of particular importance in constraining the
role of dust in the climate system.
A suitable area to test these factors and their implications is the East
European Plain (EEP), including the Black Sea - Caspian Sea region. This
area is subject to a complex set of sediment erosion, deposition,
reworking and redistribution patterns, controlled by the interaction of
uvial, marine and aeolian systems, which originate from the north
(Baltica-Fennoscandia, Urals), south (Caucasus and Iranian mountains),
east (central Asian deserts and mountains) and west (Carpathian-Alps
and Danube basins) (Fig. 1a). However, the provenance and transport
pathways of each of these sedimentary systems remain poorly con-
strained, as do their relationships and interactions. Targeted provenance
studies on loess exist only for Ukrainian deposits in the west of the EEP
(Buggle et al., 2008; Nawrocki et al., 2018; Pa´
nczyk et al., 2020) and
river provenance data is scarce. Given the complex sedimentary setting
in the Black Sea - Caspian Sea region, a provenance tracer is required
that can be applied for all these different kinds of sediments and trans-
port pathways to disentangle the origin of the material and transport
systems. Detrital zircon U
Pb geochronology represents a powerful
single grain analysis technique that has become a widely used method
for sedimentary provenance studies and may be a suitable candidate
(Fedo et al., 2004 and references therein). Detrital zircons have been
successfully used to trace the provenance of various types of siliciclastic
sediments and sedimentary rocks (e.g. sandstone, conglomerate), and
their metasedimentary equivalents (e.g. Froude et al., 1983; Nutman
et al., 1999; Horton et al., 2008; Allen et al., 2006; Vincent et al., 2013;
Cawood et al., 2003; Allen et al., 2006; Aleinikoff et al., 2008; Safonova
et al., 2010; Stevens et al., 2010; Wang et al., 2011; Garzanti et al.,
2013). In loess, single-grain U
Pb dating of detrital zircons can be very
source diagnostic in instances where multiple dust sources are expected
(Stevens et al., 2010), and overcomes ambiguities in deciphering mul-
tiple sources inherent in bulk sample geochemical data. While recycling
of zircons through multiple depositional and erosional phases can
complicate interpretations, zircons are generally highly suitable for
provenance analyses due to their high resistance to mechanical disag-
gregation and chemical weathering during erosion and transportation
(e.g. Moecher and Samson, 2006). This resistance allows examination of
crustal proto-sources and constraint of several steps in the generation
and transport pathways of loess. Due to the different timing of tectonic
events in the study area, diagnostic differences in zircon ages are ex-
pected for grains deriving from different orogens bordering the EEP.
This makes detrital zircon U
Pb geochronology a potentially powerful
tool to understand provenance and sedimentary dispersal systems in the
EEP and the Black Sea - Caspian Sea region. Furthermore, the methods
suitability for loess provenance studies on the EEP has previously been
demonstrated on a smaller scale than considered here (Nawrocki et al.,
2018; Pa´
nczyk et al., 2020). As such, here we apply detrital zircon U
analyses to river, aeolian and marine sediments over a wide area of the
EEP and south of the Caspian Sea in order to constrain the pathways and
mechanisms of sediment routing in the region, and unravel the impli-
cations for dust distribution and loess formation.
2. Study area and methodical approach
During Pleistocene cold stages, the northern part of the EEP was
covered by the Fennoscandian Ice Sheet (Velichko et al., 2011; Fig. 1a)
and the mountain regions of the Urals and Caucasus were glaciated
(Astakhov, 2017; Gobejishvili et al., 2011). In addition, periglacial
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
Fig. 1. (a) Map of the study area (EEP and entire Black Sea- Caspian Sea region) with the maximum extent of the Fennoscandian Ice Sheet at 18 ka after Hughes et al.
(2016), and showing the sites sampled for this study. The colour coding indicates the sampled material (see legend). The Lower Volga sites shown in Fig. 1b with
stratigraphy are marked by the red circle. The purple squares indicate the map details of Figs. 7, 8, 9, 10, 11 and 12. (b) Stratigraphic charts of the three Lower Volga
sites, showing the sampling depths and age of the provenance samples (after K¨
oltringer et al., 2020). The denoted published ages for SA are optically stimulated
luminescence (OSL) ages from Yanina et al. (2017) and Kurbanov et al. (2020), while the ages of the provenance samples are currently unpublished. (For inter-
pretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
inuence reached into the lowlands and far south, almost to the shores
of the Black Sea and Caspian Sea (e.g. Goretskiy, 1958; Taratunina et al.,
2020; Velichko et al., 2006). The climate in the EEP was overall cold and
dry, while precipitation appears to have increased slightly during in-
terstadials (e.g. Dodonov et al., 2006; Liang et al., 2016; K¨
oltringer et al.,
2020). Generally, precipitation decreased from north to south over the
EEP in the Pleistocene (e.g. Buggle et al., 2009), similar to the pattern
observed today, with the northern part being dominated by periglacial
forest-steppe landscape (Novenko, 2006). At present, the coastal area in
the east and north of the Black Sea, around the Azov Sea, as well as the
Northern Caspian lowland, is characterized by a dry to temperate con-
tinental climate (~300500 mm a
mean annual precipitation) and
varying types of steppe vegetation (e.g. Sirotenko and Abashina, 1992;
Kosarev et al., 2007; Buggle et al., 2009). It is suggested that similar
conditions prevailed during the Late Pleistocene in this southern part of
the EEP, but were potentially drier, favoring dust entrainment from al-
luvial, glaciouvial and coastal sediments by regional and local wind
systems and subsequent nearby deposition as loess (K¨
oltringer et al.,
2020, 2021).
In contrast to the tectonically stable EEP, the South Caspian Basin
and its surroundings (Fig. 2) represent a tectonically active region
throughout the entire Quaternary, comprising collision and subduction
of different tectonic units (e.g. Motavalli-Anbaran et al., 2011). In the
southeast of the Caspian Sea in the northeastern foothills of the Alborz
mountains and the extensive upland area along the Iranian-Turkmen
boarder (Iranian Loess Plateau; Fig. 1.), Pleistocene climate re-
constructions suggest a semiarid to subhumid climate, with a strong
precipitation gradient from north to south and also from east to west
(200700 mm a
), and generally poor vegetation cover, similar to
today (e.g. Khormali et al., 2020). The Iranian Loess Plateau is located in
the semi-arid region of North Iran, where up to >60 m thick loess de-
posits are preserved. Climatic cyclicity reecting Pleistocene glacial and
interglacial stages in the southern Caspian Sea region seems to broadly
overlap with that of the EEP (e.g. Kehl et al., 2021).
Five major river systems drain into the area of the northern Black
Sea- Caspian Sea region at present, representing a wide range of possible
pathways for sediment to enter the region. From east to west these rivers
are: the Volga, the Don, the Dnieper, the Dniester and the Danube; the
former four draining generally southwards from the EEP and the latter
draining eastwards from the Alps via the Carpathians and Carpathian-
Pannonian Basin (Fig. 1a). These and other river sediments likely
reect the geological and environmental situation of their drainage
basins and can give information about changes in drainage systems and
sediment supply, as well as responses to tectonic or climatic events (e.g.
orogeny, glaciation) (Richards, 2002). During the cold Pleistocene
phases, all of these river systems were affected by continental or
mountain glaciation, with glacial meltwater having an important con-
trol on the hydrography and sediment transport of uvial systems (e.g.
Vandenberghe, 1995; Vandenberghe and Woo, 2002). Provenance
studies of sediments from rivers from the EEP as well as from the Danube
and its tributaries suggest that these sediments generally reect the
exposed geology of the riversdrainage basins (e.g. Allen et al., 2006;
Safonova et al., 2010; Wang et al., 2011; Ujv´
ari et al., 2012; Ducea et al.,
The EEP is geologically comprised of the East European Craton
(EEC), which consists of three Late Archean to Early Proterozoic cratonic
blocks: Fennoscandia, Sarmatia and Volgo-Uralia (Fig. 2, Bogdanova,
1993). The drainage basins of the Volga (1.4 million km
), Don (0.42
million km
) and Dnieper (0.53 million km
) neighbour each other and
cover a large part of the EEP (Safonova et al., 2010). The Volga drains
into the North Caspian Sea and its catchment is bordered by the Urals to
the east, where its largest tributary, the Kama River, originates (e.g.
Golsovo and Belyaev, n.d.). In addition to this Volga input, the Caspian
Sea Basin receives signicant discharge from the Ural River in the north,
Fig. 2. Selected crustal units and sedimentary formations that are potential primary source regions for sediments in the EEP and the Black Sea- Caspian Sea. For
simplicity, the gure only shows units and formations that are most relevant for discussion of EEP sediment sources (for more information see Discussion). The
shaded areas denote certain crustal segments or sedimentary formations of the same age. Note that no Archean crust currently crops out in the EEC blocks of
Fennoscandia and Volgo-Uralia in the EEP, only in Sarmatia. The Neogene sedimentary Yergeni Formation (analysed in this study) covers the Sarmatian block in the
Yergeni uplands and Volga uplands. Also note how the north of the Black Sea is geologically comprised of one Palaeozoic girdle (orogens) and one Meso-, Neo-
proterozoic girdle (forelands). (ND-North Dobrogea, CM-Crimean Mountains, YF-Yergeni Formation).
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
the Atrek River in the southeast, and the Samur, Terek and Kura rivers
from the Caucasus (Fig. 1). Also, the Palaeo-Amu-Darya River used to
drain into the South Caspian over the Uzboy Passage until the Mid
Holocene (e.g. Hinds et al., 2004). The Don River ows into the Azov Sea
near the city Rostov-on-Don and its catchment area includes the Yergeni
uplands as well as part of the Greater Caucasus, which borders its basin
in the south. Both the Pleistocene Palaeo-Don as well as the modern Don
deeply incise into the Yergeni Formation, named after the Yergeni River,
which existed from the Late Miocene to the Early Pliocene and deposited
uvial sediments in the Yergeni uplands (hilly landscape triangularly
bordered by Don, Volga and Manych depression) and the Volga uplands
north of Volgograd (Karandeeva, 1957, Fig. 2). By contrast, the Dnieper
River drains the west of the EEP and parts of the Carpathian foreland
(72.1 thousand km
), and ows into the northwestern corner of the
Black Sea, which is comprised of a wide continental shelf with a shallow
water depth of ~20 m (e.g. Lericolais et al., 2007). The Danube River
also ows into the northwestern corner of the Black Sea, but drains a
more extensive area of the Alpine-Carpathian-Dinaric system in central
Europe (0.8 million km
). As such, the rivers draining to the Caspian Sea
and Black Sea have a range of source areas from multiple mountain
ranges to the west, north and south.
In addition to the uvial sediments deposited by these rivers, the
Black Sea and Caspian Sea area is covered by extensive marine sedi-
mentary deposits associated with past high stands. Marine deposits yield
information about the recycling and intermixing of different sediment
sources and the inuence of sea level oscillation and base level change
on sediment supply (e.g. Bridge and Demicco, 2008). Both the Black Sea
and the Caspian Sea experienced several phases of transgression and
regression, mainly controlled by hydrological changes in their water-
sheds and global ice volume changes (e.g. Deuser, 1972; Yanko-Hom-
bach and Kislov, 2018). Moreover, the Black Sea sea-level uctuations
are linked to temporary connections during the Quaternary with the
Mediterranean Sea through the shallow Bosporus Strait, and with the
Caspian Sea through the Manych depression, implying forced water
intrusion from the Caspian Sea (e.g. Mangerud et al., 2001; Leonov et al.,
2002; Badertscher et al., 2011; Krijgsman et al., 2019; Yanina, 2020;
Kurbanov et al., 2018). The Pleistocene glaciations also play a role for
these local sea level histories via direct climate forcing or control of river
discharges (e.g. Karpychev, 1993; Rychagov, 1997; Badertscher et al.,
2011). As a result of these large sea level uctuations, vast areas of dry
continental shelf are today covered by marine terraces, particularly in
the at Northern Caspian lowland (Yanina, 2014).
Aeolian sediments in the form of loess and sands are also extensively
represented in the Black Sea - Caspian Sea region (e.g. Panaiotu et al.,
2001; Gendler et al., 2006; Buggle et al., 2008; K¨
oltringer et al., 2020;
Velichko et al., 2009; Zubakov, 1988; Nawrocki et al., 2018; Pa´
et al., 2020; Khormali and Kehl, 2011a). While deposits in central and
eastern Europe are comparably well studied, loess in the southern EEP in
southern Russia is less understood and is important in developing a
wider understanding of aeolian dust and Eurasian continental atmo-
spheric circulation history. Despite their importance, located towards
the middle of the vast Eurasian loess belt, very little is known about the
nature, source, transport and accumulation of these Late Pleistocene
dust deposits, as well as their importance for understanding Eurasian
paleoclimate (K¨
oltringer et al., 2020).
