The tooth assemblage from the Sierra de Atapuerca archaeological complex (Table 1) consists of three permanent canines (two upper and one lower) from the early Pleistocene Gran Dolina (TD6) site and 32 permanent canines (16 upper and 16 lower) from the middle Pleistocene Sima de los Huesos (SH) site. A detailed description of the sample studied here can be found below. The lithostratigraphic and geochronological contexts of the sites are described in detail in the Appendix 1.

At least 160 human fossil remains belonging to H. antecessor were recovered from the Gran Dolina level TD6-2, recently assigned to MIS 21 through different geochronological techniques applied to samples from TD7 and TD6 (Arnold et al., 2014, Berger et al., 2008, Falguères et al., 1999 and Moreno et al., 2015). Based on direct ESR dating of a hominin tooth, the last published results indicate a time range from 772 to 949 kyr (Duval et al., 2018). The TD6 hominins show a certain number of derived features also present in middle Pleistocene and early Late Pleistocene populations (Bermúdez de Castro et al., 2013). In particular, the relatively high number of features exclusively shared with Neanderthals and with the Atapuerca-SH hominins are noteworthy (Bermúdez de Castro et al., 2013 and Bermúdez de Castro et al., 2015). As mentioned, this mosaic of characteristics supported the definition of the new taxon H. antecessor (Bermúdez de Castro et al., 1997 and Carbonell et al., 1995) and has originated a debate about its relationships with the last common ancestor of Neanderthals and modern humans (Bermúdez de Castro et al., 1997, Bermúdez de Castro et al., 2017, Bermúdez de Castro et al., 2015 and Bermúdez de Castro et al., 2003). The three H. antecessor canines used in this study show a good preservation state. Specifically, ATD6-1 presents a slight cervical erosion, which nonetheless does not affect the enamel, and the ATD6-69 root had not finished forming.

The middle Pleistocene sample of Sima de los Huesos consists of 16 upper and 16 lower permanent canines. The human remains from this site are represented in a single lithostratigraphic level (LU6) (Arsuaga et al., 2014), which has provided mean ages of 433 ± 15 kyr and 416 ± 12 kyr using post-infrared-stimulated luminescence (pIR-IR) and thermally transferred optically stimulated luminescence (TT-OSL), respectively (Arnold et al., 2014). Initially, this human group was assigned to H. heidelbergensis (Arsuaga, 1993, Arsuaga et al., 1991, Arsuaga et al., 1997 and Bermúdez de Castro et al., 2004). Nevertheless, in 2014, Arsuaga et al. suggested that the SH sample should be removed from the H. heidelbergensis hypodigm due to the absence of derived Neanderthal features in the Mauer mandible, the holotype of this species. Moreover, in 2016 the nuclear DNA sequenced from two SH specimens indicated that these hominins belonged to the Neanderthal evolutionary lineage, being closely related to the ancestors of Neanderthals and sharing a common ancestor with Denisovans (Meyer et al., 2016). The state of preservation of the SH canines is generally good. There is a superficial loss of enamel on the labial aspect of AT-578, and at the incisal mesial border of AT-145 the enamel is slightly chipped. The apical portion of the roots of AT-60, AT-67, AT-2783, AT-1942, AT-2151, AT-2165, AT-6729 and AT-808 are missing. There is one additional specimen from this site, AT-1144, that has not been included in this analysis because its root and crown are pathologically altered.

The TD6 and SH permanent canines from Atapuerca were compared with those belonging to the more recent (120–130 kyr) Neanderthal population sample of Krapina, Croatia (Brace, 1979, Martínez and Arsuaga, 1997, Rink et al., 1995 and Wolpoff, 1979) (Appendix 1, see Table A.1). As a whole, the Krapina dental sample comprises 282 teeth, either isolated or preserved in jaws (Wolpoff, 1979). For the purposes of the present study, we used 18 canines (12 upper and six lower), generally in good preservation conditions; in this assemblage, only eight specimens have their roots complete.

