Content uploaded by Katerina Harvati
Author content
All content in this area was uploaded by Katerina Harvati on Feb 06, 2016
Content may be subject to copyright.
Inner Ear Morphology of the Cioclovina Early Modern
European Calvaria From Romania
Alexandra Uhl,
1
Hugo Reyes-Centeno,
1,2
Dan Grigorescu,
3
Elena F. Kranioti,
4
and Katerina Harvati
1,2
*
1
Paleoanthropology, Senckenberg Center for Human Evolution and Paleoenvironment, Eberhard Karls Universit
€
at
T
€
ubingen, T
€
ubingen D-72070, Germany
2
DFG Centre for Advanced Studies ‘Words, Bones, Genes, Tools: Tracking Linguistic, Cultural and Biological
Trajectories of the Human Past’
3
Department of Paleontology, University of Bucharest, Bucharest 70111, Romania
4
Edinburgh Unit for Forensic Anthropology, School of History, Classics and Archaeology, The University of
Edinburgh, Edinburgh, Scotland EH8 9AG, UK
KEY WORDS bony labyrinth; upper paleolithic; modern humans
ABSTRACT
Objectives: The morphology of the human bony labyrinth is thought to preserve a strong phylogenetic signal and
to be minimally, if at all, affected by postnatal processes. The form of the semicircular canals is considered a derived
feature of Neanderthals and different from the modern human anatomy. Among other hominins, European Middle
Pleistocene humans have been found to be most similar to Neanderthals. Early modern humans have been proposed
to show a pattern that is distinct, but most similar to that of Holocene people. Here we examine the inner ear struc-
tures of the Cioclovina calvaria, one of the earliest reliably dated and relatively complete modern human crania
from Europe, in the context of recent and fossil human variation.
Materials and Methods: Bony labyrinths were virtually extracted from CT scans of recent Europeans and Cio-
clovina. Using univariate and multivariate methods, measurements of the semicircular canals were compared with
published measurements of other fossil specimens.
Results: Our results show that Cioclovina’s inner ear morphology falls within the range of modern variation, with
affinities to both Late Pleistocene modern humans and recent Europeans. Using discriminant functions, the sex of
the Cioclovina specimen is estimated as male.
Discussion: Results agree with previous work showing that Cioclovina exhibits fully modern cranial morphology.
Am J Phys Anthropol 000:000–000, 2016.
V
C
2016 Wiley Periodicals, Inc.
The Cioclovina calvaria is one of the earliest known
securely dated and relatively well-preserved Upper Pale-
olithic European modern human fossil specimens. As
such, it presents a rare opportunity to gain insight on
the morphology and paleobiology of the earliest modern
Europeans and to assess this specimen’s morphology in
light of the recently documented admixture with Nean-
derthals. It was discovered in the Pes¸tera Cioclovina
cave, South Transylvania during phosphate mining oper-
ations (Rainer and Simionescu, 1942; Harvati et al.,
2007), and was recently directly dated by AMS
14
Cto
29,0001-700 ka (Olariu et al., 2005) and 28,5101-170
(ultrafiltration pretreatment; Soficaru et al., 2007). Cio-
clovina is commonly assigned to the Aurignacian. Its
overall cranial morphology is unquestionably modern
(Rainer and Simionescu, 1942; Necrasov and Cristescu,
1965; Harvati et al., 2007). However, some authors have
suggested the possibility of partial Neanderthal ancestry
based on the morphology of the nuchal region (Soficaru
et al., 2007; Trinkaus, 2007). In our previous work we
have found that both the external morphology of the cra-
nium and its endocranial shape lie within the range of
modern human variation (Harvati et al., 2007; Kranioti
et al., 2011).
Here we describe the bony labyrinth morphology of
the Cioclovina specimen in the context of Neanderthal,
early modern human, Upper Paleolithic, and recent
human variability. The inner ear includes the organs for
sound perception (cochlea), as well as for movement and
orientation perception (vestibule and semicircular
canals) (McKinley and O’Loughlin, 2006). Its morphology
is considered to be species-specific among mammals
(Spoor, 1993) and has been shown to differentiate
between Neanderthals and modern humans (Hublin
et al., 1996; Spoor, 2003). Compared to modern humans,
Additional Supporting Information may be found in the online
version of this article.
Grant sponsor: the Deutsche Forschungsgemeinschaft; Grant
number: DFGINST37/706-1FUGG; Grant sponsor: the Institute for
Aegean Prehistory (INSTAP); Grant sponsor: the European
Research Council; Grant number: ERC StG 283503.
*Correspondence to: Katerina Harvati; R
€
umelinstraße 23,
D-72070 T
€
ubingen, Germany. E-mail: katerina.harvati@ifu.uni-
tuebingen.de
Received 20 June 2015; revised 18 December 2015; accepted 20
December 2015
DOI: 10.1002/ajpa.22938
Published online 00 Month 2016 in Wiley Online Library
(wileyonlinelibrary.com).
Ó 2016 WILEY PERIODICALS, INC.
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 00:00–00 (2016)
Neanderthals are characterized by (i) a smaller and nar-
rower anterior semicircular canal arc, which shows more
torsion; (ii) a smaller and more rounded posterior semi-
circular canal, which is placed inferiorly relative to the
lateral canal plane; (iii) a larger lateral semicircular
canal; and (iv) an ampullar line that is more vertically
positioned relative to the planar orientation of the lat-
eral canal (Spoor et al., 2003).
Although Neanderthals are relatively well studied, the
inner ear morphology of early modern human and Upper
Paleolithic specimens is less well known. Spoor et al.