The control of the big rivers of the EEP on the Black Sea and Caspian
Sea levels has been extensively discussed (e.g. Kvasov, 1979; Grosswald,
1998; Mangerud et al., 2001; Panin et al., 2020). While the focus has
mostly been on the inuence of rivers on the seaswater budgets, the
supply and distribution of sediment in sea basins is also expected to be
directly affected by river dynamics and sea level changes (e.g. Tudryn
et al., 2016). However, it is not only the marine sediments that are likely
to reect a connection with rivers. As mentioned earlier, loess deposits
also often show close linkages to uvial systems at different scales
(Smalley et al., 2009), and such relationships have been demonstrated
for loess deposits located in various places, including the Danube Basin,
Ukrainian and Polish, and Chinese Loess Plateau loess (e.g. Ujv´
ari et al.,
2012; Stevens et al., 2013; Nie et al., 2015; Pa´
nczyk et al., 2020; Baykal
et al., 2021). In this river-loess model, loess is transported only a short
distance by wind from relatively close by (10s to 100s of kms) deation
zones and the importance of rivers lies in the wider distribution of the
material prior to aeolian transport, potentially over thousands of kms
from source regions.
The potential role of the large EEP rivers in carrying large amounts of
silt particles from glacial outwash of the Fennoscandian Ice Sheet to the
southern EEP has been discussed previously (Jefferson et al., 2003;
Buggle et al., 2008; Smalley et al., 2009) and Pa´
nczyk et al. (2020)
reinforced this connection in their detrital zircon provenance study.
Comparable to North American loess, east Ukrainian loess represents a
classical ‘glacial loess system, in which the material produced by
northern continental glaciation is transported to the south via big rivers.
Similar is expected for south Russian loess. Indeed, glaciouvial sedi-
ments may have been more or less continuously deposited in the
southern EEP since the Early Pleistocene (Gozhik, 1995) and recent
scanning electron microscope analyses suggest glacial grinding and river
transport for Lower Volga loess (K¨
oltringer et al., 2021). Further west, in
southern Poland, Baykal et al. (2021) showed a complex history of
sediment reworking from both ice sheet and mountain sources before
loess deposition. Although untested, this scenario possibly explains also
the western and southern Russian loess deposits. Loess in the Danube
and Carpathian basins, in contrast, shows a provenance connection to
rivers from mountain regions only (Ujv´
ari et al., 2012; Pa´
nczyk et al.,
Considering this, the actual picture of loess formation in the EEP
might be complex. Not all SW-Russian loess deposits are necessarily
sourced from the large EEP rivers, and instead the Caucasus Mountains
might also function as a source area (Sergeev et al., 1986), especially if
Caucasus detritus forms a signicant component of Don sediments
entering the Black Sea. The proximity of abundant marine sediment
deposits from Caspian Sea and Black Sea shelfs, dry during parts of the
last glaciation, further complicates the situation. These deposits of
complex sediment source could represent potential loess sources too. In
addition, loose sediment products of non-glacial erosion in the active
tectonic region around the South Caspian Basin as well as the deserts to
the east of the Caspian Sea cannot be ruled out as source areas for loess
deposits around the Caspian Sea, via direct aeolian transport. This di-
versity in topography and depositional environments implies the pos-
sibility of the three major possible modes of loess genesis operating in
the region: the continental glacier provenance-river transport mode, the
mountain provenance-river transport mode, and the mountain
provenance-river transport-desert transition mode (Li et al., 2020). In
tectonically active regions, non-glacial erosion and material formation
might play a role in these mountain provenance modes. Thus, the EEP -
Black Sea - Caspian Sea area represents an ideal place to examine wider
scale controls on dust generation, transport and deposition of loess, and
to constrain continental scale spatial changes in dust pathways and
3. Sampling and analytical methods
The aim of sample collection was to obtain a wide range of Quater-
nary sediments of diverse origin, which reect the past climate and
landscape evolution of the EEP and southern Caspian-Black Sea region.
We also sample to specically examine in more detail the temporal and
spatial variability of Lower Volga loess (LVL; Fig. 1). Two loess sites
along the lower branch of the Volga River in the Northern Caspian
lowland in Southern Russia, Leninsk (LN) and Raygorod (RG) (Table 1),
were sampled at different stratigraphic depths for temporal provenance
variability analysis. Bulk samples of 12 kg were taken from optically
stimulated luminescence (OSL) dated loess layers: RG2 (~20 ka), RG3
(~30 ka), RG4 (~40 ka); LN4 (31.2 ±2.2 ka), LN5 (56.1 ±2.8 ka), LN6
(63.4 ±3.0 ka) (unpublished ages; Kurbanov et al., 2020, contains
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
limited published ages for these sections). In addition, Lower Volga loess
was sampled from one OSL dated layer at Srednyaya Akhtuba (SA, 87.6
±4.1 ka) (Fig. 1; Yanina et al., 2017). A stratigraphic description of the
three LVL sites can be found in K¨
oltringer et al. (2020), Lebedeva et al.
(2018), Makeev et al. (2021) and Taratunina et al. (2020). Samples from
six other Quaternary loess deposits located in the Northern Black/Azov
Sea - Caspian Sea region were taken from deposits of known age
wherever available for comparison and to understand wider-scale
spatial variability of aeolian dust sources in the region (Table 1,
Fig. 1). One loess site located in the western EEP (Pushkari) and one on
the northern Iranian Loess Plateau (Aghband) were also sampled (Fig. 1,
Table 1). The strategy for secondary (sedimentary) source material
sampling considers the availability of material, its exposure to deation,
the geographic distance to the sink, and topographic obstacles on the
way. Accordingly, we have sampled uvial sediments from the Palaeo-
Volga and Palaeo-Don rivers, marine sands from the Black Sea and
Caspian Sea, as well as mountain colluvium from the northern Greater
Caucasus, and desert sands from the Karakum Desert. 12 kg bulk
samples were taken from sites at each of these locations (Table 1).
Additionally, we compare our data to published U
Pb detrital zircon
data from further loess sites and potential secondary and primary source
areas in the region (Table 1).
Detrital zircon grains were separated from bulk material through
sieving (425
m), Wiley table washing, Frantz magnetic separation and
heavy liquid separation at Uppsala University, Sveriges Geologiska
okning (SGU) and the Arizona LaserChron center. To minimize
the loss of very ne-grained zircons in loess due to clay coatings and
their effects on hydraulic sorting, ultrasonic bathing of the samples was
applied following the separation technique of Hoke et al. (2014) to
disaggregate the fractions. Zircon separates were mounted on 1′′ epoxy
mounts and polished to 1
m nish. High resolution backscattered
electron imaging of each mounted sample was performed with a Hitachi
3400 N scanning electron microscope and further used to locate the
spots for laser ablation on the zircon grains. Dependent on the zircon
yield, a high analysis number (n), with ideally >300 randomly selected
zircon grains of all size fractions, was pursued for each sample. This
relatively new approach of high n detrital zircon analyses overcomes
some limitations inherent in lower n analyses of <100150 grains per
sample. Detrital zircon provenance interpretations are often substan-
tially based on the comparison of relative heights of age peaks in
probability plots and this requires large n for statistical robustness
(Pullen et al., 2014). Low n analyses (<115 grains) also harbour the risk
of missing signicant age peaks entirely (Vermeesch, 2004).
Pb dating was carried out at the Arizona LaserChron center.
Isotope ratios were measured using a Thermo Element2 high-resolution
(HR) multicollectorinductively coupled plasmamass spectrometer
(MC-ICP-MS). Laser ablation was attained for 10 s at ~7 J/cm
energy density) with a repetition rate of 7 Hz, using a 9
m laser beam
diameter for analyses (Gehrels et al., 2008; Gehrels and Pecha, 2014;
Pullen et al., 2018). The minimum size for analysed zircon grains in this
study is 12
m, in contrast to >25
m that is commonly used for other
loess provenance studies, due to laser size and count number limitations.
Table 1
Detrital zircon samples and data considered in this study, their sampling location, lithology and age.
Area Section/
Location Material Age Analysed zircons
Lower Volga Leninsk (LN) 48.7221 45.1595 Loess 6530 ka 865 This study
Lower Volga Raygorod (RG) 48.4313 44.9665 Loess ~4020 ka 865 This study
Lower Volga Srednyaya Akhtuba
48.7004 44.8937 Loess 87.6 ±4.1 ka 299 This study
Don Kalach 48.6648 43.6725 Loess Unknown Quaternary 291 This study
NE-Azov Sea Beglitsa 47.1391 38.5611 Loess 3025 ka 142 This study
E-Crimea Eltigen 45.1854 36.4049 Loess 6050 ka 222 This study
SW-EEP Pushkari 53.2351 34.1927 Loess MIS 3 78 This study
Manych depression Kalaus 45.8099 43.2314 Loess MIS 32 287 This study
Manych depression Chograi 45.4054 44.2456 Loess MIS 43 62 This study
N-Caucasus foothill Budennovsk 44.9458 44.1828 Loess MIS 32 279 This study
Iranian Loess
Aghband 37.3701 55.0903 Loess 5868 ka 257 This study
Lower Volga Raygorod 48.4313 44.9665 Volga sand ~60 ka 298 This study
Lower Volga Chorny Yar 48.0320 46.1119 Volga sand ~150130 ka 300 This study
Lower Volga Seroglazka 47.0135 47.4599 Volga sand 10085 ka 304 This study
Don Liska 48.6701 43.1693 Don alluvium MIS 43 300 This study
Yergeni uplands no described section 46.4962 43.79947 Yergeni sand Pliocene 295 This study
East Crimea Eltigen 45.1854 36.4049 Black Sea sand ~110 ka 244 This study
Turkmen Coast Cheleken 39.5274 53.1735 Caspian sand ~10 ka 275 This study
Karakum desert Choganly 38.0337 58.41192 Desert dune sand Holocene 195 This study
Caucasus no described
44.11211 41.8136 Mountain
Modern 293 This study
SW-EEP Staiki 50.0937 30.8983 Loess MIS 2 156 Pa´
nczyk et al., 2020
SW-EEP Vyazovok 49.9611 32.9215 Loess MIS 122 212 Pa´
nczyk et al., 2020
SW-EEP Dnieprovskie 46.6435 31.8938 Loess MIS 6 51 Pa´
nczyk et al., 2020
Black Sea Dnieper mouth Fluvial sand Modern 129 Wang et al., 2011
Azov Sea Don mouth Fluvial sand Modern 149 Wang et al., 2011
Lower Volga Volga Volgograd Fluvial sand Modern 100 Wang et al., 2011
Lower Volga Chorny Yar 48.03201 46.1119 Fluvial sand Modern 53 Allen et al., 2006
W-Black Sea Siret mouth 45.2306 28.00419 Fluvial sand Modern 110 Ducea et al., 2018
W-Black Sea N-Dobrogea
Diverse Neoproterozoic-
1056 Balintoni and Balica
Alborz Mountains cumulative 35.9658 52.5425 Sandstone Neoproterozoic-Cenozoic 442 Horton et al., 2008
Aspheron Peninsula Casp. Prod. series 40.4406 50.0715 Sandstone Late Miocene 219 Allen et al., 2006
Greater Caucasus Bajocian 41.0321 48.3481 Sandstone Middle Jurassic 60 Allen et al., 2006
Indolo-Kuban 45.4 40.5 Sandstone Pliocene-Quaternary 65 Vincent et al., 2013
E-Black Sea Taman 45.1847 36.5999 Sandstone Pliocene-Quaternary 66 Vincent et al., 2013
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
This allows consideration of a larger range of zircon grain sizes and
makes it less likely that ner zircon fractions will be missed, which
might be relevant for interpretation of silt-dominated loess deposits. The
following zircon standards were used as reference material for isotope
fractionation correction: FC-1 (Paces and Miller, 1993), R33 (Black
et al., 2003), and SL (Gehrels et al., 2008). The Nu Instruments Nu-TRA
software and an Arizona in-house Excel spreadsheet (E2agecalc) were
used for data normalization and reduction, uncertainty propagation and
age calculation. The
Pb measurement was used for the correction of
initial Pb, assuming its composition following Stacey and Kramers
(1975). To correct for
Pb, and
Th frac-
tionation and to account for instrumental drift, a sliding-window
average of eight reference material analyses was applied.
