lable at ScienceDirect

Quaternary Science Reviews 178 (2017) 1e13
Contents lists avai
Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev
An early Brunhes (<0.78 Ma) age for the Lower Paleolithic
tool-bearing Kozarnika cave sediments, Bulgaria

Giovanni Muttoni a, b, *, Nikolas Sirakov c, Jean-Luc Guadelli d, Dennis V. Kent e, f,
Giancarlo Scardia g, Edoardo Monesi a, b, Andrea Zerboni a, b, Enzo Ferrara h

a Dipartimento di Scienze della Terra 'Ardito Desio', Universit�a degli Studi di Milano, Via Mangiagalli 34, I-20133 Milano, Italy
b ALP - Alpine Laboratory of Paleomagnetism, Via Luigi Massa, 6, I-12016 Peveragno, Italy
c National Institute of Archaeology and Museum of Bulgarian Academy of Sciences, 2, Saborna Street, 1000 Sofia, Bulgaria
d PACEA-UMR5199 CNRS, Universit�e de Bordeaux, All�ee Geoffroy Saint-Hilaire, Bâtiment B18, CS50023, 33615 Pessac Cedex, France
e Earth and Planetary Sciences, Rutgers University, Piscataway, NJ 08854, USA
f Lamont-Doherty Earth Observatory of Columbia University, Palisades, NY 10964, USA
g Universidade Estadual Paulista (UNESP), Instituto de Geociências e Ciências Exatas, Rio Claro, SP 13506-900, Brazil
h Istituto Nazionale di Ricerca Metrologica, Nanosciences and Materials Division, INRIM, Torino 10135, Italy
a r t i c l e i n f o

Article history:
Received 13 September 2017
Accepted 25 October 2017
Available online 7 November 2017

Keywords:
Kozarnika
Bulgaria
Pleistocene
Lithic tools
Magnetostratigraphy
Brunhes
Matuyama
* Corresponding author. Dipartimento di Scienze d
versit�a degli Studi di Milano, Via Mangiagalli 34, I-20

E-mail address: giovanni.muttoni1@unimi.it (G. M

https://doi.org/10.1016/j.quascirev.2017.10.034
0277-3791/© 2017 Elsevier Ltd. All rights reserved.
a b s t r a c t

We present a new sedimentological profile and a magnetostratigraphy of the tool-bearing Kozarnika cave
sediments from Bulgaria. Modal analysis of cave infilling sedimentary texture indicates that the tool-
bearing layers contain a sizable fraction of sediment interpreted as loess. We also find evidence for a
relatively thick and well defined normal magnetic polarity in the upper-middle part of the section
interpreted as a record of the Brunhes Chron, followed down-section by reverse polarity directions
interpreted as a record of the Matuyama Chron. The lowermost levels with Lower Paleolithic tools
(Layers 13aec) lie in the early Brunhes at a nominal maximum age of ~0.75 Ma, while the Brunhes
eMatuyama boundary (0.78 Ma) falls in Layer 13 Lower immediately below. This finding represents a
conspicuous revision of previous age estimates for the same tool-bearing layers.

© 2017 Elsevier Ltd. All rights reserved.
1. Introduction

The Kozarnika cave (Sirakov et al., 2010 and references therein),
located in northwestern Bulgaria (Fig. 1) at the margin of the
Danube loess basin (Haase et al., 2007), is the repository of a 9 m-
thick stratigraphic succession containing lithological units (Ferrier
et al., 2009; Sirakov et al., 2010) labeled from Layers 3e4 at the
top to Layer 14 at the bottom (Fig. 2). This sequence hosts various
archaeological complexes spanning from the Neolithic (and
younger) at the top followed downsection by a blade industry
termed Kozarnikian, then Middle Paleolithic and Lower Paleolithic
industries (Fig. 2) (details in Guadelli et al., 2005; Sirakov et al.,
2010).

The upper Layers 3e10b yielded radiocarbon and optically
ella Terra 'Ardito Desio', Uni-
133 Milano, Italy.
uttoni).
stimulated luminescence (OSL) ages broadly spanning from 13.2 ka
to 183 ka (ka ¼ kiloannum ¼ one thousand years ago) (details in
Guadelli et al., 2005; Tillier et al., 2017; see also below). Large
mammal biostratigraphy seems to indicate that Layers 11b to 13c
should belong to Mammal Neogene/Quaternary (MNQ) zones 19 to
17 with an attributed age of ~1.4e1.6 Ma (Ma ¼ megaannum ¼ one
million years ago) (Guadelli et al., 2005; Sirakov et al., 2010; but see
Popov and Marinska, 2007 for an alternative interpretation). These
mammal age assessments are however potentially marred by the
lack of continuity of the Kozarnika sequence, which seems to
consist of an intermittent succession of sedimentary episodes and
phases of human occupation (new studies are underway to
contribute resolving these uncertainties in key Layers 10ce11c).

A preliminary magnetostratigraphic investigation of the cave
sediments (Sirakov et al., 2010) indicated that sediments down to
the middle part of Layer 11b are characterized by normal polarity
magnetization interpreted as a record of the Brunhes Chron (base at
0.78 Ma; time scale of Lourens et al., 2004). Low magnetic incli-
nation values in the lower part of Layer 11b were tentatively

mailto:giovanni.muttoni1@unimi.it
http://crossmark.crossref.org/dialog/?doi=10.1016/j.quascirev.2017.10.034&domain=pdf
www.sciencedirect.com/science/journal/02773791
http://www.elsevier.com/locate/quascirev
https://doi.org/10.1016/j.quascirev.2017.10.034
https://doi.org/10.1016/j.quascirev.2017.10.034
https://doi.org/10.1016/j.quascirev.2017.10.034


Fig. 1. Map of northwestern Bulgaria with location of the Kozarnika cave (43�390 N, 22�440 E).

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e132
interpreted as indicating transition to reverse polarity of the
Matuyama Chron, but problems of consolidation of blocks did not
allow to retrieve a continuous paleomagnetic signal (Sirakov et al.,
2010). According to these preliminary data (Sirakov et al., 2010), the
lower archaeological levels with lithic tools would predate the
BrunheseMatuyama boundary (0.78 Ma).