In the analysis we also included a recent modern human (RMH) sample comprising a whole of 125 European and African specimens (Appendix 1, see Table A.2). This material comes from the anthropological collection housed at the Escuela de Medicina Legal of Madrid (Spain), the South African modern human bone collection permanently stored at the University of Pretoria (L’Abbé et al., 2005), and from Sudanese dental extractions (Elamin and Liversidge, 2013). In total, the modern human sample consists of 58 upper and 67 lower permanent canines from 81 individuals of both sexes.

In some cases, multiple teeth from the same individual were measured in the comparative samples, although only one antimere per individual was taken into consideration. We discarded teeth with any pathological damage affecting the mineralized tissues. Tooth crown wear stage was scored following Molnar (1971). Slightly worn teeth are those with a wear stage equal or below 3, characterized by the absence of the apex in the incisal border and the emergence of a point of dentine. In order to assess the effects of occlusal wear on such kind of studies dealing with tooth tissue proportions, we also compared those teeth with a wear degree of 4 (crown with minimal dentine patch), of 5 (crown with more extensive dentine patch), and of 6 (crown in which a lateral enamel loss is observed in the wear surface).

The isolated fossil teeth from Atapuerca were scanned using the Scanco Medical AG Micro-Computed Tomography 80 housed at the ‘Centro Nacional de Investigación sobre la Evolución Humana’ (CENIEH) in Burgos. Scans were performed employing two 0.1 mm copper filters and using a voltage of 70 kV and an amperage of 114 μA. The resultant slice thickness ranged between 18 and 36 micrometres (μm). Teeth included in mandibular and/or maxillary fragments were scanned using a Phoenix v/tome/x s (GE Measurement & Control) available in the same research centre. In this case, scans were performed by two 0.1 mm Copper filters, 100–120 kV voltage and 110–140 μA amperage, resulting isometric voxel ranging between 27 and 36 μm. Micro-CT data for the Krapina specimens (voxel size: 20–40 μm) were obtained from NESPOS^© microtomographic database.

The majority of the teeth forming the recent modern human sample were scanned using the Phoenix microtomographic system housed at the CENIEH and the CTP-Mlab micro-CT equipment set at the Multidisciplinary Laboratory of the International Centre for Theoretical Physics (ICTP) of Trieste, Italy (Tuniz et al., 2013). All scans were performed by two 0.1 mm copper filters, 100–120 kV voltage and 110–140 μA amperage. The output images have a voxels size ranging between 17 and 21 μm. The South African specimens were scanned at the South African Nuclear Energy Corporation (Necsa), Pelindaba, using a Nikon XTH 225 ST equipment according to the following parameters: 100 kV voltage and 100 μA amperage. The final volumes were reconstructed with an isotropic voxel size ranging between 40.8 and 50.8 μm.

Image processing was performed using the Amira 6.0.0 software (Visage Imaging, Inc.). Dental tissues (enamel, dentine-pulp complex) were semi-automatically segmented using the Watershed Segmentation Tool and through manual editing. A non-local means filter was also applied. Small fractures and cracks were virtually filled in. We considered the cervical line as the fundamental morphological feature to isolate the crown and root (Benazzi et al., 2014). After anatomically orienting each tooth, we drew a straight line between the enamel maximum cervical extensions in the mesio-distal plane in all of the stacks of images. Finally, we made some corrections in the bucco-lingual planes. Accordingly, only the dentine contained in the enamel cap was classed as coronal dentine, which was limited to the base by a curve with a smooth surface and isolated from the root (for technical details, see García-Campos et al., 2018).

On the virtually isolated crowns, the following variables were quantified following Olejniczak et al. (2008a): volume of the enamel cap (V _e, in mm³); volume of the coronal dentine including the coronal pulp (V _cdp, in mm³); surface area of the enamel-dentine junction (S _EDJ, in mm²). These values were subsequently used to compute the 3D average enamel thickness index (3DAET = V _e/S _EDJ, in mm); the 3D relative enamel thickness index 3 D R E T = 3 D A E T / V cdp 3 × 100 , $ \left(3DRET=3DAET/\sqrt[3]{{V}_{\text{cdp}}}\right)\times 100\text{,}$ scale free), the coronal volume (V _c = V _e + V _cdp, in mm³), and the percentage of coronal volume that is dentine and pulp (V _cdp/V _c = V _cdp/V _c × 100). Following Skinner et al., 2008a and Skinner et al., 2008b, we also assessed the outer surface of the enamel cap (S _OE, in mm²) used to compute the relative outer enamel complexity ratio (OES/EDJS, free scale). Finally, the volume of the root dentine including the pulp (V _r, in mm³) was assessed.