Measurement uncertainties for
U and
Th were
attained via regression line analyses, while uncertainties for
Pb were dened by standard deviation. Internal (mea-
surement) uncertainties are reported at 1
, while external (systematic)
uncertainties are reported at 2
level. Reference material FC-1 serves to
estimate U and Th concentrations, which are accurate to ~20%. The age
cut-off was set at 900 Ma.
U ages were used for ages <900 Ma,
while for ages >900 Ma
Pb ages were used for plotting.
U age analyses with uncertainty >10% (1
) are disregarded
and for
Pb age analyses uncertainties >10% (1
) were
included only if the
U age was <400 Ma. Concordance was
dened as (206 Pb
207Pb ), and not reported for
U ages <400 Ma
Fig. 3. KDE plots of the three Raygorod samples with RG2 being the youngest and RG4 the oldest (for age and sampling position see Fig. 1b). In the left column all
zircons from age 03000 Ma are plotted, the right column shows zircons from age 0900 Ma. Every dash on the x-axis represents one zircon (this way of presentation
applies also for following KDE plots of samples in the Resultssection). The map detail shows the position of the three sites (underlined names) in respect to each
other as well as other close-by sampling sites. See Fig. 1a for the location of the LVL sites in the EEP and for the legend.
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
because of the large uncertainty in
Pb age. Results showing
>20% discordance or >5% reverse discordance were excluded from
further analyses.
All literature published U
Pb detrital zircon data, which is plotted
in this study, was reprocessed according to the same criteria as far as the
available data and their collection and processing routine allowed
(please check the respective publications for methodological details;
Table 1). The results of Kernel Density Estimation (KDE) with a xed
bandwidth of 25 Ma are plotted to display the detrital zircon age dis-
tribution for each sample (Vermeesch, 2012).
4. Results
4.1. Lower Volga loess
The three samples from Raygorod taken at different stratigraphic
depths (RG2 ~ 20 ka, RG3 ~ 30 ka, RG4 ~ 40 ka, Fig. 1b) show slight
differences, both in the presence of age fractions, as well as in their
abundance. Mesozoic ages are very rare in all three samples (n =24; <
2%). All samples have a distinct peak at 360 Ma. In total, the age fraction
between 280 and 380 Ma makes up 12% of all measured zircons in RG4
and RG3, and 7% in RG2. Ages from 900 Ma to 2000 Ma are present in
all RG samples and account for ~7076% of measured ages each,
however their distribution differs and RG2 shows less sharp individual
age peaks than RG3 and RG4. RG4 contains no zircons of age
20002300 Ma, and in RG3 zircons of age 20502500 are almost absent
(n =3), while RG2 shows a very spread age distribution between 2000
and 2300 Ma (n =10). Archean ages in RG3 and RG2 comprise ~10% of
the total, while only 5% of analysed zircons in RG4 are older than 2500
Ma. In RG2 these ages are most concentrated at around 2700 Ma (Fig. 3).
The abundance of certain age fractions across the range of zircon
ages differs to some extent in the three samples. However, direct com-
parison is complicated due to the different number of analysed zircons
and it is unclear if this variation is simply a function of sampling
uncertainty (Pullen et al., 2014) or rather reects real differences in the
population of grains at different sample depths. To test this, random
subsamples of 300 grains each were generated from all data combined
from the Raygorod site. The resulting KDEs show little variability, which
appears mostly in relative peak abundance rather than in the presence or
absence of relevant age fractions (Fig. 4).
The KDEs of the three loess samples of different depositional age
from Leninsk (LN4 ~ 31 ka, LN5 ~ 56 ka, LN6 ~ 63 ka, Fig. 1b) all show
an abundant age fraction at 360 Ma and a similar age peak distribution
between 900 Ma and 2100 Ma (Fig. 5). Mesozoic ages are also rare in all
the samples, although LN6 contains a higher number of Late Mesozoic
ages than LN4 and LN5. The younger samples LN4 and LN5 contain
Palaeoproterozoic ages of 24002500 Ma years but no Archean ages
between 2500 and 2600 Ma, as seen in LN6. Random subsamples of 300
grains were also generated from the combined set of all analysed Leninsk
ages. These Leninsk subsamples are also overall similar, showing the
same general age peaks but with some differences in age peak heights
(Fig. 4). The outcome of this test is discussed further below.
The single analysed Srednyaya Akhtuba sample (SA ~88 ka, Fig. 1b)
contains a large age fraction at 360 Ma. No Mesozoic ages younger than
Middle Triassic were measured. No ages of around 700 Ma are present,
however, this fraction is also very small in the LN and RG samples (1%).
The age distribution between 1000 and 2000 Ma is similar to the other
Volga loess samples. Only one grain older than 2800 Ma was measured
for SA (Fig. 5).
4.2. Loess from the East European Plain and the Caspian Sea region
Loess from the Kalach section on the Don River bank is of unknown
Quaternary age and shows the most prominent peak in the sample at
180 Ma. Abundant peaks of similar amplitude occur also at 100 Ma and
360 Ma. Ages between 900 and 2100 Ma, as well as Archean ages, are
present and similarly distributed to the ages in the Lower Volga loess
(Fig. 6).
Fig. 4. KDE plots of the random n =300 subsamples from the grouped Raygorod and Leninsk data.
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
Loess from the Beglitsa section at the Azov Sea (~3025 ka; Chen
et al., 2018) shows a plateau shaped zircon distribution of relatively low
abundance between 250 and 450 Ma. A fraction around 300 Ma is
dened by a few grains and does not stand out from the other Palaeozoic
age populations, which are all dened by only a few grains each and in
sum make up 8% of all analysed grains. Proterozoic ages between 1000
and 2100 Ma are present in high abundance (78%) and Early Palae-
oproterozoic and Archean ages (11%) show some ages of around 2450
Ma and a broader fraction at 27002800 Ma of similar height to the
Palaeozoic plateau of ages (Fig. 6). The loess from the Eltigen section on
the Crimean Peninsula is 6050 ka old (Kurbanov et al., 2019) and
contains abundant Palaeozoic ages (17%) with a sharp peak at 315 Ma.
The abundant age fractions at 450 and 530 Ma dene a double peak,
while several Neoproterozoic ages date to c. 600 Ma. Meso- and Palae-
oproterozoic ages between 1000 and 2000 Ma show similar abundances
as the Palaeozoic fractions for three peaks at 1030 Ma, 1630 Ma and
1890 Ma (Fig. 6). Few zircons are older than 2800 Ma (n =6).
Loess from the Manych depression, one site located at the River
Kalaus (MIS 32), the other section at the River Chograi (MIS 43), show
very similar zircon age spectra. Both show a major Palaeozoic age peak
at 300 Ma, younger than the Palaeozoic peak of 360 Ma that is abundant
in many of the previously described samples. Another Palaeozoic peak is
seen at 440 Ma. Neoproterozoic ages peak at 620 Ma. Mesoproterozoic
ages form a broad peak between 950 and 1150 Ma, while older Meso-
and Palaeoproterozoic ages are present in great abundance, peaking at
1480 Ma, 1640 Ma and 1775 Ma. This pattern is clearly dened by the
higher-n loess sample from the Kalaus section. Archean ages show a
broad double-peak at 2700 and 2800 Ma (Fig. 7). Few zircons are older
than 2800 Ma. The Budennovsk section (MIS 32) is located proximal to
the Manych sites, but in contrast to these sites, which crop out along
rivers owing through the Manych depression, the more southern
Budennovsk section is located higher up on the northern foothills of the
Greater Caucasus (Fig. 7). The Budennovsk loess sample also contains an
abundant age fraction at 300 Ma. An age peak at 450 Ma is well dened
and abundant too, while a less pronounced age fraction appears around
600 Ma. The Proterozoic and Archean ages are less frequent compared to
the Palaeozoic age fractions (51% and 32%), in contrast to the Kalaus
and Chograi samples (Fig. 7).
To the north of the study area, the loess sample from the Pushkari
section in the west EEP (MIS 3) does not show any Palaeozoic ages
Fig. 5. KDE plots of the three Leninsk samples with LN4 being the youngest and LN6 the oldest, with the oldest LVL sample here from Srednyaya Akhtuba (SA) (see
Figs. 1 and 2 for details).
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
younger than 420 Ma. Only one grain of Neoproterozoic age is present.
Meso- and Palaeoproterozoic ages are abundant, particularly between
1000 and 1300 Ma, and 1600 and 1800 Ma, dening four sharp peaks.
Only four scattered Archean zircons are present (Fig. 8). By contrast,
loess from two Ukrainian sections to the southwest of Pushkari (Staiki
and Vyazovok) published in Pa´
nczyk et al. (2020), show late Palaeozoic
age populations at around 300 Ma and 420440 Ma. Few early
Cambrian and Neoproterozoic ages are present and scattered between
520 and 620 Ma, and Meso- and Palaeoproterozoic ages are abundant
until 2000 Ma. Several Archean ages can be observed (Fig. 8). The data
from another Ukrainian loess site analysed by Pa´
nczyk et al. (2020),
Dnieprovskie at the Dniepers mouth into the Black Sea, contains
abundant and comparable high peaks for the Early Palaeozoic and Late
Neoproterozoic as well as for the Meso- and Palaeoproterozoic (Fig. 8).
Pb ages younger than 700 Ma from the Iranian loess site Aghband
located on the Iranian Loess Plateau (5868 ka, Lauer et al., 2017) are
extremely abundant (49%), with major peaks at 100 Ma, 250 Ma and
300 Ma. Less abundant age fractions are present at 450 Ma and 620 Ma.
Meso- and Palaeoproterozoic ages between 1100 and 1800 Ma, as well
as Archean ages, are essentially absent and other Proterozoic ages are
low in abundance (Fig. 9). In total 28% of all measured zircons are older
than 900 Ma.
Fig. 6. KDE plots of loess from the Kalach site at the Don River, the Beglitsa site at the Azov Sea, and the Eltigen site on the Crimean Peninsula.
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
4.3. Other sedimentary material
The Pleistocene Volga River sand sample from Chorny Yar
(~150130 ka; Butuzova et al., 2019) displays a major well-dened
peak at 360 Ma and less abundant age fractions of 250 Ma and 580
Ma. Mesoproterozoic ages form a broad plateau between 950 and 1200
Ma and show a peak of broadly the same height at 1480 Ma. Two iso-
lated and well-dened Palaeoproterozoic age fractions dene clear
peaks at 1600 and 1780 Ma. Archean ages peak at 2700 Ma (Fig. 10).
Further downstream, at the section Seroglazka (~85100 ka; Butuzova
et al., 2019; Shkatova, 2010), Volga uvial sand shows a prominent peak
at 360 Ma in its Palaeozoic age distribution. The age fractions around
250 Ma and 450 Ma are represented by fewer grains. Meso- and Palae-
oproterozoic ages are generally abundant in the sample with several
prominent, narrow peaks, except between 1300 and 1400 Ma, and
Archean ages are distributed between 2550 and 2750 Ma (Fig. 10).
Similarly, the Volga uvial sand from Raygorod, age ~ 60 ka (RG5;
Fig. 1b, Table 1), contains an abundant 360 Ma age population with a
subpopulation at around 300 Ma. Only ve spread out zircon ages are
younger than this major Palaeozoic population, while Early Palaeozoic
ages dene a more abundant fraction between 400 and 500 Ma. The few
Neoproterozoic ages in the sample occur between 530 and 760 Ma.
Meso- and Palaeproterozoic ages, by contrast, are abundant, although
again with fewer ages between 1300 and 1400 Ma. In the Archean, a
clear age peak occurs at 2700 Ma (Fig. 10).
The KDE for Pleistocene age Don River sand from the Liska section
shows a low abundance of ages between 200 and 900 Ma (4%). Meso-
and Palaeoproterozoic ages, in contrast, are abundant with several well-
Fig. 7. KDE plots of loess from the Kalaus and Chograi sites in the Manych depression, and the Budennovsk site at the northern Caucasus foothills.
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
dened single peaks (87%). Archean ages form a broad peak between
2650 and 2850 Ma (Fig. 11). By contrast, the Yergeni sand sample from
the Pliocene age Yergeni formation in the Yergeni uplands shows more
Palaeozoic ages (8%): a clearly dened but low peak at 360 Ma, and a
subpeak at 440 Ma. Neoproterozoic ages are concentrated in the Late
Neoproterozoic with no zircons dating to 700900 Ma. Mesoproterozoic
ages are abundant, especially between 1000 and 1150 Ma, dening one
major peak and an older subpeak. Palaeoproterozoic ages form two
broader age fractions with double-peaks each, at 15001700 Ma and
18001900 Ma. While Archean ages are well presented (7%), only four
zircons are older than 2800 Ma (Fig. 11). Black Sea sand from the Cri-
mean Eltigen section (~110 ka; Kurbanov et al., 2019) is directly
overlain by a sequence of loess, from which the Eltigen loess sample for
this study derives. The Black Sea sand sample contains only few, greatly
spread out Palaeozoic ages (<5%) in contrast to the overlying loess
(Fig. 6). Meso- and Palaeoproterozoic ages are very abundant by
contrast (75%), with a broader trough between 1200 and 1400 Ma and
an almost total absence of Palaeoproterozoic grains older than 2080 Ma.