Encouraged by this earlier attempt, here we report a thorough
magnetostratigraphic study of the cave sediments from Layer 5 to
Layer 14 coupled with a novel interpretation of the genesis of cave
sediments. These results will be used to assess the stratigraphic
continuity of the Kozarnika cave sequence and provide an age
assessment for the lowermost levels bearing Lower Paleolithic
tools.
2. Geological setting

2.1. Cave stratigraphy and sedimentology

The Kozarnika cave opens to the south at an altitude of 481m on
a northern hillside of a tributary valley of the Skomlia River, at
about 85 m above the valley floor (Sirakov et al., 2010). The Skomlia
valley is about 185 m deep and cuts through a Jurassic sequence
comprised of Early Jurassic red conglomerates, Middle Jurassic
yellow limestones and Late Jurassic grey limestones that host the
cave (Ferrier et al., 2009). The stratigraphy of the Pleistocene cave
infilling, updated after Sirakov et al. (2010), is comprised of the
following set of main layers described from top to bottom (Fig. 2;
see also Ferrier et al., 2009; Sirakov et al., 2010):

� Layers 3a to 4 constitute the uppermost part of the sequence and
have not been sampled for magnetostratigraphy. They are
altogether 1e1.4 m-thick and composed of calcareous clasts
(coming from the cave walls and ceiling) in a light brown to
whitish silty matrix. These layers contain archaeological levels
IVb-0I attributed to middle and recent stages of the Kozarnikien,
which is a local blade industry containing backed pieces. Cali-
brated radiocarbon ages were obtained from uppermost Layers
3a and 3b spanning from 13.2 to 24.5 ka (Guadelli et al., 2005)
(Fig. 2).

� Layers 5a to 10c are altogether about 1.2e1.5 m-thick. The upper
part of this sequence (Layers 5a to 10a), is characterized by
yellowish brown silts embedding calcareous clasts and blocks
resulting from the fragmentation of the cave walls and ceiling.
Underlying Layers 10b and 10c consist of loose dark-brown
loamy sand containing gravels and calcareous pebbles
showing variable degrees of weathering. In this sequence,
Layers 5a-c yielded archaeological levels VeVII attributed to the
early stages of the Kozarnikien, and Layer 5b yielded a calibrated
radiocarbon age of 31.2 ka (Sirakov et al., 2010) (Fig. 2). Layer 6/7
includes archaeological level VIII, which corresponds to an in-
dustry regarded as characteristic of the Middle Paleolithic and
the initial Late Paleolithic; it also yielded uncalibrated radio-
carbon ages ranging from 42.7 to 43.6 ka (Sirakov et al., 2010)
(Fig. 2). Layers 9a to 10a contain archaeological levels IX-XIII
attributed to the Middle Paleolithic (Mousterian) while Layers
10b and 10c contain the lower end of the Mousterian sequence.
Layer 10b was recently dated with optically stimulated lumi-
nescence (OSL) from 128 to 183 ka (Tillier et al., 2017) (Fig. 2).

� Layers 11a to 13c have a total thickness of about 2.5 m and are
composed of rather compact yellowish brown loamy sediments,
blotched with dark manganese-bearing nodules and more or
less rich in rock blocks and pebbles (limestone blocks, flint
pebbles from the cavewalls, andmore rarely rounded pebbles of
quartz) with high degrees of alteration. These layers contain
Lower Paleolithic core-and-flake, non-Acheulian industries.

� Layers 13 Lower to 14 represent the section base and are char-
acterized by abundant limestone boulders with occasional
yellowish brown, laminated sandy matrix in between. Archae-
ological artifacts have thus far not been recovered from these
layers.
2.2. Grain size analysis of cave sediments

Grain size analysis was performed on Layers 9e14 to elucidate
the origin of the cave sediments (Gale and Hoare, 1991). A total of
14 samples, ranging in weight from 25 to 40 g, were treated with
10 ml of 30% H2O2 in order to remove any organic matter content.
This treatment was repeated twice. After drying in an oven at 50 �C,
samples were gently disaggregated in an agatemortar and sieved at
1000 and 500 mm. Samples were weighted after each step. The
<500 mm fraction was then dispersed in 0.05% sodium meta-
phosphate solution, disaggregated in an ultrasonic bath for 15 s,
and then passed through a Malvern Laser Particle Size Analyser



Fig. 2. Lithologic log of the studied Kozarnika cave profile (updated after Sirakov et al., 2010) with photographs of the main sampled layers. Key calibrated and uncalibrated
radiocarbon ages from Layers 3 to 6/7 (Guadelli et al., 2005) and the recent OSL age from Layer 10b (Tillier et al., 2017) are reported (in ka ¼ kiloannum) together with the main lithic
industries retrieved throughout the sequence (Sirakov et al., 2010).

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e13 3



Fig. 3. Results of laser granulometry of the Kozarnika cave sediments showing a main
silt mode attributed to loess in samples from Layers from 11a to 13b while overlying
Layers 9a-b to 10c are interpreted as colluvial loess and the lowermost Layer 14 as
fluvial (alluvial) sediments.

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e134
(mod. Mastersizer 2000).
Curves of grain size distribution are reported in Fig. 3 and

described from top to bottom of the cave section.

� Samples from upper Layers 9a-b to 10c display a bimodal dis-
tribution of particles: a broad silt mode and high coarse-grained
sand (�1000 mm) content. The clay content is high in Layers 9a-b
and progressively decreases down-section, i.e., the silt mode
becomes progressively narrower.

� Grain size distributions of Layers 11a to 13b are very similar and
moderately sorted. They present a well-defined silt mode
centered at 20e60 mm and high coarse-grained sand
(�1000 mm) content. A small peak is also present at 120e140 mm
and related to fine-to medium-grained sand, while the clay
content is always very low.

� The sample from the lowermost Layer 14 is poorly sorted: it
displays a skewed distribution with a long silt tail and a sand
mode centered at 100e200 mm as well as high coarse-grained
sand (�1000 mm) content.

