Middle Pleistocene hominin teeth from Biache-Saint-Vaast, France

The study of dental morphology can be a very useful tool to understand the origin and evolution of Neanderthals in Europe during the Middle Pleistocene (MP). At present, the earliest evidence, ca. 430 ka, of a pre-Neanderthal population in Europe is the hominin sample from Atapuerca-Sima de los Huesos (SH) that present clear dental affinities with Neanderthals while other penecontemporaneous populations, such as Arago or Mala Balanica, exhibit less Neanderthal traits. We present the morphometric study of the external and internal dental structures of eleven hominin dental remains recovered from the MP, ca. 240 ka, French site of Biache-Saint-Vaast (BSV). Our analyses place the BSV hominins within the MP group, together with SH, Fontana Ranuccio, Visogliano, Steinheim or Montmaurin, that show greater morphological affinities with Neanderthals. Moreover, we identified interpopulation variability in the expression of the enamel thickness trait, with BSV hominins sharing the unique combination of thin and thick pattern in the premolars and molars with the SH population. These results further support the coexistence of two or more populations in Europe during the MP that reflect the population and settlement of human groups suggested by the Central Area of Dispersals of Eurasia (CADE) and sink and source model.


Introduction
Providing new evidence on the diversity of the European Middle Pleistocene (MP) populations is relevant for the discussion on the origin of the Neanderthal clade. The great morphological variability in the combination of dental and cranial features in the European MP fossil record suggests a non-linear evolutionary scenario characterised by genetic drift, founder effect, isolation and hybridization of the populations (e.g. Bermúdez de Castro et al. 2019;Daura et al. 2017;Martínez de Pinillos et al. 2020;Roksandic et al. 2011). Within this framework, the sink and source model with a central area of dispersal in Eurasia (Bermúdez de Castro and Martinón-Torres 2013; Dennell et al. 2011) has been proposed as a more parsimonious explanation for the non-linear evolution of the MP groups towards the classic Neanderthals. In this model, the geographical and climatic constrains of Europe constituted a driving force in the demographic dynamics (Dennell et al. 2011). Amongst European MP fossils, the Atapuerca-Sima de los Huesos (Spain), Steinheim (Germany), Montmaurin (France), Swanscombe and Pontnewydd (UK) and Fontana Ranuccio and Visogliano (Italy) exhibit in their dentitions a suite of features characteristic of the Neanderthal clade Martínez de Pinillos et al. 2017Martinón-Torres et al. 2012;Vialet et al. 2018;Zanolli et al. 2018). In other European MP specimens, such as those of Ceprano (Italy), Aroeira (Portugal), Mala Balanica (Serbia) and Arago (France), the Neanderthal affinities are less clear (e.g. Bermúdez de Castro et al. 2019;Daura et al. 2017;Manzi, 2016;Roksandic 2016;Roksandic et al. 2011;Skinner et al. 2016) attesting for the possible coexistence of at least two hominin lineages in the European MP (e.g. Bermúdez de Castro et al. 2016;Dennell et al. 2011;MacDonald et al. 2012).
This work presents the outer and inner description and comparison of the MP remains from Biache-Saint-Vaast (BSV) with the aim of actively contributing to the understanding of the European MP evolutionary scenario. In particular, its comparison with other MP key sites showing different degrees of affinities with Neanderthals, amongst others Atapuerca-Sima de los Huesos (SH), dated 430 ka (Arsuaga et al. 2014;Demuro et al. 2019), and Arago, dated 450-400 ka Yokoyama et al. 1985;Yokoyama and Nguyen 1981), will help us to explore the hypothesis of different lineages in the European MP.
The BSV site is located 17 km east of the city of Arras (Pas-de-Calais Department), in northern France, at an altitude of about 50 m above sea level. It is an open-air site, located in the river deposits of the left bank of the Scarpe River. The site was discovered in 1976 while earthworks were being carried out for the construction of a metallurgical factory. Alain Tuffreau led an emergency excavation between 1976 and 1982 to save the site (Tuffreau 1978;Tuffreau et al. 1982). Sommé (1988) studied the stratigraphic sequence and observed several archaeological levels in the sequence of the lower terrace, most dating from the same period. In 1976, the first hominin skull was found in grid 9M (Tuffreau et al. 1982), at the base of level IIa of the complex 2b (BSV1), whereas cranial fragments of a second specimen were found in 1986 when P. Auguste was reviewing the fauna remains recovered from level IIa. Those remains (BSV2) came from grid 22T and belong to a second individual (Tuffreau, 1988). The hominin remains were associated to a rich lithic and faunal context. First dating analyses, including OSL and ESR, provided ages of 175±13 ka (Huxtable and Aikten 1988) and 253 + 53/-37 ka (Yokoyama 1989), respectively. Moreover, fauna remains and pollen analyses (Sommé 1988) indicate a deposit episode for the human remains during the MIS 7 (between 240 and 190 ka; Lisiecki and Raymo 2005). A recent ESR/U-series analysis of several faunal remains associated to the paleoanthropological and archaeological remains provided a mean age of ca. 240 ka , correlating the human occupation to MIS 7c. However, an older chronology can reasonably not be excluded for the site given the dating result of 332 ± 28 ka obtained for the only tooth of the study that do not show uranium leaching .
Although Rougier (2003) performed the first description of the dental remains recovered from BSV, we aim to present a more detailed morphometric assessment of the dental remains by providing new insights on the external (OES) and internal (EDJ) morphological features as well as dental tissue proportions. With these analyses, we expect to explore the degree of affinity of BSV with other European MP hominins and Neanderthals and, ultimately, to shed light on the evolution of the Neanderthal clade.