Archean ages are divided into two fractions: one between 2500 and
2700 Ma and another with ages of around 2800 Ma and older (Fig. 11).
Local colluvium from the northern Greater Caucasus of modern depo-
sitional age shows quite different peaks, including a very abundant age
peak at 280 Ma. Two smaller age fractions peak at 450 Ma and at
600650 Ma. The abundance of Meso- and Palaeoproterozoic and
Archean ages is very low (33%) in comparison to the Palaeozoic and
Neoproterozoic ones (50%) (Fig. 11).
The Caspian marine sands from the Turkmen Cheleken Peninsula
(~10 ka; Kurbanov et al., 2014) show a high number of Palaeozoic and
Mesozoic ages (59%). Major peaks are present at 300 Ma and 450 Ma,
and a smaller but well-dened fraction also exists for the Cretaceous at
100 Ma. Neoproterozoic ages peak at around 600 Ma, while older
Proterozoic ages are almost absent, except for at around 1880 Ma.
Archean ages are spread out but with no grains older than 2700 Ma
(Fig. 9). By contrast, the KDE of the modern Karakum Desert sands at
Choganly shows a high abundance of young ages <300 Ma (29%). The
large age peaks are dened by the Mesozoic fractions at c. 100 Ma and
most abundant at around 200 Ma. Palaeozoic age peaks are well-dened
at 300 and 450 Ma (18%). Less abundant is the Precambrian population
at around 600 Ma (8%). Meso- and Palaeoproterozoic ages are compa-
rably rare (38%), although a group of ages at 1880 Ma is abundant
(Fig. 9).
In addition to our own data, we plot relevant published data from the
literature for comparison (Table 1, Fig. 12). Samples from modern Volga
alluvium (data from Allen et al., 2006; Wang et al., 2011) show one
single large Palaeozoic peak at 360 Ma (31%) and abundant but rela-
tively smaller Meso- and Palaeoproterozoic age fractions between 1000
and 2000 Ma (60%). Archean ages are present with 10% (Fig. 12). By
contrast, the modern Don sand sample from Wang et al. (2011) contains
only few grains of Palaeozoic ages (6%), whereas the Meso- and Palae-
oproterozoic fractions are abundant (82%). Archean ages are concen-
trated at around 2750 Ma and grains older than 2800 Ma are present (n
=2) (Fig. 12). The modern Dnieper sample (Wang et al., 2011) contains
more Palaeozoic and Late Proterozoic ages than the Don sample (13%).
These peaks are, however, less signicant than the abundant Meso- and
Palaeoproterozoic fractions between 1000 and 2000 Ma (82%). Archean
ages are almost absent (n =3) (Fig. 12). The detrital zircon U
Pb age
patterns of four sandstone samples from the Pliocene Caspian Productive
series (Kirmaky Suite and Balakhany Suite) sampled on the Apsheron
Peninsula are separately plotted and discussed in Allen et al. (2006),
who point out the great similarity of these Caspian Productive series
subsamples. As combined sample here, they show a large Palaeozoic age
fraction, abundant Meso- and Palaeoproterozoic ages, and several
Fig. 8. KDE plots of loess from the Pushkari site on the west Russian EEP (this study), and the Staiki, Vyazovok and Dnieprovskie sites from the Ukrainian EEP (data
from Pa´
nczyk et al., 2020).
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
zircons of Archean age (Fig. 12; Table 2). Detrital zircons from Pliocene-
Quaternary sedimentary rocks from the Taman Peninsula (sample
ILN#13_700) and from the Indolo-Kuban Basin at the foothills of the
Greater Caucasus (sample WC139/1, Table 1) are discussed in Vincent
et al. (2013). Both samples contain several Palaeozoic ages, which are
mostly concentrated around 250 Ma, and also show abundant Meso- and
Palaeoproterozoic populations in their age distributions. Archean ages
are scarce and most abundant at around 2600 Ma (Fig. 12). Allen et al.
(2006) also analysed a Middle Jurassic sandstone sample from the
eastern Greater Caucasus (Bajocian sandstone), which shows mostly
Phanerozoic zircons in its low-n analysis (n =60). The largest age
fraction peaks at 240320 Ma while ages older than 500 Ma are almost
entirely missing (Fig. 12).
A distribution of zircons mostly younger than 1000 Ma can be seen
for the detrital zircon data from different Neoproterozoic to Cenozoic
sandstone formations from the western and eastern Alborz mountains,
which are taken from Horton et al. (2008). Here, the individual low-n
samples (between 9 and 59 grains each) are grouped into one Alborz
sample in order to depict the detrital age distribution of the entire
orogen in one plot and to allow a more meaningful comparison with
other high-n data in this study. The KDE shows the largest peak at
around 600 Ma and the second highest at 360 Ma (Fig. 12). A very
Fig. 9. KDE plots of loess from the Aghband site on the north Iranian Loess Plateau as well as of marine Caspian sand from the Cheleken Peninsula and desert sand
from the Karakum Desert, Choganly site. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
different age distribution is observed for a sample from Siret River al-
luvium, taken from close to its conuence with the Danube by Ducea
et al. (2018). It shows a large peak at around 320 Ma and abundant Late
Neoproterozoic - Early Palaeozoic fractions forming two peaks at around
480 Ma and 620 Ma. Meso- and Palaeoproterozoic peaks are not as
signicant (Fig. 12). Cumulative U
Pb data from several rock samples
from the surrounding area, geologically representing the North Dobro-
gean unit (Fig. 2), come from Balintoni and Balica (2016) and show one
distinct peak at 600 Ma years, compared to which a Palaeozoic fraction
at around 320 Ma, and the numerous Meso- and Palaeoproterozoic age
populations are signicantly smaller (Fig. 12).
5. Discussion
5.1. Lower Volga loess and its temporal and spatial source variability
Overall, all LVL samples from the 3 different sites and from different
stratigraphic depths and ages show great similarity (Figs. 3 and 5). The
Fig. 10. KDE plots of Volga sand from Chorny Yar, Seroglazka and Raygorod, with the Chorny Yar sample being the oldest (~150130 ka) and the Raygorod sample
(RG5) the youngest (~60 ka) (Table 1, Fig. 1b).
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
Fig. 11. KDE plots of Don sand from the Liska site, Yergeni sands from the Yergeni uplands, Black Sea sand from the Eltigen section, and of colluvium from the
northern Greater Caucasus.
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
Fig. 12. KDE plots of relevant published detrital zircon data from sediments and sedimentary rocks in the study region. Published sources for these samples are given in Table 1. For colour coding see Fig. 1a, yellow
diamonds denote samples from diverse sedimentary rocks. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
360 Ma peak, which they all have in common, comprises zircon ages
corresponding to Variscan orogenesis. All samples also contain ages
between 460 and 420 Ma, corresponding to Caledonian orogenesis. Ages
between 540 and 630 Ma correspond to the Cadomian phase of
orogenesis and are most abundant in the samples RG4 (3%), but are
represented by a few grains in all other samples as well. Meso- and
Palaeoproterozoic as well as Archean age distributions are very similar
between all LVL samples. This indicates a similar provenance for Len-
insk, Raygorod and Srednyaya Akhtuba loess during different stages of
the same glacial epoch, which reects sources with Variscan and Cale-
donian orogenesis age zircon assemblages in addition to the dominating
Meso- and Palaeoproterozoic signal (>70% in each sample). However,
to determine whether the contributing sources changed during the time
of LVL deposition, Raygorod and Leninsk samples from different strati-
graphic depths are compared alongside random subsampling of grouped
site data (Figs. 3-5).
At Leninsk, the three samples (Fig. 1b) all show the same peaks
corresponding to Variscan and Caledonian orogenesis ages and major
Proterozoic age fractions at 1050 Ma, 1180 Ma, 13201400 Ma, 1500
Ma, 1630 Ma, 1780 Ma, 1880 Ma as well as in the Archean, suggesting
no obvious differences between them. Small changes can be observed
only in the height, rather than the presence, of these peaks. Peak height
differences, however, have to be interpreted with care due to the varying
number of analysed grains (Fig. 5). The random subsampling of grouped
Leninsk data reveals that all subsamples show a major peak at 360 Ma
and a smaller age fraction at 430 Ma (Fig. 4). These peaks are also clear
in the three original samples. This underpins the temporally stable,
signicant contribution of Variscan orogeny aged, and to a lesser extent
Caledonian orogeny aged sources to Leninsk (Figs. 4 and 5). Slight dif-
ferences are seen in the expression of Neoproterozoic ages between the
random subsamples, while the Meso- and Palaeozoic as well as the
Archean age distributions are essentially identical in terms of peak
presence. This observation suggests that a slight variability in the age
distribution of samples may be sampling induced, and comparison with
the stratigraphically different LN samples reveals that the random sub-
samples in fact show greater variability in age distributions (Figs. 4 and
5). As such, the slight age spectra differences between the Leninsk
samples by depth cannot be assigned to actual provenance changes. By
extension, this suggests that the source for loess at Leninsk did not
change signicantly from 65 to 30 ka ago (Fig. 1b), at least as is
observable from the applied technique.
Samples from Raygorod show more variation with stratigraphic
depth. Samples RG4 and RG3, deposited at ~40 ka and ~ 30 ka, are very
similar to each other, apart from the somewhat better-dened Caledo-
nian and Cadomian orogenesis age corresponding peaks in RG4 (Fig. 3).
In the Mesoproterozoic, RG4 also contains fewer grains of age c. 1300
Ma, while RG3 has a less abundant age fraction around c. 1400 Ma.
However, regardless the differing number of analysed grains between
the samples, this variability is comparable to that seen among the
random subsamples from the grouped Raygorod data, which differ only
in the peak height rather than the presence of certain age fractions
(Fig. 4). This suggests that the small differences in zircon age distribu-
tion between RG3 and RG4 are within the range of sampling induced
variation. However, RG2 (~20 ka) seems to show larger differences
compared to the older Raygorod samples. These are notable in the
absence of a 2800 Ma age fraction, the presence of a 2100 Ma age
fraction, and a less distinct 1600 Ma age peak, where instead several
large but diffuse age fractions occur between 1450 and 1900 Ma.
Furthermore, RG2 shows a lack of ages at c. 1300 Ma and a peak at c.
1400 Ma, the opposite of RG3 (Fig. 3). Thus, while these observations
may suggest an invariant dust source to Raygorod between 30 and 40 ka,
there may be some temporal variability in loess provenance during MIS
2 (RG2).
Comparing the three sites together reveals that all samples from
Leninsk and the two similar samples from Raygorod (RG3 and RG4)
share very similar characteristics. All samples have the same Variscan
and Caledonian orogenesis corresponding age fractions and the Meso-
and Palaeoproterozoic age distributions are overall very similar, with
differences lying within the range of sampling induced variation as
discussed above. The older loess sample from Srednyaya Akhtuba with
its depositional age of ~88 ka, shows an age distribution partly different
to this. Presuming a close similarity between the zircon age distributions
of the three LVL sites, the absence of Neoproterozoic ages between 640
and 800 Ma, as well as the absence of grains younger than 300 Ma and
Table 2
List of those samples that have been grouped together for further discussion, based on their similar zircon age distributions and veried by depositional age, geological
and geographic position. See sections 5.15.4 for detailed explanation.