Samples from the upper Layers 9a-b to 10c, characterized by
poor sorting and a broad silt mode, are tentatively interpreted as
mainly colluvial loess (Krumbein and Sloss, 1963) variably mixed
with coarse-grained sediments derived from the disruption of the
cavewalls and ceiling (see section 2.3). Ferrier et al. (2009) reported
the occurrence of lenses of sand and gravel with openwork texture
that supports an overall reworking by water. From these analyses
and additional field observations, we consider this facies, termed
‘mainly colluvial loess & rock fragments’ in Fig. 3, as extending
altogether from Layer 3e4 to Layer 10c.

Samples from underlying Layers 11a to 13b display a well-
defined and persistent silt mode (Fig. 3), interpreted as loess
wind-blown into the cave (Pye, 1987, 1995; Coud�e-Gaussen, 1990;
Assallay et al., 1998), associated with coarse-grained sediments
possibly derived from the cryogenic dismantling of the cave walls
and ceiling or other endokarstic processes described by Ferrier et al.
(2009). A similar bimodal grain size distribution with a dominant
loess mode has been found in cave entrances and rock shelters at
the margin of the Po Plain Loess Basin of northern Italy (Cremaschi,
1987; Castiglioni et al., 1990; Ferraris et al., 1990; Zerboni et al.,
2015). We consider this facies, termed ‘mainly cave loess & rock
fragments’ in Fig. 3, as extending altogether from Layer 11a to Layer
13c.

Finally, the grain size distribution curve of basal Layer 14 sug-
gests an alluvial origin of the sand-dominated (i.e., non-loess)
sediments; moreover, at the macro-scale, Layer 14 presents lami-
nated structure typical of water transport, in substantial agreement
with Ferrier et al. (2009). We consider this facies, termed ‘alluvial
sand’ in Fig. 3, as extending altogether from Layer 13 Lower to Layer
14.

2.3. Scanning Electron Microscope analysis of cave sediments

Representative samples from Layers 9, 11aec, and 14 were
selected for Scanning Electron Microscope (SEM) analysis. The
<63 mm and 63e250 mm fractions of 5 samples were separated by
wet sieving for observation of their surface texture (Le Ribault,
1977; Martignier et al., 2013) with a Cambridge 360 SEM equip-
ped with an energy dispersive X-ray analysis (EDS Link Isis 200
with an accelerating voltage of 20 kV, filament intensity 1.70 A, and
probe intensity of 280 pA) after carbon-coating.

The observed grains are rounded, sub-rounded, and subangular.
Chemical composition confirms the occurrence of carbonate grains,
more common in samples from colluvial Layer 9 and alluvial Layer
14, and silicate grains (quartz and subordinate igneous and



Fig. 4. Scanning electron microscope (SEM) images of quartz grains from loess units showing surface microtextures. (AeB) Quartz grains with conchoidal fractures (c) and fracture
faces (f) (sample from Layer 11a); (CeD) wind transportation V-shaped percussion marks on subangular grains (arrows) (sample from Layer 11c).

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e13 5
metamorphic grains), more common in samples from loess-
dominated Layer 11aec. In loess samples, quartz grains show
many V-shaped percussion marks and impact features (Fig. 4),
which are interpreted as the effect of wind transportation
(Martignier et al., 2013; Paveli�c et al., 2016).

EDS analysis on the bulk sediment highlighted different
elemental concentrations in the samples from colluvial Layer 9 and
alluvial Layer 14 (mean values of Si ¼ 17%, Ca ¼ 43%, Al ¼ 4%,
Fe ¼ 2%) relative to the samples from eolian Layer 11aec (mean
values of Si ¼ 50%, Ca ¼ 21%, Al ¼ 4%, Fe ¼ 2%); this could be
interpreted as the effect of a dominant igneous and metamorphic
petrography in the loess units (compatible with an allochthonous
wind input) and a dominant carbonatic petrography in the alluvial
units, reflecting an autochthonous origin of grains.
3. Paleomagnetism

3.1. Methods

Results from a previous paleomagnetic sampling were
hampered by the unconsolidated nature of the sediments (Sirakov
et al., 2010). We avoided this issue by carefully inserting ~5 cm3

plastic cylinders into oriented sediment walls, obtaining a total of
173 specimens for paleomagnetic analyses from Layer 6/7 at the top
to Layer 14 at the base (Fig. 5). The initial magnetic susceptibility
was measured on all weighed samples at room temperature with a
KLY-2 Kappabridge. A sub-set of 10 samples was subjected to rock
magnetic analyses by means of back-field isothermal remanent
magnetization (IRM) acquisition imparted at �2.5 T along the
horizontal axis in one direction (-z) and then progressively
increasing from þ0.025 T up to þ 2.5 T in the opposite direction
(þz) (Fig. 6A, samples KZ94 and KZ52; Fig. 6B). Hysteresis analyses
were conducted on 6 selected rock chips (mass ~ 40e120mg) using
a Lake Shore 7410 Vibrating Sample Magnetometer (VSM) with
maximum applied fields Hmax ¼ ± 1 T (Fig. 7A). Hysteresis mea-
surements were repeated after thermal treatment applied directly
to the samples within the VSM chamber by means of a thermo-
resistance furnace operating in Argon atmosphere. The same
apparatus was used to obtain thermomagnetic curves up to 700 �C
(applied field H ¼ 200 mT). The thermomagnetic curves, showing
magnetic moment (M) variations as a function of temperature (T),
offer general information about the magnetic mineralogy, thermal
stability, and grain size of the magnetic carriers (Fig. 7B).

