Journal of Anthropological Sciences Research articles Vol. 99 (2021), pp. 117 - 134 e-pub ahead of print doi.10.4436/jass.99017 New insights on hip bone sexual dimorphism in adolescents and adults using deformation-based geometric morphometrics Cinzia Fornai1,2,3, Nicole M. Webb1,4, Alessandro Urciuoli5, Viktoria A. Krenn1,2, Louise K. Corron6 & Martin Haeusler1 1) Institute of Evolutionary Medicine, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland e-mail: cinzia.fornai@univie.ac.at 2) Department of Evolutionary Anthropology, University of Vienna, Djerassiplatz 1, 1030 Vienna, Austria 3) Vienna School of Interdisciplinary Dentistry, Wasserzeile 35, 34000 Klosterneuburg, Austria 4) Department of Palaeoanthropology, Institute for Archaeological Sciences, Senckenberg Centre for Human Evolution and Palaeoenvironment at Eberhard Karls University of Tübingen, Ruemelinstrasse 23, 72070, Tuebingen, Germany e-mail: nicole.webb@ifu.uni-tuebingen.de 5) Institut Català de Paleontologia Miquel Crusafont, Universitat Autònoma de Barcelona, Edifici ICTA-ICP, Campus de la UAB. c/ de les columnes s/n, 08193 Cerdanyola del Vallès, Barcelona, Spain 6) Department of Anthropology, University of Nevada, Reno, 1664 N. Virginia Street, Reno, NV 89557, USA Summary - Morphological variation of the human pelvis, and particularly the hip bone, mainly results from both female-specific selective pressure related to the give birth of large-headed newborns, and constraints in both sexes for efficient bipedal locomotion, abdominal stability, and adaptation to climate. Hip bone morphology has thus been extensively investigated using several approaches, although the nuances of inter-individual and sex-related variation are still underappreciated, and the effect of sex on ontogenetic patterns is debated. Here, we employ a landmark-free, deformation-based morphometric approach to explore variation in modern human hip bone shape and size from middle adolescence to adulthood. Virtual surface models of the hip bone were obtained from 147 modern human individuals (70 females and 77 males) including adolescents, and young and mature adults. The 3D meshes were registered by rotation, translation, and uniform scaling prior to analysis in Deformetrica. The orientation and amplitude of deviations of individual specimens relative to a global mean were assessed using Principal Component Analysis, while colour maps and vectors were employed for visualisation purposes. Deformation-based morphometrics is a time-efficient and objective method free of observer-dependent biases that allows accurate shape characterisation of general and more subtle morphological variation. Here, we captured nuanced hip bone morphology revealing ontogenetic trends and sex-based variation in arcuate line curvature, greater sciatic notch shape, pubic body and rami length, acetabular expansion, and height-to-width proportions of the ilium. The observed ontogenetic trends showed a higher degree of bone modelling of the lesser pelvis of adolescent females, while male variation was mainly confined to the greater pelvis. Keywords - Pelvic morphology, Sex-related variation, Ontogeny, Deformetrica, Diffeomorphism. Introduction morphology and size (Waldeyer 1899; Krogman 1962; Ferembach et al. 1980; Brůžek 2002; The human pelvis, and especially the hip İşcan and Steyn 2013; Klales 2020). Females bone, displays clear sexual dimorphism in both have evolved a larger pelvic canal to allow for Istituto Italiano di Antropologia, founded in 1893 by Giuseppe Sergi Hip bone sexual dimorphism the passage of large-headed fetuses, while cer- Pelvic sexual dimorphism is detectable as tain dimensions like bi-acetabular width and early as the fetal period and throughout child- anteroposterior depth are constrained by the hood (Weaver 1980; Schutkowski 1993; Wilson biomechanical requirements of upright bipedal et al. 2016), but becomes more pronounced locomotion (Abitbol 1987; Rosenberg 1992; at the onset of puberty, when male and female Warrener et al. 2015; but see Washburn 1960) hip bones can be readily distinguished based on and possibly pelvic floor stability (Grunstra et their general morphology (Brůžek 2002; Corron al. 2019; Stansfield et al. 2021). These oppos- et al. 2021). Preliminary evidence suggests that ing selective pressures are commonly referred female pelvic morphology diverges from that of to as the “obstetrical dilemma” by Washburn males before adolescence (Huseynov et al. 2016), (1960, see also Haeusler et al. 2021). This evo- while further shape changes occur in both sexes lutionary trade-off is often used to explain the before the adult shape is attained. In particular, high degree of sexual dimorphism observed in Coleman (1969) noted that in females, both the the human pelvis, with females typically hav- ischial spine and the ilium undergo a medial ing larger overall pelvic canal dimensions and a to lateral reorientation and the superior pubic markedly wider subpubic angle when compared ramus elongates. to males. Furthermore, pelvic shape has been Despite these findings, pelvic ontogeny and shown to vary in relation to body size (Fischer the mechanisms guiding pelvic