Subsample Material Age Reference Combined sample
LN4 Loess 31.2 ±2.2 ka
This study LVL_2065 ka
LN5 Loess 56.1 ±2.8 ka
LN6 Loess 63.4 ±3.0 ka
Raygorod RG3 Loess ~30 ka
RG4 Loess ~40 ka
Kalaus Kalaus Loess MIS 32 This study Manych
Chograi Chograi Loess MIS 43
Staiki Loess MIS 2 Pa´
nczyk et al., 2020 EEP_Ukr
Vyazovok Loess MIS 2
Raygorod RG5 Fluvial sand ~60 ka This study Volga_15060 ka
Seroglazka Seroglazka Fluvial sand ~10085 ka
Chorny Yar Chorny
Yar Fluvial sand ~150130 ka
Chorny Yar Fluvial sand Modern Allen et al., 2006 Volga_modern
Volgograd downstream Fluvial sand Modern Wang et al., 2011
Liska Liska Fluvial sand Unknown Quaternary This study Don_Yerg Yergeni Yergeni Fluvial sand Pliocene
Don mouth Fluvial sand Modern Wang et al., 2011
Indolo-Kuban ILN#13_700 Sandstone Pliocene-Quaternary Vincent et al., 2013
Taman WC139/1 Sandstone Pliocene-Quaternary
Casp. Prod. series
Sandstone Pliocene Allen et al., 2006
Cauc_Coll Colluvium Modern this study Cauc_Jur
Bajocian Sandstone Middle Jurassic Allen et al., 2006
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
older than 2800 Ma, despite their overall scarcity in LVL samples,
indicate that the source input might have changed from MIS 5 to MIS 4.
Based on the analysis above, we combine the LVL data into three
groups showing similar provenance signature: 1) LVL younger than 20
ka (sample RG2); 2) LVL between 60 and 20 ka (samples RG3, RG4, LN4,
LN5, LN6), and; 3) LVL older than c. 80 ka (sample SA) (Table 2, Fig. 13).
We note that these groupings are approximate due to low sample
number, and we emphasize that we separate samples into different
groups only when there are clear differences in the age spectra (i.e.,
presence of zircon age populations) that seem to exceed the variation
seen in the subsampling experiments. In any case, overall, LVL zircon
Pb age distributions are very similar between samples, sites and
through time, indicating relatively constant sources. However, some
small changes are seen, particularly for Raygorod loess deposited earlier
than 20 ka and Srednyaya Akhtuba loess deposited before 80 ka,
compared to all other analysed LVL samples.
5.2. Loess from the East European Plain and the Caspian Sea region
Overall, remarkable variation in loess source is seen over the region
of the southern and western EEP, in the area of the Black Sea and the
Caspian Sea. This variation in zircon age distribution is apparent even
for closely located sites, suggesting strong variability in zircon age as-
semblages in different source regions, several processes of trans-
portation and reworking prior to loess deposition, and highly site/area
specic controls on aeolian sediment origins. Based on the characteris-
tics of their zircon age KDEs, and thus their sources, the analysed loess
samples are allocated to four EEP loess provinces and one Southeast
Caspian loess province (discussed below). While the exact extent of
these provinces can only be estimated from the results of this study,
these loess provinces will be used to assign specic sources in the further
discussion (5.6): The Southwest EEP province, South EEP province,
North Caucasus province, Southeast EEP province and Southeast Cas-
pian province (Fig. 14). The provenance of these loess provinces can be
clearly separated from that of ‘Carpathian foreland and lower Danube
Basin loess(CDB loess). A more detailed subdivision of this loess and its
difference to EEP loess based on provenance and transport modes is
discussed in e.g. Pa´
nczyk et al. (2020). The comparison of our results
with their data and with data from other European loess provenance
studies (Ujv´
ari et al., 2012) reinforce their ndings and allow separation
of the Southwest EEP province from this highly generalised ‘CDB loess.
This ‘CDBloess, particularly in the Danube Basin, shows abundant
Mesozoic to Neoproterozoic zircon ages, and at least south of the Car-
pathians and Bohemian Massif generally contains only minor or even
misses Meso- and Palaeoproterozoic age fractions (Ujv´
ari et al., 2012),
which are the signature age populations that all loess provinces of the
EEP, except the North Caucasus province, have in common. Below, the
groupings are discussed in terms of age spectra of the constituent sam-
ples, while later on in chapter 5.6, specic sources and source variability
are discussed.
5.2.1. The Southwest EEP province
This loess province comprises Russian and Ukrainian loess sites on
the southwestern EEP. Its characteristic zircon age assemblage widely
misses Mesozoic ages, shows small Palaeozoic age populations and
abundant Meso- and Palaeoproterozoic ages (>65%). From the loess
samples discussed in this study, the samples from the Pushkari, Staiki,
Vyazovok and Dnieprovskie sites all show this characteristic age distri-
bution pattern, and are relatively closely and similarly situated
geographically (Fig. 8). In addition, detrital zircon U
Pb data from the
Fig. 13. KDE plots of LVL, grouped according to their similarities and differences in age distribution. Sample RG2 is treated as LVL <20 ka, samples RG3, RG4, LN4,
LN5, LN6 are combined to LVL_2065 ka and sample SA represents LVL >80 ka.
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
Ukrainian loess site Cherepyn (Nawrocki et al., 2019) ts this prove-
nance signal too. The Ukrainian sections along the Dnieper (Staiki, MIS
2; Dnieprovskie, MIS 6) and close to its tributaries (Vyazovok, MIS 122;
Cherepyn, MIS 2), as well as the Russian Pushkari site (MIS 3) located
close to a tributary river of the Dnieper (Desna River), lie within the
drainage basin of the Dnieper, into which the Fennoscandian Ice Sheet
repeatedly penetrated during the Quaternary (e.g. Chugunny and
Matoshko, 1995; Gozhik, 1995). All four Southwest EEP province sites
show a similar zircon age distribution, particularly in regard to Meso-,
Palaeoproterozoic and Archean ages, but some differences are also seen.
While the Pushkari and Dnieprovskie loess samples do not show a higher
number of grains forming any broader Palaeozoic fraction and are
entirely missing ages corresponding to Variscan orogenesis, the northern
Ukrainian sections contain such an age fraction. Age populations cor-
responding to Caledonian orogenesis ages are present in all four sam-
ples, while ages corresponding to Cadomian orogenesis are nearly
missing from the Pushkari sample (Fig. 8). While these differences
suggest some provenance differences for the Pushkari loess compared to
the three Ukrainian sites and also for the northern Ukrainian sites
compared to the Black Sea site, the low number of analysed zircons for
most of these samples limits their direct comparison and the pinpointing
of specic sources of minor age populations. A comparison is further
complicated by the somewhat different depositional age of the discussed
samples. As such, they all are evaluated as representatives for the
Southwest EEP loess province (Fig. 14). The two strikingly similar
Ukrainian sites Staiki and Vyazovok are combined into one larger-n
sample for further discussion (Table 2). A similar zircon age KDE
pattern, characteristic for this loess province, is displayed in loess from
Beglitsa (~3025 ka). This site is located at the shore of the Azov Sea
close to the inlet of the Don. Its loess sample shows small Palaeozoic and
Late Neoproterozoic age populations, while its Meso- and Palae-
oproterozoic ages are abundant (79%) and dene several large peaks
(Fig. 6). As such, Beglitsa loess ts most closely into the Southwest EEP
loess province (Fig. 14).
5.2.2. The South EEP province
The main characteristics of loess from the South EEP province are a
large Palaeozoic double peak at 300 Ma and 440 Ma, the presence of an
age fraction at 600 Ma, abundant Meso- and Palaeoproterozoic ages, and
the presence of Mesoarchean ages. This zircon age distribution pattern is
shown by Eltigen loess, which is located at the tip of the Crimean
Peninsula right at the Kerch Strait, which connects the Azov and Black
seas. Despite the geographical proximity and the similar nature of the
Beglitsa and Eltigen sections, both cropping out along the coast, the
more southerly situated Eltigen loess section shows large Palaeozoic and
Neoproterozoic age fractions in contrast to the Beglitsa loess, although
the Meso- and Palaeoproterozoic age distributions are similar. At Elti-
gen, two peaks, dened by age populations at 300 and 450530 Ma
correspondent to Variscan (6%) and Caledonian (9%) orogenesis ages
respectively, are well expressed. There is also a signicant peak at c. 600
Ma, together with abundant Neo- and Mesoarchean ages, pointing to
differences to loess from the Southwest EEP province (Figs. 6 and 14).
The loess samples from the two sites Kalaus (MIS 32) and Chograi (MIS
43), located in the Manych depression, also t into the South EEP
province group, showing the characteristic abundant Palaeozoic peaks
at 300 Ma and 450 Ma, a well-dened age fraction at 600630 Ma, and
abundant Archean ages, in addition to the characteristic abundance of
Meso- and Palaeoproterozoic ages (Fig. 7). This allows us to outline the
South EEP province as shown in Fig. 14. Due to their very similar zircon
age distribution and the low-n analysis of the Chograi sample, these two
Manych depression loess samples are combined to one Manych sample
for further consideration (Table 2).
5.2.3. The North Caucasus province
This province differs from the previous ones mainly due to the low
abundance of Meso-, and Palaeoproterozoic aged zircons in its loess. Its
main characteristics also include a large peak at 300 Ma and a compa-
rably smaller major age population at 440 Ma. Located on the northern
foothills of the Caucasus and at a higher elevation than the close by loess
sites of the South EEP province, loess from the Budennovsk section
shows this characteristic zircon age distribution. While the large Varis-
can orogenesis aged peak at 300 Ma, as well as the age populations at
445 Ma and 600 Ma respectively, are similar to those observed for loess
from the Southeast EEP province, the Meso- and Palaeoproterozoic age
populations are small and Archean ages are nearly absent, which is
clearly different from all the other discussed loess samples from the EEP
Fig. 14. Loess provinces of the EEP and in the South Caspian Sea region classied based on the detrital zircon age provenance signal of the loess samples discussed in
this study. Note that the shape and extent of the loess provinces is only roughly estimated (yellow shade) and that the ‘CDBloess province is not further subdivided
(please see e.g. Pa´
nczyk et al. (2020), Ujv´
ari et al. (2012) for provenance discussions of the subdivision of this loess province). The locations of samples considered in
this study are marked, for legend see Fig. 1a. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of
this article.)
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
(Figs. 7 and 14).
5.2.4. The Southeast EEP province
The most remarkable difference between the Southeast EEP and the
Southwest and South EEP loess provinces is found in the zircon age
pattern of the Palaeozoic, while the Meso- and Palaeoproterozoic age
distributions are similar. Representatives of the Southeast EEP province
show a large Variscan orogenesis aged fraction that, in contrast to loess
from the South EEP and North Caucasus provinces, peaks at 360 Ma. The
LVL sites show this characteristic well (see 5.1; Fig. 14).
5.2.5. Southeast Caspian province
The age distribution of the Iranian Aghband loess section is clearly
different to all analysed EEP loess samples. Mesozoic, Palaeozoic and
Neoproterozoic ages are abundant, while Meso- and Palaeoproterozoic
ages are nearly absent (Fig. 9). Thus, and also due to its geographical
separation from the loess sites on the EEP, this section is taken to
represent a separated loess province southeast of the Caspian Sea
(Fig. 14).
5.2.6. Non-allocated loess sites
Loess from the Kalach site on the Don River (unknown Quaternary
age) represents a particular case, which does not allow its allocation to
any of the above described loess provinces. Despite its close proximity to
the LVL sections, and cropping out along the Don River, it shows some
remarkable differences in its zircon distribution. While the 360 Ma age
peak in the sample corresponds to the timing of the Variscan orogeny
just as for the LVL samples, and Meso- and Palaeoproterozoic, and
Archean ages are also similarly distributed, the Mesozoic age peaks
detected in the Kalach loess are absent in all other discussed samples,
and indicate a different source input at this site (Figs. 6 and 14). While
its Palaeozoic, Proterozoic, and Archean age distributions suggest an
afliation to the Southeast EEP province, its Mesozoic ages are not
compatible. As such, it is possible that the Kalach site is a representative
of yet another EEP loess province, or that some specic local input is
recorded at that particular site, and that it is not representative of any
larger province (Fig. 14).
5.3. River and sea sediments
In order to simplify the comparisons between regions for the later
discussion (5.5, 5.6) and to enable more meaningful comparisons where
samples have relatively low n, here we combine samples of similar types
of material from comparable catchments in cases where their zircon age
distributions are very similar. This applies to both original and pub-
lished data and is undertaken independently from the grouping of loess
data into loess provinces discussed above, and for which the river and
sea sediments might represent sediment sources.
All three Palaeo-Volga samples (Chorny Yar, Seroglazka, RG5;
Table 1) display highly comparable zircon age distributions, most
notably in the large Variscan orogeny aged population at around 360 Ma
(~10%) and the abundant Meso-, Palaeoproterozoic age fractions
(~75%). Here, only slight differences occur in the prominence and
relative abundance of some age peaks, for example the diffuse peaks of
the Chorny Yar sample between 1000 and 1350 Ma in comparison to the
well-dened double peak in the Seroglazka and RG5 samples (Fig. 10).
