The Qesem Cave hominin material (part 2): A morphometric analysis of dm2-QC2 deciduous lower second molar

Quaternary International 398 (2016) 175e189

Contents lists available at ScienceDirect

Quaternary International journal homepage: www.elsevier.com/locate/quaint

The Qesem Cave hominin material (part 2): A morphometric analysis of dm2-QC2 deciduous lower second molar Cinzia Fornai a, b, *, Stefano Benazzi c, d, Avi Gopher e, Ran Barkai e, Rachel Sarig f, g, Fred L. Bookstein a, h, Israel Hershkovitz f, i, Gerhard W. Weber a, j a

Department of Anthropology, University of Vienna, Austria Institute of Evolutionary Medicine, University of Zurich, Switzerland Department of Cultural Heritage, University of Bologna, Italy d Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology, Germany e Institute of Archaeology, Tel Aviv University, Israel f Dan David Center for Human Evolution and Biohistory, The Steinhardt Museum of Natural History and National Research Center, Tel Aviv University, Israel g The Department of Orthodontics, The Maurice and Gabriela Goldschleger School of Dental Medicine, Israel h Department of Statistics, University of Washington, USA i The Department of Anatomy and Anthropology, Sackler Faculty of Medicine, Tel Aviv University, Israel j Core Facility for Micro-Computed Tomography, University of Vienna, Austria b c

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 13 January 2016

The Qesem Cave Middle Pleistocene hominin site has yielded a well preserved lower second deciduous molar (dm2-QC2), among several other human dental remains. It has been previously described by Hershkovitz et al. using traditional methods. In this study, we used multiple approaches in order to characterize the outer and inner morphology of dm2-QC2, namely a descriptive investigation of the inner morphology, analysis of the dental tissues, and comparative 3D geometric morphometric investigation of various aspects of the dental crown based on data gathered from mCT images. Dm2-QC2 was compared to a sample of 44 specimens, including recent and fossil modern humans, Neanderthals, and Homo erectus. Our comprehensive quantitative investigation agrees with Hershkovitz et al. with regard to the mixed morphology of this specimen. Dm2-QC2 allies morphologically with Neanderthals and Skhul X for its squared cervical outline, but is intermediate between modern humans and Neanderthals for its mildly distally expanded crown outline. Dm2-QC2 falls within Neanderthal variability in having relatively high dentine horns with inwardly bent tips. It is peculiar for its mesio-distal elongation of the occlusal marginal ridge at the enameledentine junction. The relative enamel thickness delivers different results if measured at the mesial section (intermediate but closer to modern humans) or at the entire crown (close to the Neanderthal distribution). In terms of size, dm2-QC2 and Qafzeh 15 are among the largest specimens in our sample. Dm2-QC2 shows a mosaic of features, with a prevalence of those typical or at least frequent in Neanderthals. Among the latter, the mid-trigonid crest and taurodontism are only slightly expressed. On the basis of this isolated dental evidence, and considering the paucity of information available on human evolution in the Levantine Middle Pleistocene, we do not attempt a taxonomic classification of this single tooth. However, dm2-QC2 indicates that Neanderthal features were already present in the Middle Pleistocene Levant. We cannot rule out the possibility that the dm2-QC2 population contributed to later Levantine or other Eurasian populations, but this is too speculative given the information currently available. © 2015 Elsevier Ltd and INQUA. All rights reserved.

Keywords: Relative enamel thickness Middle Pleistocene Levant Milk tooth Homo Neanderthals

1. Introduction

* Corresponding author. E-mail address: [email protected] (C. Fornai). http://dx.doi.org/10.1016/j.quaint.2015.11.102 1040-6182/© 2015 Elsevier Ltd and INQUA. All rights reserved.

The Qesem Cave (QC) Middle Pleistocene hominin site has yielded several deciduous and permanent teeth (Hershkovitz et al., 2011, 2016; Weber et al., 2016) associated to the Acheulo-Yabrudian

176

C. Fornai et al. / Quaternary International 398 (2016) 175e189

Cultural Complex (AYCC) and dated to about 420e220 ka (Barkai
res et al., 2003; Gopher et al., 2010; Mercier et al., 2013; Falgue et al., 2015). A complete Homo sp. lower left second deciduous molar (dm2) (hereinafter dm2-QC2; Fig. 1) was found in sub-square I/12a of the upper part of the Lower Sequence of the QC deposits which contained an Amudian lithic industry, dated to approximately 300 ka, or older (Hershkovitz et al., 2011). Dm2-QC2 has not been found in association with other QC human remains. It has been previously described by Hershkovitz et al. (2011; therein referred to as “Qesem dm2” or simply “dm2”) as a large, taurodontic tooth, with mesial cusp tips inclined inward and expanded talonid. These authors also stressed the presence of an anterior fovea and the possible lack of a trigonid crest (no image data of the enameledentine junction were available at that time). They pointed out that some of these features, although frequent in Neanderthals (NEA), were also shared with other Late Pleistocene Homo (e.g. Qafzeh and Skhul). In terms of linear size measurements (i.e., bucco-lingual distance) Hershkovitz et al. found dm2-QC2 to exceed the range of their recent modern human (MH) sample and allied dm2-QC2 to Qafzeh 15 early MH for its occlusal area, both specimens showing larger crown areas than some later NEA (i.e., , and La Ferrassie 8). On the basis of these Kebara 1, Pech de l'Aze observations, Hershkovitz et al. (2011) considered a possible taxonomic affinity of dm2-QC2 to the Skhul/Qafzeh group. Further quantitative studies on other QC and Levantine teeth showed a complex picture for the dental remains, in terms of both size and morphology (Weber et al., 2016). In spite of previous works (Hershkovitz et al., 2011, 2016; Sarig et al., 2016; Weber et al., 2016), there is still much uncertainty about the taxonomic nature of the QC inhabitants. First, they are represented by isolated teeth only. Second, dental variability of Middle Pleistocene hominins is largely unexplored, especially for the Levant, where human remains are

very scarce. Likewise, the dm2 hypodigm comparable to dm2-QC2 is small. In this contribution, we provide a description of dm2-QC2 using multiple approaches based on the investigation of 3D image data. We carried out a qualitative description of the dm2-QC2 enameledentine junction (EDJ), an investigation of the dental tissue proportions, and an analysis of the enamel thickness in both 2D and 3D, aspects yet unexplored. In fact, these features have been claimed to be a taxonomically distinctive factor within Homo (Kono et al., 2002; Smith et al., 2012), and between MH and NEA, both in permanent (e.g., Zilberman et al., 1992; Olejniczak et al., 2008a,b; Kupczik and Hublin, 2010) and deciduous teeth (e.g., Zilberman and Smith, 1992; Zanolli et al., 2010, 2012; Benazzi et al., 2011a,b, 2015; Macchiarelli et al., 2013; Fornai et al., 2014). Moreover, we performed a geometric morphometric (GM) investigation of both inner and outer aspects of the crown based on 3D coordinate data. In particular, we considered the occlusal ridge at the EDJ since this feature might carry a taxonomic signal (e.g., Corruccini, 1987, 1998; Olejniczak et al., 2004, 2007; Macchiarelli et al., 2006; Skinner et al., 2009; Fornai et al., 2015). From the outer aspect, we analyzed the crown and cervical outlines of the dental crown (Benazzi et al., 2012, 2014a) in order also to include specimens showing moderate degree of occlusal wear in our comparative sample. 2. Materials and methods 2.1. Sample Our sample is composed of 3D surface models from MicroComputed Tomography (mCT) data of dm2-QC2 and 44 comparative dm2s from Holocene and Pleistocene MH (n ¼ 22 and n ¼ 6,

Fig. 1. 3D surface models of dm2-QC2 crown. Enamel (in grey) and dentine (in yellow) in various views. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

C. Fornai et al. / Quaternary International 398 (2016) 175e189

respectively), NEA (n ¼ 15), and Homo erectus (n ¼ 1) (details in Table 1). The recent MH (RMH), a subsample of the MH sample, includes individuals from Europe, as well as Bedouins and Epipaleolithic specimens from Israel, i.e. Natufians. Ten MH and seven NEA specimens showing a comparable degree of wear to dm2-QC2 at the mesial section were considered for the analysis of the enamel thickness (namely, stage 3 of Molnar's classification, 1971). For analysis of the 3D enamel thickness, the comparative sample (18 MH, 7 NEA, and 1 H. erectus) included specimens with wear stages 1e3 (Molnar, 1971) since the volumetric analysis is less sensitive to minor enamel variation (i.e., enamel loss) than the analysis of a single section.

