The bony labyrinth of Neanderthals

Transcription

1 Journal of Human Evolution 44 (2003) The bony labyrinth of Neanderthals Fred Spoor a*, Jean-Jacques Hublin b, Marc Braun c, Frans Zonneveld d a Evolutionary Anatomy Unit, Dept. of Anatomy and Developmental Biology, University College London, Rockefeller Building, University Street, London WC1E 6JJ, UK b Laboratoire d Anthropologie, Université de Bordeaux 1, Avenue des facultés, Talence, France c Dept. of Anatomy, University of Nancy I, France d Department of Radiology, Utrecht University Hospital, The Netherlands Received 3 July 2002; accepted 7 October 2002 Abstract This paper presents a comprehensive comparative analysis of the Neanderthal bony labyrinth, a structure located inside the petrous temporal bone. Fifteen Neanderthal specimens are compared with a Holocene human sample, as well as with a small number of European Middle Pleistocene hominins, and early anatomically modern and European Upper Palaeolithic humans. Compared with Holocene humans the bony labyrinth of Neanderthals can be characterized by an anterior semicircular canal arc which is smaller in absolute and relative size, is relatively narrow, and shows more torsion. The posterior semicircular canal arc is smaller in absolute and relative size as well, it is more circular in shape, and is positioned more inferiorly relative to the lateral canal plane. The lateral semicircular canal arc is absolutely and relatively larger. Finally, the Neanderthal ampullar line is more vertically inclined relative to the planar orientation of the lateral canal. The European Upper Palaeolithic and early modern humans are most similar, although not fully identical to Holocene humans in labyrinthine morphology. The European Middle Pleistocene hominins show the typical semicircular canal morphology of Neanderthals, with the exception of the arc shape and inferiorly position of the posterior canal and the strongly inclined ampullar line. The marked difference between the labyrinths of Neanderthals and modern humans can be used to assess the phylogenetic affinities of fragmentary temporal bone fossils. However, this application is limited by a degree of overlap between the morphologies. The typical shape of the Neanderthal labyrinth appears to mirror aspects of the surrounding petrous pyramid, and both may follow from the phylogenetic impact of Neanderthal brain morphology moulding the shape of the posterior cranial fossa. The functionally important arc sizes of the Neanderthal semicircular canals may reflect a pattern of head movements different from that of modern humans, possibly related to aspects of locomotor behaviour and the kinematic properties of their head and neck Elsevier Science Ltd. All rights reserved. Keywords: bony labyrinth; temporal bone; Neanderthals; Pleistocene hominins Introduction * Corresponding author address: f.spoor@ucl.ac.uk (F. Spoor). The temporal bone of Neanderthals shows a suite of derived morphological features (e.g., /03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved. doi: /s (02)

2 142 F. Spoor et al. / Journal of Human Evolution 44 (2003) Vallois, 1969; Hublin, 1978; Santa Luca, 1978; Condemi, 1988), and it is, therefore, among the most diagnostic parts of the Neanderthal skull. Efforts to explore the inside of the Neanderthal temporal bone have employed radiological techniques to visualize mastoid pneumatization (Kindler, 1960; Kindler and Kiefer, 1963) and, inside the petrous part, the bony labyrinth (Delattre et al., 1967; Fenart and Empereur- Buisson, 1970; Wind and Zonneveld 1985; Zonneveld and Wind 1985; Silipo et al., 1991; Zollikofer et al., 1995). The latter structure houses the inner ear, which includes the sense organs for the perception of sound in the cochlea, and of movement and spatial orientation in the vestibule and semicircular canals. Using computed tomography, Hublin et al. (1996) provided a first comparative analysis of the bony labyrinth of Neanderthals. The study identified a number of characters which appear to distinguish Neanderthals from both modern humans and Homo erectus, and used these to establish the phylogenetic affinities of the infant Châtelperronian temporal bone from Arcy-sur-Cure (France). These findings confirmed previous observations that the mammalian bony labyrinth tends to show a consistent, species-specific morphology (Hyrtl, 1845; Gray, 1907, 1908; Spoor, 1993), and can help identify a fossil s affiliation (Spoor, 1993; Spoor et al., 1994). Hublin et al. (1996) found that in Neanderthals the arc sizes of the vertical (anterior and posterior) semicircular canals are smaller than in modern humans and H. erectus, whereas its lateral canal is larger-arced. Moreover, the position of the posterior canal was described as markedly inferior relative to the plane of the lateral canal. It was observed that among great apes and a number of hominin species the position and size of the posterior canal are correlated: the larger the canal the more inferiorly it is positioned. However, Neanderthals do not follow this general trend because their posterior canal is inferiorly positioned, but relatively small. Given that H. erectus is similar to modern humans in any of the traits that characterize the Neanderthal labyrinth (Spoor, 1993; Spoor and Zonneveld, 1994; Spoor et al., 1994), Hublin et al. (1996) concluded that the Neanderthal morphology is likely derived relative to both H. erectus and modern humans. Subsequent studies of Neanderthal specimens have established that Dederiyeh from Syria shows the typical Neanderthal labyrinthine morphology (Spoor et al., 2003), whereas Le Moustier 1 appears to have a morphology closer to that of modern humans (Thompson and Illerhaus, 1998; Ponce de León and Zollikofer, 1999). Spoor and Zonneveld (1998) describe two key influences on the morphology of the labyrinth that likely underlie differences between primate species. The arc size and planar orientation of the semicircular canals are directly linked to their sensory function of perceiving angular head motion, whereas other aspects of labyrinthine shape are correlated with cranial base morphology, such as the degree of sagittal flexion. Apparently characterized by different canal arc sizes and an inferiorly positioned posterior canal, the Neanderthal labyrinth appears to show features related to both function and cranial base morphology. Expanding upon the initial findings of Hublin et al. (1996) this paper presents a more comprehensive comparative analysis of the Neanderthal bony labyrinth. The full morphology of the structure is considered, the Neanderthal sample is increased, and preliminary comparisons are made with small samples of European Middle Pleistocene hominins, as well as early modern and European Upper Palaeolithic humans. Materials and Methods The sample investigated comprises 15 Neanderthals, four European Upper Palaeolithic modern humans, two early anatomically modern humans, and three European Middle Pleistocene hominins (Table 1), as well as 54 Holocene humans with a worldwide, geographically diverse origin (see Appendix 5.1 of Spoor, 1993). All specimens in the comparative sample are adult, but some Neanderthal specimens are immature. The latter can be compared directly with the adult specimens because the bony labyrinth reaches adult size and shape well before birth (Bast, 1930; Spoor, 1993). The Neanderthals correlate with oxygen isotope