The striking overall similarity suggests no signicant changes in the
sedimentary load of the Palaeo-Volga through time and allows grouping
these three Palaeo-Volga samples into one Palaeo-Volga sample
(Volga_15060 ka; Table 2, Fig. 15). Differences between the Caledonian
orogeny aged fractions of the three Paleo-Volga samples (4% in RG5; 3%
in Seroglazka; 2% in Chorny Yar) are unlikely to be source diagnostic.
Due to the overall small number of Caledonian aged grains (n =6 in
RG5, 4 in Seroglazka, 2 in Chorny Yar), the probability of overlooking
this age population is high and dependent on the number of analysed
zircons per sample. Modern Volga sediment samples come from Wang
et al. (2011) and Allen et al. (2006), collected from the lower reaches of
the Volga downstream from Volgograd and the present oodplain near
Chorny Yar, the same outcrop from which our ~150130 ka old Palaeo-
Volga sand sample was taken (Table 1). The zircon age distributions of
these two modern Volga samples are very similar (Fig. 12), which are
therefore combined into one modern Volga sample (Table 2, Fig. 15).
The modern Volga shows some differences compared to the Palaeo-
Volga discussed above, most notably the low peak heights of Meso-
and Palaeoproterozoic age populations (64%) in comparison to the large
Variscan orogeny age fraction (20%) (Fig. 15). To compare, these age
populations account for 75% and 10% in the Palaeo-Volga samples.
Potentially, the observed variability from the Palaeo- to modern Volga
signals changes in the catchment or the course of the Volga from the
Pleistocene to today, but could also be driven by human modication of
the modern Volga River.
The KDE of the Palaeo-Don alluvium from the Liska site (MIS 43)
shows strong overall similarity to the Pliocene Yergeni sand sample.
However, apart from some minor variation in the prominence and
abundance of Proterozoic and Archean age populations, there is one
main difference; the well-dened age fraction at c. 360 Ma among the
overall low abundance Palaeozoic populations in the Yergeni sample is
missing from the low abundance Palaeozoic fraction of the Palaeo-Don
sample, where only 3 grains reect this age (Fig. 11). However, the
abundance of these Palaeozoic ages is still small in comparison to the
Meso- and Palaeoproterozoic age fractions. This pattern is clearly
different to Volga material and suggests that the provenance of the
Yergeni sample is probably more closely related to the source of the
Palaeo-Don than the Palaeo-Volga. It should also be considered that the
analysed Yergeni sample is considerably older (Pliocene age) than the
Palaeo-Don and Palaeo-Volga samples (late Quaternary). Furthermore,
while the Palaeo-Volga material does not show much variation regard-
less of its age and sampling site (Fig. 10), the Palaeo-Don sediment may
show some variability in zircon age distribution if it were of different age
or sampled further downstream. The analysed modern Don sample
published by Wang et al. (2011) comes from further downstream than
the Palaeo-Don sample, which allows some testing of this possible
geographical and temporal variation. Both Don samples show a similar
zircon age distribution and considering the overall rather small differ-
ences to the Yergeni sample, these three samples are grouped together
for further analysis (Don_Yerg; Table 2, Fig. 15).
The zircon age distribution of the modern Dnieper alluvium is similar
to the Don and Yergeni samples, but differs by the lack of Archean ages.
Furthermore, the age fraction coeval to Cadomian orogenesis is well
dened (Fig. 12). Further west, and different to all other discussed river
sediments, the detrital zircon U
Pb age distribution from the Siret River
sample from close to its inlet to the Danube shows strong peaks in the
Neoproterozoic and Palaeozoic, reecting the Cadomian, Caledonian
and Variscan phases of orogenesis in well-dened peaks (Table 1,
Fig. 12). All this signals starkly different sources for the Siret River
sediment compared to those supplying the rivers on the EEP. The Siret
River drains into the Black Sea via the Danube, yet the ~110 ka old
Black Sea sand from the Crimean Eltigen section shows little similarity
to the Siret sample and rather reects EEP input. The KDE of Eltigen
sand zircon ages is comparable to the Liska Palaeo-Don alluvium, with
small Phanerozoic and Neoproterozoic zircon components (12%), and
abundant primarily Meso- and Palaeoproterozoic ages (75%). However,
there are differences in the Palaeoproterozoic fractions older than 1800
Ma (Fig. 11). The Caspian Sea sand from the Turkmen Cheleken
Peninsula, in contrast, does not show similarities with any of the dis-
cussed EEP rivers, with Mesoproterozoic ages essentially absent, and
Palaeoproterozoic and Archean fractions greatly reduced, signalling a
provenance history different from the Black Sea and detached from the
EEP. This is in line with what is suggested from the grouping of the
Southeast Caspian loess province (Fig. 15).
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
Fig. 15. Comparison of the KDE plots of all samples that are discussed in this study, divided into geographical sections independent from their potential provenance (i.e. loess provinces, section 5.2). Some plots
represent combined samples from multiple samples/sites (see Table 2). For colour coding see Fig. 1a, yellow diamonds and curves denote samples from diverse sedimentary rocks. (For interpretation of the references to
colour in this gure legend, the reader is referred to the web version of this article.)
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
5.4. Other sedimentary material
As for the river sediments, zircon U
Pb data from the other samples
discussed in this study, including published data, is combined where
possible to facilitate sediment provenance analyses in the Black Sea -
Caspian Sea region. The modern Karakum Desert sand sample at Cho-
ganly and the Cheleken Caspian Sea sand sample (~10 ka) both show
very reduced Meso-, Palaeoproterozoic and Archean zircon age abun-
dance, which clearly contrasts to what is observed in EEP samples.
However, the two samples exhibit differences in their Palaeozoic and
Mesozoic age distribution (Fig. 9), further indicating differences in the
provenance of these two samples (Fig. 15). Located northwest of the
South Caspian Basin, a Middle Jurassic sandstone sample from the
eastern Greater Caucasus (Bajocian sandstone) displays a similar age
distribution to the high-n KDE of the local colluvium sample from the
Greater Caucasus (Cauc_coll). Both show a large late Variscan peak at
240320 Ma and contain very few grains of Meso- and Palae-
oproterozoic ages. While the two samples also share ages of 440 Ma,
differences occur as represented by a Mesozoic age fraction at 170 Ma in
the Bajocian sandstone, and by an age fraction of 600 Ma in the Caucasus
colluvium (Figs. 11 and 12). Given the great discrepancy in the number
of analysed grains for the two samples, these small differences may be an
artefact of the low zircon grain number of the Bajocian sandstone and
are thus not clearly indicative of real provenance differences. These two
Greater Caucasus samples are therefore combined to the sample
Cauc_Jurfor further discussion (Table 2, Fig. 15).
Further to the northwest in the northern foreland of the Greater
Caucasus and geographically close to the sampling location of the
Caucasus colluvium, detrital zircons from Pliocene-Quaternary sedi-
mentary rocks from the Taman Peninsula and from the Indolo-Kuban
Basin (samples ILN#13_700, WC139/1; Table 1) show similar KDEs to
each other, as well as to the combined sample from the Pliocene Caspian
Productive series (Kirmaky Suite and Balakhany Suite) from the
Apsheron Peninsula, located on the eastern lower reaches of the Greater
Caucasus, geographically close to the sampling location of the Bajocian
sandstone (Fig. 12). The only notable differences are observed for
Archean ages, which are present in the Indolo-Kuban and Caspian Pro-
ductive series sample but nearly absent in the Taman sample. The
Indolo-Kuban and Caspian Productive series samples also show a
somewhat greater abundance of Palaeozoic ages compared to Meso-,
Palaeoproterozoic age fractions, while Palaeozoic ages are relatively
sparse in the Taman sample. However, despite the small number of
analysed grains in the Indolo-Kuban and Taman samples, the zircon age
distributions are comparable. Given this, and also due to the samples
similar formation age, all three samples are combined to a set of 350
zircons (sample Plio-Quat; Table 2, Fig. 15). The sample derived from
the grouped Alborz sandstones, representing the detritus from the
Alborz mountain range south of the Caspian Sea, as opposed to the
Caucasus and its foreland, shows its largest age fraction corresponding
to the timing of Cadomian orogenesis. It also contains abundant Palae-
ozoic ages peaking at 360 Ma. As such, it signals a different provenance
to other discussed samples and is not further combined with any other
data (Table 1, Fig. 15).
5.5. River sediment provenance
While the similarities and differences in the zircon age distributions
of the samples discussed above already yield insights into their common
or diverse provenance, we now try to trace the specic proto-sources of
these river sediments. This can be done by correlating the zircon for-
mation ages reected in our samples with specic tectonic events in
Eurasia. To do so, it is crucial to consider the geographic relationships
between the discussed sediment deposits in the Pleistocene and the
environmental conditions during their deposition, as well as the geology
of their catchment. The Fennoscandian Ice Sheet as well as mountain
glaciations of the Urals, Caucasus, and to some extent the Carpathians
affected the drainage and likely also the sediment supply of the large
rivers of the EEP draining into the Black Sea- Caspian Sea region (e.g.
Grosswald, 1980; Fedorov, 1971; Tudryn et al., 2016). As the largest of
the EEP rivers, the Volga mostly drains the Archean EEC block of Volgo-
Uralia, eroding its exposed basement and sedimentary cover, but also
small parts of the Archean EEC block of Sarmatia and Fennoscandia and
the Ural Mountains. Potentially some drainage also comes from areas
underlain by the Timan basement (Fig. 2). The course of the Volga
stayed broadly the same in the Quaternary, with no remarkable changes
in the drainage basin of the Upper Volga, although the lower reach of the
palaeoriver, south of Volgograd, is believed to have migrated further
west than today (Kroonenberg et al., 1997). Not much is known about
the drainage area of the Pliocene Yergeni River, but based on the dis-
tribution of the Yergeni formation within the drainage basin of the lower
reaches of the Don, it is assumed that its lower course might have been
similar to the one of the Pleistocene modern Don (Zastrozhnov, 1991;
Fig. 2). The Don River mostly drains Sarmatian basement and its sedi-
mentary cover, including the exposed Archean crust of the Voronezh
Massif (e.g. Shchipansky and Bogdanova, 1996). Parts of the Scythian
Platform basement and the Northern Caucasus are also within its
drainage area. The drainage basin of the Dnieper comprises mostly
Fennoscandian and Sarmatian basement and sedimentary cover, where
it erodes into the extensive exposure of Archean crust in the Ukrainian
Shield on its way to the Black Sea (Fig. 2). With this background in mind,
the specic age peaks seen in the different river systems (Figs. 10, 11 and
12, Table 1) are now discussed in chronological order, starting with
Archean peaks. Proto-source provenance assignments for the river sed-
iments are made on that basis.
Old Archean crust crops out in the Voronezh Massif and Ukrainian
Shield of the Sarmatian province, but exposed Archean crust older than
2800 Ma is absent from Volgo-Uralia and the part of Fennoscandia that
comprises the EEP. Indeed, the scarcity of this age fraction in the Palaeo-
and modern Volga samples therefore reects well the geology of its
drainage basin (Figs. 2 and 15; e.g. Bogdanova, 1986; Gorbatschev and
Bogdanova, 1993; Puchtel et al., 1998; Bogdanova et al., 2008). This is
also the case for the modern and Palaeo-Don (Liska) samples, which
contain more grains of >2800 Ma than the Volga sediment (11 out of
300 compared to 9 out of 908) and reect the presence of Mesoarchean
crust (Voronezh Massif) within the Don River basin (Fig. 11). Even the
metamorphic overprinting of this old crust at 2.8 Ga (Shchipansky and
Bogdanova, 1996) might be discernible in the Palaeo-Don alluvium
sample. By contrast, only 4 grains show such old ages in the Yergeni
sample, spread between 2800 and 3250 Ma. This suggests that the
Voronezh Massif does not lie within the Yergeni drainage basin (Fig. 11).
Surprisingly, and as already noted by Wang et al. (2011), the modern
Dnieper sample contains only few grains of Archean age (n =3), despite
draining the Ukrainian Shield (Figs. 2 and 12). While the low-n detrital
zircon analysis of only one sample limits meaningful interpretation, this
can potentially be explained as a result of articial sediment trapping.
The construction of dams along the river might cause larger grains from
a closer source to settle preferentially in reservoirs upstream of the
dams. To test whether this is the case, a comparison with Palaeo-Dnieper
material from a similar location as the modern sample would be needed.