Materials and methods
The dental sample under study comprises eleven maxillary teeth ( Fig. 1; Table 1) including four isolated specimens (a left I 2 , right and left P 3 s, and right and left P 4 s) and the right and left molar series (M 1 -M 3 ) included in two maxillary fragments. The comparative sample comprises 224 teeth (including original data and from the literature) belonging to European MP hominins, Neanderthals, fossil H. sapiens and modern humans (Table 1). Due to limited mCT data access, we could not include the same sample for all the analyses (e.g. Arago). In addition, since some of the analyses are wear-dependent the number of specimens may vary for each analysis (Table 1 describes the sample included in each analysis).

Scanning of the samples
For this study, high-resolution μCT scanning of the fossil and modern material was performed in two laboratories. The BSV teeth were scanned at the Muséum National d'Histoire Naturelle, France (AST-RX Platform), and the modern human collections at CENIEH, Spain. The BSV maxillary remains and isolated teeth were scanned using a GE 103 Phoenix v/tome/x_L 450 instrument. The scanning of the isolated remains (I 2 , P 3 s and P 4 ) was performed using the following parameters: 120 kV and 180 μA, 0.5 mm Cu filter and isometric voxel size of 22 μm, while the scanning of the right and left maxillary fragments including M 1 , M 2 and M 3 was performed at 100 kV and 410 μA, 0.4 mm Cu filter and resulting isometric voxel size of 41 μm. Finally, the modern human sample was scanned with a GE 103 Phoenix v/tome/x_s 240 instrument (CENIEH) with the following parameters: 100 kV and 100 μA, 0.2 mm Cu filter and isometric voxel size of 18μm.

Virtual segmentation
3D virtual segmentation of the dental tissues (enamel, dentine and pulp) was performed in Amira (6.3.0, FEI Inc.) using the semiautomatic tool, threshold-based segmentation, for the characterisation of the enamel-dentine junction (EDJ) morphology, tissue proportions and enamel thickness topographic distribution.