177

Details about the mCT scanning procedures applied to the other specimens can be found in Toussaint et al. (2010), Benazzi et al. (2011b, 2012), Zanolli et al., 2012, and the Neanderthal Studies Professional Online Service, NESPOS Database 2011 (www.nespos. org). The mCT data were virtually segmented through thresholding (according to the protocol of Spoor et al., 1993) and were manually corrected using the software package Amira 5.6 (FEI, http://www. vsg3d.com/) so as to isolate the dental crown and produce 3D models of its outer enamel surface and EDJ. In case of minor wear affecting one or more of the horn tips, the latter were virtually reconstructed using the “brush” tool to select and extend the

Table 1 Dm2 sample compared to dm2-QC2. Taxon

Specimen

Abbreviation

Provenience

Source for mCT data

Crown outline

Cervical outline

EDJ

Homo erectus

PCG.2

PCG2

Indonesia

x

x

x

Neanderthals

Abri Suard S14-5* Abri Suard S37b* Abri Suard S42* Archi 1 Cavallo A* Couvin 1* Engis 2* Krapina d62* Krapina d63* Krapina d64* Krapina d65* Krapina d66* Krapina d68* Roc de Marsal* Scladina 4A 13* stonice 36* Dolní Ve Lagar Velho 1a,* Madeleine 4* Qafzeh 15 (L) Skhul I (L) Skhul X Natufians n ¼ 6

AS_S14-5 AS_S37b AS_S42 Archi1 CavA Cou1 Engis2 Kr_d62 Kr_d63 Kr_d64 Kr_d65 Kr_d66 Kr_d68 RdM Sclad DolniVest36 LagarVel1 Mad4 Qafzeh15 SkhulI SkhulX AM_H57, AM_H88, Hay_H12, Hay_H13, Hay_H18, Hay_H28 BLZ_004, BLZ_273, BLZ_279, BLZ_294 RHS

France France France Italy Italy Belgium Belgium Croatia Croatia Croatia Croatia Croatia Croatia France Belgium Czech Republic Portugal France Israel Israel Israel Israel

Centre de Micro-tomographie, University of Poitiers NESPOS NESPOS NESPOS NESPOS Vienna mCT Lab Toussaint et al., 2010 Toussaint et al., 2010 NESPOS NESPOS NESPOS NESPOS NESPOS NESPOS NESPOS Toussaint et al., 2010 Vienna mCT Lab NESPOS NESPOS Vienna mCT Lab Vienna mCT Lab Vienna mCT Lab Vienna mCT Lab

x x x x x x x x x x x x x x

x x x

x x x

x x x

x x x

x

x

x

x

x

x

x x

x x

x x

x x x x x n¼6

x x x x x n¼6

x x x x x x n¼2

x

x x x

x

x

n¼2

n¼1

Israel

Vienna mCT Lab

n¼4

n¼4

n¼4

n¼1

n¼2

Austria

Vienna mCT Lab

n ¼ 12

n ¼ 12

n¼9

n¼5

n ¼ 11

Total

n ¼ 42

n ¼ 41

n ¼ 29

n ¼ 17

n ¼ 26

Modern humans

Bedouins n ¼ 4

Contemporary individuals*, n ¼ 12

2D ET

3D ET x

x x x x x x x x x x

a

For the taxonomic attribution of Lagar Velho 1 we referred to Benazzi et al. (2012); * ¼ crown and cervical outlines from Benazzi et al. (2012); x ¼ comparative individuals considered for each of the analyses performed; EDJ ¼ enameledentine junction; ET ¼ enamel thickness; L ¼ left.

We considered 44 specimens for the crown outline analysis, of which 41 were used for the cervical outline analysis. Among these individuals, only 29 teeth (21 MH, 7 NEA, and 1 H. erectus) could be compared to dm2-QC2 for the EDJ analysis, since the others presented damage or moderate to heavy dental wear affecting the areas of interest. When both antimeres were available we considered the one in the better state of preservation. Specimens from the right side were mirrored to the left. 2.2. Image data acquisition We used mCT scans at isotropic voxel size between 15 and 55 mm. All specimens from Israel were scanned at the Vienna mCT Lab by means of a Viscom X8060 mCT scanner (130 kV, 100 mA).

contour of the dentine to the missing area. This selection was performed in each case on two individual slices at a distance of a few slices from each other. The volume between these two slices was then delineated by interpolation. 2.3. Analysis of the inner structure We provided descriptive information on the features of dm2QC2 inner structure that became accessible with the mCT scans, and integrated this description with the one already given in Hershkovitz et al. (2011). In particular, we described the morphology of the EDJ, pulp chamber and root canals (referring to: Korenhof, 1982; Turner et al., 1991; Kupczik and Hublin, 2010; nBailey et al., 2011; Martínez de Pinillos et al., 2014; Martino

178

C. Fornai et al. / Quaternary International 398 (2016) 175e189

Torres et al., 2014). In addition, we provided a tentative assessment of the total volume for the dentine (although root tips are incomplete) and root canals after removal of the matrix encrustation from roots, pulp chamber, and root canals. We assessed presence and degree of taurodontism based on Keene (1966). 2.4. 2D and 3D analyses of the enamel thickness and dental tissue proportions The 2D enamel thickness was evaluated on the mesial section following the indications given in Benazzi et al. (2014b). We briefly describe below the protocol used. After the dental crown was segmented and the mCT image stack was realigned according to the cervical plane, we marked the mesial horn tips and identified the section perpendicular to the cervical plane and passing through the mesial horn tips (in Amira 5.6). Because wear affected the protoconid horn tip in part of the sample (but not in dm2-QC2), we identified the point through which the section should pass as explained in Fornai et al. (2014). Sections other than the mesial were not considered because the buccal side is often the most worn in lower molars, and the lingual side, which on the contrary is less affected by wear, was not suitable for the very variable position of the lingual horn tips along the bucco-lingual direction (see section 3.6 and Fig. SM2). The distal section has been considered in very few studies (Smith et al., 2005; Mahoney, 2011) aimed at evaluating the variation in enamel tissue proportions in the dental row. Here we focused only on the mesial section since our main aim is a comparative assessment of dm2-QC2 to known Homo groups. The mesial section was segmented and virtual models of the enamel and dentine surfaces were created using a constrained smoothing. The surface was then imported in Rapidform XOR and cut by a line passing through the most apical points of the enamel cap on the section. The area of the enamel, the area of the coronal dentine (including the pulp chamber), and the length of the EDJ were gathered from the section. These values were used to calculate the relative enamel thickness (RET ¼ (enamel area/EDJ length)/Square root of the dentine area; scale free index) (Martin, 1985; Olejniczak et al., 2008a,b). A 3D investigation of the enamel thickness was performed in spite of the erosion present on the dm2-QC2 outer surface. The small inlay of enamel missing at the lingual cervical third of the metaconid was virtually restored for this analysis (which however represents less than 0.2% of the total volume). We followed methods 3D-a by Benazzi et al. (2014b), where the molar crown is separated by the roots at the best-fit plane of the cervical margin. Similarly to the 2D approach, the enamel was virtually segmented from the dentine and the following values were gathered: enamel volume, coronal dentine volume (including the pulp chamber), and area of the EDJ in order to calculate the 3D relative enamel thickness (RET ¼ (enamel volume/EDJ area)/Cube root of the dentine volume; scale free index) (Martin, 1985; Olejniczak et al., 2008a,b). We report also the enamel thickness values for dm2-QC2 after virtual correction of the enamel from the erosional pits and patches, which led us to a coronal enamel volume 5% larger than before this restoration. For the computation of dm2-QC2 RET index, we used the coronal enamel volume after virtual correction. We used a permutation test (n ¼ 1000) for the difference between MH and NEA group mean RET indices. We performed a quadratic discriminant analysis (QDA; see for example, Bookstein, 2014, section 6.5.2) in order to assess the likelihood that dm2QC2 could be classified to either MH or NEA. QDA is appropriate because the subsamples have different variances and sample sizes.