Statistical analyses were performed using the SPSS software (v. 18.0, SPSS Science, Inc.). Several statistical tests were employed to examine inter-taxon differences in tissue measurements, between the upper and lower canines, and by wear stages. In order to compare the crown variables, only slightly worn teeth (wear stages 1–3) were considered. Likewise, only the teeth bearing a complete root were used for root volume comparisons. Statistical tests were only applied when groups were represented by samples consisting of at least three cases. The normal distribution was assessed using the Shapiro–Wilk Test (when sample size was smaller than 50) and the Kolmogorov–Smirnov one-sample test with Lilliefors correction (when sample size was greater than 50). When normality could be assumed, we applied a Student's t-test; otherwise, we used the non-parametric Mann–Whitney U-test. Means were determined to be significantly different at the α = 0.05 level.

In order to estimate the intra- and inter-observer error, a sub-sample of seven teeth was analysed by two researchers (C.G.C. and M.M.M.). Each tooth was measured three times by both observers, who independently orientated, segmented, and isolated each specimen, and assessed each variable and index following the methods described above. The degree of error was estimated by calculating the inter- and intra-observer mean differences for each crown component measurement. The average intra-observer error was 1.59% (0.69–2.48%), whereas the average inter-observer error was 2.43% (1.72–3.15%). All values are within the accepted range.

The canines of Krapina display the largest crown volume (V _c) values, whereas those of recent modern humans exhibit the lowest ones (Table 2 and Table 3). Among the remainder groups, a marked reduction in V _c is noticeable in SH, whose values clearly overlap the extant human range of variation. In fact, the differences between SH and RMH for the upper canines are not significant (p-value = 0.357), while for the lower canines SH shows significantly larger crown dimensions (p-value = 0.006) (Table 4). The V _c values of the canines of the early Pleistocene sample of Gran Dolina TD6 fall mid-way between the middle Pleistocene sample of Atapuerca-SH (with smaller crowns) and Krapina (with larger crowns), near Krapina's lower range.

For the root volume (V _r), a marked size reduction compared to the European early and middle Pleistocene groups is noticeable in recent humans (Table 2 and Table 3). Both maxillary and mandibular canine roots are large in H. antecessor and overlap the SH and Krapina ranges. In spite of such similarity, SH root dimensions fall mid-way between the Neanderthal (larger values) and the recent modern human (lower values) estimates and are significantly different to both (p-values < 0.05), even if it is closer to the former (Table 4). In some specific cases, the SH root volumes overtake the Krapina's variation range (spec. AT-591 and AT-145).

With the exception of the Neanderthal sample, in all groups examined in this study, the V _c values of the maxillary canines are significantly larger than those measured in the mandibular ones, and the same is true for the V _r (Table 5).

The marked canine size variation is also reflected in the absolute tissue dimensions (Table 4 and Fig. 1 and Fig. 2). Here again, the enamel volume (V _e) and the coronal dentine-pulp complex volume (V _cdp) values of RMH are smaller, partially due to the smaller of the canine crown size, whereas H. neanderthalensis have larger magnitudes of both dental tissues (Table 4). With the exception of the lower canine V _e, both variables (V _e and V _cdp) differ significantly between the SH and Krapina samples (p-values < 0.05), exhibiting the first population's lower values (Table 4). Conversely, in SH, the V _e values of the maxillary canine approach those of the extant human estimates. Once again, H. antecessor's V _e and V _cdp values are intermediate between SH (lower volumes) and H. neanderthalensis (greater dimensions) in both tissues, but overlap with the variation range expressed by both samples (Fig. 1 and Fig. 2). The outer enamel surface (OES) and the enamel–dentine junction surface (EDJS) areas show distribution patterns comparable to those described for V _e and V _cdp, respectively (Table 4).