Metrics
Metric analyses comprised several approaches, including (a) mesio-distal (MD) and bucco-lingual (BL) diameter, crown index (CI) and total computed crown base area (TCBA) of the BSV dental sample (Table 2), and (b) relative occlusal polygon area (ROPA) and cusp angles of the BVS right M 1 (Tables 3 and 4). For the MD and BL diameters as well as for CI and TCBA (Tables 5, 6, 7 and 8), we compared the BSV results to those of SH, Arago, Neanderthals and Krapina samples (Bermúdez de Castro 1993;Bermúdez de Castro et al. 1999, 2019Wolpoff 1979). Finally, for the ROPA variable and cusp angles, we compared the BSV estimates to those published for H. antecessor, SH, H. heidelbergensis, Neanderthals and H. sapiens (Bailey 2004;Martinón-Torres et al. 2013).

Morphological characterisation of the OES and EDJ
For the characterisation of the BSV enamel (OES) morphological traits and its comparison with other hominin samples, we employed the modified version of the Arizona State University Dental Anthropological System (Turner et al. 1991) by Martinón-Torres et al. (2012 and references therein).
Since there is not a scoring system for the non-metric traits at the dentine level, we employed the modified ASU-DAS system for OES (Martinón-Torres et al. 2012) to assess the morphological traits of the BSV and comparative sample (Table 9; SI Tables S1-S3). Moreover, for the Carabelli's trait, we followed the score system by Ortiz et al. (2012). Comparative sample includes the original dental remains of SH, Neanderthals and modern humans, as well as published data on Visogliano, Neanderthals and fossil H. sapiens (Zanolli et al. , 2019.

Enamel thickness topographic distribution
Three-dimensional topographic maps were generated by using the surface distance (between the enamel and dentine) module (SDM) in Amira (6.3.0, FEI Inc.). The computed distances between the OES and the EDJ are defined by a chromatic scale from thinnest (blue) to thickest (red) (Bayle et al. 2017;Macchiarelli et al. 2013). In addition to the BSV sample, we selected a sample exhibiting minimal enamel wear, including Atapuerca H. antecessor and SH, Neanderthals from Krapina and La Quina as well as modern humans. For representation purposes, left premolars and molars were mirror-imaged.

Geometric morphometrics of the EDJ
Geometric morphometric (GM) analyses of the EDJ were performed on the virtual surfaces of an original sample including BSV, SH, Neanderthals (NESPOS website) and modern humans (Table 1). Regarding BSV sample, those teeth exhibiting large dentine patches, such as the left P 3 , were excluded from the analysis. We selected the right BSV teeth due to less dental wear. When necessary, we performed Original data minimal dentine reconstruction, using the filling-holes tool in Geomagic (3D Systems, Inc.). Finally, we mirrored the comparative specimens according to the BSV teeth. For the premolars, we placed one landmark on each of the dentine horn tips of the P 3 and P 4 (protocone and metacone) and 49 semilandmarks along the marginal ridges (SI Fig. 1A). For the M 1 , we placed landmarks on the four cusps: protocone, paracone, metacone and hypocone.
However, due to variability in the cusp number in M 2 and M 3 , we placed landmarks on the protocone, paracone and metacone. In addition, we placed 95 semilandmarks along the marginal ridges of all molars (M 1 , M 2 and M 3 ; see SI Fig. 1B and C). We performed the weighted between-group principal component analysis (bgPCA) based on the Procrustes and deformation-based shape residuals (Mitteroecker and Bookstein 2011) but see Bookstein (2019) and Cardini et al. (2019) for a revision of the bgPCA analysis. Finally, we tested for allometry on the landmark-based analyses using the coefficient of determination (R 2 ) of a multiple regression (Bookstein, 1991), in which the independent variable is centroid size and the dependent variables are the bgPC scores (Mitteroecker et al. 2013).