3 F. Spoor et al. / Journal of Human Evolution 44 (2003) Table 1 The fossil sample analyzed in this study Neanderthals European Upper Palaeolithic Early anatomically modern European Middle Pleistocene Dederiyeh (r; im) Abri Pataud 1 (r; ad) Qafzeh 6 (l; ad) Abri Suard (l; ad) Gibraltar 1 (r; ad) Abri Pataud 3 (l; im) Skhul 5 (r; ad) Reilingen (r; ad) Gibraltar 2 (l; im) Cro-Magnon 1 (r; ad) Steinheim (l,r; ad) La Chapelle-aux-Saints (l; ad) Laugerie Basse 1 (l; ad) La Ferrassie 1 (r; ad) La Ferrassie 2 (r, ad) La Ferrassie 3 (l,r; im) La Quina 5 (l; ad) La Quina H27 (r; ad) Le Moustier 1 (l,r; im) Pech de l Azé 1 (l; im) Petit-Puymoyen 5 (r; ad) Spy 1 (r; ad) Spy 2 (r; ad) Tabun C1 (l; ad) Abbreviations: l. left side; r. right side; ad. adult; im. immature. stages 3 and 4, with the exception of Tabun C1 which likely correlates with oxygen isotope stage 5 or 6 (Grün and Stringer, 2000, but see Schwarcz et al., 1998). The term modern human, when used without any further qualification, refers to the combined Holocene, Upper Palaeolithic and early anatomically modern human samples, or the populations they represent. The bony labyrinths were visualized by computed tomography (CT), following the procedures described in Spoor and Zonneveld (1995). The following medical CT scanners were used: Philips Tomoscan 310/350 (Holocene humans, Gibraltar 2, Reilingen, Steinheim, Tabun C1), Siemens Somatom plus 4 (Gibraltar 1, Skhul 5, Spy 1&2), Toshiba Xvigor (Dederiyeh 93002), General Electric Highspeed (all other specimens). The scans were made in the sagittal plane, and in a transverse plane parallel with the arc of the lateral semicircular canal. The slice thickness is 1.0 or 1.5 mm and the slice increment ranges from 0.5 to 1.5 mm. Images were reconstructed with a pixel size of between 0.1 and 0.3 mm. One specimen, Le Moustier 1, was scanned with a nonmedical CT scanner, and its image dataset has an isotropic voxel size of 0.1 mm (Thompson and Illerhaus, 1998). One labyrinth of each specimen was investigated, with the exception of three fossils of which both sides were assessed (Table 1). In the latter case measurements of the left and right side were averaged, noting that bilateral differences are small compared with inter-individual ones (Spoor, 1993). Measurements, taken from the digital images, are those used in Spoor and Zonneveld (1998), a comprehensive review of extant primate labyrinthine morphology. They are shown in Figure 1, summarized in Table 2, and detailed definitions are given in Spoor and Zonneveld (1995). After reduction of the pixel size through weighted interpolation, linear measurements were taken to the nearest tenth of a millimeter, and angles to the nearest degree. For the CT images with the lowest resolution used in this study (Philips Tomoscan 310/350), the maximum error of these measurements was established experimentally as 0.1 mm and 4 degrees, respectively (Spoor and Zonneveld, 1995). The images with higher resolutions will have at least this accuracy and precision. The linear measurements of the labyrinth include the height and width of the arc of each semicircular canal and of the basal turn of the cochlea [Figure 1(a),(b)]. The height of each canal arc is measured to the point furthest away from the vestibule (the vertex), and the width perpendicular to the height. The radius of curvature (R) of each

4 144 F. Spoor et al. / Journal of Human Evolution 44 (2003) Fig. 1. Superior (a), (c) and lateral (b), (d) aspects of a left human labyrinth, as reconstructed from transverse and sagittal CT scans respectively, showing the measurements used in this study. Measurement abbreviations are listed in Table 2. canal arc and of the cochlear basal turn was calculated by taking half the average of the height and width measurements (0.5[h + w]/2). Interspecifically, the size of the semicircular canals and the cochlea correlates with body mass (Watts, 1924; Jones and Spells, 1963; Spoor and Zonneveld, 1998). Ideally, this phenomenon should be taken into account when comparing the hominin groups, even though the increase of labyrinth size with body mass is small (e.g., a Gorilla labyrinth is about 2.8 times larger than a Microcebus labyrinth), and body mass estimates for the fossil groups assessed here do not vary widely (Ruff et al., 1997). Cochlea size scales uniformly among extant primate species (Spoor and Zonneveld, 1998), and the reduced major axis