sexual dimor- and Mitteroecker 2015, 2017), climate and sto- phism are still debated as the irregular morphol- chastic evolutionary processes (see for example, ogy of the hip bone is also influenced by factors Kurki 2007; Betti et al. 2013; Betti and Manica other than sex, including age, hormonal regula- 2018; and further discussion in Mitteroecker et tion, and body size, as well as ecological, nutri- al. 2021). tional, and climatic factors (e.g., Betti et al. 2013; These sex-related differences are also reflected Huseynov et al. 2016; Warrener et al. 2015; in the shape and size of the isolated hip bone, Wells 2017; Grunstra et al. 2019; Dunsworth with females showing a wider greater sciatic 2020; Mitteroecker et al. 2021). Accordingly, notch, a composite arch, an outward rotation the high degree of morphological variation of of the ischium, and a medio-laterally elongated the hip bone is expected to manifest in increas- pubis (Genovés 1959). Pubic morphology is ing morphological overlap between the sexes. particularly dimorphic, with females usually pos- Moreover, it has been shown that morphologi- sessing a ventral arc on the anterior aspect of the cal variation between the sexes is typical of many pubic body, a subpubic concavity, and a narrower mammals. Thus, great apes display a similar pat- medial aspect of the ischiopubic ramus accompa- tern of sexual dimorphism in the pelvis, though nied by a discrete ridge (Phenice 1969; Klales et at a different magnitude compared to humans al. 2012;). Additionally, females are character- (Huseynov et al. 2016; Fischer et al. 2021; Webb ised by more triangular obturator foramina, rela- et al. 2021). Even species not subject to direct tively lower and more laterally divergent ilia, and obstetric selective pressure, such as the marsupial smaller and more anterolaterally directed acetab- Virginia opossum (Didelphis virginiana) show ula compared to males (Genovés 1959; İşcan and sexual dimorphism in certain pelvic features Steyn 2013. Additional sex characteristics of the (Tague 2003). hip bone include the relative size of the auricular The range of hip bone morphological vari- surface and its elevation with respect to the rest of ation has been extensively investigated using the iliac surface (Bass 1995) and the shape of the conventional approaches based on qualitative preauricular sulcus, although the mere presence trait assessment and linear metrics (Buikstra of a groove in this region has been considered and Ubelaker 1994; Rogers and Saunders less reliable as an indicator of sex, particularly if 1994; Brůžek 2002), as well as landmark-based the groove is faint (Brůžek 2002; Karsten 2018). geometric morphometrics (Steyn et al. 2004; 118 Hip bone sexual dimorphism Bytheway and Ross 2010; Charles 2010; Fischer In this study, we aim to explore the mor- and Mitteroecker 2017; Rmoutilová et al. 2017; phological variation of the hip bone by applying Robertson et al. 2019; Cox 2021). Geometric deformation-based geometric morphometrics morphometrics has become a preferred method to 3D surface models from a sample of modern in anthropology and other biological sciences humans of known sex including both adolescents for its ability to capture and retain the spatial and adults. Using this novel approach, we expect relationship between landmark points, thereby to obtain detailed information regarding the providing more information about shape than role of sexual dimorphism in shaping hip bone previously unveiled by linear metrics (Bookstein morphology, and how it ultimately contributes 1991; Slice 2005; Weber and Bookstein 2011; to pelvic variation. Moreover, we explore ontoge- Bookstein 2018). However, these methods rest netic shape changes occurring in the male and on the assumption that the points used in the female hip bones between middle adolescence analyses are homologous, which means they are and adulthood with reference to previous quali- anatomically or ontogenetically correspondent. tative observations (Coleman 1969; Corron et Accordingly, semilandmarks, which are geo- al. 2021) and geometric morphometric findings metrically homologous points, are often used (Bilfeld et al. 2013; Huseynov et al. 2016). for interlandmark shape analysis (Gunz et al. 2005; Gunz and Mitteroecker 2013). Although the concept of homology is at the very foun- Materials and Methods dation of anthropometry, there are inherent limitations to its application, especially when Study sample and specimen preparation comparing specimens from different genera The study sample was composed of 147 male (Zelditch et al. 1995). Rigorous landmark col- (n = 76) and female (n = 71) modern human hip lection consequently requires highly experi- bones, 46 of which were of adolescents between enced operators, and even adequate training or the age of 15 and 18 years (25 males and 21 automated methods for landmark placement do females) (Tab. 1). Chronological age was veri- not reduce the time-consuming nature of this fied via associated medical records for 54 living process or its susceptibility to error (Fischer et individuals scanned at the Assistance Publique al. 2019). Moreover, an extremely high number