To delineate the magnetostratigraphy, a complete set of samples
was subjected to progressive alternating field (AF) demagnetization
up to 80 mT. The natural remanent magnetization (NRM) was
measured after each demagnetization step with a 2G Enterprises
DC-SQUID cryogenic magnetometer located in a shielded room.
Standard least-square analysis was used to calculate magnetic
component directions from vector end-point demagnetization di-
agrams (Fig. 8), and standard Fisher statistical analysis was used to
analyze the mean component directions.
3.2. Results

The magnetic susceptibility fluctuates around a mean value of
~20*10�8 m3/kg throughout most of the section (with one isolated
anomalously high value at 6.48 m), while the section base is
characterized by higher values of ~40e60*10�8 m3/kg (Fig. 5A). The
intensity of the NRM is comprised between 20*10�7 Am2/kg and
120*10�7 Am2/kg (punctuated by an isolated anomalously high
value again at 6.48 m) (Fig. 5B). During IRM acquisition experi-
ments, a �2.5 T field imparted along the -z axis produced a parallel
IRM, and subsequent fields imparted along the þz axis
from þ0.025 T up to þ2.5 T generated a progressively increasing
antiparallel IRM aligned along the þz axis (Fig. 6A, samples KZ94
and KZ52). Only sample KA10 (at 6.48 m) was highly anisotropic
whereby the IRM generated by the�2.5 T field along the -z axis was
found to lie at high angle relative to the -z axis, and subsequent
fields along the þz axis produced IRMs again oriented at high an-
gles relative to the þz axis (Fig. 6A, sample KA10). In any case, IRM
acquisition curves reveal the presence of a dominant low to mod-
erate coercivity ferromagnetic mineral that tends to saturate in
fields of only 0.06 T (60 mT) variably associated with a higher
coercivity mineral that shows no tendency to saturate up to 2.5 T
fields (Fig. 6B). The proportion of soft/hard minerals, isolated from



Fig. 5. The Kozarnika lithostratigraphic profile placed aside paleomagnetic data from this study; key radiocarbon (Guadelli et al., 2005) and OSL (Tillier et al., 2017) ages (in
ka ¼ kiloannum) and the vertical distribution of Lower Paleolithic tools are indicated. From left to the right: (A) magnetic susceptibility, (B) natural remanent magnetization (NRM)
intensity, (C) mean angular deviation of the characteristic C component, (D) virtual geomagnetic pole (VGP) latitudes and (E) magnetic polarity where black is normal polarity and
white is reverse polarity. The BrunheseMatuyama boundary (0.78 Ma) falls at ~8.3 m within the mid part of Layer 13 Lower. The thin normal polarity interval centered at ~8.5 m
could be a record of the Jaramillo subchron (0.99e1.07 Ma) while the basalmost normal polarity interval could represent a partial record of the Olduvai subchron with top placed at
1.78 Ma. M is Matuyama Chron. J? ¼ Jaramillo (uncertain), O? ¼ Olduvai (uncertain).

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e136
the IRM acquisition curves following Kruiver et al. (2001), varies
throughout the section whereby the hard coercivity fraction is
confined to 5e10% of the total IRM from the the top of the section
down to Layer 13b, and then increases to values of up to 20% in
Layers 13 Lower and Layer 13/14 (Fig. 6C).

Hysteresis loops obtained before thermal treatment show the
ubiquitous presence of a soft magnetic phase with very low coer-
cive field (Hc ~ 10 mT) associated with a prevalent paramagnetic
phase (Fig. 7A). After thermal treatment, magnetization and coer-
cive force of the samples increase drastically, resulting in more
clearly defined hysteresis loops. As well, magnetic susceptibility
values of samples before thermal treatment are about
10e40*10�8 m3/kg, and increase more than ten times after heating
treatment. The heating branches of the thermomagnetic experi-
ments show the presence of a Curie temperature of about 580 �C
(Fig. 7B) indicative of magnetite (Dunlop and €Ozdemir, 1997), in
agreement with the results obtained from the IRM experiments.
The cooling branches show much higher magnetizations (Fig. 7B)
probably due to the growth of iron oxides from paramagnetic
precursors during heating. No evidence is provided through hys-
teresis or thermomagnetic experiments of the harder magnetic
phase (hematite and/or goethite) revealed by the IRM experiments.

Vector end-point demagnetization diagrams show the presence
of soft components of the remanent magnetization termed A that
are easily removed between 0 mT and 10e20 mT (Fig. 8). These A
components are oriented downward (positive inclinations) with



Fig. 6. (A) Examples of isothermal remanent magnetization (IRM) acquisition experiments consisting of an initial �2.5 T field imparted along the sample -z axis followed by fields
progressively increasing from þ0.025 T up to þ2.5 T imparted along the þz axis; all samples responded isotropically producing IRMs aligned along the imparted fields except for
sample KA10 at 6.48 m that resulted highly anisotropic with IRMs lying at high angles relative to the imparted fields. (B) IRM acquisition curves for representative samples from the
Kozarnika cave sediments showing the presence of a dominant low coercitivity magnetic fraction variably associated with a subsidiary hard coercitivity fraction. (C) The soft/hard
coercivity ratios of samples reveal that he hard coercivity fraction is higher (up to 20% of the total IRM) in Layers 13 Lower and Layer 13/14.

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e13 7



Fig. 7. (A) Hysteresis loops before thermal treatment (red curves) and after thermal treatment (black curves) of samples KZ85, KZ68, KZ94, and KF03 showing a paramagnetic phase
(positive slope of magnetic moment vs. applied field above 100 mT) coexisting with a soft ferromagnetic phase, which becomes more evident after thermal treatment. (B)
Thermomagnetic (heating and cooling) curves for the same samples, showing a Curie point of about 580 �C interpreted as due to magnetite. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e138



Fig. 8. Vector end-point demagnetization diagrams of representative samples from the Kozarnika cave sediments. Full symbols are projections on the horizontal plane and open
symbols on the vertical plane. Demagnetization temperatures are expressed in �C.

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e13 9



Fig. 9. Equal area projection of the A and characteristic C component vectors in in situ coordinates from the Kozarnika cave sediments. Full symbols represent down-pointing
vectors, open symbols represent up-pointing vectors.