5 Table 2 The abbreviations in alphabetical order of the measurements of the labyrinth and the petrous pyramid Linear dimensions and their derivatives ASCh Height of the anterior semicircular canal [Figure 1(b)]. ASCw Width of the anterior semicircular canal [Figure 1(a)]. COh Height of the basal turn of the cochlea [Figure 1(b)]. COw Width of the basal turn of the cochlea [Figure 1(a)]. h/w Shape index of the arc of a semicircular canal or of the basal turn of the cochlea (height/width 100) LSCh Height of the lateral semicircular canal [Figure 1(a)]. LSCw Width of the lateral semicircular canal [Figure 1(a)]. PSCh Height of the posterior semicircular canal [Figure 1(a)]. PSCw Width of the posterior semicircular canal [Figure 1(b)]. R Radius of curvature of a semicircular canal or the cochlear basal turn, measured to the centre of the lumen (R = 0.5 [height + width]/2) SLI The sagittal labyrinthine index, calculated from the dimensions SLIs and SLIi as SLIi/(SLIs + SLIi) 100 [Figure 1(b)]. Quantifies how the arc of the posterior canal is positioned relative to the plane of the lateral canal (LSCm) Orientations APA The ampullar line, connecting the centres of the anterior and posterior ampullae, and projected onto the sagittal plane [Figures 1(d), 7]. Reflects how the vertical canals are set onto the vestibule in lateromedial view. ASCi The inferior most part of the anterior semicircular canal, defined by the line in the transverse plane connecting the apertures of the anterior ampulla and the common crus into the vestibule [Figure 1(c)]. Labelled as V in Spoor and Zonneveld (1995) ASCm The arc of the anterior semicircular canal at its greatest width in the transverse plane [Figure 1(c)]. ASCs The superior most part of the anterior semicircular canal in the transverse plane [Figure 1(c)]. CCR The common crus in the sagittal plane [Figure 1(d)]. COs The basal turn of the cochlea in the sagittal plane [Figure 1(d)]. COt The basal turn of the cochlea in the transverse plane [Figure 1(c)]. FC3 The third part of the facial canal in the sagittal plane [Figure 7]. LSCI The lateral most part of the lateral semicircular canal in the sagittal plane [Figure 1(d)]. LSCm The arc of the lateral semicircular canal at its greatest width in the sagittal plane [Figure 1(d), 7]. LSCt The axis of symmetry of the lateral semicircular canal in transverse plane [Figure 1(c)]. PPp The posterior petrosal surface in the sagittal plane at the level of the common crus [Figure 7]. PSCi The inferior limb of the posterior semicircular canal in the transverse plane [Figure 1(c)]. PSCm The arc of the posterior semicircular canal at its greatest width in the transverse plane [Figure 1(c)]. PSCs The superior limb of the posterior semicircular canal in the transverse plane [Figure 1(c)]. SG Intersection of the (mid)sagittal plane of the cranium in the transverse plane. VC The vestibulocochlear line, connecting the centre of the arc of the lateral semicircular canal and the lateral most point of the second cochlear turn, projected onto the sagittal plane [Figure 1(d)]. Reflects the infero-superior position of the cochlea relative to the vestibule. VSC Reference line in the transverse plane based on the vertical semicircular canals. It bisects the angle between the arc orientations of these two canals that opens anteriorly or posteriorly [ASCm, PSCm; Figure 1(c)]. F. Spoor et al. / Journal of Human Evolution 44 (2003)

6 146 F. Spoor et al. / Journal of Human Evolution 44 (2003) (RMA) regression is used here to correct for body mass. Semicircular canal size, on the other hand, does not scale uniformly among primates (Spoor and Zonneveld, 1998). The regressions most appropriate to correct for body mass in the current study are those for the extant great apes, as these appear to represent the primitive condition for hominins (Spoor, 1996). However, based on just four extant species the correlations of canal size with body mass do not reach statistical significance, and the possible impact of scaling on the comparisons of the three canals is only considered visually on the basis of bivariate plots. Estimated body masses used follow Smith and Jungers (1997) for all extant primates, including Holocene humans, and Ruff et al. (1997) for the Upper Palaeolithic and early modern humans and Neanderthals. For the European Middle Pleistocene a rough minimum estimate of 80 Kg was used (C. Ruff, personal communication). In addition to the height and width measurements of the canals and the cochlea two further linear measurements were taken to calculate the sagittal labyrinthine index (SLI). This index expresses the percentage of the posterior semicircular canal that is located inferiorly to the level of the lateral semicircular canal [Figure 1(b): SLIs, SLIi]. Angles were taken to quantify the spatial orientation of labyrinthine structures in relation to each other and in relation to aspects of the cranium. All orientations are defined in the transverse or sagittal plane [Figure 1(c),(d)], and angles calculated between two such orientations are therefore projected onto either of these planes. Angles are indicated by the abbreviations of the two orientations on which they are based, separated by the < symbol. For example, CCR<LSCm is the angle between the common crus and the lateral canal in the sagittal plane [Figure 1(d): Table 2]. Two aspects of each of the three semicircular canals are quantified using angles, the degree of torsion, and the planar orientation. The arc of a semicircular canal is rarely entirely planar; rather it is nearly always somewhat twisted, showing a degree of torsion. Here the torsion of the anterior and posterior canals is quantified as the difference between the orientations of their superior-most and inferior-most parts [Figure 1(c): ASCs<ASCi; PSCs<PSCi], and that of the lateral canal as the difference between orientations of the lateral-most part and at the greatest arc width [Figure 1(d): LSCl<LSCm]. The planar orientation of each canal is approximated by measuring the arc at its widest part, for the vertical semicircular canals in the transverse plane [Figure 1(c): ASCm, PSCm], and for the lateral canal in the sagittal plane [Figure 1(d): LSCm]. These three canal orientations constitute the most stable, i.e., least variable, feature of the labyrinth among extant primate species, most likely because they reflect the functionally important physiological plane of optimum perceptive sensitivity (Spoor and Zonneveld, 1998). They are, therefore, used as reference orientations in the comparison of more diverse aspects of labyrinthine shape. The measurements of the anterior and posterior canals in the transverse plane are combined into a single reference orientation by taking the line that bisects the angle between the two canal orientations [Figure 1(c): VSC, the bisector of the angle ASCm<PSCm opening anteriorly and posteriorly]. Other aspects of the labyrinth of which the orientation is measured are the axis of symmetry of the lateral canal, the common crus, the ampullar line, and the cochlea [Figure 1, Table 2: LSCt, CCR, APA, COt, COs and VC, respectively]. The labyrinth is well preserved in all fossils included in this study. The only postmortem damage concerns the inferoposterior part of the posterior semicircular canal of Tabun Cl. The planar orientation (PSCm) and height (PSCh) of this canal can nevertheless be estimated on the basis of the preserved bone. However, the orientation of the inferior limb of this canal (PSCi) cannot be measured and the torsion (PSCtor) can therefore not be calculated. Angles involving structures other than the labyrinth itself were considered for adult specimens only. The planar orientations of the anterior and posterior semicircular canals to the cranial midsagittal plane (ASCm<SG; PSCm<SG) are obviously not available for isolated temporal bones, and neither could they be measured for Spy 1, Spy 2 and Cro-Magnon 1 because the required