Hôpitaux de Marseille and 57 specimens belong- of points (including landmarks and semiland- ing to the Weisbach osteological collection of the marks) are necessary to represent an object in National History Museum Vienna. The rest of its entirety, particularly for a complex structure the sample (n = 36) was comprised of young and like the hip bone (e.g., Torres-Tamayo et al. mature adults from skeletal collections (as speci- 2018). fied in Table1) without degenerative changes, for Developments within geometric morpho- whom exact age was unknown. All incomplete or metrics have led to a landmark-free, deforma- pathological individuals were excluded from this tion-based geometric morphometric approach, study. The right hip bone was arbitrarily selected recently introduced into the paleoanthropologi- for all individuals. The pelvis shows fluctuat- cal literature (Beaudet et al. 2020; Urciuoli et ing rather than directional asymmetry, ensuring al. 2020; 2021). This novel method offers the its developmental stability (Brown et al. 2008; opportunity to conveniently study objects with Tobolsky et al. 2016), and suggesting that later- irregular shape. The hip bone thus represents a ality or asymmetry should not significantly affect suitable test case for landmark-free approaches the expression of sexual dimorphism or influence (Kuchař et al. 2021) since its challenging anat- the outcomes of our analyses. Three-dimensional omy commonly requires the use of less reliable surface models of the hip bones were obtained Type II and Type III landmarks (Bookstein 1991, using an optical 3D-surface scanner (QT Weber and Bookstein 2011). Sculptor PT-M4c, https://www.polymetric.de; 119 Hip bone sexual dimorphism Tab. 1 - Demographic information for the study compromise the accuracy of the virtual models. sample, including geographic origin, reposi- tory, age and sex. a: Anthropological Institute The meshes were subsequently aligned to a ran- and Museum, University of Zurich, Zurich; b: domly chosen reference specimen (see Bookstein Laboratory of Prehistoric Archaeology and 1991) by means of rigid and uniform transfor- Anthropology, University of Geneva, Geneva; mations with scaling, using the ‘align surfaces’ c: Department of Evolutionary Anthropology, University of Vienna, Vienna; d: Department of module of Amira and its ‘rigid + uniform’ option, Anthropology, Natural History Museum Vienna; which equates to a classical Procrustes superim- e: Assistance Publique Hôpitaux de Marseille, position. The pre-alignment was performed to Marseille; f: Musée de l’Homme, Paris ensure that all the specimens were translated and oriented to the same global reference (Durrleman ORIGIN REPOSITORY ADOLESCENTS ADULTS et al. 2012b; Beaudet et al. 2016; 2021; Zanolli M F M F et al. 2018). The scaling factors were recorded for use in subsequent size analyses. The aligned sur- Africa a / / 1 1 faces were then converted into legacy VTK files Europe a; b; c; d; e 22 21 42 37 using the open-source software Paraview v. 5.6.0 (www.paraview.org). South- a; f 3 1 9 10 East Asia and Deformation-based geometric morphometrics analysis Oceania The legacy VTK files were imported into Total 25 22 52 48 Deformetrica v. 4.2 (www.deformetrica.org; Bône et al. 2018) for the deformation-based shape analysis. This landmark-free method computes deformations of each specimen from Haeusler et al. 2004) for the osteological collec- the sample’s mean shape, a theoretical specimen tion subsets, and segmenting the medical com- calculated from all the specimens in the sample, puted tomography scans using the Amira soft- based on the geometric correspondence between ware (www.fei.com) for the clinical subjects. As continuous surfaces (Glaunès and Joshi 2006; several authors have demonstrated, the inclusion Durrleman et al. 2012b; Bône et al. 2018). The of 3D surface models obtained from different deformations are then described by means of dif- imaging techniques does not affect study out- feomorphisms, which are invertible and continu- comes (Waltenberger et al. 2021). Instances of ous mathematical functions used to map differ- minor taphonomic damage of the skeletal hip entiable manifolds, a manifold being a topologi- bones were virtually repaired using the Geomagic cal space that locally approximates a linear space Design X software (https://it.3dsystems.com). (Krantz et al. 2013). Unlike landmark-based geo- As a preprocessing step for deformation- metric morphometrics, deformation-based mor- based geometric morphometric analysis, each phometrics allows for direct assessment of shape surface model was smoothed and cleaned, which differences between surfaces. Moreover, shape eliminated minor imperfections such as holes variation is quantified based on momentum vec- or inverted triangles to restore surface continu- tors exemplifying the magnitude and direction ity. The meshes were also decimated to 50,000 of surface deformations attached to previously (± 100) triangles to reduce computational process- identified control points rather than based on the ing time without compromising surface accuracy distances between paired homologous landmark and enable comparability in