When the interpopulation differences are considered for each index (3DAET, 3DRET, V _cdp/V _c and OES/EDJS), we observe that, although RMH canines have absolutely smaller enamel cap volumes, their relative thickness (3DRET) is significantly greater than measured in the other samples (p-values < 0.05) (Table 4 and Fig. 1 and Fig. 2). Likewise, the fossil Homo samples have smaller OES/EDJS values than those measured in extant humans. Lastly, the canines from TD6, SH and Krapina display a higher percentage of crown volume that is a dentine-pulp complex (V _cdp/V _c), whereas this index is lower in the extant human sample (Fig. 1 and Fig. 2).

The differences in upper and lower canine dimensions are also reflected in their tissue components (V _e and V _cdp), being absolutely larger in the upper canines. The same is true for EDJS and OES. Contrary to the remainder, in the Neanderthal sample, the mean values of the absolute measurements assessed on the upper and lower canines are not significantly different, except for the values of V _e (Table 5). On the other hand, the average and relative enamel thickness (3DAET, 3DRET) is higher in the upper canines. SH is the only population in which the 3DRET is similar in both canines. Differently from the enamel, in all groups the V _cdp/ V _c values are higher in the mandibular canines, even if the differences are not significant, except in the case of H. neanderthalensis (p-value = 0.011). The OES/EDJS values are similar in the upper and lower teeth of the SH and Krapina samples. However, this index is greater in the H. antecessor and modern human upper canines, but statistically different only in the latter group.

The colour maps shown in Fig. 3 and Fig. 4 reflect patterns of enamel distribution, in which the thickest enamel is represented in red and the thinnest in blue. These maps support the 3DAET and 3DRET results noted. Overall, in both maxillary and mandibular canines of all population examined, the enamel cap is thicker on the buccal aspect, especially on the cuspal half of the crown; conversely in the cervical portion is relatively and absolutely thinner. However, whereas in the upper canines the thicker enamel is homogeneously spread out (involving both distal and mesial marginal ridges), in the lower ones the thickest values are more distally placed. In the H. antecessor, SH and Krapina mandibular canines, the enamel cap is thicker on the distal than on the mesial crown aspect, and is especially thick at the level of the distal marginal ridge. Among the fossil samples, Krapina shows the lowest thickness values. However, although the RMH lower canines also exhibit thicker enamel on the distal aspect, thickened enamel is found in the rest of the incisal half of the buccal surface, till the marginal ridge. Likewise, although the enamel thickness patterns of upper canines (Fig. 3) seem to be common in all groups, in RMH the greater thickness spreads to the half of the crown.

The descriptive statistics of the indices obtained from the measures performed on the subset of more extensively worn teeth (degrees 4 to 6; Molnar, 1971) are shown in the Appendices, Tables B.1 and B.2. As expected from the enamel thickness distribution maps (Fig. 3 and Fig. 4), occlusal wear affects the dental tissue proportions and a greater variation is observed as wear increases. As it was predictable, in all samples the enamel thickness (3DAET, 3DRET) values are lower in the more extensively worn crowns. In general, the percentage of variation of the indices is greater in the upper canines. On the other hand, the V _cdp/V _c values are larger in more worn teeth, mainly due to enamel volume reduction. Each species represented in this study is affected by wear in a different way, so their indices neither increase nor decrease in the same way.

In contrast to the posterior dentition, latter Homo species anterior tooth crowns and roots have been described as increasing in size in comparison with previous hominin taxa, being this increment more pronounced in Neanderthals, who even exceed the dimensions of H. sapiens (Buti et al., 2017 and Wolpoff, 1999). The Neanderthal incisors are known to be especially large relative to the rest of dentition (Wolpoff, 1979). Our results show that Neanderthal canine crowns were also larger than those of H. sapiens, which agrees with Buti et al. (2017). This increase in size is especially pronounced in the lower canines, which reach similar dimensions to those of the upper canines, a characteristic which is not observed in any other among the species included in this study (Table 5). Neanderthal and recent modern human anterior teeth also differ for their root size. Le Cabec et al. (2013) observed that Neanderthals have larger anterior roots dimensions than recent modern humans, which is supported by our results. These authors also noted that the anterior roots of the early and middle Pleistocene specimens were at least as large as in Neanderthals, thus concluding that Neanderthal root dimensions could result from the retention of an ancestral condition.