Considering the RET for the analysis of the enamel thickness, the models were built leaving out dm2-QC2 data. Each QDA analysis is reported simply as one single likelihood ration (LR) for the hypothesis that dm2-QC2 arises from the NEA sample. The p ~0.05 threshold for this is 6.83, and that for the hypothesis that dm2-QC2 arises from MH is 0.146. The data were analyzed using software routines written in “R” software (R Development Core Team, 2013). The dental tissue proportions for both mesial enamel and dentine areas, and coronal enamel and dentine volumes were calculated in Excel spreadsheets. In addition, we rendered enamel distribution (Amira 5.6) through a 3D topographic mapping of the crown which uses a chromatic scale from dark-blue to red indicating increasing thickness. 2.5. Data collection from crown and cervical outlines The image data was further processed following the instructions given in Benazzi et al. (2012), with some modifications as explained below. The models of the dental crowns (outer surfaces) were imported in RapidForm XOR2 (INUS Technology) and realigned so that the best fit plane at the cervical margin was parallel to the XY plane and the lingual margin was parallel to the X axis (Fig. 2). The crown and cervical outlines were collected and projected onto the cervical plane. The crown outline corresponded to the silhouette of the oriented tooth as seen in occlusal view. The cervical outline consisted of the perimeter of the dental crown at the intersection with the cervical plane. The outlines were further processed in Rhinoceros 4.0 Beta CAD environment (Robert McNeel & Associates, Seattle, WA; www.rhino3d.com), as follows. Each outline was intersected by a fan of 24 equiangularly spaced radii originating from the centroid of its area and was resampled using 24 pseudolandmarks located at the intersection with the fan (Benazzi and colleagues used, instead, 16 pseudolandmarks). As detailed in Table 1, a great part of the sample was already used in Benazzi et al. (2012), therefore for those individuals we used the outlines already gathered for that study, and performed the collections of the landmark points ex novo. 2.6. Landmark collection from the EDJ surface A total of 8 landmarks (LM) and 23 sliding semilandmarks (sLM) were used in order to represent the EDJ marginal edge (Fig. 3). The real landmarks were placed at the horn tips of the main cusps (5 LM) and at the indentations between the buccal (2 LM) and the lingual (1 LM) horn tips. Since dm2s might present accessory cusps, we circumvented this potential lack of homology by always tracing the occlusal curve using a spline curve, as if accessory cusps were not present. The landmark collection and sliding of the semilandmarks on the marginal edge were carried out by means of the EVAN Toolbox (ET; http://evan-society.org), which uses the minimum bending energy technique for sliding semilandmarks (Bookstein, 1997; Gunz et al., 2005; Gunz and Mitteroecker, 2013). 2.7. GM analysis of the landmark configurations from crown and cervical outlines The configurations of pseudolandmarks gathered from cervical and crown outlines were considered separately in all phases of the GM analysis. Each set of landmark configurations had to be scaled through a General Procrustes Analysis (GPA; Gower, 1975), while translation and rotation were already performed on the

C. Fornai et al. / Quaternary International 398 (2016) 175e189

179

Fig. 2. a. Crown and cervical outlines are collected on the aligned model. b. The outlines are superimposed onto a fan of 24 equiangularly spaced radii departing from the centroid of the area. The landmarks are identified at the intersection between the outlines and the fan. c. Landmark data for the crown outlines and d. for the cervical outlines.

outlines (as explained in 2.2). A Principal Component Analysis (PCA) of the Procrustes shape coordinates was carried out both in shape space and in form space, the latter by adding to the matrix of shape coordinates size information expressed by the natural logarithm of Centroid Size to the Procrustes shape space (Mitteroecker et al., 2004, 2005). Afterwards, we visualized the shape changes between landmark configurations by using deformation grids based on the thin-plate spline (TPS) function (Bookstein, 1989, 1991). These tasks were performed using ET software, while box plots for representing the distribution of the

natural logarithm of Centroid Size were produced with “R” software. 2.8. GM analysis of the landmark configurations from EDJ The GM analysis of the landmark configurations representing the marginal ridge of the EDJ was performed as described in Section 2.7, but differed in the process of GPA in that translation and rotation were carried out. 2.9. Quadratic discriminant analysis A QDA was performed on ranges of principal component scores for cervical and crown outlines, and EDJ in order to assess the likelihood that dm2-QC2 could be classified to either MH or NEA (the justification for this is the same as is section 2.4). Prior to this analysis, PCG.2 H. erectus was excluded since it does not belong to either of the groups. Of course we are not claiming that dm2-QC2 was a member of one of these species or the other (it is much older than any of the specimens included in our sample), but only that the QDA provides a better answer to the question of likely ancestry than does any method based on computed similarities or distances. We performed the QDA also for Qafzeh 15, Skhul I and Skhul X, leaving out dm2-QC2, PCG.2 and, alternatively, Qafzeh 15, Skhul I and Skhul X. 3. Results 3.1. Description of the inner morphology

Fig. 3. Occlusal view of dm2-QC2 EDJ showing the landmarks (red dots) and curve on the marginal edge along which semilandmarks were slid. The red arrow indicates that the curve was traced following the curvature of the distal margin, and ignoring the relief of an additional cusp. M ¼ mesial; B ¼ buccal. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

For the description of the outer tooth morphology the reader is referred to Hershkovitz et al. (2011). With regard to the EDJ surface, dm2-QC2 is complete, in excellent state of preservation, and unaffected by wear. The pinpoint patch of dentine exposed at the protoconid might originate from an erosional pit at a score 2 (Molnar, 1971) wear facet, rather than from attrition only. In this case dm2QC2 wear stage would not exceed score 2, but it might also be an

180

C. Fornai et al. / Quaternary International 398 (2016) 175e189

Fig. 4. Images a.ee. show different views (a., mesial; b., buccal; c., distal; d., lingual; e., occlusal) of the topographic rendering of the enamel local variation with increasing thickness from blue to red. The thickest enamel is found at the occlusal marginal ridge, and has its maximum extension buccally and distally. Pictures f1e3 show the pulp chamber and root canals in lingual view (f1e2) and mesio-lingual view (f3). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

incipient stage 3. A small and round interproximal contact facet is visible on the mesial aspect of the outer surface. The horn tips are rather high, acute and inwardly directed, which reflects the morphology of the cusps at the outer enamel cap. Hershkovitz et al. (2011) noted a possible absence of the midtrigonid crest on the outer enamel surface which is indeed difficult to identify if the underlying EDJ is not available. Examining the mCT images, we could detect a continuous trigonid crest on the EDJ (grade 2 according to Bailey et al., 2011) with continuous distal trigonid crest originating from the metaconid and corresponding to configuration type 6 as per Martínez de Pinillos et al. (2014). The presence of a mid-trigonid crest on the outer enamel

surface is debatable since the mesial cusps are very close to each other (see Fig. 1, occlusal view), and only a very short crest segment is visible (type A, Martínez de Pinillos et al., 2014). A n-Torres et al., 2014) is not talonid crest (Korenhof, 1982; Martino present, but we point out the presence of three short distal wrinkles that originate from the distal margin and run parallel to each other. Two wrinkles are also visible mesially to the midtrigonid crest. This feature does not correspond to a mesial trigonid crest (sensu Martínez de Pinillos et al., 2014), since it does not connect the mesial margin to the metaconid. A very small tuberculum sextum (C6) is also present on the OES (score 1 as per Turner et al., 1991).

Fig. 5. Neanderthal (NEA) and modern human (MH) 2D and 3D relative enamel thickness (RET) distributions are non-overlapping. Dm2-QC2 2D RET is intermediate between NEA and MH. Instead, the 3D RET measured on dm2-QC2 is within the NEA distribution, while its corrected 3D RET (after virtual restoration of the eroded parts; *dm2-QC2) is slightly above it. Skhul I, which represents the lowermost value of MH distribution, is represented to ease the comparison.

C. Fornai et al. / Quaternary International 398 (2016) 175e189

181

Fig. 6. Dental tissue proportions for the mesial section and the dental crown. Dm2-QC2 proportions for the dental areas are intermediate between Neanderthal (NEA) and modern humans (MH, including Skhul I), while its proportions for the dental volumes are comparable to that of NEA. Possibly, these different outcomes for the 2D and 3D dental tissue proportions can be attributed to a varying enamel distribution in dm2-QC2 and NEA, where the latter shows prevalence in the amount of dentine than dm2-QC2. Skhul I shows very similar values to MH, while PCG.2 Homo erectus presents proportionally even higher enamel than MH.

182

C. Fornai et al. / Quaternary International 398 (2016) 175e189

Mesial and distal roots are well separated (Fig. 4, f1e3). The mesial roots are fused with two distinct root canals, while the distal roots, also separated, present a sheath-like root canal. The pulp chamber extends apically to approximately one third of the root length (measured linearly by means of planes parallel to the cervical plane and passing through the most apical point of the pulp chamber and the most apical point of the root canals). At the crown, the volume of the pulp chamber is in a 1:10 relationship to dentine volume (about 21 mm3e211 mm3). We calculated a taurodont index of 25.6 for dm2-QC, which corresponds to a hypotaurodontic class according to Keene (1966), confirming the previous diagnosis of taurodontism by Hershkovitz et al. (2011). In addition, we noted concretions covering the root canals and the pulp chamber (Fig. SM1). We also estimated the total dentine volume (~508 mm3) after virtual removal of the matrix still present on the outer surface of the roots. This value is to be taken with caution because the roots are apically incomplete and the matrix has infiltrated the dentine. We estimated that approximately 5% of the root volume is missing; therefore, a corrected volume of approximately 533 mm3 might be closer to the original dentine volume. The volume of the pulp chamber and root canals is in a 1:4 relationship to the total dentine volume.