terms of number of points (Durrleman et al. 2012a, b). faces. Possible detrimental effects of the decima- Previous work has shown that deformation- tion process to the models (Veneziano et al. 2018) based morphometrics produces comparable were limited by the high quality and smoothness outcomes to landmark-based 3D geometric of the original meshes and, therefore, did not morphometrics (Urciuoli et al. 2018, 2020). As 120 Hip bone sexual dimorphism a first step, Deformetrica computes a template In addition, we performed a Wilcoxon-Mann- surface representing the sample’s mean shape Whitney test with Bonferroni correction on and identifies a set of control points placed at each of the first three PCs, using the ‘pairwise. the most variable regions of that template. The wilcox.test’ function of the stats R package. The template is then warped to match the other scaling factors used for standardising the models included specimens (or targets), and deforma- were normalised (i.e., inverted) prior to visualisa- tions are quantified by attaching momentum tion in boxplots and analysed using the Mann- vectors to the control points describing the mag- Whitney U test. To evaluate ontogenetically nitude and direction of each deformation. The derived variation, and the appearance and expres- number of control points is automatically identi- sion of sexual dimorphism, cumulative displace- fied and optimised by the software based on the ment variations of male and female adult hip Kernel width specified by the user (Bône et al. bones from the adolescent global mean were ren- 2018). The Kernel width is a sensitivity param- dered with colour maps. Additionally, momen- eter that must be optimised to ensure that the tum vectors representing deformation from the deformation from the global mean to the target global mean were used to represent maximum specimens is accurate, namely that the deformed local displacements at the most variable regions global mean surface perfectly matches the model of the hip bone. Accuracies of classification into of each of the target specimens. Test runs on male and female sexes were assessed using linear sample subsets are necessary for the user to iden- Discriminant Function Analysis (DFA) of the tify the correct Kernel width. The shape analysis PC scores, an approach widely adopted in bio- was performed at the Barcelona Supercomputing logical anthropology (e.g., Franklin et al. 2005; Center to reduce computation time using the Krishan et al. 2016; Oikonomopoulou et al. MinoTauro GPU cluster computer (www.bsc.es/ 2017). Principal components were incrementally marenostrum/minotauro). added to the model until classification accuracy Patterns of shape variation were inspected was maximised. Thus, the addition of further using principal component analysis (PCA) on PCs after that would have consistently provided the deformation fields (i.e., the raw shape data lower classification accuracies. DFA models for obtained by combining the momenta and con- male and female adults only, and male and female trol points stemming from the deformation adolescents only were calculated after running analysis; Supplementary Data 1 and 2) using separate PCAs for the adult and the adolescent the ade4 (v.1.7-16) package (Dray and Dufour subsamples. We then contrasted our results with 2007) for the R statistical environment (R those obtained by Robertson et al. (2019), who Development Core Team 2021). To assess allo- also used DFA to evaluate classification accuracy metric trends, we performed major axis regres- of male and female adult hip bones, based on sions of each of the first three principal compo- traditional landmark-based geometric morpho- nent (PC) scores against the natural logarithm metrics. This offered the possibility to directly of the inverted scaling factor (computed during compare the outcomes of their classic geometric the alignment of the surfaces) using the func- morphometric approach using 32 and 17 pelvic tion ‘lmodel2’ of the lmodel2 R package (v.1.7-3) landmarks with our deformation-based geomet- (Legendre 2018). The significant PCs were iden- ric morphometrics approach. tified using the ‘getMeaningfulPCs’ function of the Morpho R package (v.2.9) (Schlager 2013). The differences between the four groups (i.e., Results adult males, adult females, adolescent males, and adolescent females) were statistically assessed Hip bone shape variation using a MANOVA test with Bonferroni correc- The results of the PCA on the momen- tion in PAST (https://palaeo-electronica.org). tum vectors, representing deformations from 121 Hip bone sexual dimorphism the global mean, are presented in Figure 1 and adolescents showing a straighter arcuate line Supplementary Videos 1-4. PC1 (15% of total and a shorter pubis than adults, resulting in a variance explained) showed a general grouping of relatively medio-laterally narrower pelvic inlet. males vs. females based on several morphological Ontogenetic shape changes also involved a rear- features. For instance, variation in the robusticity rangement of the ilium, which was more verti- (thickness) of the hip bone was noticeable, with cal in adults, while the superior