(2003) found that their sample of Upper Paleolithic
(n 5 4) and anatomically modern humans from the
Levant (n 5 2) were most similar, though not identical,
to their geographically diverse Holocene material. Like-
wise, the inner ear morphology of the Lagar Velho 1
specimen—another purported modern human-
Neanderthal hybrid—was assessed to have affinities to
other Upper Paleolithic specimens to a greater degree
than to Neanderthals (Spoor et al., 2002a). A more
recent study by Bouchneb and Crevecoeur (2009) ana-
lyzed an expanded sample of early modern and Upper
Paleolithic humans. These authors proposed that early
modern human inner ear morphology is distinct from
that of either Neanderthals or recent modern humans.
Furthermore, they reported a close similarity between
the Nazlet Khater 2 specimen from Egypt and their
European Upper Paleolithic sample on the basis of dis-
criminant analysis.
In light of the species specific, highly diagnostic inner
ear morphology of Neanderthals and modern humans,
the evaluation of this feature in the early Upper Paleo-
lithic cranium from Cioclovina is particularly interest-
ing. The study of this specimen will increase the small
sample of Upper Paleolithic specimens for which this
morphology is known and will help further evaluate the
question of partial Neanderthal ancestry of this individ-
ual. This question has become increasingly relevant
after the recent publication of a genomic analysis of the
Oase 1 mandible, which found that that individual had
a Neanderthal ancestor as recently as 4–6 generations
previously (Fu et al., 2015), raising the possibility, pro-
posed earlier also on morphological grounds, of a high
level of Neanderthal admixture in the Romanian early
Upper Paleolithic samples. Although the effects of
admixture in the skeletal phenotype are not always clear
(see e.g. Ackermann et al., 2010; Ackermann, 2006; Har-
vati and Roksandic, Accepted), one of its possible conse-
quences is intermediate phenotype (e.g. Frost et al.,
2003; Harvati et al., 2007). An inner ear morphology
that is intermediate between that of Neanderthals and
modern humans, therefore, might support the previously
proposed hypothesis of hybrid status for this specimen
(Soficaru et al., 2007; Trinkaus 2007).
MATERIALS AND METHODS
Although Cioclovina’s left temporal bone shows some
damage, the inner ear structures are completely pre-
served bilaterally. The specimen was scanned with a Sie-
mens Sensation 64 medical computed tomography (CT)
scanner in the facilities of the Centrul De Sanatate Pro-
Life SRL, Bucharest. The scanning direction was coronal
(transverse). Slice thickness of 0.625 mm, slice incre-
ment of 0.5 mm, X-ray tube voltage 120 kV, and tube
current 304 mA were used. All slices were formatted to
the same size of 512 3 512 pixels. Our comparative
extant human (ExH) sample consisted of 110 Europeans
from Crete, Greece [n 5 79] and Baden-W
€
urttemberg,
Germany [n 5 31]. Fifty-two of the crania from Greece
were scanned with an Electric LightSpeed 64 Slice CT
(General Electric) in the facilities of the Iatriko Kritis
Medical Centre, Heraklion, Crete, using 0.625 mm slice
thickness. Voxel size was 0.5 3 0.5 3 0.5 mm. The
remaining Greek crania were scanned using a Siemens
SOMATOM Sensation 16 at the Heraklion University
Hospital, Crete, using a tube current of 320 mA, tube
voltage of 120 kV, slice thickness of 0.75 mm, and slice
increment of 0.6 mm. Scans with a field of view of 227 3
227 mm
2
(matrix 512 3 512) were made in the coronal
(transverse) plane. Voxel size was 0.4 3 0.4 3 0.6 mm.
The sample from Germany was scanned at the Paleoan-
thropology High-Resolution CT Laboratory with a Phoe-
nix v|tome|x mCT scanner (General Electric) using 180
kV, 120 mA, 2500 images per scan with an exposure of
200 milliseconds (ms) per image and a resolution
between approximately 101 and 116 mm. Slices had a
uniform pixel size per individual between 0.101 3
0.101 mm and 0.116 3 0.116 mm for the German
sample.
Temporal labyrinth virtual endocasts of the German
sample were reconstructed by A.U. using commercial
medical imaging software (Amira/Avizo: FEI Company,
Hillsboro, OR, USA), following procedures outlined in
Osipov et al. (2013). The Cretan sample reconstructions
were from Osipov et al. (2013), who determined thresh-
old values separating bone and air cavity throughout the
bony labyrinth structure using the half maximum height
(HMH) protocol (Spoor 1993; Spoor & Zonneveld, 1995).
Endocast reconstructions allowed us to visualize overall
labyrinth morphology prior to collecting biometrical
data. Currently established methods (e.g. Spoor, 1993;