Skhul I is relatively thicker at the mesial section than at the crown, with its 3D RET at the lowest range of MH distribution. Skhul I provided higher 2D and 3D RET than dm2-QC2. The PCG.2 3D RET is in the MH range. Here, we point out a difference in our 3D RET for PCG.2 (¼ 14.69) with respect to that produced by Zanolli et al. (2012) (¼ 13.46; also reported in Macchiarelli et al. (2013)). We think this discrepancy resulted from the application of two different protocols for the identification of the dental crown (Benazzi et al., 2014 in our case, Olejniczak et al., 2008a,b in Zanolli and colleagues). The dental tissue proportions at the mesial section and those for the entire crown delivered slightly different pictures for dm2QC2. Its mesial section tissue proportions (28% enamel, 72% dentine) are intermediate between MH and NEA, being closer to MH (30% enamele70% dentine versus 23% enamele77% dentine for NEA). Instead, dm2-QC2 tissue proportions for the whole dental crown (35% enamele65% dentine) are closer to those of NEA (33% enamele67% dentine) and differ more from MH whose enamel volume is relatively higher (42% enamele58% dentine). Skhul I dental tissue proportions are comparable to those of MH, and PCG.2 enamel quantity is relatively higher than MH.

3.2. 2D and 3D relative enamel thickness and dental tissue proportions

3.3. Full Procrustes distances

The results for the analysis of the enamel thickness on mesial sections and on the crown volume, and those for the dental tissue proportions are presented in Figs. 5 and 6, and in Table 2. Additional information is shown in Fig. SM2. Acknowledging the small size of our comparative sample, we point out that both the 2D and 3D RET indices distinguished clearly between MH and NEA (p < 0.001), which do not overlap. The 2D RET index for dm2-QC2 is intermediate between MH and NEA. The corrected 3D RET index is slightly above the NEA distribution, but the uncorrected one is within the upper NEA distribution. The QDA dm2-QC2 is classified either as MH (LR ¼ 0.002 for being a NEA) based on the 2D RET or as NEA based on the 3D RET (LR ¼ 719).

Based on the full Procrustes distances for each of the shape spaces, the 5 nearest neighbors of dm2-QC2, SkhuleQafzeh specimens, and PCG.2 are presented in Table 3. The cervical outline provides the most consistent results, since, among the five nearest neighbors to dm2-QC2, four are NEA specimens, while the other is Skhul X. The results for the crown outline are less consistent and show morphological closeness to both NEA (n ¼ 2) and MH (n ¼ 3), the latter including again Skhul X. Based on the EDJ marginal ridge, the nearest neighbor of dm2QC2 is a NEA specimen, followed by four MH. The SkhuleQafzeh specimens and PCG.2 have almost exclusively MH as neighbors, except for Qafzeh 15 crown outline (two NEA, two MH and PCG.2).

Table 2 2D and 3D enamel thickness components, average enamel thickness (AET), and relative enamel thickness (RET). Means and ranges are provided for Neanderthals (NEA) and modern humans (MH). 3D dental crown

Enamel volume

Dentine volume (including pulp chamber)

EDJ area

3D AET

3D RET

dm2-QC2 dm2-QC2a Skhul I PCG.2 NEA (n ¼ 7)

129.24 135.70 114.12 158.12 113.69

198.87 198.87 162.93 181.80 190.95 e

0.65 0.68 0.70 0.87 0.60 e

10.28 10.80 12.45 14.69 9.71 e

MHb (n ¼ 18)

e (97.84e128.66) 126.96

252.28 252.28 178.20 207.75 230.02 e

(178.59e207.84) 154.14 e

(1.54e0.66) 0.82 e

(9.14e10.51) 14.80 e

2D mesial section dm2-QC2 Skhul I NEA (n ¼ 7)

e (98.74e155.52) Enamel area 10.88 9.75 9.64

(202.67e252.61) 177.10 e

(136.36e175.16) EDJ length 18.15 16.02 19.05 e

(0.70e0.95) 2D AET 0.60 0.61 0.51 e

(12.99e17.27) 2D RET 11.35 12.60 9.01 e

MHb (n ¼ 10)

e (8.42e10.52) 10.56

(148.69e219.24) Dentine area (including pulp chamber) 27.86 23.33 31.85 e (27.91e36.05) 25.12 e

(16.47e21.74) 16.03 e

(0.45e0.54) 0.64 e

(8.34e10.10) 12.76 e

(21.83e29.24)

(14.72e16.87)

(0.49e0.75)

(9.95e14.73)

e (8.67e12.50) a b

dm2-QC2 enamel volume augmented by 5% to compensate for erosion. Skhul I excluded.

C. Fornai et al. / Quaternary International 398 (2016) 175e189

183

Table 3 Full Procrustes distances for the cervical outline, crown outline and enameledentine junction (EDJ) between some specimens of particular interest to our study and their first five nearest neighbors.

Cervical outline

Crown outline

EDJ

dm2-QC2

Qafzeh_15

SkhulI

SkhulX

PCG2

Kr_d68 (0.0233) Kr_d62 (0.0247) SkhulX (0.0261) Kr_d66 (0.0270) CavA (0.0274) Kr_d62 (0.0190) SkhulX (0.0203) BLZ_294 (0.0207) BLZ_279 (0.0209) Kr_d68 (0.0214) Kr_d68 (0.0977) BLZ_004 (0.1050) Mad4 (0.1084) RHS_105 (0.1124) BLZ_279 (0.1182)

RHS_113 (0.0150) Hay_H28 (0.0197) RdM (0.02590) Hay_H13 (0.0268) AM_H57 (0.0269) Engis2 (0.0187) Hay_H13 (0.0201) AM_H57 (0.0213) Cou1 (0.0214) PCG2 (0.0228) AM_H88 (0.0860) BLZ_294 (0.0914) RHS_96 (0.0933) PCG2 (0.0954) SkhulI (0.1017)

SkhulX (0.0299) Hay_H13 (0.0300) AM_H88 (0.0303) BLZ_294 (0.0325) BLZ_279 (0.0327) BLZ_294 (0.0182) RHS_44 (0.0190) BLZ_279 (0.0194) SkhulX (0.0204) BLZ_004 (0.0206) DolniVest36 (0.0659) Mad4 (0.0676) RHS_52 (0.0691) RHS_105 (0.0695) Kr_d68 (0.0705)

BLZ_279 (0.0219) BLZ_294 (0.0234) dm2-QC2 (0.0261) Kr_d64 (0.0280) SkhulI (0.0299) BLZ_004 (0.0168) BLZ_279 (0.0178) BLZ_294 (0.0191) RHS_73 (0.0200) dm2-QC2 (0.0203) RHS_93 (0.1105) BLZ_273 (0.1177) RHS_57 (0.1183) RHS_116 (0.1197) RHS_44 (0.1206)

Hay_H13 (0.0211) Hay_H12 (0.0229) AS_S37b (0.0236) RHS_75 (0.0246) RHS_113 (0.0276) RHS_96 (0.0182) AM_H57 (0.0183) RHS_44 (0.0194) RHS_113 (0.0200) Hay_H13 (0.0208) SkhulI (0.0941) Qafzeh15 (0.0954) RHS_96 (0.1027) RHS_52 (0.1035) DolniVest36 (0.1088)

Fig. 7. PCA plot for the cervical outlines in shape space (PC1 ¼ 36%; PC2 ¼ 21%), and warps of the landmark configurations at the extremities of the range of distribution of the specimens. In dm2-QC2 (red star) the cervical margin is rather squared, which separates it from MH and put it within the NEA cluster. Refer to Table 1 for the individuals' labels. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.4. PCA, cervical outlines The PCA plot for the cervical outlines is shown in Fig. 7. The cervical outline delivers a clear picture, neatly distinguishing MH from NEA. In agreement with Benazzi et al. (2012), along PC1 (36% of the total variance explained) specimens vary in the relative expansion of the bucco-distal corner of the cervical outline, which is reduced in MH, and more expanded in NEA, giving the cervical outline a more regular, squared shape in the latter. Dm2-QC2 plots within the NEA

distribution for this feature. Qafzeh 15, Skhul X, and PCG.2 are found at the fringes of the two distributions. Along PC2 (21%), the cervical outlines vary in terms of relative expansion of their mesial and distal portions, which does not distinguish between taxa. The QDA classifies dm2-QC as NEA (LR > 11) based on the cervical outline for any range of PCs from 1e2 to 1e7 (Table 4). Skhul I is classified as MH, while the LRs for Qafzeh 15 and Skhul X provide equivocal results for the range PC1e2 and classify these specimens as NEA for subsequent PCs.