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e1310
scattered, generally northerly declinations (Figs. 8 and 9). At higher
AF demagnetization steps, the NRM becomes harder to release; a
characteristic remanent magnetization component, termed C, was
isolated in 119 samples, up to a maximum AF field of 80 mT (Fig. 8).
These C components are either oriented northerly and down, and
hence they constitute a sort of a prosecution at higher AF values of
the A components, or show a tendency to turn to southerly decli-
nations and upward (negative) inclinations (Figs. 8 and 9). The bi-
polar C components, with mean angular deviation (MAD) values of
usually less than 10� (Fig. 5C), are grouped along an axis with a
mean (northerly) declination ¼ 358.3�E and inclination ¼ 62.5�

(n ¼ 119, k ¼ 9.8, a95 ¼ 4.4�; Fig. 9).
Both the A and C components of the NRM are released in AF field

intensities that reside well within the IRM acquisition spectrum of
the soft coercivity magnetite phase (Figs. 6 and 7). The (subsidiary)
hard coercivity phase (hematite and/or goethite), which shows up
only in the IRM acquisition experiments and in fields above 100mT,
does not seem to carry any sizable portion of the NRM, which is
virtually completely unblocked by 80 mT. This suggests that the C
components and the normal and reverse magnetic polarities that
they define are carried by the same ferromagnetic (magnetite)
phase, irrespective of polarity, which supports the primary nature
of these magnetic components.

The declination and inclination values of the C components
were used to calculate virtual geomagnetic pole (VGP) latitudes and
plotted versus stratigraphic depth (Fig. 5D). VGP latitudes
approaching þ90� are interpreted as normal polarity and VGP lat-
itudes approaching �90� as reverse polarity, thus defining a
sequence of stratigraphically superposed normal and reverse
magnetozones (Fig. 5E). The section is characterized by consistent
normal magnetic polarity from upper Layers 9a-b at 4.7 m down to
8.3 m within Layer 13 Lower. This relatively thick normal polarity
interval is interrupted by anomalous sample KA10 at 6.48 m in the
lower part of Layer 11b displaying a southerly-and-up C component
direction that is associated with anomalously high NRM and sus-
ceptibility values (Fig. 5D and E); this sample was also highly
anisotropic during IRM experiments whereby the induced
magnetization was at a high angle relative to the applied fields
(Fig. 6A). Below meter level 8.3, and still within Layer 13 Lower, as
well as in underlying Layer 13/14, reverse polarity C component
directions have been observed after removal of a pervasive initial
overprint component of normal polarity. These reverse polarity C
component directions are characterized by a relatively high degree
of scattering possibly due to incomplete cleaning of the normal
polarity overprint. At meter level 8.5 at the base of Layer 13 Lower,
two superposed samples showed hard-coercivity component di-
rections of exclusive normal polarity, while at the section bottom,
Layer 14 is characterized again by hard-coercivity component di-
rections of exclusive normal magnetic polarity (Fig. 5D and E).
4. Discussion

The Kozarnika magnetostratigraphic profile was tentatively
correlated to the reference geomagnetic polarity time scale (GPTS;
Lourens et al., 2004) to assess the age of the cave sediments
(Fig. 10). We interpret the thick normal magnetic polarity interval
extending from Layers 9a-b at 4.7 m down to 8.3 mwithin Layer 13
Lower (Fig. 5D and E) as a record of the Brunhes Chron (�0.78 Ma)
(Fig. 10). This interpretation agrees with two independent lines of
evidence:

(i) The radiocarbon and OSL ages in the upper layers (e.g., 31.2
ka in Layer 5b; 128 ka to 183 ka in Layer 10b) clearly indicate
that the section top formed since the very end of the Middle
Pleistocene and especially in the Late Pleistocene (¼ late
Brunhes) (Fig. 10).

(ii) As previously stated, the presence in most of the analyzed
samples of a narrow grain size distribution in the silt range
indicates loess as a dominant textural component in the
Kozarnika sedimentary infill down at least to Layers 13aec
(layers below are alluvial in origin). The Kozarnika cave is
located at the south-western margin of the Bulgarian lower
Danube loess belt (Haase et al., 2007; Buggle et al., 2009,
2013) where loess deposition generally started sometime
between the Jaramillo Subchron (0.99 Ma) and the Brun-
heseMatuyama boundary (0.78 Ma) (Sartori et al., 1999;
Jordanova et al., 2008; Fitzsimmons et al., 2012; Markovi�c
et al., 2012, 2015; Muttoni et al., 2015). This age window
(0.78e0.99 Ma) thus represents a maximum likely age of
onset of loess accumulation in the Kozarnika cave, implying
that the lowermost cave loess of Layers 13aec should not be
older than 0.99 Ma, in agreement with our magnetostrati-
graphic interpretation indicating that these layers are
younger than 0.78 Ma (Fig. 10; see also below).



Fig. 10. A summary of the main results of this study: the Kozarnika archaeologic (Sirakov et al., 2010), lithostratigraphic, and magnetostratigraphic profile calibrated with key
radiocarbon (Guadelli et al., 2005) and OSL (Tillier et al., 2017) ages has been correlated to the reference geomagnetic polarity time scale (GPTS; Lourens et al., 2004) to assess the
age of the lowermost levels bearing Lower Paleolithic tools (Layers 13aec; Sirakov et al., 2010). According to the proposed correlation, Layers 13aec are securely placed in the early
Brunhes (<0.78 Ma); taking into account also the inferred ages of the main Danube loess intervals (L1eL10; Markovi�c et al., 2012, 2015), Layers 13aec could correspond to Loess L6
to L7 in the ~0.6 to ~0. 75 Ma time interval.

G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e13 11



G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e1312
According to the above, we regard our placement of the Brun-
heseMatuyama boundary at ~8.3 mwithin the mid part of Layer 13
Lower as robust (Fig. 10). The lowermost levels with Lower Paleo-
lithic tools are in Layers 13aec, which therefore appear to lie in the
early Brunhes close in age to the BrunheseMatuyama boundary
(Fig. 10). Assuming that Layers 13aec correspond to Loess L6 to L7
of the Danube loess-paleosol sequence (e.g., Markovi�c et al., 2012,
2015), a tentative age of ~0.6 to ~0.75 Ma for these lowermost
tool-bearing layers can be derived (Fig. 10).