Spoor and Zonneveld, 1995) for collecting linear meas-
urements from CT scans rely on identifying landmarks
defined by cranial anatomical orientation. For most med-
ical CTs, which produce nonisotropic voxel sizes, scan-
ning specimens at standardized planes is necessary in
order to prevent measurement error associated with the
different planes scanned. For example, measurements
along the transverse plane have been reported to incur
larger error because the resolution at this plane is
smaller (Richtsmeier et al., 1995; L’Engle Williams and
Richtsmeier, 2003; Park et al., 2006). This issue is neu-
tralized with the use of mCT scanning technology, as
resolution is equal (isometric) across all planes. Since
specimens were scanned using different machines at dif-
ferent resolutions, it was necessary to amend currently
established methods with the following procedure. First,
the endocasts were segmented using the same threshold
values for a given specimen, ensuring that the edges
between bone and air cavity were reliably demarcated
throughout the bony labyrinth structure. We note that
threshold values were different for each specimen
because the level of conservation and bone density differ
between individuals. Discriminating between surface
and cavity using high-resolution CT scans with a small
voxel size can be done visually and informed by the
HMH protocol (Gunz et al., 2012). In order to verify our
threshold values, we also used the Ray Casting Algo-
rithm (RCA) (Scherf and Tilgner, 2009). Second, the mid-
point of the lumen of the canals was identified with a
wire skeleton, calculated by thinning the encased seg-
mented volume of the endocast, following the procedure
outlined in Gunz et al. (2012). The segmentation and
2 A. UHL ET AL.
American Journal of Physical Anthropology
skeletonization procedures are a substitute for the
thresholding procedures adopted for lower resolution CT
scans (Spoor, 1993; Spoor and Zonneveld, 1995). Third,
using the three-dimensional endocast and its corre-
sponding wire skeleton, landmarks defined by anatomi-
cal orientations were identified by one of us (A.U.). We
note that these procedures are similar to those adopted
for fragmentary specimens (Braga et al., 2013; Hill
et al., 2014) and that the combined use of surface recon-
structions and axial slices increases the accuracy of
landmark identification (L’Engle Williams and Richtsme-
ier, 2003). Finally, labyrinth canal measurements, as
defined by Spoor and colleagues (Spoor, 1993; Spoor and
Zonneveld, 1995), were calculated from the identified
landmarks. To assess intra-observer error for all raw
measurements (prior to conversion as ratios and indices
used here) five individuals were measured three times
each. The error for the 8 variables ranged from lowest at
0.76% (anterior canal width) to highest at 3.04% (poste-
rior canal height) error (see Supporting Information
Table S1). In order to confirm that measurements
obtained from lower quality scans (Crete sample) and
higher quality scans (German sample) were compatible,
downsampling was performed on the scans of three Ger-
man individuals. The scans were then segmented and
measured using both the 0.1 and 0.6 resolutions. Differ-
ences in measurements were minimal (less than 2%
deviation).
All measurements were collected by A.U. Raw meas-
urements were transformed into variables that reflect
the shape indices of the arcs of the semicircular canals
(ASCh/w, PSCh/w, and LSCh/w), the sagittal labyrinth
index (SLI), the radius of the curvature of the semicircu-
lar canals (ASC-R, PSC-R, and LSC-R), and the radius
of curvature of each semicircular canal relative to the
total canal radii (ASC%R, PSC%R, and LSC%R). In
specimens where both the right and left temporal laby-
rinths were conserved, we used the bilateral mean meas-
urements, following Bouchneb and Crevecoeur (2009).
We compared our measurements to those of Pleisto-
cene specimens reported in a number of sources in the
literature. These included 25 Neanderthals (Spoor et al.,
2002b; Spoor et al., 2003; Glantz et al., 2008; Hill et al.,
2014; G
omez-Olivencia et al., 2015), seven Late Pleisto-
cene modern humans (LPM) (Spoor et al., 2002a; Spoor
et al., 2003; Bouchneb and Crevecoeur, 2009; Dobos¸
et al., 2010; Ponce de Le
on and Zollikofer, 2013), two
early modern specimens from Qafzeh and Skhul (QSK)
(Spoor et al., 2003), and three Middle Pleistocene Euro-
peans (MPE) (Spoor et al., 2003). Our dataset thus com-
prised 148 specimens in total, reported in Table 1.
For each variable, differences between the sampled
groups were compared by conducting a Student’s t-test
between Cioclovina, extant humans, Neanderthals, and
Late Pleistocene modern humans (Table 2). Following
Bouchneb and Crevecoeur (2009), we performed a princi-
pal components analysis (PCA) on the correlation matrix
using all 10 variables in order to assess the affinities of
Cioclovina in multivariate space. Nearest neighbors were
calculated using the minimum spanning tree, based on a
Euclidean distance measure among the data points. Miss-
ing values were imputed for the Oase 2 specimen, for
which only five of the variables were available. For each
missing variable, values were imputed by calculating the
mean of the total dataset for that variable. In order to
assess the influence of data imputation for this specimen
on our results, the PCA was repeated using only the five
variables available for Oase 2 [ASCh/w, LSCh/w, SLI,
ASC-R, and LSC-R]. We used the 10 variables to conduct
a linear discriminant analysis (DA). Cioclovina was
treated as unknown in the DA in order to assess its affin-
ities with the groups used. Lastly, we calculated the prob-
ability that this specimen would be classified into other
groups. All multivariate analyses and missing value
imputation were conducted using the PAST software, ver-
sion 2.17c (Hammer et al., 2001), and the JMP software,
version 11.2 (SAS Institute, Cary, NC, USA).
Finally, we used the multivariate equations published
in Osipov et al. (2013) to assess the sex of the Cioclovina
specimen. That study developed discriminant functions,
which can be used to estimate sex for an individual of
unknown sex with up to 82.4% accuracy from bony laby-
rinth dimensions. Positive values indicate male sex
whereas negative values indicate female sex. The equa-
tion with highest classification accuracy (82.4%) was
used here:
TABLE 1. Inner ear comparative sample
Sample
a
Specimens References
Cioclovina This study
ExH (n 5 110) Crete (n 5 79) This study
German (n 5 31) This study
LPM (n 5 7) Abri Pataud 1 Spoor et al., 2003
Abri Pataud 3 Spoor et al., 2003
Cro Magnon 1 Spoor et al., 2003
Laugerie Basse 1 Spoor et al., 2003
Nazlet Khater 2 Bouchneb &
Crevecoeur, 2009
Oase 2 Ponce de Le
on and
Zollikofer, 2013
Muierii 2 Dobos¸ et al., 2010
MPE (n 5 3) Abri Suard Spoor et al., 2003
Reilingen Spoor et al., 2003
Steinheim Spoor et al., 2003
QSK (n 5 2) Qafzeh 6 Spoor et al., 2003
Skhul 5 Spoor et al., 2003
NEA (n 5 25) Dederiyeh Spoor et al., 2002
Gibraltar 1 Spoor et al., 2003
Gibraltar 2 Spoor et al., 2003
Krapina 38.1 Hill et al., 2014
Krapina 38.12 Hill et al., 2014
Krapina 38.13 Hill et al., 2014
Krapina 39.13 Hill et al., 2014
Krapina 39.18 Hill et al., 2014
Krapina 39.20 Hill et al., 2014
Krapina 39.4 Hill et al., 2014
Krapina 39.8 Hill et al., 2014
La Chapelle-
aux-Saints
Spoor et al., 2003
La Ferrassie 1 Spoor et al., 2003
La Ferrassie 2 Spoor et al., 2003
La Ferrassie 3 Spoor et al., 2003
La Ferrassie 8 G
omez-Olivencia
et al., 2015
La Quina 5 Spoor et al., 2003
La Quina 27 Spoor et al., 2003
Le Moustier 1 Spoor et al., 2003
Obi Rakhmat 1 Glantz et al., 2008
Pech-de-l’Az
e 1 Spoor et al., 2003
Petit-Puymoyen 5 Spoor et al., 2003
Spy 1 Spoor et al., 2003
Spy 2 Spoor et al., 2003
Tabun C1 Spoor et al., 2003
a
Sample group abbreviations: extant modern humans (ExH),
Neanderthals (NEA), Late Pleistocene modern humans (LPM),
Qafzeh/Skhul (QSK), Middle Pleistocene Europeans (MPE).