184

C. Fornai et al. / Quaternary International 398 (2016) 175e189

Table 4 Likelihood ratios (LRs) from the quadratic discriminant analysis on ranges of principal component scores for cervical outline, crown outline, and enameledentine junction marginal edge (EDJ). LRs higher than 6.3 (in bold) indicate likely affiliation to Neanderthal, lower than 0.146 (normal font) to modern humans, and between these values (in italic) provide equivocal signals. Only the PC ranges up to 1e4 are reported. Cervical outline

dm2-QC2 Qafzeh15 SkhulI SkhulX

Crown outline

EDJ

PC1-2

PC1-3

PC1-4

PC1-2

PC1-3

PC1-4

PC1-2

PC1-3

PC1e4

27.056 0.260 0.080 1.572

184.506 0.035 0.000 0.050

2131.410 0.000 0.000 4.392

1.779 23.602 0.094 0.005

60.186 11.745 0.003 0.102

70.522 2.486 0.005 0.041

939.339 0.036 0.094 0.031

0.389 0.000 0.003 0.041

0.353 0.000 0.005 0.012

3.5. PCA, crown outlines

3.6. PCA, EDJ marginal ridge

The shape PCA for the crown outline is shown in Fig. 8. The crown outline distinguishes between MH and NEA rather well along PC1 (35% of the total variance explained). As already observed by Benazzi et al. (2012), the main factor driving the betweenspecies separation reflects the relative expansion of the buccodistal portion of the crown outline, NEA having crown outlines relatively expanded bucco-distally. Along PC2 (25%) specimens vary in the mesio-distal elongation of their crowns, but this factor does not distinguish between taxa. The dm2-QC2 crown outline is moderately expanded distally (as described in Hershkovitz et al., 2011). Skhul I, Skhul X and Qafzeh 15 are moderately expanded distally too. Like Skhul I and Skhul X, dm2-QC2 is rather elongated, but this feature does not discriminate taxonomically. Although Qafzeh 15 plots within the Neanderthal distribution, it differs from dm2-QC2 for its rounder crown outline. The QDA signal for dm2-QC2 (Table 4) for PC1e2 is equivocal (LR ¼ 1.8), but dm2-QC2 is more likely to be a NEA for PC ranges from 1e3 to 1e5 (LR > 60). The QDA provides equivocal results for Qafzeh 15 which is likely NEA for PC1e3 (LR > 11), while subsequent PC ranges provide contrasting LRs. Skhul I and Skhul X are likely MHs.

The results for the EDJ analysis in shape space are shown in Fig. 9. There is a perfect separation between the two Homo taxa in our small sample that combines both PCs (in contrast to the outlines, where only the first PC distinguishes the taxa). The EDJ shape separates dm2-QC2 from MH and sets it clearly within the distribution of NEA. The variation along PC1 (27%) reflects the relative expansion of the distal to the mesial aspects, where RMH from Central Europe are relatively less expanded than NEA and the other MHs (fossils, Natufians and Bedouins). Along PC2 (16%), the shape varies in terms of the horn height and inclination along the buccolingual direction. This feature distinguishes MH from NEA, and sets dm2-QC2 within NEA distribution. Dm2-QC2, Skhul X, and especially Engis 2 show the most extreme inward bending within our sample. Along PC3 (11%; not shown here), the TPS warping shows that the variation pertains mainly to the mesio-distal elongation of the EDJ margin, which does not distinguish NEA from MH, but sets dm2-QC2 apart from the rest of the sample, being the most elongated. Qafzeh 15, Skhul I, and Skhul X lie on a diagonal, with Qafzeh 15 and Skhul X at the opposite extremes of the distribution, and Skhul I between them. Skhul X is comparable to dm2-QC2 in terms of height of the dentine horns, but differs from it having relatively

Fig. 8. PCA plot for the crown outlines in shape space (PC1 ¼ 35%; PC2 ¼ 25%), and warps of the landmark configurations at the extremities of the range of distribution of the specimens. Dm2-QC2 plots in the range of NEA distribution. Dm2-QC2 crown outline is intermediate in terms of relative expansion of the distal portion, but it is rather elongated. Refer to Table 1 for the individuals' labels.

C. Fornai et al. / Quaternary International 398 (2016) 175e189

185

Fig. 9. PCA plot for the EDJ marginal edge in shape space (PC1 ¼ 27%; PC2 ¼ 16%), and warped surfaces at the extremities of the range of distribution. Dm2-QC2 separates from MH (grey diamonds) and plots within NEA distribution (black circles), for the height and position of the horns which bend inwardly. Refer to Table 1 for the individuals' labels.

narrower distal aspect. Skhul I, and especially Qafzeh 15, possess a different configuration of the horns, which are shorter and flare out with respect to dm2-QC2, but are similar for the relative expansion of the distal side. Based on the EDJ, the QDA classifies dm2-QC2 as extremely likely to be NEA (LR ¼ 939) for PC1e2. However, when subsequent PCs are considered dm2-QC2 is classified as MH. We explain this inconsistency by the fact that dm2-QC2 is distinguished from both comparative groups along PC3 (for its marked mesio-distal elongation). The Late Pleistocene Qafzeh 15, Skhul I and Skhul X are consistently classified as MHs for any range of PCs up to at least PC1e4. 3.7. GM analyses in form space The results of all analyses in form space are illustrated in Fig. 10 together with the box plots for the natural logarithm of centroid sizes. If size is considered in the analyses of cervical and crown outlines, MH and NEA overlap greatly along PC1 (81% and 87% of variance explained, respectively). Dm2-QC2 and Qafzeh 15 crown and cervical outlines are among the largest (dm2-QC2 has the largest cervical outline and Qafzeh 15 has the largest crown outline) together with some NEA and Natufian specimens, while PCG.2 and both Skhul dm2s are intermediate in size. In the EDJ analysis in form space, PC1 (55% of the total variance explained) separates MH from NEA well. Size does not contribute as much as in the previous analyses to the total variance explained. Dm2-QC2, Qafzeh 15 and all other fossil specimens are the largest, and recent MH the smallest specimens within the sample. The contribution of size to the analysis in form space does not obscure the difference of the Skhul X EDJ marginal ridge, which is still separated from the rest of the sample along PC2 (11%).

4. Discussion Dm2-QC2 is a lower left deciduous second molar. It is part of the QC dental record and was found within Amudian deposits possibly older than 300 ka (Hershkovitz et al., 2011, 2016). A first description of this deciduous tooth, accompanied by a preliminary assessment of dm2-QC2 dimensions, was carried out by Hershkovitz et al. (2011) who stressed the mixed presence of similarities to the Late Pleistocene modern humans (i.e., Qafzeh/Skhul specimens) on the one hand and Neanderthal-like features on the other. Our qualitative and quantitative investigation of dm2-QC2 outer and inner 3D morphology generally supports the previous morphological assessments of the outer aspect of the tooth by Hershkovitz et al. (2011), and documents a high correspondence between the EDJ and the outer occlusal surface (see Guy et al., 2015). We could also confirm earlier observations on dm2-QC2's large size and the presence of taurodontism (Hershkovitz et al., 2011). Our results highlighted a mixed morphology for dm2-QC2, with the occurrence of several Neanderthal-like features together with other features intermediate between Neanderthal and modern humans, along with some peculiar to dm2-QC2. Dm2-QC2 is close to Neanderthals with regard to some geometric large scale factors and shows some discrete traits found in high frequency in Neanderthals. In particular, its squared cervical outline and the high and inwardly bent dentine horns are considered typical Neanderthal features (Bailey and Hublin, 2006; Benazzi et al., 2012, 2014a). Its mid-trigonid crest on the EDJ is weakly expressed, but clearly present. However, while this trait is frequently observed in Neanderthals and rarely found in modern humans (Bailey, 2002; Bailey et al., 2011), it also appears in Sangiran (Kaifu et al., 2005), Homo n-Torres et al., 2007), or Tighenif (Zanolli and antecessor (Martino Mazurier, 2013). Likewise, the analysis of the dental tissues highlights an intermediate state.

186

C. Fornai et al. / Quaternary International 398 (2016) 175e189

Fig. 10. PCA plots for the cervical and crown outlines and for the EDJ in form space, and box plots for the distribution of the natural logarithm centroid sizes. MH and NEA, overlap greatly in the form of their crown and cervical outlines, while the EDJ form is distinctive. Dm2-QC2 and Qafzeh 15 are among the largest specimens. Refer to Table 1 for the individuals' labels.