Despite the shortage of comparative data of anterior teeth tissue proportions currently limits the possibility of assessing evolutionary patterns in a reliable way, the available evidence suggests that the changes in canine crown size of Neanderthals can be preferentially attributed to shifts in the volume of the dentine–pulp complex, whereas enamel volume represents a minor role (Buti et al., 2017). Our results support this statement. Although it has been observed that H. neanderthalensis canines have a significantly larger enamel volume (V _e) than H. sapiens, the increment in their dentine-pulp complex (V _cdp, V _cdp/V _c) is even greater, which, together with larger and more complex EDJ surfaces (EDJS, OES/EDJS), results into relatively lower enamel thickness (3DRET) values. Such differences are also expressed in the patterns of enamel distribution showed in Fig. 3 and Fig. 4. The tooth canine assemblage from Krapina shows lower enamel thickness values than measured in our recent modern human sample, which is reflected in a less intense red colour. It also can be observed that in modern humans the areas with greater enamel thickness values are more widely spread through the buccal surface of the crown, especially in lower canines.

Therefore, these results support the first hypothesis and show that, together with a number of characteristic morphological features (Gómez-Robles, 2010, Gómez-Robles et al., 2012, Kupczik and Hublin, 2010, Martínez de Pinillos et al., 2014, Martinón-Torres et al., 2012 and Martinón-Torres et al., 2014), crown and root size, as well as dental tissue proportions, represent useful indicators to discriminate between the Neanderthal and H. sapiens lineages.

In an evolutionary perspective, a point of major interest concerns the assessment of the time when such tooth structural differences between the two lineages emerged. With this respect, Buti et al. (2017) assessed dental tissue proportions of early modern human (EMH), Upper Palaeolithic modern human (UPMH) and recent modern human (RMH) canine teeth and separately compared them with those of Neanderthals. In contrast to later H. sapiens populations, they observed that EMH canines are within the range of variation of Neanderthals for all crown tissue components (V _e, V _cdp, EDJS). Similar results were obtained by Smith et al. (2012) in a study on 2D dental tissue measurements variation within the genus Homo, where it has been concluded that differences across the dentition are less pronounced between Neanderthals and fossil H. sapiens than between Neanderthals and recent H. sapiens. The same conclusion was reached by Le Cabec et al. (2013) about root dimensions. As a whole, these results provide evidence that the anterior teeth tissue proportions and root size typical of extant populations appeared late in time. Specifically, this suggests that enamel thickness in modern humans may not be the result of a plesiomorphic trait retention. However, whereas it has been suggested that large root can be considered a primitive condition (Le Cabec et al., 2013 and Le Cabec et al., 2012), the question of when the overall thinly enameled pattern emerged in the Neanderthal lineage is still unresolved.

The dentition of the middle Pleistocene population of Sima de los Huesos presents most of the morphological traits usually considered characteristic of the Neanderthals (Martinón-Torres et al., 2012), displaying an expression of ‘mass additive’ features in their anterior dentition (Irish, 1998 and Martinón-Torres et al., 2012). The derived conformation of SH teeth contrasts with that of other penecontemporaneous populations such as that of Arago (Bermúdez de Castro et al., 2018), even overtaking some classic Neanderthals like those of Krapina, Hortus and Le Moustier (Gómez-Robles, 2010, Gómez-Robles et al., 2007, Gómez-Robles et al., 2008 and Gómez-Robles et al., 2011). Regarding the tooth size variation, the metric features of the Atapuerca–SH dental remains correspond to the evolutionary pattern characterizing the middle Pleistocene populations. Although SH canine and incisor sizes are large in relation with molar size, showing an expansion in their anterior teeth, their dentition is characterized by their small dimensions (Bermúdez de Castro, 1986). The results from this study reassert the reduced crown volume of their canines in comparison to those of Neanderthals, reaching a similar size to that of modern humans. This reduction is less marked in their roots, which, despite having volumes statistically smaller than those of Neanderthals, clearly falls into their range of variation, showing, perhaps, its more conservative nature. However, in spite of their size, SH canine crowns comprise similar percentages of dentine (upper canines: 55.10 ± 2.08%; lower canines: 56.11 ± 2.40%) than those of Neanderthal populations (upper canines: 57.83 ± 1.16%; lower canines: 60.38 ± 1.39%), as well as similar EDJ surface complexity (OES/EDJS values). As a result, the 3DRET of their upper and lower canines falls into the variation range of H. neanderthalensis, being statistically similar. Likewise, the enamel thickness coloured maps (Fig. 3 and Fig. 4) show that SH and Krapina populations share the general pattern of enamel thickness distribution, both in the upper and lower canines.