7 F. Spoor et al. / Journal of Human Evolution 44 (2003) Fig. 2. Lateral (a) (c) and superior (d) (f) aspects of the right bony labyrinths of a Holocene human (a), (d), and the Neanderthal specimens Gibraltar 1 (b), (e) and Petit Puymoyen 5 (c), (f), reconstructed from sagittal CT scans. The lateral views are aligned according to the plane of the lateral semicircular canal. S. superior, A, anterior, and L, lateral. Scale bar is 5 mm. transverse overview images were not available. Spoor and Zonneveld (1998) assess the planar orientation of the lateral semicircular canal relative to aspects of the midline cranial base. However, these basicranial parts are not preserved in the examined Neanderthal specimens other than Gibraltar 1, and such angles are thus not considered here. The lateral canal is, however, compared relative to the orientations of the posterior surface of the petrous pyramid (PPp) and the third part of the facial canal (FC3). Both are defined in Spoor and Zonneveld (1995), and schematically shown in Figure 7. Differences between the means of the five hominin groups were compared using ANOVA and t-tests. Bonferroni adjustments for multiplicity were made using the sequential procedure described in Rice (1989). Both the protected (i.e., Bonferroni adjusted) and unprotected probabilities are given. In reporting the quantitative analyses the emphasis is on the comparisons between Neanderthals and Holocene humans. For the preliminary comparisons involving the other three hominin groups only significant differences of means will be highlighted, as nonsignificance can easily reflect the small sample sizes rather than similarity. Descriptions and comparisons CT-based three-dimensional reconstructions of the bony labyrinths of two Neanderthal specimens and a representative Holocene human are shown in Figure 2. Furthermore, CT slices through the arcs of the three semicircular canals of a Holocene human and a Neanderthal are shown in Figure 3. The absolute and relative radii of curvature of the semicircular canals are given in Table 3 (R and %R). The statistical significance of the difference between sample means is indicated. Neanderthals

8 148 F. Spoor et al. / Journal of Human Evolution 44 (2003) Fig. 3. CT slices through the arcs of the posterior (a), (d), anterior (b), (e) and lateral (c), (f) semicircular canals of a Holocene human (a) (c) and the Neanderthal specimen Petit Puymoyen 5 (d) (f). Scale bar is 5 mm. Notice that the more irregular appearance of the Petit Puymoyen canals is the consequence of the presence of matrix in parts of their lumen. have absolutely and relatively smaller vertical (anterior and posterior) canal arcs, and a larger lateral canal arc than Holocene humans. In these features they are closest to the Middle Pleistocene hominins. The Upper Palaeolithic and early modern humans have larger anterior, and the former relatively smaller lateral canals than Neanderthals. Compared with Holocene humans both have an absolutely and relatively larger lateral canal, and a relatively smaller posterior canal. Figure 4 plots the mean canal arc radii against estimated body mass for the hominin groups and a sample of extant primates. The plots of the anterior and posterior canals show the grouping of Neanderthals and Middle Pleistocene hominins on the one hand, and Holocene, Upper Palaeolithic and early modern humans on the other [Figure 4(a),(b)]. The former are larger-bodied, but have smaller canals than the latter. Thus, correcting for body mass will increase the differences between the two groupings observed for absolute anterior and posterior canal radii. The Neanderthals and Middle Pleistocene hominins fall close to the assumed great ape regression, whereas the Holocene, Upper Palaeolithic and early modern humans fall well above it, and thus have larger vertical canals for their body mass. In the plot for the lateral canal the hominins form a cluster just below and mostly parallel with the assumed great ape regression. This pattern suggests that differences in lateral canal radius between the hominin groups largely follow from differences in body mass, but the exact extent cannot be established owing to the uncertain slope of the great ape regression. Particularly the size difference between the lateral canals of early modern and Holocene humans could be more than the effect of size alone. The indices expressing the arc shape of the three semicircular canals are given in Table 4 (h/w). The shape of the anterior canal of both Neanderthals and the Middle Pleistocene hominins is relatively narrow in width compared with Holocene and early modern humans. The relatively greater width of the anterior canal arc in Holocene humans corresponds with a morphology where the ampulla bulges out laterally, whereas it points more superiorly in Neanderthals [Figure 3(a),(d)]. The Upper Palaeolithic sample mean for the shape index of the anterior canal is close to that of the Holocene humans, but statistically not significantly different from that of the Neanderthals. The average arc shape of the posterior canal of Neanderthals is close to circular (i.e., shape index = 100), whereas it is taller than it is wide in Holocene and Upper Palaeolithic humans [Figure 3(b),(e)]. Given the orientations of the heights and widths of the anterior and posterior canals (Figure 1) these results imply that in Neanderthals the arc shape of both anterior and posterior canals is foreshortened anteroposteriorly when compared with Holocene humans. The arc shapes of the lateral canal are not significantly different (Table 4). The anterior semicircular canal of Neanderthals and the Middle Pleistocene hominins shows more torsion, and the Upper Palaeolithic humans less torsion than Holocene humans (Table 4: ASCtor). The torsion of the posterior canal is similar in Neanderthals and Holocene humans, but the Upper Palaeolithic sample shows more torsion than Holocene humans (Table 4: PSCtor). The degrees of torsion of the lateral canal are not significantly different (Table 4: LSCtor).