C. Fornai et al. / Quaternary International 398 (2016) 175e189

Dm2-QC2 enamel volume is relatively thinner than in modern humans and only slightly above the Neanderthal range of distribution (Macchiarelli et al., 2006; Olejniczak et al., 2008a,b). The enamel thickness measured at the mesial section is also thinner than, but closer to modern humans (including Skhul I). The relative expansion of the pulp cavity is indicative of slight taurodontism. The latter feature is not taxonomically diagnostic, but it is frequent in Neanderthals (e.g., Trinkaus, 1983; Kupczik and Hublin, 2010). Furthermore, dm2-QC2 presents an intermediate morphology of the crown outline with its only mildly expanded distal side. The latter feature manifests on the EDJ marginal ridge as well which is, therefore, intermediate in relative expansion of its distal portion. Additionally, the mesio-distal elongation of the EDJ sets dm2-QC2 apart from both modern humans and Neanderthals. Our study also shows that the SkhuleQafzeh teeth are morphologically quite heterogeneous, presenting major intragroup variation. Although the affiliation of these Homo specimens is not the focus of this paper, we observed that they variably ally with modern humans (mostly), Neanderthals, or with none of them on the basis of the morphology of the various features considered. Within the SkhuleQafzeh sample, Skhul X is the closest specimen to QC in several aspects, namely the squared cervical outline, elongated and distally expanded crown outline, and the configuration of the dentine horns. Nevertheless, it differs from dm2-QC2 in its reduced distal portion of the EDJ marginal edge, which is considered a modern trait. Conversely, Qafzeh 15 differs the most from dm2-QC2, especially in its short and flaringout dentine horns, and the round crown outline. However, they are similar in terms of distal expansion of the EDJ (but this feature does not separate the Homo groups in our study). Skhul I, instead, is intermediate in most of the features considered, and allies with modern humans. Both Skhul I and Skhul X present affinities to the other modern humans in our comparative sample. Nonetheless, Skhul I and Skhul X differ from each other especially in terms of EDJ shape. The Sangiran Dome specimen PCG.2, the oldest within our sample (Zanolli et al., 2012), presents a clear morphological affinity to modern humans in all of the features considered here. This affinity corroborates the finding by Bailey et al. (2014) obtained for other Pleistocene Asian individuals based on crown and cervical outlines. In terms of size, dm2-QC2 is comparable to the largest Neanderthals and Qafzeh 15, the latter featuring an exceptionally large size among the modern human sample. Both Skhul I and Skhul X are intermediate in size. Weber et al. (2016) used similar methods to investigate some of the QC lower permanent teeth (third and fourth premolars, P3 and P4, and second molar, M2) from the older middle part of the lower sequence. They also observed morphological affinities to Neanderthal, particularly for the M2, and a mixed morphology for the premolars. Moreover, the molar was quite large while both P3 and P4 were within the recent modern human size range. We acknowledge that our comparative sample lacks specimens penecontemporaneous to dm2-QC2. Nonetheless, we could compare the 3D RET to that of the Middle Pleistocene European Homo heidelbergensis Arago 5 dm2 (Caune de l'Arago at Tautavel, France) published in Macchiarelli et al. (2013), which they calculated to be 11.58. This value is higher than that of dm2-QC2, but still indicates relatively thinner enamel than our modern human sample. If we could correct for the methodological differences between our work and that of Macchiarelli et al. (2013) (using, respectively, the Benazzi et al. (2014b) protocol and the Olejniczak et al. (2008a,b) protocol) we would expect a slightly higher 3D RET index for Arago 5 dm2, probably still lower than our modern human sample. But dm2-QC2 presents relatively thinner enamel than the earlier Arago 5 dm2 from Western Europe.

187

Unfortunately, the early Middle Pleistocene deciduous remains from Tighenif, Algeria, investigated by Zanolli et al. (2010), are not directly comparable to dm2-QC2 since they consist of upper molars. Dm2-QC2 is comparable to a dm2 specimen from the European  n-Torres Middle Pleistocene Sima de los Huesos analyzed in Martino et al. (2014) in the lack of a talonid crest. However, the biological or taxonomic value of this feature is not yet understood. In the light of the mosaic of features observed in dm2-QC2 and generally of the range of variability in Middle Pleistocene hominins, in the QC individuals (Hershkovitz et al., 2011, 2016; Weber et al., 2016), and in other Late Pleistocene Levantine fossil specimens, we must be extremely cautious in interpreting dm2-QC2 morphology. A taxonomic affiliation is not obvious. The Levant represents a cross-roads between North Africa, Europe and Asia. However, the peopling of the Levant and the Early and Middle Pleistocene population history are poorly known, and the closest reference specimens are represented by the African and European Middle Pleistocene Homo hypodigm. It is currently debated how to interpret the mosaic of features and variably expressed Neanderthal-like characteristics of the Middle Pleistocene European Homo in a taxonomic context. The Sima de los Huesos (permanent) dental hypodigm has been considered variable and yet mostly showing Neanderthal-like dental features in all tooth types n-Torres et al., 2014). Recently, their attribution to (Martino H. heidelbergensis has been questioned in favor of early Neanderthal (see Stringer, 2012; Buck and Stringer, 2014). A definite taxonomic classification based on a single tooth would be too speculative, especially considering the paucity of human remains and information available. Notably, QC is the first site in the Levant that delivered at least a handful of hominin dental material from the Middle Pleistocene. There is much uncertainty as to how to interpret human evolution during this period of extensive climatic changes and complex population dynamics (Dennell et al., 2011). A mosaic of features is not unexpected for a MiddlenPleistocene member of Homo (e.g., Stringer, 2012; Martino Torres et al., 2013; Buck and Stringer, 2014). If we take into account the chronology and geographical location of dm2-QC2, the results of our study would be in agreement with the following. Dm2-QC2 could represent a Levantine population that, typically for other penecontemporaneous European populations (e.g. Arago, Steinheim, Bilzingsleben) variably expresses Neanderthal features. We also have to consider that an early-Neanderthal form was already present in Western Europe at least 100 ka before dm2-QC2,  n-Torres as indicated by the remains at Sima de los Huesos (Martino et al., 2012; Stringer, 2012; Arsuaga et al., 2014). Moreover, the lower molar M2-QC12 (Weber et al., 2016) provided evidence that Neanderthal-like features had been long present at QC. Whether there was continuity between the dm2-QC2 people and the Levantine Late Pleistocene humans remains an unresolved issue. A possible continuity leading into Levantine Late Pleistocene modern humans is an alternative, but based on our data, it seems a less likely scenario. Nonetheless, the possibility that dm2-QC2 people might have left traces in Levantine Late Pleistocene modern humans cannot be ruled out if we bear in mind the mosaic of features observed in both dm2-QC2 and the very heterogeneous SkhuleQafzeh specimens. It is to be noted, however, that the dm2QC2 people and the Levantine Late Pleistocene modern humans were separated by roughly 200 ka. 5. Conclusions We undertook a detailed morphological and endostructural comparative description of various outer and inner dental aspects of the lower left second deciduous molar labelled dm2-QC2 from the Middle Pleistocene fossil site of Qesem Cave, Israel. For this

188

C. Fornai et al. / Quaternary International 398 (2016) 175e189

isolated dental specimen we highlighted a mosaic of features in part shared with modern humans and in part suggesting a morphological affinity to Neanderthals (Hershkovitz et al., 2011). Some of the features considered typical or frequent in Neanderthals are not as markedly expressed in dm2-QC2 as in some Late Pleistocene Neanderthals. The analysis of the Late Pleistocene SkhuleQafzeh sample shows very diverse morphologies of these dm2 specimens. Skhul X is the specimen from this subsample that resembles dm2-QC2 (and some Neanderthal individuals) most closely, while Qafzeh 15 differs the most. Our findings are in line with Weber et al. (2016) who found similar results in morphospace for some of the QC permanent lower premolars and second molar, identifying Neanderthal characteristics within the older deposits at Qesem Cave. Yet, the paucity of Middle Pleistocene remains and the scant knowledge of population history for the Middle Pleistocene humans in the Levant, suggests caution in interpreting the results from one isolated tooth. Also, the possible contribution of Qesem Cave people to the Late Pleistocene Levantine Homo from Skhul and Qafzeh deserves further investigation. A better understanding of the taxonomic nature of the QC population from which this tooth came will require more comprehensive knowledge of the (dental) morphological variability of the Middle Pleistocene humans. Acknowledgements The authors thank Michel Toussaint for access to Couvin 1 and Scladina 4 A 13; Jirí Svoboda and the Academy of Sciences of the stonice 36. We are grateful to Czech Republic for access to Dolní Ve rio da Cultura, the French Muse e National de the Portuguese Ministe histoire, and the UMR 5199 Universite  Bordeaux 1. Access to the Pre image data was made possible also by the NESPOS (Neanderthal Studies Professional Online Service) Database 2011 (https://www. nespos.org/display/openspace/Home). We thank the Soprintendenza per i Beni Archeologici della Puglia for Cavallo A. We thank Harry Widianto and the Balai Pelestarian Situs Manusia Purba of Sangiran for granting permission to include the Sangiran Dome ment Zanolli for PCG.2 specimen, and Roberto Macchiarelli and Cle providing access to the image data which were produced at the Centre de Microtomographie, University of Poitiers. ment Zanolli also facilitated the comparison with the results Cle published in his works. Marina Martínez de Pinillos helped describing some traits of the EDJ surface. We thank an anonymous reviewer for comments which helped to improve the manuscript. This research was financially supported by the Siegfried LudwigeRudolf Slavicek Foundation, Vienna, Austria, Project number FA547016. The anthropological study of the Qesem Cave hominids is supported by the Dan David Foundation. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.quaint.2015.11.102. References llez, A., Sharp, W.D., Arsuaga, J.L., Martínez, I., Arnold, L.J., Aranburu, A., Gracia-Te
res, C., Pantoja-Pe rez, A., Bischoff, J., Poza-Rey, E., Pare s, J.M., Quam, R.M., Falgue n-Torres, M., García, N., Carretero, J.M., Demuro, M., Lorenzo, C., Sala, N., Martino zar de Velasco, A., Cuenca-Besco  s, G., Go mez-Olivencia, A., Moreno, D., Alca Pablos, A., Shen, C.-C., Rodríguez, L., Ortega, A.I., García, R., Bonmatí, A., Bermúdez de Castro, J.M., Carbonell, E., 2014. Neandertal roots: cranial and chronological evidence from Sima de los Huesos. Science 344, 1358e1363. Bailey, S.E., 2002. A closer look at Neanderthal postcanine dental morphology: the mandibular dentition. The Anatomical Record 269, 148e156.