On the other hand, H. antecessor is defined by a unique mosaic of: primitive traits, shared with early Homo species, and derived features, some of them present in modern humans and others shared with Neanderthals and SH (Bermúdez de Castro et al., 2015, Bermúdez de Castro et al., 1999a and Martinón-Torres et al., 2012). Despite the fact that most of the dental traits of the Atapuerca early Pleistocene homonins are primitive, which align them with H. ergaster, H. erectus and some African early and middle Pleistocene specimens (Bermúdez de Castro et al., 2015), the permanent lower and upper canines of H. antecessor are morphologically derived with regard to the Homo clade (Martinón-Torres et al., 2008). Moreover, H. antecessor shares with European middle Pleistocene groups the expansion of the anterior teeth, particularly in the upper canines, which are near to the maximum range recorded for the genus Homo (Bermúdez de Castro et al., 1999a). Our results corroborate the large canine crown and root volumes of Gran Dolina population, being encompassed into the range of Neanderthals (near the minimum in the case of the crowns) but greater than SH teeth. Their absolute dental tissue measurement values (V _e, V _cdp, EDJS and OES) are also intermediate between those of H. neanderthalensis and SH hominins, but similar to both (Fig. 1 and Fig. 2). As a result, H. antecessor shares with these species the dental tissue proportions of its permanent canines (Table 2 and Table 3). Likewise, their pattern of enamel thickness distribution shown in the colour maps (Fig. 3 and Fig. 4) is closer to that of H. neanderthalensis, although not as much to that of SH.

Canine dental tissue proportions results showed above support the second hypothesis. The noticeable similarities between SH and Krapina dental tissue proportions indicate that thinly enamelled pattern cannot be considered as an autapomorphic Neanderthal trait but a homologous trait shared with at least the middle Pleistocene hominins from Atapuerca. On the contrary, our results appear to reject the third hypothesis, since Gran Dolina canines seem to exhibit thinly enamelled pattern, a feature that is shared with SH hominins and Neanderthals anterior dentition. This would moves backwards the timing for the appearance of the thinner enamel derived feature to around 800 kyr ago, suggesting that the relatively thin enamel might be a primitive trait in Neanderthals. However, the scarce H. antecessor sample size forces us to take these results with caution, and wait for new studies on the rest of their dentition to be published. On the other hand, these results seem to support the existence to certain similarity between the Atapuerca populations and the African H. erectus/ergaster, distinguishing them from the thickly-enamelled Asiatic H. erectus (Smith et al., 2012 and Zanolli, 2014). Once again, future studies about posterior dentition tissue patterns will be those which solve this issue.

Lastly, the large size of the canine roots of both H. antecessor and SH population supports the primitive polarity of this trait, highlighting that big root dimensions seems to be associated in some way with the large dentine-pulp complex percentages in their crowns.

Our study shows that wear affects the taxonomical comparison of enamel and dentine dimensions (in agreement with Buti et al., 2017), supporting the last hypothesis proposed in this study. The enamel thickness coloured maps show that in all species examined enamel is thicker in the incisal area of the buccal view. Therefore, the more pronounced the wear is, the less noticeable are the inter-specific differences. Additionally, the patterns of enamel thickness distribution change within species so they are affected in a different manner by this erosive process, making it difficult compare them. So although in many cases there is no other option than to compare teeth with similar stages of wear, due to the lack of slightly worn teeth samples, the conclusions obtained from this comparison should be taken with caution.