9 F. Spoor et al. / Journal of Human Evolution 44 (2003) Table 3 The radii of curvature to the centre of the lumen (R) of the semicircular canals (ASC, PSC, LSC) given in millimeters, and the relative radii of the semicircular canals in percent (%R: sum of the three radii is 100%) ASC-R PSC-R LSC-R ASC %R PSC %R LSC %R Dederiyeh Gibraltar Gibraltar La Chapelle aux Saints La Ferrassie La Ferrassie La Ferrassie La Quina La Quina H Le Moustier Pech de l Azé Petit Puymoyen Spy Spy Tabun C Abri Pataud Abri Pataud Cro Magnon Laugerie Basse Qafzeh Skhul Abri Suard Reilingen Steinheim Holocene (54) Mean Range 2.6/ / /2.8 34/41 32/40 23/31 S.D Upper Palaeolithic (4) Mean S.D Early modern (2) Mean S.D Neanderthals (15) Mean S.D Middle Pleistocene (3) Mean S.D ANOVA x xx xxx xxx xxx xxx t-test: Neanderthal Holocene xx xx xxx xxx xxx xxx Neanderthal Upper Palaeolithic x ns ns ns ns xx Neanderthal early modern x ns ns ns ns ns Neanderthal Middle Pleistocene ns x ns ns ns ns Upper Palaeolithic Holocene ns ns x ns x x Early modern Holocene ns ns xx ns x x Middle Pleistocene Holocene ns xxx ns ns xx xxx The statistical significance of the difference between the means is indicated by x: P<0.05, xx: P<0.01 and xxx: P< Bonferroni adjusted ( protected ) probabilities are underlined.

10 150 F. Spoor et al. / Journal of Human Evolution 44 (2003) Fig. 4. Bivariate double logarithmic plots between estimated body mass and the radii of curvature to the centre of the lumen of (a) the anterior semicircular canal (ASC-R), (b). the posterior canal (PSC-R) and (c) the lateral canal (LSC-R). N. Neanderthals, H. Holocene modern humans, U. European Upper Palaeolithic humans, M, European Middle Pleistocene hominins, o. great apes, +. other extant primate species. Data of extant species after Spoor and Zonneveld (1998). The body mass of fossil hominins follows Ruff et al. (1997), and that of other species Smith and Jungers (1997). The reduced major axis regressions for the great ape species are indicated, but the correlations are not statistically significant (r rank = 0.800, P > 0.05, for all three canals). The planar orientations of the anterior and posterior semicircular canals in the cranium, i.e., relative to the midsagittal plane, are not different in Neanderthals and the other hominins for which these angles could be measured (Table 5: ASCm<SG, PSCm<SG). This could be the result of the small sample sizes of the fossil groups for which these measurements are available. However, similarity is supported by the observation that the angle between the planar orientations of the two canals, which could be measured for all specimens considered, shows no significant differences either (Table 6: ASCm<PSCm). The angle between the planar orientation of the lateral canal and the posterior petrosal surface is larger in Neanderthals than in Holocene humans (Table 5: LSCm<PPp). Likewise, the angle between this canal s orientation and the third portion of the facial canal is larger in Neanderthals than in Holocene and Upper Palaeolithic humans (Table 5: LSCm< FC3). This means that in Neanderthals the posterior petrosal surface and the facial canal portion are oriented more upright relative to the plane of the lateral canal. Neither the angle between the axis of symmetry of the lateral canal arc and the vertical canal orientations, nor that between the common crus and the lateral canal orientation are significantly different between the groups (Table 6: LSCt<VSC, CCR<LSCm, respectively). On the other hand, the ampullar line is more vertically inclined relative to the lateral canal orientation in Neanderthals than in any of the other groups, whereas it is less vertically inclined in early modern humans than in Holocene humans (Table 6: APA<LSCm). Moreover, the arc of the posterior canal is positioned more inferiorly, relative to the lateral canal in Neanderthals than in all other groups, whereas it is positioned more superiorly in Upper Palaeolithic humans than in Holocene humans (Table 6: SLI). As a consequence of this morphology the common crus of Neanderthals tends to be unusually short compared with Holocene humans, and any other extant primate species investigated thus far [Figure 2(a) (c)]. Figure 5(a) demonstrates the interspecific correlation among hominid species between the position (SLI) and size (R) of the posterior canal; the