^telperronian assemBailey, S.E., Hublin, J.J., 2006. Did Neanderthals make the Cha blage from La Grotte du Renne (Arcy-sur-Cure, France)? In: Hublin, J.-J., Harvati, K., Harrison, T. (Eds.), Neanderthals Revisited: New Approaches and Perspectives. Springer Netherlands, pp. 191e209. Bailey, S.E., Benazzi, S., Souday, C., Astorino, C., Paul, K., Hublin, J.-J., 2014. Taxonomic differences in deciduous upper second molar crown outlines of Homo sapiens, Homo neanderthalensis and Homo erectus. Journal of Human Evolution 72, 1e9. Bailey, S.E., Skinner, M.M., Hublin, J.-J., 2011. What lies beneath? An evaluation of lower molar trigonid crest patterns based on both dentine and enamel expression. American Journal of Physical Anthropology 145, 505e518. Barkai, R., Gopher, A., Lauritzen, S.E., Frumkin, A., 2003. Uranium series dates from Qesem Cave, Israel, and the end of the Lower Palaeolithic. Nature 423, 977e979. Benazzi, S., Douka, K., Fornai, C., Bauer, C.C., Kullmer, O., Svoboda, J., Pap, I., Mallegni, F., Bayle, P., Coquerelle, M., Condemi, S., Ronchitelli, A., Harvati, K., Weber, G.W., 2011a. Early dispersal of modern humans in Europe and implications for Neanderthal behaviour. Nature 479, 525e528. Benazzi, S., Fornai, C., Bayle, P., Coquerelle, M., Kullmer, O., Mallegni, F., Weber, G.W., 2011b. Comparison of dental measurement systems for taxonomic assignment of Neanderthal and modern human lower second deciduous molars. Journal of Human Evolution 61, 320e326. Benazzi, S., Fornai, C., Buti, L., Toussaint, M., Mallegni, F., Ricci, S., Gruppioni, G., Weber, G.W., Condemi, S., Ronchitelli, A., 2012. Cervical and crown outline analysis of worn Neanderthal and modern human lower second deciduous molars. American Journal of Physical Anthropology 149, 537e546. Benazzi, S., Bailey, S.E., Peresani, M., Mannino, M.A., Romandini, M., Richards, M.P., Hublin, J.-J., 2014a. Middle Paleolithic and Uluzzian human remains from Fumane Cave, Italy. Journal of Human Evolution 70, 61e68. Benazzi, S., Panetta, D., Fornai, C., Toussaint, M., Gruppioni, G., Hublin, J.-J., 2014b. Technical note: guidelines for the digital computation of 2D and 3D enamel thickness in hominoid teeth. American Journal of Physical Anthropology 153, 305e313. Benazzi, S., Slon, S., Talamo, S., Negrino, F., Peresani, M., Bailey, S.E., Sawyer, S., Panetta, D., Vicino, G., Starnini, E., Mannino, M.A., Salvadori, P.A., Meyer, M., €€ Pa abo, S., Hublin, J.-J., 2015. The makers of the protoaurignacian and implications for Neandertal extinction. Science 348, 793e796. Bookstein, F.L., 1989. Principal warps: thin-plate splines and the decomposition of deformations. Pattern Analysis and Machine Intelligence, IEEE Transactions on 11, 567e585. Bookstein, F.L., 1991. Morphometric Tools for Landmark Data: Geometry and Biology. Cambridge University Press, New York. Bookstein, F.L., 1997. Landmark methods for forms without landmarks: localizing group differences in outline shape. Medical Image Analysis 1, 225e243. Bookstein, F.L., 2014. Measuring and Reasoning. Numerical Inference in the Sciences. Cambridge University Press, Cambridge. Buck, L.T., Stringer, C.B., 2014. Homo heidelbergensis. Current Biology 24, R214eR215. Corruccini, R.S., 1987. The dentinoenamel junction in primates. International Journal of Primatology 8, 99e114. Corruccini, R.S., 1998. The dentino-enamel junction in primate mandibular molars. In: Lukacs, J.R. (Ed.), Human Dental Development, Morphology, and Pathology: a Tribute to Albert A. Dahlberg. University of Oregon, Portland, pp. 1e16. n-Torres, M., Bermúdez de Castro, J.M., 2011. Hominin variDennell, R.W., Martino ability, climatic instability and population demography in Middle Pleistocene Europe. Quaternary Science Reviews 30, 1511e1524. Fornai, C., Benazzi, S., Svoboda, J., Pap, I., Harvati, K., Weber, G.W., 2014. Enamel thickness variation of deciduous first and second upper molars in modern humans and Neanderthals. Journal of Human Evolution 76, 83e91. Fornai, C., Bookstein, F.L., Weber, G.W., 2015. Variability of Australopithecus second maxillary molars from Sterkfontein Member 4. Journal of Human Evolution 85, 181e192.
res, C., Richard, M., Tombret, O., Shao, Q., Bahain, J.J., Gopher, A., Barkai, R., Falgue 2015. New ESR/U-series dates in Yabrudian and Amudian layers at Qesem Cave, Israel. Quaternary International 398, 6e12. http://dx.doi.org/10.1016/ j.quaint.2015.02.006. Gopher, A., Ayalon, A., Bar-Matthews, M., Barkai, R., Frumkin, A., Karkanas, P., Shahack-Gross, R., 2010. The chronology of the late Lower Paleolithic in the Levant based on UeTh ages of speleothems from Qesem Cave, Israel. Quaternary Geochronology 5, 644e656. Gower, J.C., 1975. Generalized procrustes analysis. Psychometrika 40, 33e51. Gunz, P., Mitteroecker, P., 2013. Semilandmarks: a Method for Quantifying Curves and Surfaces. Gunz, P., Mitteroecker, P., Bookstein, F.L., 2005. Semilandmarks in three dimensions. In: Slice, D.E. (Ed.), Modern Morphometrics in Physical Anthropology. Kluwer Academic/Plenum Publisher, New York. Guy, F., Lazzari, V., Gilissen, E., Thiery, G., 2015. To what extent is primate second molar enamel occlusal morphology shaped by the enamel-dentine junction? PLoS One 10, e0138802. Hershkovitz, I., Smith, P., Sarig, R., Quam, R., Rodríguez, L., García, R., Arsuaga, J.L., Barkai, R., Gopher, A., 2011. Middle pleistocene dental remains from Qesem Cave (Israel). American Journal of Physical Anthropology 144, 575e592. Hershkovitz, I., Weber, G.W., Gopher, A., Bar-Matthews, M., Slon, V., Quam, R., Sarig, R., 2016. New Middle Pleistocene dental remains from Qesem Cave (Israel). Quaternary International 398, 148e158. http://dx.doi.org/10.1016/ j.quaint.2015.08.059.