Farther downsection, the thin interval with positive inclinations
centered at 8.5 m at the base of Layer 13 Lower (Fig. 5D and E) could
be speculatively interpreted as a record of the Jaramillo subchron
(0.99e1.07 Ma) (Fig. 10), whereas the lowermost thin interval with
positive inclinations at 8.7 m in alluvial Layer 14 (Fig. 5D and E)
could be a partial record of the Olduvai subchron, for which,
however, we lack a record of its base (Fig. 10). While the estimated
position of the BrunheseMatuyama boundary is relatively robust in
the section, the interpretation of these thin short normal polarity
intervals embedded in secure reverse polarity (Matuyama) is sub-
ject to uncertainty due to the sedimentological complexity of these
layers that may have experienced post-depositional clay illuviation
or sediments redistribution from upper layers during the Brunhes
Chron, or may have even experienced localized normal polarity
overprints due to retarded lock-in of the NRM (Fig. 10).

According to previous magnetostratigraphic analyses reported
in Sirakov et al. (2010), Layer 11b contains lowmagnetic inclination
values tentatively interpreted as indicating the occurrence therein
of the BrunheseMatuyama boundary, in apparent contradiction
with our findings. Based on this initial report, we paid particular
attention to Layer 11b where paleomagnetic sampling resolution
was particularly high. We found one sample at 6.48 m in the lower
part of Layer 11b displaying upward and southerly directions;
however, the sample magnetic properties were unusual, associated
with high NRM values and a highly anisotropic response to IRM
experiments. Moreover, there is no evidence in this densely
sampled interval of a clear and persistent normal-to-reverse po-
larity reversal (Fig. 5D and E). Considering all the data, we are
confident in the placement of the BrunheseMatuyama boundary
(0.78 Ma) at ~8.3 m within Layer 13 Lower (Fig. 5D and E).

As stressed in the Introduction, stratigraphic gaps may be pre-
sent in cave entrance settings like Kozarnika (Farrand, 2001); for
example, there is evidence within Layer 11a of burrowing with
associated erosion of sediment. However, our magnetostratigraphic
data with a thick and persistent normal polarity interval (Brunhes
Chron) underlain by reverse polarity (Matuyama Chron), used in
conjunction with the temporally ordered sequence of lithic in-
dustries (Guadelli et al., 2005; Sirakov et al., 2010), seem to exclude
the presence of major stratigraphic gaps on the order of 105 years.

These new magnetostratigraphic data prompt the necessity to
continue investigating the complex mammal associations retrieved
throughout the sequence. According to Sirakov et al. (2010), the
lower Layers 11be13 should belong to MNQ zones 19 to 17, which,
according to a recent radiochronologic (40Ar/39Ar) reassessment of
classic mammalian localities from France, should correspond to a
time interval comprised between ~1.2 and ~2.6 Ma (Nomade et al.,
2014). This time interval corresponds to the early Matuyama Chron
and the embedded Olduvai subchron (1.78e1.94 Ma), in contra-
diction with our findings that place the BrunheseMatuyama
boundary (0.78 Ma) in Layer 13 Lower. Indeed, alluvial Layer 14 at
the very base of the sequence, below Layer 13 Lower, could contain
a record of the Olduvai, but the taphonomic analysis of the fauna in
Layers 12 and 13aec, and the fact that Layer 14 is entirely sterile
seem to preclude that certain mammal bone fragments attributed
by Sirakov et al. (2010) to MNQ 19e17 could have been reworked at
the base of Layer 13. Indeed, the disagreement between mammal
biostratigraphy and magnetostratigraphy will need further
investigation.

5. Conclusions

We presented a complete magnetostratigraphic profile of the
Kozarnika cave sediments that constrains the age of the lowermost
levels with Lower Paleolithic tools (Layers 13aec) to the early
Brunhes at nominal ages of ~0.6 to ~0. 75 Ma (Fig. 10). Our data
represent a conspicuous revision of previous estimates based on
mammal biostratigraphy of hominin presence at Kozarnika as early
as 1.4e1.6 Ma (Sirakov et al., 2010; see also Michel et al., 2017). This
age revision brings Kozarnika into better agreement with a recent
critical reviews indicating that the earliest main peopling of Europe
occurred in a chronological window comprised between the top of
the Jaramillo subchron (0.99 Ma) and the BrunheseMatuyama
boundary (0.78 Ma) (Muttoni et al., 2017; see also Muttoni et al.,
2014). Although Kozarnika seems to have been first occupied at
later times, the pre-Brunhes but post-Jaramillo time interval rep-
resents a prime target for surveys in search of additional sites in the
Danube area with mammal immigrants from Asia and Africa (e.g.,
Kostolac, Serbia; Muttoni et al., 2015) possibly including early
hominins.

Acknowledgements

Aleta Guadelli is warmly thanked for assistance in the field.
Agostino Rizzi is thanked for helping with SEM-EDS analysis. G.M.
and E.M. were supported by grant from Fondo Scavi Archeologici e
Universit�a degli Studi di Milano 2015. This is Lamont-Doherty
contribution #8159.

References

Assallay, A.M., Rogers, C.D.F., Smalley, I.J., Jefferson, I.F., 1998. Silt: 2e62 mm, 9e4f.
Earth Sci. Rev. 45, 61e88.

Buggle, B., Hambach, U., Glaser, B., Gerasimenko, N., Markovi�c, S.B., Glaser, I.,
Z€oller, L., 2009. Magnetic susceptibility stratigraphy and spatial and temporal
palaeoclimatic trends in East European loess paleosol sequences. Quat. Int. 196,
86e106.

Buggle, B., Hambach, U., Kehl, M., Markovi�c, S.B., Z€oller, L., Glaser, B., 2013. The
progressive evolution of a continental climate in SE-Central European lowlands
during the Middle Pleistocene recorded in loess paleosol sequences. Geology 41,
771e774.

Castiglioni, G.B., Cremaschi, M., Guerreschi, A., Meneghel, M., Sauro, U., Van
Vliet, L.B., 1990. The loess deposits in the Lessini plateau. In: Cremaschi, M. (Ed.),
The Loess in Northern and Central Italy: a Loess Basin between the Alps and the
Mediterranean Region. CNR, Milano, pp. 41e59.