CIOCLOVINA INNER EAR 3
American Journal of Physical Anthropology
PSChMðÞ0:728ðÞ1 PSCwMðÞ1:407ðÞ1 LSCh=wMðÞ
6:217ðÞ1218:749ðÞ:
RESULTS
Figure 1 shows the virtual endocast of the Cioclovina
bony labyrinth in comparison to an extant human (a
German female). The anterior canal does not appear to
have a high level of torsion, consistent with the condi-
tion found in Holocene humans. The posterior canal
position is slightly superiorly placed relative to the lat-
eral canal plane, in contrast to the Neanderthal
condition.
The variables of the right and left bony labyrinths of
Cioclovina, as well as the mean values between right
and left side, are reported in Table 2. The raw measure-
ments for Cioclovina are reported in Supporting Infor-
mation Table S2. The measurement values for our ExH
sample from Crete and Germany are comparable to the
ranges reported by Wu et al. (2014) and Spoor (1993) for
other extant human populations. Notably, the ASC-R
value for Cioclovina places it at the upper end of the
recent human range of variation and outside the Nean-
derthal range. Le Moustier 1 is closest to Cioclovina, fol-
lowed by early and Late Pleistocene modern human
specimens. Results of the Student t-test are also
reported in Table 2. The test comparing the means of
the ExH and NEA shows that nearly all of the variables
examined are significantly different (a 5 0.05) between
these samples. Notably, the SLI, LSC-R, PSC%R,
LSC%R variables are highly significantly different
(a 5 0.0001). The t-test comparing the means of Cioclo-
vina and either the ExH or NEA groups are less
straightforward, with only two variables (ASC-R,
LSC%R) at significant levels for the Cioclovina-NEA
comparison and one variable (PSC%R) for the
Cioclovina-ExH comparison. No variables were found to
be significant for the Cioclovina-LPM comparison.
For the PCA, the first three components account for
approximately 68% of the total variance. PC1 (28% of
variance) tends to separate the Neanderthal and MPE
specimens on the negative side, with some overlap with
ExHs (Fig. 2). The MPE specimens fall mostly within
the NEA variation, with the exception of Abri Suard,
which falls just outside the NEA convex hull and within
that of ExH. The LPM specimens are fully within the
sampled range of the ExH variation. The pattern along
PC1 is largely driven by LSC%R (loading negatively),
PSC%R (loading positively), as well as SLI (loading neg-
atively), and PSC-R (loading positively), consistent with
the univariate results (see Supporting Information Table
3). PC2 (24% of variance) is mainly driven by LSC-R,
PSC%R, and PSC-R, all loading positively (Supporting
Information Table 3). Along PC1 and PC2, Cioclovina
plots in the area of overlap of the NEA and ExH range,
just outside the edge of the LPM convex hull and near
one of the other two Romanian early modern humans,
Oase 2. While Oase 2 as well as Qafzeh 6 fall within the
area of overlap between NEA and LPM/ExH, Muierii 2,
the third Romanian LPM specimen, shows more positive
TABLE 2. Inner ear measurements of Cioclovina and comparative sample
a
ASCh/w PSCh/w LSCh/w SLI ASC-R PSC-R LSC-R ASC%R PSC%R LSC%R
Cioclovina Mean 90.4 102.7 95.8 54.9 3.5 3.1 2.6 37.8 33.9 28.4
Left 94.4 103.0 95.3 54.1 3.6 3.2 2.6 38.4 34.0 27.5
Right 86.4 102.3 96.3 55.7 3.4 3.1 2.6 37.1 33.7 29.2
ExH Mean 90.9 105.6 94.7 47.8 3.2 3.1 2.3 37.0 35.9 27.2
Min 80.7 90.9 80.3 36.8 2.5 2.2 1.6 32.9 31.2 21.5
Max 105.7 121.7 116.3 62.4 3.8 3.7 3.0 41.1 39.1 31.0
SD 4.2 7.1 6.6 6.4 0.3 0.3 0.2 1.3 1.4 1.6
NEA Mean 93.1 102.0 92.0 62.9 3.0 2.8 2.6 36.1 33.6 30.3
Min 84.0 87.0 83.0 52.5 2.7 2.2 2.3 34.0 29.0 28.0
Max 103.0 115.0 105.0 76.0 3.4 3.4 2.9 41.0 37.0 32.1
SD 5.9 7.7 5.8 6.1 0.2 0.3 0.2 1.6 1.8 1.1
LPM Mean 90.0 108.1 93.7 45.0 3.3 3.1 2.5 37.5 34.4 28.1
Min 80.0 96.0 86.0 33.0 3.1 2.8 2.4 36.0 32.5 27.0
Max 98.4 118.0 104.3 55.1 3.6 3.3 2.7 39.5 36.0 29.0
SD 7.1 9.8 3.7 7.7 0.2 0.2 0.1 1.5 1.4 0.8
QSK Mean 81.5 96.5 84.5 43.5 3.4 3.1 2.7 37.0 34.0 29.0
Min 72.0 88.0 82.0 40.0 3.2 3.1 2.6 36.0 33.0 29.0
Max 91.0 105.0 87.0 47.0 3.6 3.1 2.8 38.0 35.0 29.0
MPE Mean 94.3 112.3 87.3 50.3 2.9 2.7 2.4 36.3 33.7 30.3
Min 90.0 96.0 84.0 40.0 2.8 2.7 2.3 35.0 32.0 29.0
Max 98.0 132.0 91.0 60.0 3.1 2.7 2.6 37.0 35.0 31.0
SD 4.0 18.2 3.5 10.0 0.2 0.0 0.2 1.2 1.6 1.2