C. Fornai et al. / Quaternary International 398 (2016) 175e189 Kaifu, Y., Aziz, F., Baba, H., 2005. Hominid mandibular remains from Sangiran: 1952e1986 collection. American Journal of Physical Anthropology 128, 497e519. Keene, H.J., 1966. A morphologic and biometric study of taurodontism in a contemporary population. American Journal of Physical Anthropology 25, 208e209. Kono, R.T., Suwa, G., Tanijiri, T., 2002. A three-dimensional analysis of enamel distribution patterns in human permanent first molars. Archives of Oral Biology 47, 867e875. Korenhof, C.A.W., 1982. Evolutionary trends of the inner enamel anatomy of den, B. (Ed.), Teeth: form ciduous molars from Sangiran (Java, Indonesia). In: Kurte Function and Evolution. Columbia University Press, New York, pp. 350e365. Kupczik, K., Hublin, J.-J., 2010. Mandibular molar root morphology in Neanderthals and Late Pleistocene and recent Homo sapiens. Journal of Human Evolution 59, 525e541. Macchiarelli, R., Bayle, P., Bondioli, L., Mazurier, A., Zanolli, C., 2013. From outer to inner structural morphology in dental anthropology. The integration of the third dimension in the visualization and quantitative analysis of fossil remains. In: Scott, G.R., Irish, J.D. (Eds.), Anthropological Perspectives on Tooth Morphology. Genetics, Evolution,Variation. Cambridge University Press, Cambridge, pp. 250e277. Macchiarelli, R., Bondioli, L., Debenath, A., Mazurier, A., Tournepiche, J.-F., Birch, W., Dean, M.C., 2006. How Neanderthal molar teeth grew. Nature 444, 748e751. Mahoney, P., 2011. Human deciduous mandibular molar incremental enamel development. American Journal of Physical Anthropology 144, 204e214. Martin, L.B., 1985. Significance of enamel thickness in hominoid evolution. Nature 314, 260e263. n-Torres, M., Skinner, M.M., Arsuaga, J.L., GraciaMartínez de Pinillos, M., Martino llez, A., Martínez, I., Martín-France s, L., Bermúdez de Castro, J.M., 2014. Te Trigonid crests expression in Atapuerca-Sima de los Huesos lower molars: Internal and external morphological expression and evolutionary inferences. Comptes Rendus Palevol 13, 205e221. n-Torres, M., Bermúdez de Castro, J.M., Go  mez-Robles, A., Arsuaga, J.L., Martino Carbonell, E., Lordkipanidze, D., Manzi, G., Margvelashvili, A., 2007. Dental evidence on the hominin dispersals during the Pleistocene. Proceedings of the National Academy of Sciences 104, 13279e13282. n-Torres, M., Bermúdez De Castro, J.M., Go  mez-Robles, A., Prado-Simo  n, L., Martino Arsuaga, J.L., 2012. . Morphological description and comparison of the dental remains from Atapuerca-Sima de los Huesos site (Spain). Journal of Human Evolution 62, 7e58. n-Torres, M., Martínez de Pinillos, M., Skinner, M.M., Martín-France s, L., Martino llez, A., Martínez, I., Arsuaga, J.L., Bermúdez de Castro, J.M., 2014. Gracia-Te Talonid crests expression at the enameledentine junction of hominin lower permanent and deciduous molars. Comptes Rendus Palevol 13, 223e234. n-Torres, M., Spe va  , P., Gracia-Te llez, A., Martínez, I., Bruner, E., Martino ckova Arsuaga, J.L., Bermúdez de Castro, J.M., 2013. Morphometric analysis of molars in a Middle Pleistocene population shows a mosaic of ‘modern’ and Neanderthal features. Journal of Anatomy 223, 353e363.
res, C., Shao, Q., Gopher, A., Barkai, R., Bahain, J.-J., Mercier, N., Valladas, H., Falgue Vialettes, L., Joron, J.-L., Reyss, J.-L., 2013. New datings of Amudian layers at Qesem Cave (Israel): results of TL applied to burnt flints and ESR/U-series to teeth. Journal of Archaeological Science 40, 3011e3020. Mitteroecker, P., Gunz, P., Bernhard, M., Schaefer, K., Bookstein, F.L., 2004. Comparison of cranial ontogenetic trajectories among great apes and humans. Journal of Human Evolution 46, 679e698. Mitteroecker, P., Gunz, P., Bookstein, F.L., 2005. Heterochrony and geometric morphometrics: a comparison of cranial growth in Pan paniscus versus Pan troglodytes. Evolution & Development 7, 244e258. Molnar, S., 1971. Human tooth wear, tooth function and cultural variability. American Journal of Physical Anthropology 34, 175e189. Olejniczak, A.J., Gilbert, C.C., Martin, L.B., Smith, T.M., Ulhaas, L., Grine, F.E., 2007. Morphology of the enamel-dentine junction in sections of anthropoid primate maxillary molars. Journal of Human Evolution 53, 292e301.

189

Olejniczak, A.J., Martin, L.B., Ulhaas, L., 2004. Quantification of dentine shape in anthropoid primates. Annals of Anatomy 186, 479e485. Olejniczak, A.J., Smith, T.M., Feeney, R.N.M., Macchiarelli, R., Mazurier, A., Bondioli, L., Rosas, A., Fortea, J., de la Rasilla, M., Garcia-Tabernero, A., Radov ci c, J., Skinner, M.M., Toussaint, M., Hublin, J.J., 2008a. Dental tissue proportions and enamel thickness in Neandertal and modern human molars. Journal of Human Evolution 55, 12e23. Olejniczak, A.J., Smith, T.M., Skinner, M.M., Grine, F.E., Feeney, R.N.M., Thackeray, J.F., Hublin, J.-J., 2008b. Three-dimensional molar enamel distribution and thickness in Australopithecus and Paranthropus. Biology Letters 4, 406e410. R Core Team, 2013. R: a Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Sarig, R., Gopher, A., Barkai, R., Weber, G., Fornai, C., Sella-Tunis, T., Hershkovitz, I., 2016. How did the Qesem people use their teeth? Analysis of dental wear patterns. Quaternary International 398, 136e147. http://dx.doi.org/10.1016/ j.quaint.2015.10.033. Skinner, M.M., Gunz, P., Wood, B.A., Boesch, C., Hublin, J.-J., 2009. Discrimination of extant Pan species and subspecies using the enamel-dentine junction morphology of lower molars. American Journal of Physical Anthropology 140, 234e243. Smith, T.M., Olejniczak, A.J., Martin, L.B., Reid, D.J., 2005. Variation in hominoid molar enamel thickness. Journal of Human Evolution 48, 575e592. Smith, T.M., Olejniczak, A.J., Zermeno, J.P., Tafforeau, P., Skinner, M.M., Hoffmann, A., Radov ci c, J., Toussaint, M., Kruszynski, R., Menter, C., Moggi-Cecchi, J., Glasmacher, U.A., Kullmer, O., Schrenk, F., Stringer, C., Hublin, J.-J., 2012. Variation in enamel thickness within the genus Homo. Journal of Human Evolution 62, 395e411. Spoor, C.F., Zonneveld, F.W., Macho, G.A., 1993. Linear measurements of cortical bone and dental enamel by computed tomography: applications and problems. American Journal of Physical Anthropology 91, 469e484. Stringer, C., 2012. The status of Homo heidelbergensis (Schoetensack 1908). Evolutionary Anthropology: Issues, News, and Reviews 21, 101e107. Toussaint, M., Olejniczak, A.J., El Zaatari, S., Cattelain, P., Flas, D., Letourneux, C., Pirson, S., 2010. The Neandertal lower right deciduous second molar from Trou de l'Abîme at Couvin, Belgium. Journal of Human Evolution 58, 56e67. Trinkaus, E., 1983. The Shanidar Neanderthals. Academic Press, USA. Turner, C.G.I., Nichol, C.R., Scott, G.R., 1991. Scoring procedures for key morphological traits of the permanent dentition: the Arizona State University Dental Anthropology System. In: Kelley, M.A., Larsen, C.S. (Eds.), Advances in Dental Anthropology. Wiley-Liss, New York, pp. 13e31. Weber, G.W., Fornai, C., Gopher, A., Barkai, R., Sarig, R., Hershkovitz, I., 2016. The Qesem Cave hominin material (part 1): a morphometric analysis of the mandibular premolars and molar. Quaternary International 398, 159e174. http://dx.doi.org/10.1016/j.quaint.2015.10.027. Zanolli, C., Bayle, P., Macchiarelli, R., 2010. Tissue proportions and enamel thickness distribution in the early Middle Pleistocene human deciduous molars from Tighenif, Algeria. Comptes Rendus Palevol 9, 341e348. Zanolli, C., Bondioli, L., Mancini, L., Mazurier, A., Widianto, H., Macchiarelli, R., 2012. Brief communication: two human fossil deciduous molars from the Sangiran dome (Java, Indonesia): outer and inner morphology. American Journal of Physical Anthropology 147, 472e481. Zanolli, C., Mazurier, A., 2013. Endostructural characterization of the H. heidelbergensis dental remains from the early Middle Pleistocene site of Tighenif, Algeria. Comptes Rendus Palevol 12, 293e304. Zilberman, U., Skinner, M., Smith, P., 1992. Tooth components of mandibular deciduous molars of Homo sapiens sapiens and Homo sapiens neanderthalensis: a radiographic study. American Journal of Physical Anthropology 87, 255e262. Zilberman, U., Smith, P., 1992. A comparison of tooth structure in Neanderthals and early Homo sapiens sapiens: a radiographic study. Journal of Anatomy 180 (Pt 3), 387e393.