Statistical analyses
For representation purposes, we computed standard boxplots of the three enamel thickness variables, including AET, RET and percentage of dentine and pulp. Due to the small BSV sample size, we performed adjusted Z-score, comparing individually each specimen against the SH, Neanderthal and modern human groups. The adjusted Z-score test (Maureille et al. 2001;Scolan et al. 2012) allows the comparison of unbalanced and reduced samples (one specimen against a group) by using Student's inverse t distribution. Adjusted Z-scores of AET, RET and percentage of dentine variables were computed to compare 3D dental tissue proportions and enamel thickness values of the BSV specimens to the means and standard deviations of the SH, Neanderthal and MH groups. In these Z-scores, the −1.0 to +1.0 interval comprises the 95% of the variation in the reference samples.

Metrics
Tables 5 and 6 present the MD and BL dimensions of the right BSV teeth. The MD and BL dimensions of BSV right       (Table 7) dental remains are closer to the mean value of SH sample than to any other group, except for the right M 2 and M 3 that are closer to Krapina mean values (Bermúdez de Castro et al. 1993, 1999, 2019Wolpoff 1979).
The ROPA of the BSV right M 1 represents the 24.0% of the TCBA (Tables 3 and 8). This value is within the variation range of SH population  and Neanderthals (Bailey 2004) and out of the range of variation reported for early and contemporary H. sapiens (Bailey 2004). As for the cusp angles (Table 4), BSV right M 1 follows the pattern A > C > B > D, which is the pattern described for H. antecessor ).

Morphological characterisation of the OES and EDJ
Following, we will describe the main morphological traits observed in the BSV teeth, OES and EDJ, as well as its comparison with the rest of the sample. A detailed description of the OES and EDJ morphology of the BSV and comparative sample can be found in SI and SI Tables S1-S3.
Overall, we observed that BSV displays the so-called typical Neanderthal morphology, in both the presence and degree of expression of traits. In addition, we found a good correspondence between the BSV OES and EDJ surfaces, although in some instances the expression of features in the EDJ is higher than what we observed in the OES. As such, we only included in the description those EDJ features that did not show correspondence with those observed in the OES ( Table 9 for the frequencies of the degree of expression of main morphological traits in BSV and comparative samples).
Left I 2 (Figs. 1 and 2) exhibits a wear category of 3-4 ( Molnar's classification, 1971). At the OES, this tooth shows a pronounced labial convexity, strong shovel shape and tuberculum dentale with a free apex. This suite of traits is commonly observed in SH, Neanderthals and Arago compared to modern humans (Martinón-Torres et al. 2012;Zanolli et al. 2018).
Right and left P 3 s (Figs. 1 and 3) exhibit wear categories 2 and 3 (Molnar 1971), respectively. While the right P 3 exhibits a single buccal and lingual essential ridges, these are bifurcated in the left P 3 , morphologies frequently seen in SH and Neanderthals (Martinón-Torres et al. 2012). Moreover, the left P 3 presents a continuous transverse crest, a trait commonly recorded in MP samples .
Right and left P 4 s ( Figs. 1 and 4) exhibit wear category 2 (Molnar 1971). We recorded a continuous transverse crest as well as a single buccal and a bifurcated lingual essential crest in the right P 4 . In the left P 4 , the essential crests are bifurcated, a common trait recorded in MP hominins ), but not in SH (Martinón-Torres) and Neanderthals (Martinón-Torres et al. 2012).
Right and left M 1 s (Figs. 1 and 5) exhibit wear category 3 (Molnar 1971). The bulging and well-developed hypocone is responsible for the distobuccal protrusion of the crown in