ExH-NEA P-value 0.023 0.025 0.055 < 0.0001 0.019 0.0002 < 0.0001 0.003 < 0.0001 < 0.0001
ExH-LPM P-value 0.046 0.476 0.293 0.011 0.099 0.880 0.059 0.229 0.006 0.152
NEA-LPM P-value 0.082 0.082 0.503 < 0.0001 0.007 0.069 0.527 0.023 0.234 0.0011
Cio-ExH P-value 0.861 0.555 0.813 0.064 0.069 0.756 0.080 0.398 0.040 0.283
Cio-NEA P-value 0.525 0.910 0.366 0.077 0.001 0.133 0.644 0.147 0.842 0.025
Cio-LPM P-value 0.581 0.500 0.187 0.102 0.424 0.607 0.324 0.958 0.826 0.819
a
Variable abbreviations as reported by Spoor 1993. Brief definitions as follows: ASC: Anterior Semi-circular Canal; PSC: Posterior
Semi-circular Canal; LSC: Lateral Semi-circular Canal; h/w: height divided by width of canal; SLI: Sagittal Labyrinthine Index; R:
Radius of curvature; and %R: Radius of curvature divided by the sum of the radii of all three canals, multiplied by 100 to form a
percentage (Spoor et al. 2003).
The h/w measurements are in mm, SLI and R are indices, %R are percentages.
4 A. UHL ET AL.
American Journal of Physical Anthropology
Fig. 1. Right inner ear virtual endocasts. Cioclovina is shown in purple (left) and an extant German is shown in gray (right).
Lateral (upper) and superior (lower) views.
Fig. 2. Principal Component Analysis, PC1 and PC2 ( 52% of variance). Cioclovina: black filled circle; ExH: Germans in
inverted grey triangles; Cretans in filled grey diamonds; NEA: red crosses (Krapina specimens abbreviated as “K”); LPM: green
crosses; QSK: green diamonds; MPE: turquoise triangles.
CIOCLOVINA INNER EAR 5
PC 1 and PC 2 scores and falls outside the NEA convex
hull and well within the LPM and ExH ranges. Along
these axes Skhul 5 falls within the ExH range, and just
outside the LPM one. PC 3 (16% of variance) tends to
separate the MPE and LPM humans. It is influenced by
ASC%R, ASC-R (loading positively), and PSC%R (load-
ing negatively). Along PC1 and PC3 (Fig. 3), Cioclovina
and Muierii 2 fall within the sampled ExH and LPM
ranges and outside the Neanderthal convex hull. Qafzeh
6 plots away from the Neanderthal convex hull, and
within the ExH range. On the other hand, Oase 2
remains in the area of overlap between the NEA and
LPM/ExH convex hulls, as does Skhul 5. In this plot, all
MPE specimens are within the Neanderthal range, with
Steinheim positioned at the edge of the convex hull. Cio-
clovina’s nearest neighbors in the minimum spanning
tree (plotted on the PCA graphs in Supporting Informa-
tion Figs. 1 and 2) are a Cretan and a German individ-
ual. Muierii 2 has 2 individuals from Crete as nearest
neighbors. Oase 2, on the other hand, is nearest neigh-
bor with both a modern human (Cretan) as well as a
Neanderthal (Krapina 39.8; see Supporting Information
Figs. 1 and 2).
Because five of the variables used in the PCA are not
available for Oase 2 and were therefore imputed, we
repeated the PCA using only the five variables reported
for this specimen (ASCh/w, LSCh/w, SLI, ASC-R, and
LSC-R), in order to assess the influence of the imputed
measurements on our results (Supporting Information
Fig. 3). In the five variable PCA, PC 1 accounts for
35% of the variance and PC 2 for 25%. As might be
expected, the overlap between NEA and ExH/LPM, as
well as between MPE and ExH, is greater in this
reduced-variable analysis. However, Cioclovina is no lon-
ger in the NEA overlap space but clearly within the
ExH convex hull and on the edge of the LPM convex
hull. The other two Romanian specimens, Muierii 2 and
Oase 2 remain within the ExH-NEA-LPM overlap space.
Qafzeh 6 falls with the ExH ranges and just outside the
LPM convex hull, while Skhul 5 plots outside the range
of any of the samples along these axes, although closest
to the ExH and LPM.
Results of the discriminant analysis procedure yielded
an overall 80.19% classification accuracy with cross-
validation (Table 3). Cioclovina was classified in the ExH
group (57% probability), with a 32% chance of being
classified as LPM and 10% as NEA.
Using the Osipov et al (2013) discriminant function
with the highest classification accuracy on the Cioclo-
vina data, we obtained the value of 0.45, indicating male
sex. However, this value is relatively low, giving a 68%
confidence for this sex estimation according to the poste-
rior probabilities from Osipov et al. (2013).