The sediments of all these EEP rivers have large ranges of Proterozoic
ages in common, marking the polyphase evolution of the EEC during the
assemblage of the three Archean provinces noted above (Bogdanova,
1993). The KDE of the Paleo-Don alluvium shows a weak but well-
dened peak at 2000 Ma, which is absent from the Volga, the Yergeni
and the Dnieper samples (Figs. 10, 11 and 12). These ages might reect
the collision of Volgo-Uralia and Sarmatia, affecting the north-eastern
margin of Sarmatia, an area within the Don River basin (Fig. 2; Bog-
danova et al., 2008). The large age fraction at 1800 Ma, distinguishable
in all river samples, is likely to represent the collision of Volgo-Uralia -
Sarmatia with Fennoscandia and formation of the EEC (Bogdanova
et al., 2006, 2008; Oczlon, 2006). A smaller but still well-dened age
population around 1900 Ma is present in the Volga, Don and Yergeni
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
material but absent from the Dnieper, and likely to derive from the
orogeny that joined the two provinces Kola and Karelia in building
Fennoscandia (Lapland-Kola orogen; Kuznetsov et al., 2014), the re-
mains of which seem to be more abundant in basement of the north-
eastern EEP. The abundant age populations between 1600 and 1500
Ma in all samples probably represent a phase of multiple orogenic events
and related intracratonic magmatism affecting the EEC (Bogdanova
et al., 2008). Ages between 1400 and 1300 Ma are abundant in the Volga
and Yergeni sediments but less evident in the Don and Dnieper samples
(Figs. 10, 11 and 12). These ages coincide with sill intrusions and
volcanism in the South Urals (Alekseev, 1984; Bogdanova et al., 2008).
Given that the Urals are within the drainage basin of the Volga, but not
drained by Don and Dnieper, this suggests that this age fraction repre-
sents Uralian input. As such, the Yergeni River basin might also have
included parts of the Urals. The broad age fraction between 900 and
1200 Ma, found in all river samples, matches reasonably well with the
Grenville-Sveconorwegian orogeny (e.g. Bingen et al., 2003; Bogdanova
et al., 2008). However, there are no geological units of this age in the
catchments of the EEP rivers, so zircon grains of this age would poten-
tially have been transported into their drainage basin by the Fenno-
scandian Ice Sheet during the Pleistocene glaciations. The scarcity of
grains between 900 and 600 Ma for all these rivers matches with the
general absence of these ages in their EEP catchment basins and shows
that grains of this age were also not transported into their drainage areas
from any surrounding geological terranes. In contrast, some river sam-
ples contain ages corresponding in age to Cadomian orogenesis, while
others are lacking such a signal. In the Palaeo-Volga samples, an age
fraction around 580 Ma might represent grains deriving from the
Timanides in the area of the north-eastern margin of the EEC, an area
unique on the EEP in yielding Cadomian age grains (Kuznetsov et al.,
2007, 2010). That this signal is almost absent from the modern Volga
samples can possibly be explained with the control of the Fennoscandian
Ice Sheet on sediment reworking and transport, e.g. resulting in higher
Volga river discharge than today (Fig. 15). The Timan origin of these
ages appears possible also because the Don, Yergeni and Dnieper sedi-
ments, which do not drain the Timanides, contain considerably fewer
corresponding ages (510580 Ma) (Figs. 2, 11 and 12).
The Caledonian and coeval phases of orogenesis are not very well
reected in any of the EEP river sediments. Of all samples, the Palaeo-
Volga and the Yergeni contain the most zircons of corresponding ages
(Figs. 10 and 11). Safonova et al. (2010) argue that all Palaeozoic
detrital zircon ages in their Volga sample derive from the Urals and
suggest their formation in the collision of the Kazakhstan block with the
Uralian arc-trench system in the Early Carboniferous, with magmatism
onset as early as 500 Ma ago (Nikishin et al., 1996). Alternatively, the
origin of these grains could also lie in the northwest, where the Fen-
noscandian Ice Sheet penetrated into the drainage basins of the EEP
rivers during Pleistocene cold stages, supplying reworked sedimentary
material of the EEC, potentially including the Baltic Shield (Figs. 2 and
15). However, Caledonian orogenesis aged zircons are also relatively
sparse in Danish till sediments derived in part from the Scandinavian
Caledonides, despite these tills lying much closer to this proto-source
than the Lower Volga (Knudsen et al., 2009).
A large single peak corresponding in age to Variscan orogenesis
stands out for the Volga samples. While this peak is as large as the
abundant Meso- and Palaeoproterozoic ages for the Palaeo-Volga, the
modern Volga material contains fewer Proterozoic zircons in compari-
son to this remarkably well dened Variscan peak (Fig. 15). Variscan age
zircons are widespread in volcanic rocks in the southern Urals (Puchkov,
1997, 2009 and references therein). Since the Variscan signal is not
signicant in the age distribution of the other EEP rivers, which do not
include the Ural Mountains in their drainage basins, this can be seen as
strong evidence for Uralian sediment input to the Volga. The only river
material for which a Variscan orogenesis age fraction is slightly
enhanced compared to the other Palaeozoic age populations is the
Yergeni sample. This once again indicates some subordinate drainage
from the Urals by the Pliocene Yergeni River (Fig. 11). The observed
difference in the abundance of Meso- to Palaeoproterozoic and Variscan
ages in respect to each other between modern and Palaeo-Volga, might
be explained with the construction of large dams Volga upstream, which
may lead to modern sediment samples from the lower reaches of the
Volga not being representative of the natural drainage contribution. It is
unclear how this may have affected the palaeo and modern sediment
distributions but if the Ural Mountains-derived grains pass fewer dams
compared to EEC derived material, then these grains may be over-
represented in Volga sediment from lower reaches. Post-Variscan zircon
ages are widely missing from the samples of all EEP rivers. Given the
above discussed evidence for partly Uralian provenance of the Volga
sediments, this is somewhat surprising for the Volga samples, since the
later stages of the Uralian orogeny occurred in the Mid-Carboniferous to
Permian. This stage resulted in the formation of the Uralian Main
Granite Axis, which extensively crops out especially in the central and
southern Urals (Puchkov, 1997). Allen et al. (2006) explain the lack of
these Late Uralian orogeny age signals in their Volga sample with a
drainage pattern effect. Rivers draining the Main Granite Axis are
directed eastwards rather than towards the Volga basin to the west of the
Urals. Safonova et al. (2010), however, argue for the contribution of
these granites to the sedimentary load of the Volga based on their
detrital zircon age data. Our data suggest that this is the case at least for
the Palaeo-Volga, for which several mid-Carboniferous to Permian ages
are present (Fig. 10). Whether these differences represent shifts in up-
stream river drainage is currently unclear.
Overall, this analysis shows that the zircon age composition of the
Volga samples is consistent with derivation from sediments of the Urals
and especially the EEP, the latter in turn being fed by sediment sources
in the Volga-Uralic, Sarmatian and Fennoscandian basement. A similar
geological catchment can be inferred for the Yergeni River sample but
the Uralian input appears signicantly smaller. The Don does not seem
to carry any Uralian material and the main source for its sedimentary
load lies on the EEP as well, including more recycling of material
deriving from the Samartian block than seen for the Volga and the
Yergeni rivers. The Dnieper sample also shows reworking of the EEP
basement and cover sequences, reecting mostly Fennoscandian sources
and surprisingly little Archean material from Sarmatia. In contrast to all
these rivers, which have the largest part of their drainage basins on the
EEP, the drainage basin of the Siret River mostly comprises the Carpa-
thians (Ducea et al., 2018). This difference is also reected in its zircon
age distribution, with a diagnostic Neoproterozoic to Palaeozoic age
pattern, in contrast to the zircon age spectra of the EEP rivers, where
instead Neoproterozoic age fractions are widely missing (Fig. 15).
5.6. Loess provenance and Black Sea - Caspian Sea sink regions
While for the loess deposits of the three EEP loess provinces
(Southwest, South and Southeast) a provenance connection to the EEP
and its large rivers seems likely, the situation for the North Caucasus and
Southeast Caspian loess provinces is complicated by the inuence of the
surrounding mountains and seas and the active tectonics in the region
(Fig. 14). Particularly in respect to the topography of the Caspian Sea,
the provenance of the depositional environments in the Southeast Cas-
pian region needs to be discussed separately from those on the EEP.
5.6.1. The EEP and Caucasus
The Southwest, South, Southeast EEP loess provinces, the uncate-
gorized province around the Kalach site, and the North Caucasus prov-
ince, all indicate stark source variability across the EEP (Fig. 14). The
LVL samples (MIS 5- MIS 2) represent the Southeast EEP province region
dened in chapter 5.2, and these samples all show the same distinct
Variscan orogeny aged Palaeozoic zircon population and wide range of
Precambrian ages as the analysed Palaeo-Volga material (Fig. 15). The
detrital zircon ages therefore suggest Volga River sand to be the direct
near-source material for LVL. This is supported by the locality of the
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
loess sections along the Volga River and their stratigraphy showing
Volga sands directly underlying the loess sequence in the section
(Fig. 1a). Furthermore, such a relationship is also suggested by the
character of the loess being relatively coarse grained with similar grain
surface features as Volga River sand (K¨
oltringer et al., 2021). Thus, LVL
is dominated by glacially derived material transported in a multi-step
mode via the Volga River. In addition to the contribution of EEC ma-
terial, a proportion of the sediment has a mountain origin in the Urals,
transported to the Caspian lowland via tributaries to the Volga River and
this signals the importance of both continental and mountain glaciations
in the formation of LVL. However, as discussed above, there is a degree
of variability between age distributions of some LVL samples potentially
indicating the inuence of other sources additional to the Volga, or that
the Volga was more variable in its sediment load than is indicated by the
samples analysed here. For example, the Khazarian transgression of the
Caspian Sea (MIS 5; Tudryn et al., 2013; Shkatova, 2010) could have
provided a pathway for additional detritus sourcing the LVL after its
reworking by the post Khazarian Lower Volga (sample RG5). Stronger
westerlies during the cold phase of loess deposition (K¨
oltringer et al.,
2021could be another contributing factor for the observed slight vari-
ability (Fig. 15). Despite any differences in the age distribution of LVL of
different depositional age, no major shift in provenance is detected as no
exotic and specic source-diagnostic age populations are found. That
some phases of the multiphase EEC evolution are more abundant in
samples of different age in the LVL could simply reect ice sheet uc-
tuations and changes in drainage patterns or wind strength.
In addition to the loess deposits along the Lower Volga, the North
Caspian Basin represents another sedimentary sink in this region and
ought to show a similar provenance to the Volga River, which directly
drains into the shallow North Caspian Sea. Unfortunately, this cannot be
tested here, as no marine coarse siliciclastic material from the North
Caspian Sea could be analysed for this study, and no data is found in
While the LVL is clearly Volga derived, the source of loess to the
Kalach site on the Don River is less clear. As such, we do not include it in
the Southeast EEP loess province (Fig. 14). The Proterozoic and Archean
age distribution in the loess from the Kalach site indicates reworking of
material from the EEC and the clearly dened and abundant Variscan
orogeny aged Palaeozoic fraction is identical to the one found in Volga
River sand and LVL, which suggests Uralian input via the Volga. This
points to the Volga as a likely source but because Don and Volga River
material display similar age distributions for their Proterozoic and
Archean grains, it could also be that Don material provides the Pre-
Cambrian ages, and the Palaeozoic and Mesozoic fractions reect the
input of an unknown source. Alternatively, both the Volga and the Don
could contribute as sources to the Pre-Cambrian age populations, while
only the Volga contribution brings the Variscan orogeny aged signal
from the Urals, and the Mesozoic age populations derive from an un-
known third source. As such, it is hard to tell how many and which
sources contribute to the Kalach loess site, especially since the origin of
its Mesozoic age fractions remains enigmatic. These striking age pop-
ulations at 100 and 180 Ma cannot be found in any other of the analysed
loess or secondary source material in the Northern Caspian lowland
(Fig. 15), but might reect some local yet unknown sediment input,
which interestingly enough did not inuence the closely located LVL
(70 km east; Fig. 3). In the wider region, ages of this range are only
known from the southern Transcaucasus and Lesser Caucasus (Rolland,
2017 and references therein) but transport from these far-off regions
seems implausible. The location of the section with respect to the course
of Palaeo-Volga and Palaeo-Don makes provenance from both rivers
likely, yet the example of Kalach loess shows that it is not only large
rivers that appear to have control on loess sediment sources in the region
(Fig. 14). In fact, the stark spatial variation of loess on the EEP seems to
be a function of the input from multiple diverse sources, as discussed
further below.