Enamel thickness
3D estimates of enamel thickness and crown tissue proportions of BSV1 specimens are described in Table 6 together with the mean, S.D. and range values of the comparative samples. Overall, BSV1 individual exhibits a mixed pattern characterised by thinly enamelled premolars and thickly enamelled molars (Fig. 8).
Although BSV1 right P 4 values for absolute and relative enamel thickness (AET and RET) are within the range of variation of all comparative samples, the BSV1 P 4 shows the lowest absolute and relative (AET and RET) values of enamel thickness in relation to the mean values of the comparative sample. BSV1 P 4 aligns with SH and Neanderthal samples for the AET, RET and Vcdp/Vc and discriminates from both the fossil H. sapiens and modern humans (Table 10) These results emphasise the affinity of BSV with SH and its departure from the Neanderthal condition (Table 10).
Although BSV right P 4 falls within the 95% of variation range of all comparative groups, the BSV specimen closer resembles the Neanderthal thin condition, shared also with Fig. 3 3D reconstruction of the P 3 outer enamel surface (OES) and enamel-dentine junction (EDJ) views in BSV1 compared to those of Sima de los Huesos, Neanderthal and modern human. SH = Sima de los Huesos (AT-2399), NEA = Neanderthal (La Quina), MH = modern human (UCL20). O = occlusal, B = buccal, L = lingual. (When needed, specimens have been mirrored to the right to match the BSV1 specimen) SH, and differentiates it from the modern human thick condition (Fig. 9). The BSV right M 2 closer resembles the modern human condition although the values for the AET, RET and Vcdp/Vc fall within the 95% of variation range of all comparative groups. The BSV1 right M 3 shows closer affinity with SH sample, thick condition, for all three variables, although BSV also falls within the 95% of modern human variation range. On the contrary, the BSV1 right M 3 clearly departs from the Neanderthal condition as it falls outside the 95% of the variation range.

Enamel thickness distribution
Overall, BSV1 dental remains show closer affinity with the pattern of enamel distribution shown by SH specimens (Figs. 10, 11 and 12). The pattern of enamel thickness distribution in the peripheral areas, mostly on the buccal and lingual surfaces of the two main cusps, is observed in BSV1, TD6, SH and Neanderthal specimens, and different from the modern human specimen showing thicker enamel on the occlusal basin (Fig. 10).
Although the wear exhibited by the BSV1 M 2 may conceal some aspects of the enamel distribution, in general BSV shares with SH and modern human specimen the larger areas of enamel thickness on the buccal and the lingual surface of the cusps. In TD6, the enamel is thicker and more widespread in these areas, while in Neanderthals, the thickness is more subtle in bucco-lingual cusps (Fig. 11).
BSV1 M 3 shares with the modern human specimen thicker enamel distributed in the lingual cusp (protocone) as well as in the two buccal cusps and the occlusal basin.
The distribution exhibited by SH specimen also resembles this. In contrast, in the Neanderthal specimen thickness is more peripherally distributed (Fig. 12).

Geometric morphometrics
The results of the bgPCA analyses of the BSV specimens are shown in Figs. 13, 14, 15, 16 and 17. Figure 13 illustrates the bgPCA analysis for the BSV1 right P 3 . The two principal components account for the 99.14% of the total variation. BSV1 specimen shares the negative PC1 morphospace with SH and Neanderthal specimens characterised by a more round-shaped outline conferred by the shorter M-D diameter. Similarly, BSV specimen clusters with all SH specimens, and few Neanderthal and modern human specimens, on the positive axis of the PC2, characterised by rounded, versus a rectangular, outline.
In the bgPCA analysis of BSV1 right P 4 (Fig. 14), the two principal components account for the 98.97% of the total variation, with the PC1 accounting for the 89.78%. BSV1 shares the morphospace with SH specimens and half of modern humans, characterised by a more rounded-shaped outline, with slightly larger mesio-distal diameter, and the protocone and paracone slightly distally placed in comparison to Neanderthals and the rest of the modern human sample. Figure 15 illustrates the bgPCA analysis for the BSV1 right M 1 . The two principal components account for the 84.26% of the total variation. BSV1 specimen shares the positive PC1 morphospace with the two specimens from SH and the majority of Neanderthals characterised by a rectangular outline conferred by the larger area of the distal cusps, specially the distally placed hypocone, and the shorter height of the dentine horns, in contrast to the fossil H. sapiens and modern human samples.
The results of the bgPCA analysis of the BSV1 right M 2 is shown in Fig. 16. The two principal components account for the 99.78% of the total variation. BSV1 specimen falls within the positive PC1 together with SH specimens and half of the modern human sample. These are characterised by a more squared-shaped outline consequence of hypocone reduction, and in contrast to the Neanderthal morphology characterised by a rectangular-shaped outline resulting from a distally displaced and protruding hypocone. Figure 17 illustrates the bgPCA for the right M 3 . The two principal components account for the 99.92% of the total variation. BSV1 specimen shares the positive morphospaces with SH specimens, three Neanderthals from Krapina and some modern human specimens. These are characterised by Adjusted Z-score of the 3D variables: AET (black circles), RET (red triangles) and percentage of dentine (green squares) assessed in P 4 , M 2 and M 3 from BSV and compared to the variation expressed by Sima de los Huesos (SH), Neanderthals (NEA) and modern humans (MH). The solid line passing through zero represents the mean, and the other two lines correspond to the estimated 95% limit of variation expressed for the two comparative samples an oval-shaped outline elongated on bucco-lingual direction resulting from the more distally placed protocone and the reduction of the hypocone, in contrast to the rest of the Neanderthal sample and majority of modern human specimens.