Many hypotheses have been proposed to explain the adaptive significance of dental tissue variation. Some authors have interpreted the variation of molar enamel and dentine amounts as a response to changes in functional demands, associated with the increase of the resistance as a result of the consumption of more abrasive foods (Lucas et al., 2008). However, there do not appear to be clear differences in the relative enamel thickness thresholds between hard- and soft-object feeders, which suggests that dietary inferences based on enamel thickness are not necessarily accurate (Dumont, 1995 and Smith et al., 2012). Moreover, these inferences may be complicated by non-masticatory behaviours (Constantino et al., 2011, Hall and Schaller, 1964 and Martin et al., 2003). On the other hand, the attrition of the thinly enamelled Neanderthal anterior teeth has been often compared with that of current groups such as the Inuit and Australian Aborigines (Hinton, 1981 and Molnar, 1971), who show the highest rates of attrition within modern human populations. Other authors have proposed that the amount of enamel is the result of both the enamel secretion rates and the formation times of teeth (Grine and Martin, 1988). Nevertheless, once again, Neanderthal molars, as well as canines, seem to have secretion rates and crown formation times that do not significantly differ from those of modern human (Dean, 2009 and Dean et al., 2001). However, these authors might on the right track since a combination of similar secretion rates and formation times with larger EDJ surfaces areas might be behind the decrease in enamel thickness in Neanderthal teeth. In this framework, relative enamel thickness variation may be, therefore, interpreted primarily as a consequence of changes in the coronal amount of dentine more than in the enamel cap.

Following this reasoning, it is important to notice that the degree of variation of molars dentine usually appears correlated with changes in root dimensions and craniofacial morphology (Kupczik and Hublin, 2010 and Lieberman, 2011). Likewise, some studies have linked the large anterior dentition of Neanderthals with the increase of their craniofacial skeleton (Anton, 1990, Hublin, 1998 and Trinkaus, 1987). This association may suggest similarities in dentine and bone tissue response to different physiological changes, possibly as a result of a common ontogenetic development and similar organic composition, which differentiates them from the enamel component (Plavcan, 2012 and Smith et al., 2012). The Growth Hormone (GH)/Insulin-like growth factor I (IGF-I) axis might to be behind both tissues regulation. This hormonal axis is a major regulator of postnatal growth and development (Giustina and Veldhuis, 1998), which has a strong influence on the metabolism of bone and oral tissues (Slootweg, 1993). GH directly acts on the osteoblast to stimulate bone formation (Nishiyama et al., 1996). Likewise, this hormone is able to induce the proliferation of epithelial steam cells and Hertwig's epithelial root sheath, the preodontoblast differentiation and the dentine matrix formation (Young et al., 1992).

According to the GH/IGF-I axis influence hypothesis, the increase in the amount of dentine should appear associated with larger cranial and postcranial robustness. SH specimens share derived facial, dental, mandibular, and glenoid features with Neanderthals, which overall lead to a functional masticatory complex similar in both groups, more robust to those from modern humans (Arsuaga et al., 2014 and Harvati et al., 2011). On the other hand, in a study by Arsuaga et al. in 2015, in which SH postcranial morphology was compared to that of other Homo species, the authors concluded that SH may be included in ‘wide Homo’ bauplan, showing wide and heavy bodies whose parameters (stature and body mass) were largely maintained in European middle Pleistocene groups, including Neanderthals (Arsuaga et al., 2015). Estimates of body proportions in H. antecessor suggest that TD6 males presented similar stature and body mass to Sima de los Huesos male individuals (Pablos et al., 2012 and Pablos et al., 2017), reflecting a marked robustness, which contrasts with their face gracility (Bermúdez de Castro et al., 2008 and Rosas and Bermúdez de Castro, 1999). Therefore, although it could be very risky to try to directly relate changes in the dentine-pulp complex with shifts in bone robustness, all these evidences seem to show the existence of parallel responses in dentine and bone tissues in the face of identical physiological signals.