11 F. Spoor et al. / Journal of Human Evolution 44 (2003) Table 4 The shape indices (h/w 100) and torsions (tor) of the semicircular canals (ASC, PSC, LSC) ASCh/w PSCh/w LSCh/w ASCtor * PSCtor LSCtor Dederiyeh Gibraltar Gibraltar La Chapelle aux Saints La Ferrassie La Ferrassie La Ferrassie La Quina La Quina H Le Moustier Pech de l Aze Petit Puymoyen Spy Spy Tabun C Abri Pataud Abri Pataud Cro Magnon Laugerie Basse Qafzeh Skhul Abri Suard Reilingen Steinheim Holocene (54) Mean Range 74/97 94/128 67/100 5/28 22/0 7/14 S.D Upper Palaeolithic (4) Mean S.D Early modern (2) Mean S.D Neanderthals (15 ) Mean S.D Middle Pleistocene (3) Mean S.D ANOVA xxx xx ns xx x ns t-test: Neanderthal Holocene xxx xxx ns xxx ns ns Neanderthal Upper Palaeolithic ns x ns xxx ns ns Neanderthal early modern x ns ns ns ns ns Neanderthal Middle Pleistocene ns ns ns ns ns ns Upper Palaeolithic Holocene ns ns ns xxx xx ns Early modern Holocene ns ns ns ns ns ns Middle Pleistocene Holocene x ns ns xxx ns ns Statistical significance as indicated in the caption of Table 3. * ASCs<ASCi; positive when ASCs is more coronally oriented than ASCi. PSCs<PSCi; positive when PSCs is more sagitally oriented than PSCi. LSCl<LSCm; positive when LSCl is more inclined than LSCm. PSCtor based on 14 specimens.

12 152 F. Spoor et al. / Journal of Human Evolution 44 (2003) Table 5 Angles in degrees describing the planar orientation of the semicircular canals in the cranium ASCm<SG PSCm<SG LSCm<PPp LSCm<FC3 Gibraltar La Chapelle aux Saints La Ferrassie La Ferrassie La Quina La Quina H Petit Puymoyen Spy Spy Tabun C1 62 Abri Pataud Cro Magnon Laugerie Basse Qafzeh Skhul Abri Suard Reilingen Steinheim Holocene (50,50,53,52) Mean Range 26/47 129/150 42/80 58/90 S.D Upper Palaeolithic (3) Mean S.D Early modern (2) Mean S.D Neanderthals (4,4,10,9) Mean S.D Middle Pleistocene (3) Mean S.D ANOVA ns ns x xxx t-test: Neanderthal Holocene ns ns xx xxx Neanderthal Upper Palaeolithic ns x Neanderthal early modern ns ns ns ns Neanderthal Middle Pleistocene ns ns ns ns Upper Palaeolithic Holocene ns ns Early modern Holocene ns ns ns ns Middle Pleistocene Holocene ns ns ns ns Explanation of the measurement codes and symbols in Figure 1 and Table 2. The first two angles open anteriorly, the other two antero-superiorly. Statistical significance as indicated in Table 3.

13 F. Spoor et al. / Journal of Human Evolution 44 (2003) Table 6 Angles in degrees and index (SLI) in per cent of the semicircular canals ASCm<PSCm LSCt<VSC CCR<LSCm APA<LSCm SLI Dederiyeh Gibraltar Gibraltar La Chapelle aux Saints La Ferrassie La Ferrassie La Ferrassie La Quina La Quina H Le Moustier Pech de l Azé Petit Puymoyen Spy Spy Tabun C Abri Pataud Abri Pataud Cro Magnon Laugerie Basse Qafzeh Skhul Abri Suard Reilingen Steinheim Holocene (54) Mean Range 90/ / /133 32/57 34/69 S.D Upper Palaeolithic (4) Mean S.D Early modern (2) Mean S.D Neanderthals (15) Mean S.D Middle Pleistocene (3) Mean S.D ANOVA ns ns ns xxx xxx t-test: Neanderthal Holocene ns ns ns xxx xxx Neanderthal Upper Palaeolithic ns ns ns xxx xxx Neanderthal early modern ns ns ns xxx xxx Neanderthal Middle Pleistocene ns ns ns xx xx Upper Palaeolithic Holocene ns ns ns ns x Early modern Holocene ns ns ns x ns Middle Pleistocene Holocene ns x ns ns ns Explanation of the measurement codes and symbols in Figure 1 and Table 2. The first two angles open laterally; the other two antero-superiorly. Statistical significance as indicated in Table 3.

14 154 F. Spoor et al. / Journal of Human Evolution 44 (2003) Fig. 5. The relationship between the radius of curvature of the posterior semicircular canal (PSC-R, in millimeters) and the Sagittal Labyrinthine Index (SLI, in percentages). (a) Mean values of N, Neanderthals; H, Holocene modern humans; U, Upper Palaeolithic modern humans; E, early modern humans; M, Middle Pleistocene hominins; Δ, H. erectus (OH 9, Sangiran 2, Sangiran 4); Australopithecus africanus (Taung, Sts 5, Sts 19, MLD 37/38); Paranthropus robustus (SK 46, SK 47, SK 879); Dryopithecus brancoi (RUD 77); Pan troglodytes (n = 7); Pan paniscus (n = 6); Gorilla gorilla (n = 6) and Pongo pygmaeus (n = 7). The RMA regression (y = ) is given for the sample excluding Neanderthals, Upper Palaeolithic and early modern humans, and Middle Pleistocene hominins (r rank = 0.917, P<0.001). (b) Mean values N and H as in (a) and specimen values of o, Neanderthals; +, Holocene modern humans; x, Upper Palaeolithic modern humans; #, early modern humans and, Middle Pleistocene hominins. larger the canal the more inferior its position. The posterior canal in Neanderthals has a more inferior position (higher SLI) for its size than predicted by the regression (Table 7). Likewise, the canal of Middle Pleistocene hominins has a somewhat more inferior position than predicted. In contrast, that of the Upper Palaeolithic and early modern humans is positioned slightly more superiorly (lower SLI) than predicted. However, the deviations from the general hominid regression shown by the latter three groups are small compared to the degree of inter-individual variation among Holocene humans and Neanderthals [Figure 5(b)]. Among individuals of each of these two groups posterior canal size and position are not correlated (P > 0.05; r rank and 0.470, respectively). The bivariate plot of Figure 5(b) shows a good separation of Neanderthals and Holocene humans, but there is a degree of overlap, with Le Moustier 1 and Spy 1 falling well within Table 7 Values of the sagittal labyrinthine index (SLI) and the cochlear basal turn size (CO-R) as predicted by posterior canal size (PSC-R) and body mass, respectively, using the RMA regressions shown in Figures 5 and 7 X i Y pred SE L1 L2 Y obs SLI (%) predicted from PSC-R (mm) Holocene Upper Palaeolithic Early Modern Neanderthals Middle Pleistocene CO-R (mm) predicted from body mass (g) Holocene Upper Palaeolithic Early Modern Neanderthals Middle Pleistocene The standard error (SE) and 95% confidence interval (L1, L2) of the predicted value (Y pred ) are given. Observed values (Y obs ) outside the confidence interval are given in bold.