Discussion
In the last years, the discovery of new fossils as well as the reassessment of old findings is providing evidence of the high morphometric variation of the European populations during the MP, where populations with different geographic and chronological settings present different Neanderthal affinities. In this context, the variability of the MP populations could be better explained by less linear models such as the ebb and flow model (Hublin and Roebroeks 2009) -Robles et al. 2007-Robles et al. , 2011Roksandic 2016;Skinner et al. 2016). While, the second group would be characterised by closer morphological dental affinities with the classic Neanderthals, including the Atapuerca-SH, Pontnewydd, Fontana Ranuccio, Visogliano, Steinheim and Montmaurin hominins (e.g. Gómez-Robles et al. 2007, 2011Hanegraef et al. 2018;Martínez de Pinillos et al. 2020;Martinón-Torres et al. 2012;Zanolli et al. 2018). However, even within these groups there is variability. While Mauer and Mala Balanica hominins lack the thin enamel pattern characteristic of Neanderthals (Skinner et al. 2016;Smith et al. 2012), the Arago dentition exhibits thin enamelled crowns (Macchiarelli et al. 2013). Similarly, Fontana Ranuccio, Visogliano and Steinheim exhibit the enamel thickness pattern typical of Neanderthals, but Atapuerca-SH and Montmaurin molars are characterised by thick enamelled crowns.
Previous assessments of the BSV crania remains assigned these hominins to a population between late H. heidelbergensis and the first H. neanderthalensis (Vandermeersch, 1978(Vandermeersch, , 1982Rougier, 2003). Arsuaga and Martínez (1997) analysis of the temporal bone of BSV2 concluded that this individual closely resembled the morphology of the "Neanderthal 1" group than those of the European MP. Finally, Guipert et al. (2011) concluded that BSV2 individual belonged to the first European Neanderthals due to the combination of Neanderthal apomorphies and several plesiomorphies. Within this scenario, the study of BSV fossil teeth aims to explore the BSV affinities with other MP groups and Neanderthals as well as contributing to the discussion about the MP population variability. Previous metric data such as the CI, ROPA, TCBA, cusp angles or enamel thickness confirmed the differences between Neanderthals and H. sapiens (e.g., Bailey 2004;Bermúdez de Castro 1993;Buti et al. 2017;Martinón-Torres et al. 2013;Olejniczak et al. 2008). In this study, the performed metric analyses resulted in BSV showing greater affinities with SH and Neanderthals than with any other group for all variables except the cusp angles. BSV right M 1 shares the pattern with H. antecessor species , thus displaying a less Neanderthal-like condition than SH but more derived than in H. erectus. Similarly, the enamel thickness pattern shown by BSV dental remains, combination of thin enamel in the premolars and thick enamel in the molars, is exclusively shared with the MP population from SH. We can tentatively relate the enamel thickness similarities between BSV and SH to their morphometric similarities; however, enamel thickness variation results from the interplay of multiple factors, including genetic (Horvath et al. 2014), developmental and life history features (Dean et al. 2001;Grine 2002Grine , 2005Smith et al. 2007), the structural organisation of the mineralized dental tissues (Macchiarelli et al. 2006;Olejniczak et al. 2008), dental and body size reduction (Kupczik and Hublin 2010;Lieberman 2011;Smith et al. 2012) and dietary adaptions (Lucas et al. 2008;Olejniczak et al. 2008), that should also be explored. Regarding the external (OES) and internal (EDJ) morphology, BSV exhibits a suite of features characteristic of some MP populations (including SH, Pontnewydd and Vigliano) as well as Neanderthals (e.g. Bailey 2004Bailey , 2006Martinón-Torres et al. 2012;Zanolli et al. 2018Zanolli et al. , 2019. The concomitant expression