Less source diversity is indicated by the Southwest EEP loess
province (Fig. 14), which characteristically shows small Palaeozoic age
populations and abundant Meso- and Palaeoproterozoic ages. The
Beglitsa loess section, located on the shore of the Azov Sea, close to the
mouth of the Don River, shows great similarity to the Don-Yergeni
material, which is likely transported to the Azov Sea by the Don River
(Fig. 15). Mesozoic ages are missing from both samples, and the Palae-
ozoic age distribution is almost identical, as are the Proterozoic age
populations within the range of 9002000 Ma. The peak at 1380 Ma,
which indicates a Uralian source signal, suggests additional input from
the Yergeni formation and likely not only through erosion by the Don.
Smaller rivers, which drain the Yergeni uplands and contribute to the
Don closer to its mouth than where the Palaeo-Don sample in this study
is located (Liska section), likely supplied more Yergeni material for
aeolian transport to the Beglitsa site. As such, this suggests that Palaeo-
Don alluvium, transporting material from the EEC and eroding then
transporting Yergeni sediment, represents the main aeolian near-source
for Beglitsa loess. Furthermore, Mesoarchean ages (>2800 Ma, n =4) in
the Beglitsa sample are likely to come from the Voronezh Massif in the
Don basin (Fig. 2). However, ages at around 2100 Ma require another
explanation. These ages correlate with the timing of the Volgo-
Sarmatian orogeny, which is not reected in the Don-Yergeni sample
(Fig. 15).
The loess from the Ukrainian sites Staiki and Vyazovok (combined to
sample EEP_Ukr; Table 2), Dnieprovskie, and the Russian Pushkari site
on the southwestern EEP are also mainly fed by the EEC and its sedi-
mentary cover, as suggested by their great similarity to the modern
Dnieper sediment, with mostly abundant Meso- and Palaeoproterozoic
ages (Fig. 15). This observation reinforces the previously proposed
provenance of Ukrainian loess in this province (Pa´
nczyk et al., 2020).
The low number of zircons in the Pushkari sample does not allow for
detailed comparison, but its location along one of the Dniepers tribu-
taries suggests sediment input from a similar geological catchment. The
most notable difference compared to the modern Dnieper River sedi-
ment and the Ukrainian loess (including Cherepyn; Nawrocki et al.,
2019), however, is that the youngest age fraction in the Pushkari loess is
of Silurian age, while the other samples also contain a limited amount
(~3%) of younger zircon populations (Fig. 15). This Silurian age frac-
tion might arguably be of Scandinavian source, however, these ages are
also present in the Podolia sedimentary succession of Lower Devonian
depositional age that largely crops out in the valley of the Dniester River
and its tributaries (Schito et al., 2018; Kozłowska, 2019). Considering
that a Carpathian origin of these Palaeozoic ages is not plausible for the
Pushkari sample, the succession of the Ukrainian Podolia sedimentary
basin represents a more likely contributing provenance to the loess sites
within the Dnieper drainage basin. Regarding younger Palaeozoic ages,
their number in the Ukrainian loess samples and the Dnieper River
sample are low and in all cases these ages are rather spread and do not
dene larger age fractions (Figs. 8 and 12). However, the presence of
young Palaeozoic ages, entirely missing in the Pushkari loess sample
located further to the northeast, might indicate some reworking from the
Carpathians, where Variscan metamorphism and subordinate magma-
tism in the range of ~350250 Ma occurs in different tectonic units (e.g.
agus¸anu and Tanaka, 1999; Medaris et al., 2003; Ducea et al., 2018).
Additionally, the Siret River sample, the Danubes largest tributary,
which is sourced by the Carpathians and East European foreland (Ducea
et al., 2018), shows a great abundance of Variscan age grains (~325 Ma;
Fig. 15). Overall, while the abundant Meso- and Palaeoproterozoic ages
indicate that reworked glaciouvial Fennoscandian Ice Sheet deposits
from Dnieper alluvium are likely to represent the biggest source to the
northern Ukrainian loess sites (Buggle et al., 2008; Pa´
nczyk et al., 2020),
additional input from the Podolia sedimentary basin and the Carpa-
thians seems to also occur, and indeed might be greater to the Dnie-
provskie site on the Black Sea due to its relatively high Cadomian
orogeny age peak, an age fraction, which is also abundant in the Siret
sample (Fig. 15). Furthermore, Podolia and the Carpathians also
represent a major sediment catchment for loess on the Ukrainian Black
C. K¨
oltringer et al.
Global and Planetary Change 209 (2022) 103736
Sea west coast (Roxolany; Nawrocki et al., 2018).
In summary, loess from the Southwest EEP province is sourced
almost entirely from the EEC and its sedimentary cover, including some
Palaeozoic input (Urals in the case of Beglitsa loess, Carpathians and
Podolia for the more western sites), and no larger additional sources
from other geological domains can be identied. The two large rivers,
the Don and the Dnieper, provide the vast majority of material to loess
sites in the Southwest EEP province.
Rivers are important also for the provenance of loess sites in the
South EEP province, located in the northern foreland of the Crimea-
Caucasus mountain belt between the Black Sea and Caspian Sea area
(Manych and Eltigen; Figs. 1a and 14), although a greater source di-
versity than for the two previously discussed loess provinces is indi-
cated. The main differences to these provinces are the large Palaeozoic
peaks and a smaller but well-dened late Neoproterozoic to early
Cambrian peak (Fig. 15). The Meso- and Palaeproterozoic age distri-
butions instead show strong similarities to LVL and as such to the large
EEP rivers such as the Volga, Don and Dnieper, which in turn indicates
reworking of EEC material. The potential supply route for this EEC
material may be the Paleo-Don, considering the geographic relationship
between the discussed loess sites of the South EEP province and the
courses of the rivers, as well as Archean ages >2800 Ma shown in both
the Manych (combined) and Eltigen loess (Fig. 15). These ages indicate
Sarmatian crust as proto-source. The Ukrainian Shield borders the Cri-
mea to the north and is likely to supply these Archean zircons to the
Eltigen site. The Manych sites, however, lie further away from where
this Archean crust crops out so zircons of these ages may have been
incorporated into the loess via the Palaeo-Don River through drainage of
the Voronezh Massif instead (Fig. 2). However, as mentioned above, the
distribution of ages <900 Ma is characteristic to loess from the South
EEP province and cannot be linked to the EEC (Fig. 15). While the
Palaeozoic ages have been explained with provenance from the Urals
and Carpathians for the Southeast and the Southwest EEP loess prov-
inces above, these sources are unlikely to provide such strong, well
dened peaks for the South EEP province.
In order to be able to explain the origin of these age populations for
the South EEP province, the geological set-up in its area needs to be
discussed rst. The Caucasus and the Crimean Mountains are part of the
same Variscan orogen, with parts of it cropping out on the west coast of
the Black Sea in the form of the North Dobrogean orogen (Fig. 2;
Balintoni and Balica, 2016). The geological basement, which was
uplifted during this Variscan orogenesis, is dominated by Neo-
proterozoic U
Pb ages (560610 Ma; Ducea et al., 2018). This is re-
ected in the zircon age distribution in the cumulative sample from
North Dobrogea (Fig. 15). Apart from the Variscan ages, there is also
Lower Jurassic magmatism reported at least for the Crimean Mountains
(Meijers et al., 2010). The northern foreland to the North Dobrogea-
Crimea-Caucasus mountain range is geologically represented by the
Scythian Platform, which is composed of undeformed Palaeozoic sedi-
mentary rocks (Fig. 2; Okay and Topuz, 2017). Little is known about the
basement of the Scythian Platform but e.g. Kazantsev (1982) report
mainly Meso- and Neoproterozoic ages. The basement of the Scythian
platform is widely covered by sedimentary rocks such as the ones
sampled from the Indolo-Kuban Basin and the Taman Peninsula
(Table 1, Fig. 12; part of the Plio_Quat sample, Table 2), for which the
Scythian Platform and EEC represent the main sediment sources (Vin-
cent et al., 2013). For the southwest of the Scythian Platform, the
occurrence of some Early-Middle Triassic volcanic rocks are reported,
and these are also sporadically found along the eastern zone of the
Manych depression (Tikhomirov et al., 2004).
The Eltigen loess section is located on the southern tip of the Crimean
Peninsula, and the Manych sites on the EEC block of Sarmatia, right at
the border of the Scythian Platform. Ages of 550670 Ma are likely
derived from the Neoproterozoic basement of the Scythian Platform
(Figs. 2 and 15). Other potential sources for these ages to the loess are
absent in the proximal area. Some plutonic rocks of corresponding age
are indeed reported from the Caucasus, but are mostly found in the
Transcaucasus, which drains to the south of the orogen (Gamkrelidze
et al., 2011). The abundant Palaeozoic ages, in contrast, are likely to be
of Caucasian origin. As discussed earlier, there is no proximal source for
ages corresponding to the Caledonian orogenesis on the EEP and the
occurrence of such ages in its sedimentary sequence is explained by
input from either the Scandinavian Caledonides or the Urals. Such an
origin, however, seems unlikely for the large corresponding Palaeozoic
age population in the loess in the South EEP province. Also the Scythian
Platform was not affected by the Caledonian or any coeval orogeny. By
contrast, corresponding Palaeozoic ages are widely reported from the
Caucasus (Somin et al., 2007) and the c. 460 Ma peak observed for
Eltigen and Manych loess is also present in the sedimentary rocks
derived from the Caucasus (Cauc_Jur; Fig. 15).
The large Variscan orogenesis aged Palaeozoic peak in the Eltigen
and Manych loess is younger than the corresponding Uralian signal
observed in the Volga River and LVL samples. These Palaeozoic ages
more closely correspond to the crystallization of granites in the Greater
Caucasus, which are dated to ~300 Ma (Hanel et al., 1992; Somin, 2011;
Fig. 15). While this reects provenance from the Greater Caucasus, the
relatively abundant ~235250 Ma ages, poorly expressed in the Cau-
casus, in both samples might represent the Early-Middle Triassic vol-
canic rocks from the East Manych and southwestern zone of the Scythian
Platform (Tikhomirov et al., 2004). However, the sporadic occurrence of
these rocks in the region is inconsistent with the number of corre-
sponding ages in the loess samples and corresponding Early-Middle
Triassic ages are also largely present in the Budennovsk loess sample
(Fig. 7). Even though Lower Jurassic magmatism is reported from the
Crimean Mountains, Eltigen loess does not contain zircons of this age, in
contrast to the Manych and Budennovsk loess, which contain a few
Lower Jurassic aged grains. This suggests that these Mesozoic ages might
nevertheless derive from the Caucasus, where several thermal events are
recorded from the Permian to the Late Cretaceous (e.g. Rolland et al.,
2016) Indeed, another sample with a signicant number of these
Mesozoic ages is the Pliocene age Caspian Productive Series (Fig. 12).
Allen et al. (2006) inferred that this sedimentary succession is sourced
from a combination of EEC material, supplied by the Palaeo-Volga, and
material from the Greater Caucasus. Since no Mesozoic ages are supplied
by the Palaeo-Volga, this signal therefore seems to derive from the
Caucasus (Fig. 15).
Based on the discussion above, it can be concluded that Manych and
Eltigen loess from the South EEP loess province (Fig. 14) are mainly
derived from material eroded from the EEC via the Don River, as is re-
ected by the Archean and Meso-, Palaeoproterozoic age fractions. In
addition, they are also partially sourced from the Greater Caucasus,
which contributes the Palaeozoic and Mesozoic ages. Also, the Scythian
Platform represents an additional source of material to the loess sites;
mainly Neoproterozoic ages. As such, the South EEP province also seems
to be controlled by multiple diverse sources.
The Northern Caucasus loess province might have a less diverse
provenance than suggested for the EEP loess provinces. The sample from
the Budennovsk section appears to have its main sediment source in the
Caucasus. This is indicated by its Palaeozoic age peaks at ~445 and ~
300 Ma, which correspond to the formation of magmatic rocks in the
Greater Caucasus, as discussed above for loess from the South EEP
province (Fig. 15). Also, the low abundance of Precambrian ages in-
dicates a dominant Caucasus provenance since the samples from the
Greater Caucasus (Cauc_Jur) do not show Meso-, Palaeproterozoic and
Archean ages (Table 2, Figs. 12 and 15). This is a distinct difference to
the EEP-sourced loess and indicates short-distance aeolian transport in
the region, likely controlled by local winds. Even though mountain
glaciation plays a role in the Greater Caucasus, tectonic inuence might
enhance erosion and sediment supply in the Caucasus as well. Despite its
close proximity to the Manych sites, the age spectra at Budennovsk is
remarkably different but matches its different topographic position up
on the northern Caucasian foothills, whereas the Manych sites are
C. K¨
oltringer et al.