DISCUSSION
As the t-tests show, the variables sampled here are
useful in differentiating between extant humans and
Neanderthals, consistent with previous studies (Spoor
et al., 2002a,b,2003; Bouchneb and Crevecoeur, 2009).
The t-test results, however, are not unequivocal when
assessing the differences between Cioclovina and either
ExH or NEA. However, the fact that none of the varia-
bles were significantly different between Cioclovina and
the LPM sample suggests that the former may be best
placed in the latter group, as also expected according to
its cranial morphology (Harvati et al., 2007) and its
Fig. 3. Principal Component Analysis, PC1 and PC3 (44% of variance). Symbols and colors as in Fig. 2.
6 A. UHL ET AL.
American Journal of Physical Anthropology
geological age. SLI, a measurement which quantifies the
position of the posterior canal relative to the lateral
canal, has been found by Spoor et al. (2003) to signifi-
cantly differentiate Neanderthal from Holocene, Upper
Paleolithic, early modern, and Middle Pleistocene speci-
mens. However, there is overlap of the SLI values of the
different samples used here and Cioclovina falls within
the overlap zone. It has an SLI of 54.9 for the bilateral
mean (54.1 for left and 55.7 for the right side, Table 2),
which places it within the ExH, NEA, QSK, and MPE
SLI range and at the maximum SLI value for the LPM.
In our PCA, Neanderthals and LPM overlap somewhat
along PCs 1 and 2 as well as PCs 1 and 3. The overlap
is driven primarily by the inclusion of the Le Moustier 1
specimen, which is the only Neanderthal with a positive
PC1 score. This is consistent with previous univariate
reports which found Le Moustier 1’s bony labyrinth
measurements to be within the range of modern humans
(Spoor et al., 2003) and Homo erectus (Ponce De Le
on
and Zollikofer, 1999). This finding has been interpreted
to represent either the lower range of Neanderthal vari-
ation or, alternatively, gene flow between modern
humans and Neanderthals (Spoor et al., 2003). Cioclo-
vina falls well within the ExH, at the edge of the NEA,
and very near the LPM convex hulls along PC1 and PC2
(Fig. 2). It would not overlap with the Neanderthal con-
vex hull if Le Moustier 1 were excluded from the analy-
sis. Indeed, along PC1 and PC3 (Fig. 3), Cioclovina is
fully within the convex hulls of both ExH and LPM and
outside the Neanderthal range of variation. Here, too,
the overlap between NEA and modern humans is driven
by Le Moustier 1. Our discriminant analysis classified
Cioclovina as a likely ExH, with the next likeliest classi-
fication as LPM and only very low probability (10%) of
being a Neanderthal.
In light of the genomic findings of recent Neanderthal
ancestry for Oase 1 (Fu et al., 2015), the position of the
two other Romanian Upper Paleolithic specimens
included in our analyses is also interesting. For SLI,
Oase 2, a different individual from the Oase 1 mandible,
falls within the ExH, MPE, and LPM range, while
Muierii 2 is below the NEA range but within the ExH,
MPE and LPM ranges. While Muierii 2 is well within
the ExH range and plots away from the Neanderthal
convex hull, Oase 2 consistently falls within the area of
overlap between Neanderthals and modern humans in
the PCA (both PC1-2 as well as PC1-3)—although it
should be noted that this is because of the inclusion of
Le Moustier 1 in the Neanderthal sample. Furthermore,
while both Cioclovina and Muierii 2 have ExH speci-
mens as their nearest MST neighbors, Oase 2 is nearest
neighbor with both a NEA and an ExH specimen. Since
Oase 2 is missing five variables, these were imputed
using mean substitution across the entire sample.
Because the majority of our sample represents ExH
specimens, the imputed averages used for Oase 2 should
also be most similar to ExH. The above results for Oase
2, therefore, are all the more remarkable for showing an
inner ear morphology with some Neanderthal affinities.
When the PCA is repeated with only the five variables
preserved in Oase 2, both it and Muierii 2 are the only
LPM specimens falling within the Neanderthal-modern
human overlap area (Supporting Information Fig. 3).
The discriminant analysis results classify Muierii 2 as
ExH to 90%. Unfortunately, Oase 2 could not be reliably
classified because of the lack of variables. When the dis-
criminant analysis was repeated using only the five vari-
ables available for Oase 2, this specimen was classified
as 41% MPE, 26% ExH, and 22% NEA. However, the
overall classification success of the reduced-variable DA
is very low at 59.86%.
The sex of Cioclovina has been a matter of discussion.
While Rainer and Simionescu (1941) considered it a
female, Smith (1984) and Soficaru et al. (2007) proposed
that its robust morphology suggests it is a male. Harvati
et al. (2007), on the other hand, found that Cioclovina’s
nearest neighbor in terms of its cranial morphology was
Abri Pataud, a specimen described as female. Similarly,
in the analysis reported here its two nearest neighbors
were also female. Cioclovina’s large superciliary arches
and mastoid process, when compared to females such as
Mladec
ˇ
1 and 2 and Muierii 1 appear masculine, but
when compared to males such as Mladec
ˇ
5 and 6, are
not as robust (Soficaru et al., 2007). When applying a
discriminant function developed by Osipov et al. (2013)
to estimate sex on the basis of its inner ear dimensions,
we obtained value of 0.45, indicating that Cioclovina is
male. Nevertheless, this estimation has relatively low
confidence, consistent with the previously reported
ambiguous results. However, the Osipov et al. (2013)
equation was developed on a recent Greek sample, which
may not be an appropriate model for sexing Upper Pale-
olithic individuals. More research on geographically
diverse populations and time periods is needed before
this approach can be applied with greater confidence to
fossil skeletal remains.
CONCLUSIONS
In summary, our multivariate results show that the
Cioclovina inner ear morphology lies within the range of
variation of ExH and is also similar with that of LPM.