in the I 2 of a marked shovel shape, a pronounced labial convexity and a well-developed tuberculum dentale represents the typical Neanderthal morphology. The M 1 s have a characteristic morphology, in which the occlusal contour is rhomboidal, with a distal displacement of the lingual cusps and a rounded protrusion of the distobuccal corner due to a large hypocone. In particular, this morphology was already present in the Early Pleistocene hominins from the Gran Dolina-TD6 site (Gómez-Robles et al. 2007) and thus, it may represent a primitive morphology retained by some MP populations (such as SH, Pontnewydd and Steinheim but not Arago) and Neanderthals Compton and Stringer 2015;Gómez-Robles et al. 2007). Although previous studies showed the limited morphological discrimination of the premolars between Neanderthals and H. sapiens (e.g. Martinón-Torres et al. 2019;Zanolli et al. 2019), the BSV premolars display some features that are prevalent in MP hominins and Neanderthals but quite uncommon in modern humans. As such, the presence of a transverse crest and lingual essential crest in the BSV P 3 s and P 4 s is also expressed in other MP populations (such as SH and Visogliano), and the majority of Neanderthals (including the recently described specimen from Wezhem) but its presence is quite limited in the modern human sample (Martinón-Torres et al. 2012;Zanolli et al. 2018Zanolli et al. , 2019; and this study). Even though the BSV M 2 and M 3 do not display any diagnostic feature shared exclusively with MP hominins and Neanderthals, the GM analysis of the EDJ shape contour in these molars, as well as in the premolars, shows that BSV dental remains share greater affinities with the MP sample from SH and modern humans than with Neanderthals. In particular, at the EDJ, the degree of reduction of the metacone and the hypocone in M 2 and M 3 aligns BSV with SH hominins and modern humans. Thus, the morphometric analysis of BSV dentition places these hominins within the MP group that exhibits more Neanderthal affinities in their dentitions, in particular, closer to the SH population due to the expression of the morphological traits, EDJ shape and enamel thickness. Still, and despite the younger chronology of the BSV hominins with respect to SH, in some instances BSV dental remains do not exhibit the derived condition for some traits that are indeed present in SH population. For instance, BSV exhibits the primitive condition for the cusp angle trait, following the pattern described for H. antecessor, while both SH and Neanderthals exhibit the derived condition.
When studying MP populations, one of the greatest challenges comes from the scarcity of the fossil record, which is mainly represented by isolated remains that are geographically and chronologically scattered. For decades, the discussion on the origin of the Neanderthal clade was dominated by the accretion theoretical model. This model explains the neanderthalization process as the gradual accumulation of Neanderthal traits through time (Dean et al. 1998;Hublin 2009). However, the MP fossil record, prominently the Atapuerca-SH hominins, questioned its validity (Dennell et al. 2011;Martinón-Torres et al. 2012). Despite its early chronology, ca. 430 ka, the SH dentition is morphologically "more Neanderthal" than other penecontemporaneous MP samples and even more derived than some classic Neanderthals, thus contradicting the gradual and ordered neanderthalization process defended by the accretion model, where specimens would align from less to more Neanderthal along a chronological scale (Martinón-Torres et al. 2012). This new description of the BSV dental remains reinforces the evidence of the high morphological variation of the European populations during the MP, where populations with different geographic and chronological settings present different Neanderthal affinities. As such, based on dental analyses, two groups