Coud�e-Gaussen, G., 1990. The loess and loess-like deposits along the sides of the
western Mediterranean Sea: genetic and palaeoclimatic significance. Quat. Int.
5, 1e8.

Cremaschi, M., 1987. Loess deposits of the plain of the Po and the adjoining Adriatic
basin (Northern Italy). In: Pecsi, M., French, H.M. (Eds.), Loess and Periglacial
Phenomena. Akademiai Kiado, Budapest.

Dunlop, D.J., €Ozdemir, €O., 1997. Rock Magnetism. Fundamentals and Frontiers.
Cambridge Studies in Magnetism Series. Cambridge University Press, 573 pp.

Farrand, W.R., 2001. Sediments and stratigraphy in rockshelters and caves: a per-
sonal perspective on principles and pragmatics. Geoarchaeology 16 (5),
537e557.

Ferraris, M., Sala, B., Scola, V., 1990. The late Pleistocene fauna with Pliomys lenki
from the Ghiacciaia cave loess (Northern Italy). Quat. Int. 5, 71e79.

Ferrier, C., Leblanc, J.-Cl, Berthet, A.-L., Delfour, G., 2009. La grotte Kozarnika (Gara
Oreshets, Bulgarie): donn�ees stratigraphiques, g�eologiques et
s�edimentologiques. In: Gatsov, I., Guadelli, J.-L. (Eds.), Saxa Loquuntur! “The
Stone Will Speak!”, pp. 13e27. Avalon, Sofia.

Fitzsimmons, K., Markovi�c, S., Hambach, U., 2012. Pleistocene environmental dy-
namics recorded in the loess of the middle and lower Danube basin. Quat. Sci.
Rev. 41, 104e118.

Gale, S.J., Hoare, P.G., 1991. Quaternary Sediments. Belhaven Press, London.
Guadelli, J.L., Sirakov, N., Ivanova, S., Sirakova, S., Anastassova, E., Courtaud, P.,

Dimitrova, I., Djabarska, N., Fernandez, P., Ferrier, C., Fontugne, M., Gambier, D.,
Guadelli, A., Iordanova, D., Iordanova, N., Kovatcheva, M., Krumov, I.,
Leblanc, J.C., Mallye, J.B., Marinska, M., Miteva, V., Popov, V., Spassov, R.,
Taneva, S., Tisterat-Laborde, N., Tsanova, T., 2005. Une s�equence du

http://refhub.elsevier.com/S0277-3791(17)30738-2/sref1
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref1
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref1
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref1
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref1
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref2
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref2
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref2
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref2
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref2
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref2
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref2
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref3
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref3
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref3
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref3
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref3
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref3
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref3
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref4
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref4
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref4
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref4
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref4
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref5
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref5
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref5
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref5
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref5
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref6
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref6
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref6
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref7
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref7
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref7
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref7
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref8
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref8
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref8
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref8
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref9
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref9
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref9
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref10
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref10
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref10
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref10
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref10
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref10
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref10
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref10
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref11
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref11
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref11
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref11
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref11
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref12
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13


G. Muttoni et al. / Quaternary Science Reviews 178 (2017) 1e13 13
Pal�eolithique inf�erieur au Pal�eolithique r�ecent dans les Balkans: la grotte
Kozarnika �a Orechets (nord-ouest de la Bulgarie). In: Molines, N., Moncel, M.-H.,
Monnier, J.-L. (Eds.), Les Premiers Peuplements en Europe, British Archaeolog-
ical Reports, International Series, vol. 1364, pp. 87e103.

Haase, D., Fink, J., Haase, G., Ruske, R., P�ecsi, M., Richter, H., Altermann, M., J€ager, K.-
D., 2007. Loess in Europe e its spatial distribution based on a European loess
map, scale 1:2,500,000. Quat. Sci. Rev. 26, 1301e1312.

Jordanova, D., Hus, J., Evlogiev, J., Geeraerts, R., 2008. Paleomagnetism of the loess/
palaeosol sequence in Viatovo (NE Bulgaria) in the Danube basin. Phys. Earth
Planet. Interiors 167 (1), 71e83.

Kruiver, P.P., Dekkers, M.J., Heslop, D., 2001. Quantification of magnetic coercivity
components by the analysis of acquisition curves of isothermal remanent
magnetisation. Earth Planet. Sci. Lett. 189 (3), 269e276.

Krumbein, W.C., Sloss, L.L., 1963. Stratigraphy and Sedimentation. Freeman & Co.,
San Francisco, USA.

Le Ribault, L., 1977. L’exoscopie des Quartz. Masson, Paris.
Lourens, L.J., Hilgen, F.J., Laskar, J., Shackleton, N.J., Wilson, D., 2004. The Neogene

period. In: Gradstein, F.M., Ogg, J.G., Smith, A.G. (Eds.), A Geologic Time Scale
2004. Cambridge University Press, Cambridge, pp. 409e440.

Markovi�c, S.B., Hambach, U., Stevens, T., Jovanovi�c, M., O'Hara-Dhand, K., Basarin, B.,
Lu, H., Smalley, I., Buggle, B., Zech, M., 2012. Loess in the Vojvodina region
(Northern Serbia): an essential link between European and asian Pleistocene
environments. Neth. J. Geosciences 91, 173e188.

Markovi�c, S.B., Stevens, T., Kukla, G.J., Hambach, U., Fitzsimmons, K.E., Gibbard, F.,
Buggle, B., Zech, M., Guo, Z., Hao, Q., Wu, H., O'Hara Dhand, H., Smalley, I.J.,
Újv�ari, G., Sümegi, P., Timar-Gabor, A., Veres, D., Sirocko, F., Vasiljevi�c, D.A.,
Jary, Z., Svensson, A., Jovi�c, V., Lehmkuhl, F., Kov�acs, J., Svir�cev, Z., 2015. Danube
loess stratigraphy d towards a pan-European loess stratigraphic model. Earth-
Science Rev. 148, 228e258.