15 F. Spoor et al. / Journal of Human Evolution 44 (2003) the Holocene human range. The two Holocene human specimens that fall within the core Neanderthal range (i.e., excluding Le Moustier 1 and Spy 1) originate from Central Asia (Kalmuck) and Mozambique. The size of the cochlear basal turn of Neanderthals is not different from that in Holocene humans, whereas that of early modern humans, and to a lesser extent that of Upper Palaeolithic humans is larger (Table 8: CO-R). When taking body mass into account, by calculating the residuals from the extant primate regression, Holocene and Upper Palaeolithic humans are not significantly different in cochlea size, whereas the latter do have a larger cochlea than Neanderthals (Figure 6; Table 8). Only the cochlea of the early modern humans is larger than predicted by body mass on the basis of the extant primate regression (Table 7). No significant differences are observed for the shape index and orientation in the transverse plane of the cochlear basal turn (Table 8: COh/w, COt<VSC, respectively). The position and orientation of the cochlea, relative to the plane of the lateral canal, is not different in Neanderthals and Holocene humans (Table 8: VC<LSCm, COs< LSCm, respectively). On the other hand, both early modern and Upper Palaeolithic humans have a more superiorly positioned cochlea than Holocene humans (Table 8: VC<LSCm), and the apex of the cochlea faces more inferiorly in Middle Pleistocene hominins than in Neanderthals and Holocene humans (Table 8: COs<LSCm). In summary, compared with Holocene humans the bony labyrinth of Neanderthals can be characterized as follows. The anterior semicircular canal arc is smaller in absolute and relative size, is narrow in width compared to its height, and shows more torsion. The posterior semicircular canal arc is smaller in absolute and relative size, is less high relative to its width, and is positioned more inferiorly relative to the lateral canal plane. The lateral semicircular canal arc is absolutely and relatively larger. The ampullar line is more vertically inclined. The European Upper Palaeolithic and early modern humans are most similar, although not fully identical to Holocene humans in labyrinthine morphology. The European Middle Pleistocene hominins show the typical semicircular canal morphology of Neanderthals, with the exception of the arc shape and inferiorly position of the posterior canal and the strongly inclined ampullar line. Whereas sample means of individual traits differ significantly between modern humans and Neanderthals, their ranges largely overlap. In fact, among the discriminating traits all Neanderthal specimens fall within the Holocene human range for ASC-R, PSC-R, ASC-%R, LSC<PPp and APA<LSCm. La Chapelle aux Saints, and La Quina H27 are only outside the Holocene range for the angle describing the facial canal orientation (FC3<LSCm), which is not a character of the labyrinth itself. Spy 1 is only outside the range for the shape index of the lateral canal (LSCh/w), a character that does not discriminate between Neanderthals and Holocene humans. Of all Neanderthals examined the labyrinth of Le Moustier 1 is the closest in morphology to that of Holocene humans, falling within the latter s range for all traits. Among Neanderthals it has the relatively smallest lateral canal (LSC-%R), the lowest anterior canal arc (ASCh/w), the least vertically inclined ampullar line (APA<LSCm), and almost the lowest sagittal labyrinthine index (SLI). In contrast, the La Ferrassie 1 labyrinth shows three traits outside the Holocene range, that are all related to its posterior canal (PSC-%R, PSCh/w, PSCtor). The labyrinths of La Quina 5, Pech de 1 Azé and Petit Puymoyen each show two traits outside Holocene range. Discussion The results of the present study confirm the initial findings of Hublin et al. (1996) that the bony labyrinth of Neanderthals is distinct in morphology from that of Holocene and Late Pleistocene modern humans. The increase of the Neanderthal sample, from nine in Hublin et al. (1996) to 15 here, has not led to changes in the mean values of the canal radii, other than an increase from 2.5 to

16 156 F. Spoor et al. / Journal of Human Evolution 44 (2003) Table 8 Linear dimensions and angles of the cochlea CO-R residual COh/w COt<VSC VC<LSCm COs<LSCm Dederiyeh Gibraltar Gibraltar La Chapelle aux Saints La Ferrassie La Ferrassie La Ferrassie La Quina La Quina H Le Moustier Pech de l Azé Petit Puymoyen Spy Spy Tabun C Abri Pataud Abri Pataud Cro Magnon Laugerie Basse Qafzeh Skhul Abri Suard Reilingen Steinheim Holocene (54) Mean Range 2.0/ / / /170 46/69 S.D Upper Palaeolithic (4) Mean S.D Early modern (2) Mean S.D Neanderthals (15) Mean S.D Middle Pleistocene (3) Mean S.D ANOVA xx ns ns xx x t-test: Neanderthal Holocene ns ns ns ns ns ns Neanderthal Upper Palaeolithic ns x ns ns x ns Neanderthal early modern xx xx ns ns ns ns Neanderthal Middle Pleistocene ns ns ns ns ns x Upper Palaeolithic Holocene x ns ns ns xx ns Early modern Holocene xxx xx ns ns x ns Middle Pleistocene Holocene ns ns ns ns ns xxx Explanation of the measurement codes and symbols in Figure 1 and Table 2. The first angle opens antero-laterally; the other two antero-superiorly. Statistical significance as indicated in Table 3.