are recognised; the first one clusters specimens characterised by a more primitive morphology with less to none Neanderthal affinities such as Mala Balanica (BH-1), Mauer or Arago (e.g. Bailey 2002;Bermúdez de Castro et al. 2019;Gómez-Robles et al. 2007, 2011Roksandic, 2016;Skinner et al. 2016). The second group includes those hominins exhibiting most (if not all) the dental features that are considered typical of the Neanderthal species, including the Atapuerca-SH, Pontnewydd, Fontana Ranuccio, Visogliano, Steinheim, Montmaurin and BSV hominins (e.g. Gómez-Robles et al. 2007, 2011Hanegraef et al. 2018;Martínez de Pinillos et al. 2020;Martinón-Torres et al. 2012;Zanolli et al. 2018). However, when considering other skeletal parts we still observe intrapopulation variability within these two groups. While BSV cranial remains exhibit clear Neanderthal features, the number of primitive features retained by the SH population is higher (Rougier 2003;Arsuaga and Martínez 1997). Similarly, while SH mandibles exhibit a clear Neanderthal pattern (Rosas 2001), the Montmaurin mandible is characterised by a particularly primitive conformation ). The question is whether these two MP groups represent (i) a single population with a high degree of variability or (ii) two distinct paleodemes, understood as "representative of prehistoric populations or lineages acting as portions of dynamic evolutionary units" (Trinkaus 1990) where only the one with clear Neanderthal dental affinities (including SH, Pontnewydd, Fontana Ranuccio, Visogliano, Steinheim, Montmaurin and BSV) can be considered active contributors for the Neanderthal gene pool. In contrast to the linearity proposed by the accretion model, we feel that the high variability within the MP populations fits better with the sink and source model (Dennell et al. 2011), where instead of an anagenetic in situ evolution of the MP populations into Neanderthals, the settlement of Europe is seen as the result of intermittent dispersals into Europe from a source population located outside the continent. The source population would evolve in what we named as the Central Area of Dispersals of Eurasia (CADE) giving rise to daughter populations that disperse to the West and the East of Eurasia when environmental conditions allow (Bermúdez de Castro and Martinón-Torres 2013; Dennell et al. 2011). The recent publication of the Nesher Ramla fossils in the Levantine Corridor (Hershkovitz et al. 2021) would support the idea of the Near East representing the residential area of a population that was likely involved in the evolution of the MP Homo and Neanderthals in Europe through discontinuous migrations.

Conclusion
We examined the external and internal dental structures of the maxillary dental remains recovered from Biache-Saint-Vaast site in order to contribute to the current debate on the variability of the European MP hominins and the evolution of the Neanderthal clade. Our study shows that BSV teeth cluster with other MP groups that show greater dental affinities with Neanderthals such as Atapuerca-SH, Fontana Ranuccio, Visogliano, Steinheim and Montmaurin, and in contrast to the specimens showing less Neanderthal features, such as Arago, Mauer or Mala Balanica. Within the Neanderthal-like group, we also observed variability related to the enamel thickness variation. In this aspect, the BSV hominins are closer to the Atapuerca-SH population, than to any other group (i.e. Fontana Ranuccio, Visogliano or Steinheim), since they show a unique combination of thin (premolars) and thick (molars) enamelled dentition. The results on BSV dental remains together with previous evidence of the MP fossil record indicate the coexistence of two or more populations in Europe that may result from an intermittent settlement of human groups originated in the CADE as proposed in the sink and source model.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.