However, it also falls in the area of overlap between
NEA and ExH ranges of variation, which suggests an
intermediate morphology between the two taxa. Never-
theless, this overlap is driven by one specimen, Le
Moustier 1, which shows a modern human like inner ear
morphology uncharacteristic of other NEA specimens.
Cioclovina’s nearest neighbors, calculated through the
TABLE 3. Discriminant analysis results
a
ExH NEA LPM QSK MPE Total
ExH 90 (82%) 3 (2.7%) 14 (13%) 0 (–) 3 (2.7%) 110
NEA 0 (–) 21 (84%) 0 (–) 1 (4%) 3 (12%) 25
LPM 1 (16.7%) 0 (–) 5 (83.3%) 0 (–) 0 (–) 6
b
QSK 0 (–) 0 (–) 2 (100%) 0 (–) 0 (–) 2
MPE 0 (–) 1 (33%) 0 (–) 0 (–) 2 (67%) 3
Classifications 91 25 21 1 8 146
a
Cross-validated results and classification percentages in parentheses, rounded to the nearest whole number.
b
Oase 2 is missing too many variables to be included in the DA.
CIOCLOVINA INNER EAR 7
American Journal of Physical Anthropology
minimum spanning tree, are both ExH specimens, a
German female and a Cretan female. Our discriminant
analysis results are generally consistent with both our
univariate and multivariate analyses. Our results
broadly agree with the conclusions drawn from previous
work on the Cioclovina external cranial (Harvati et al.,
2007) as well as endocranial (Kranioti et al., 2011) mor-
phology, which show a modern human anatomy in many
cases similar to other Upper Paleolithic specimens.
Although partial Neanderthal ancestry cannot be conclu-
sively excluded on the basis of morphology alone, no
clear evidence for recent admixture in this specimen
could be suggested from our results. Our analyses found
similar results for the Muierii 2 Romanian Upper Paleo-
lithic individual. However, the inner ear anatomy of
Oase 2, the last early Upper Paleolithic Romanian speci-
men included here, although not completely preserved,
is more ambiguous and appears to show some Neander-
thal affinities. This could be consistent with the hypothe-
sis of a higher percentage of Neanderthal ancestry for
this individual, as previously proposed on the basis of its
cranial anatomy (Rougier et al., 2007) and as recently
determined for Oase 1 on the basis of aDNA analysis
(Fu et al., 2015). Further work on these specimens,
including paleogenetic analyses, will help shed light on
this question.
ACKOWLEDGMENTS
We thank Wieland Binczik, Michael Franken, and Lisa
Kellner for their assistance with the German skeletal col-
lection, as well as Dr. T. Ciprut and staff from the scanning
facilities in Bucharest and Crete.
LITERATURE CITED
Ackermann RR. 2010. Phenotypic traits of primate hybrids: rec-
ognizing admixture in the fossil record. Evol Anthropol 19:
258–270.
Ackermann RR, Rogers J, Cheverud JM. 2006. Identifying the
morphological signatures of hybridization in primate and
human evolution. J Hum E 51:632–645.
Bouchneb L, Crevecoeur I. 2009. The inner ear of Nazlet Khater
2 (Upper Paleolithic, Egypt). J Hum E 56:257–262.
Braga J, Thackeray JF, Dumoncel J, Descouens D, Bruxelles L,
Loubes J-M, Kahn J-L, Stampanoni M, Bam L, Hoffman J,
Frikkie de Beer, Spoor F. 2013. A new partial temporal bone
of a juvenile hominin from the site of Kromdraai B (South
Africa). J Hum E 65:447–456.
Dobos¸ A, Soficaru A, and Trinkaus E. 2010. The prehistory and
paleontology of the Pes¸tera Muierii (Romania): Universit
ede
Lie
`
ge, Service de Pr
ehistoire.
Frost SR, Marcus LF, Bookstein FL, Reddy DP, Delson E. 2003.
Cranial allometry, phylogeography, and systematics of large-
bodied papionins (Primates: cercopithecinae) inferred from
geometric morphometric analysis of landmark data. Anat Rec
275A:1048–1072.
Fu Q, Hajdinjak M, Moldovan OT, Constantin S, Mallick S,
Skoglund P, Patterson N, Rohland N, Lazaridis I, Nickel B,
Viola B. 2015. An early modern human from Romania with a
recent Neanderthal ancestor. Nature 524:216–219.
Glantz M, Viola B, Wrinn P, Chikisheva T, Derevianko A,
Krivoshapkin A, Islamov U, Suleimanov R, Ritzman T. 2008.
New hominin remains from Uzbekistan. J Hum E 55:223–
237.
G
omez-Olivencia A, Crevecoeur I, Balzeau A. 2015. La Ferras-
sie 8 Neandertal child reloaded: New remains and re-
assessment of the original collection. J Hum E 82:107–126.
Gunz P, Ramsier M, Kuhrig M, Hublin J-J, Spoor F. 2012. The
mammalian bony labyrinth reconsidered, introducing a com-
prehensive geometric morphometric approach. J Anat 220:
529–543.
Hammer Ø, Harpet D, Ryan P. 2001. Paleontological Statistics
software package for education and data analysis. Palaeontol
Electron 4:9
Harvati K and Roksandic M. Accepted. The human fossil record
from Romania: Early upper Paleolithic European Mandibles
and Neanderthal Admixture. In: Harvati K, Roksandic M,
editors. Paleoanthropology of the Balkans and Anatolia:
Human Evolution and its Context. Dordrecht: Springer.
Harvati K, Gunz P, Grigorescu D. 2007. Cioclovina (Romania):
Affinities of an early modern European. J Hum E 53:732–746.
Hill CA, Radovc
ˇ
ic
´
J, Frayer DW. 2014. Brief communication:
Investigation of the semicircular canal variation in the Kra-
pina Neandertals. Amer J Phys Anthrop 154:302–306.