Martignier, L., Adatte, T., Verrecchia, E.P., 2013. Bedrock versus superficial deposits
in the Swiss Jura Mountains: what is the legitimate soil parent material? Earth
Surf. Process. Landforms 38, 331e345.

Michel, V., Shen, C.C., Woodhead, J., Hu, H.M., Wu, C.C., Moull�e, P.�E., Khatib, S.,
Cauche, D., Moncel, M.H., Valensi, P., Chou, Y.M., Gallet, S., Echassoux, A.,
Orange, F., de Lumley, H., 2017. New dating evidence of the early presence of
hominins in Southern Europe. Sci. Rep. 7, 10074.

Muttoni, G., Kent, D.V., Scardia, G., Monesi, E., 2014. Migration of hominins with
megaherbivores into Europe via the Danube-Po gateway in the late Matuyama
climate revolution. Riv. Ital. Paleontol. Stratigr. 120, 351e365.
Muttoni, G., Scardia, G., Dimitrijevi�c, V., Kent, D.V., Monesi, E., Mrdjic, N., Kora�c, M.,

2015. Age of Mammthus trogontherii from Kostolac, Serbia, and the entry of
megaherbivores into Europe during late Matuyama climate revolution. Quat.
Res. 84, 439e447.

Muttoni, G., Scardia, G., Kent, D.V., 2017. Early hominins in Europe: the Galerian
migration hypothesis. Quat. Sci. Rev. in press.

Nomade, S., Pastre, J.F., Guillou, H., Faure, M., Gu�erin, C., Delson, E., Debard, E.,
Voinchet, P., Messager, E., 2014. 40 Ar/39 Ar constraints on some French land-
mark Late Pliocene to Early Pleistocene large mammalian paleofaunas: paleo-
environmental and paleoecological implications. Quat. Geochronol. 21, 2e15.

Paveli�c, D., Kova�ci�c, M., Banak, A., Jim�enez-Moreno, G., Markovi�c, F., Pikelj, K.,
Vranjkovi�c, A., Premu�zak, L., Tiblja�s, D., Belak, M., 2016. Early Miocene European
loess: a new record of aridity in southern Europe. Geol. Soc. Am. Bull. 128,
110e121.

Popov, V.V., Marinska, M., 2007. An almost one million year long (Early to Late
Pleistocene) small mammal succession from the archaeological layers of
Kozarnika cave in Northern Bulgaria. Cour. Forschungsinstitut Senckenberg 259,
79e92.

Pye, K., 1987. Aeolian Dust and Dust Deposits. Academic press, London, 334 pp.
Pye, K., 1995. The nature, origin and accumulation of loess. Quat. Sci. Rev. 14,

653e667.
Sartori, M., Heller, F., Forster, T., Borkovec, M., Hammann, J., Vincent, E., 1999.

Magnetic properties of loess grain size fractions from the section at Paks
(Hungary). Phys. Earth Planet. Interiors 116, 53e64.

Sirakov, N., Guadelli, J.L., Ivanova, S., Sirakova, S., Boudadi-Maligne, M., Dimitrova, I.,
Ph, F., Ferrier, C., Guadelli, A., Iordanova, D., Iordanova, N., 2010. An ancient
continuous human presence in the Balkans and the beginnings of human set-
tlement in western Eurasia: a Lower Pleistocene example of the Lower Paleo-
lithic levels in Kozarnika cave (North-western Bulgaria). Quat. Int. 223, 94e106.

Tillier, A.-M., Sirakov, N., Guadelli, A., Fernandez, P., Sirakova, S., Dimitrova, I.,
Ferrier, C., Krumov, I., Leblanc, J.-C., Miteva, V., Popov, V., Taneva, S., Guadelli, J.-
L., 2017. Evidence of Neanderthals in the Balkans: the infant radius from
Kozarnika Cave, Bulgaria. J. Hum. Evol. 111, 54e62.

Zerboni, A., Trombino, L., Frigerio, C., Livio, F., Berlusconi, A., Michetti, A.M.,
Rodnight, H., Sp€otl, C., 2015. The loess-paleosol sequence at Monte Netto: a
record of climate change in the Upper Pleistocene of the central Po Plain,
northern Italy. J. Soils Sediments 15, 1329e1350.

http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref13
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref14
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref14
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref14
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref14
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref14
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref14
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref14
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref15
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref15
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref15
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref15
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref16
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref16
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref16
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref16
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref17
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref17
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref17
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref18
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref19
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref19
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref19
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref19
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref20
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref20
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref20
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref20
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref20
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref20
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref20
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref21
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref22
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref22
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref22
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref22
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref23
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref23
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref23
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref23
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref23
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref23
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref24
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref24
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref24
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref24
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref25
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref25
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref25
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref25
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref25
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref25
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref25
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref26
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref26
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref27
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref27
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref27
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref27
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref27
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref27
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref28
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref29
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref29
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref29
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref29
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref29
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref30
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref31
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref31
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref31
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref32
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref32
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref32
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref32
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref33
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref33
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref33
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref33
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref33
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref33
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref34
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref34
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref34
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref34
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref34
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref35
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref35
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref35
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref35
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref35
http://refhub.elsevier.com/S0277-3791(17)30738-2/sref35

	An early Brunhes (<0.78 Ma) age for the Lower Paleolithic tool-bearing Kozarnika cave sediments, Bulgaria
	1. Introduction
	2. Geological setting
	2.1. Cave stratigraphy and sedimentology
	2.2. Grain size analysis of cave sediments
	2.3. Scanning Electron Microscope analysis of cave sediments

	3. Paleomagnetism
	3.1. Methods
	3.2. Results

	4. Discussion
	5. Conclusions
	Acknowledgements
	References