17 F. Spoor et al. / Journal of Human Evolution 44 (2003) Fig. 6. Bivariate double logarithmic plots between estimated body mass and the radii of curvature to the basal turn of the cochlea (CO-R). Symbols as in Figure 4. Sample of 25 extant primate species as in Spoor and Zonneveld, 1998, plus Galago moholi. Solid line is the reduced major axis regression for the extant primates (r rank 0.878, P<0.001), RMA slope 0.139, intercept ). 2.6 mm for the lateral canal. Furthermore, the mean value of the sagittal labyrinthine index (SLI) has changed from 68 to 65. Importantly, the present, more comprehensive comparative analysis has identified a number of additional traits that distinguish the labyrinths of the two hominin groups. The labyrinth of only one Neanderthal, Le Moustier 1, has been examined quantitatively in studies other than Hublin et al. (1996). Thompson and Illerhaus (1998) report anterior, posterior and lateral canal radii of 3.2, 3.3 and 2.6 mm, and an SLI of 64 for its right labyrinth. These values differ from those obtained for the right labyrinth in this study (3.4, 3.4, 2.7 mm and 55, respectively), even though they are based on the same CT images. The discrepancies appear to follow from the different ways in which the measurements were taken. Thompson and Illerhaus (1998) used 3D reconstructions derived from the CT images, similar to those shown in Figure 2, rather than the crosssectional images themselves [J. Thompson personal communication; assistance was given by one of us (FS)]. Most landmarks involved are located in the centre of the lumen of the canals (Figure 1; Spoor and Zonneveld, 1995), and these have to be estimated from the surface contours when taking measurements from 3D images. In practice, it is particularly difficult to estimate the planar orientation of the lateral canal, as defined in Spoor and Zonneveld (1995: LSCm), whereas this is essential to obtain compatible SLI values. Ponce de León and Zollikofer (1999) also measured 3D reconstructions of the labyrinths of Le Moustier 1, but based on CT images different from the ones used here. The values of 3.4 mm for the posterior canal radius and 54 for the SLI, plotted in their Figure 7B, are very close to those obtained here (3.4 mm and 55). Unlike Thompson and Illerhaus (1998), however, Ponce de León and Zollikofer displayed the 3D reconstructions on a stereo screen (C. Zollikofer personal communication), and the added perception of depth undoubtedly improved their ability to accurately locate the internal landmarks. The identification of an expanded suite of labyrinthine traits characterizing Neanderthals underlines the potential of using this structure to assess the phylogenetic affinities of fragmentary Late Pleistocene hominin fossils. That the labyrinth

18 158 F. Spoor et al. / Journal of Human Evolution 44 (2003) may also reflect the genotypic make-up of an individual to a greater degree than do most other skeletal parts because postnatal influences on the morphology by environmental or behavioral factors are minimal or absent. On the other hand, the present study highlights that each of the identified traits shows considerable overlap between modern humans and Neanderthals. Consequently, statistically conclusive attributions will almost always require multivariate analyses, as exemplified by studies of the Arcy sur Cure and Dederiyeh labyrinths (Hublin et al., 1996; Spoor et al., 2003). Moreover, conclusive attribution will be impossible when dealing with labyrinths similar to that of Le Moustier 1, which entirely fall in the morphological overlap zone of Neanderthals and modern humans. Fig. 7. Lateral view of the left bony labyrinths of (a) Pan paniscus, (b) a Holocene modern human and (c) the La Ferassie 1 Neanderthal. The labyrinths are aligned according to the planes of their lateral semicircular canal (LSCm; dashed line), and the course of the second and third parts of the facial nerve canal (thick line) and the endocranial petrosal contour at the level of the common crus are indicated. Single headed arrows indicate morphological differences comparing (b) to (a), and (c) to (b). The ampullar line (APA, thick dotted line), the third part of the facial nerve canal (FC3) and the posterior petrosal surface (PPp) are increasingly vertically inclined from (a) to (c), and the inferior component of the sagittal labyrinthine index increases (SLIi, double headed arrows). Neanderthals and Holocene humans are similar in having a common crus that is tilted posteriorly (CCR), and a cochlea that is positioned more superiorly (COs and VC) than in nonhominin primates. reaches adult morphology well before birth not only has the practical advantage that adult and immature specimens can be compared directly. Given its unusual ontogeny the bony labyrinth Shape of the labyrinth The most striking aspect of the Neanderthal labyrinth, as yet not found in any other hominoid species (Spoor, 1993; Spoor and Zonneveld, 1998), is the particularly inferior position of its posterior semicircular canal. Individual traits that are associated with this morphology are not only the high SLI value, but also the more vertically inclined ampullar line (APA<LSCm), which follows from the inferiorly positioned posterior ampulla, and the unusually shortened common crus (Figure 2). Moreover, a particularly inferior position of the ampullar limb of the posterior canal is spatially consistent with a relatively large arc width, and thus the lower shape index seen in Neanderthals (PSCh/w). Among extant primate species the SLI and the ampullar line angle are correlated, and both are also correlated with the orientation of the posterior petrosal surface and the third part of the facial canal (Spoor, 1993; Spoor and Zonneveld, 1998). These correlations imply that a more inferiorly positioned posterior canal arc corresponds with a more vertically inclined ampullar line, posterior petrosal surface and facial canal. When viewing the lateral aspect of the left temporal bone and labyrinth this morphology can be seen as a joint clockwise rotation, relative to both the plane of the lateral semicircular canal (LSCm) and the midline anterior cranial base (Spoor and Zonneveld,