Hublin J-J, Spoor F, Braun M, Zonneveld F, Condemi S. 1996.
A late Neanderthal associated with Upper Palaeolithic arte-
facts. Nature 381:224–226.
Kranioti EF, Holloway R, Senck S, Ciprut T, Grigorescu D,
Harvati K. 2011. Virtual assessment of the endocranial mor-
phology of the early modern European fossil calvaria from
Cioclovina, Romania. Anat Rec 294:1083–1092.
L’Engle Williams F, Richtsmeier JT. 2003. Comparison of man-
dibular landmarks from computed tomography and 3D digit-
izer data. Clin Anat 16:494–500.
McKinley MP, and O’Loughlin VD. 2006. Human Anatomy. New
York: McGraw-Hill.
Necrasov O, Cristescu M. 1965. Donn
ees anthropologiques sur
les populations de l’age de la pierre en Roumanie. Homo 16:
129–161.
Olariu A, Skog G, Hellborg R, Stenstr
€
om K, Faarinen M,
Persson P, Alexandrescu E. 2005. Dating of two Paleolithic
human fossil bones from Romania by accelerator mass spec-
trometry. Applications of High-Precision Atomic and Nuclear
Methods. Editura Academiei Romaine, Bucharest 222–226.
Osipov B, Harvati K, Nathena D, Spanakis K, Karantanas A,
Kranioti EF. 2013. Sexual dimorphism of the bony labyrinth:
A new age-independent method. Amer J Phys Anthrop 151:
290–301.
Park S-H, Yu H-S, Kim K-D, Lee K-J, Baik H-S. 2006. A pro-
posal for a new analysis of craniofacial morphology by 3-
dimensional computed tomography. Am J Orthod Dentofacial
Orthop 129:600.e623–600.e634.
Ponce De Le
on MS, Zollikofer CPE. 1999. New evidence from
Le Moustier 1: Computer-assisted reconstruction and mor-
phometry of the skull. Anat Rec 254:474–489.
Ponce de Le
on M, and Zollikofer C. 2013. The internal cranial
morphology of Oase 2. Life and Death at the Pes¸tera cu Oase.
In: Trinkaus E, Constantin S, Zilh
~
ao J, editors. A Setting for
Modern Human Emergence in Europe. New York: Oxford
Univ Press. p 332–347.
Rainer F, Simionescu I. 1942. Sur le premier cr
^
ane d’homme
Pal
eolithique trouv
e en Roumanie. Analele Academiei
Rom
^
ane, Memoriile Sect
¸iunii S¸tiint¸ifice 3:489–503.
Richtsmeier JT, Paik CH, Elfert PC, Cole TM, Dahlman HR.
1995. Precision, repeatability, and validation of the localiza-
tion of cranial landmarks using computed tomography scans.
Cleft Palate Craniofac J 32:217–227.
Rougier H, Milota S, Rodrigo R, Gherase M, Sarcin
aL,
Moldovan O, Zilh
~
ao J, Constantin S, Franciscus RG,
Zollikofer CP, de Le
on MP. 2007. Pes¸tera cu Oase 2 and the
cranial morphology of early modern Europeans. Proc Natl
Acad Sci USA 104:1165–1170.
Scherf H, Tilgner R. 2009. A new high-resolution computed
tomography (CT) segmentation method for trabecular bone
architectural analysis. Amer J Phys Anthrop 140:39–51.
Smith, FH. 1984. Fossil hominids from the Upper Pleistocene of
central Europe and the origin of modern Europeans. The ori-
gins of modern humans. In: Smith FH, Spencer F, editors.
New York: Alan R. Liss. p 137–209.
Soficaru A, Petrea C, Dobos¸ A, Trinkaus E. 2007. The human
cranium from the Pes¸tera Cioclovina Uscat
a, Romania: Con-
text, age, taphonomy, morphology, and paleopathology. Curr
Anthrop 48:611–619.
8 A. UHL ET AL.
American Journal of Physical Anthropology
Spoor CF. 1993. The comparative morphology and phylogeny of
the human bony labyrinth: Universiteit Utrecht, Faculteit
Geneeskunde.
Spoor F. 2003. The semicircular canal system and locomotor
behaviour, with special reference to hominin evolution. Cou-
rier-Forschungsinstitut Senckenberg 93–104.
Spoor F, Esteves F, Tecel
~
ao Silva F, Pacheco Dias R. 2002a. The
bony labyrinth. In: Zilh
~
ao J, Trinkaus E, editors. Portrait of
the Artist as a Child: The Gravettian Human Skeleton from
the Abrigo do Lagar Velho and its Archaeological Context.
Lisbon: Instituto Portugu
^
es de Arqueologia. p 610.
Spoor F, Hublin J-J, Kondo O. 2002b. The bony labyrinth
of the Dederiyeh child. In: Akazawa T, Muhesen S,
editors. Neanderthal Burials. Excavation of the Dederiyeh
Cave, Afrin, Syria. Tokyo: The Tokyo University Press. p 215–
220.
Spoor F, Hublin J-J, Braun M, Zonneveld F. 2003. The bony lab-
yrinth of Neanderthals. J Hum E 44:141–165.
Spoor F, Zonneveld F. 1995. Morphometry of the primate bony
labyrinth: a new method based on high-resolution computed
tomography. J Anat 186:271
Trinkaus E. 2007. European early modern humans and the fate
of the Neandertals. Proc Natl Acad Sci U S A 104:7367–7372.
Wu X-J, Crevecoeur I, Liu W, Xing S, Trinkaus E. 2014. Tempo-
ral labyrinths of eastern Eurasian Pleistocene humans. Proc
Natl Acad Sci U S A.
CIOCLOVINA INNER EAR 9
American Journal of Physical Anthropology