pubic ramus was males being more robust than females except for more posteriorly oriented near the symphysis (see the posterior region of the ilium and the iliac crest, PC2 warpings in Figure 1). Adolescent and adult which were relatively more robust in females. males were more similar in shape compared to In females, a wider greater sciatic notch, a more adolescent and adult females. Shape differences everted ischium, longer and more slender pubic between adolescent and adult means were signif- rami, and a wider pubic body collectively resulted icant for both males and females, while none of in a broader pelvic inlet and outlet. The auricu- the differences between male and female adoles- lar surface varied from being depressed in males cents, adults or both were statistically significant to elevated in females (see PC1 warpings in Figure (Tab. 2). A significant but weak allometric trend 1), contributing to increased medio-lateral width was observed along PC2 (adjusted R2: 0.0345; of the pelvic inlet in females by accentuating the F-statistic: 6.22; p-value = 0.0138). curvature of the linea terminalis. The widening of Males and females were best separated in the the greater sciatic notch corresponded to a cranio- PC1-PC3 plot. Shape changes along PC3 (8% caudally shorter posterior region of the ilium, as of total variance) reflected the expansion of the if resorption had occurred at its inferior border iliac blade in both breadth and height, and the between the posterior inferior iliac spine and the concomitant medio-lateral shortening of the deepest point of the greater sciatic notch. Thus, pubis. For the same PC1 scores, males tended to with respect to males, females presented with a rel- have higher PC3 scores than females, which, in atively broader iliac blade. The broader pubic body terms of morphology, reflected a greater expan- and elongated pubic rami in females produced a sion of the ilium in males for similar pubic typically wide subpubic angle and contributed to length to females (see PC3 warpings in Figure the concave appearance of the lower border of the 1). Along PC3, males and females (considering inferior pubic ramus as well as the more triangular adolescents only, adults only, or all individuals shape of the foramen obturatum. The pubic sym- together) differed significantly. The differences physis was of similar cranio-caudal length for both between adolescent and adult females were also sexes. The characteristic pre-auricular sulcus could significant, but adolescent males did not differ be discerned in females on the inferior side of the significantly from adult males or adult females sacroiliac joint, and a retro-auricular depression (Tab. 2). Allometry was significant along PC3 was visible between the protruding auriculum and (adjusted R2: 0.1861; F-statistic: 34.38; p-value the most posterior region of the iliac tuberosity < 0.0001), where the observed shape changes (see PC1 warpings in Figure 1). Furthermore, vari- were positively correlated with the size of the ation was detected in the relative expansion of the specimens. acetabulum, which was wider for males than for females. Along PC1, differences between all group Group mean differences means were significant except for those between The surface deviations of the male and adolescent and adult males (Tab. 2). Moreover, a female adult means and the male and female statistically significant, albeit weak, negative allo- adolescent means from the global adolescent metric trend was found along PC1 (adjusted R2: mean are illustrated in Figures 2 and 3 (see also 0.1063; F-statistic: 18.36; p-value < 0.0001). Supplementary Videos 5-16). The MANOVA Principal Component 2 (12% of total was performed using the first 26 PCs corre- variance) reflected ontogenetic trends with sponding to 90% of total variance explained. 122 Hip bone sexual dimorphism Fig. 1 – Score plots derived from PCA performed on the deformation fields from the global mean. The associated warped models are shown at the extremes of the range of distribution along the first three principal components (PCs). Positive deformations are mapped in warm (red) gradient colours, while negative deformations are in cold (blue) gradient colours. Given a certain region, a positive deformation signifies a volume increase to the surface reference. Accordingly, a negative deformation signifies a decrease in volume. The male (closed symbols) and female (open symbols) adolescents (squares) and adults (circles) are enclosed within 95% prediction ellipses in the plots. a. PC1-PC2 plot showing sex-related variation along PC1 and an ontogenetic trend along PC2. Along PC1, hip bones vary mostly in terms of relative width of the greater sciatic notch, length of the pubis and outward rotation of the ischium. The retro-auricular surface, pronounced in females, is indicated by a black arrow. Variation along PC2 is driven by a medio-lateral expansion of the pelvic inlet. b. The PC1-PC3 plot distinguishes best between males and females by combining the most commonly described sex-related shape changes observed along PC1 with the relative expansion of the ilium along PC3. Colour bar values in mm. 123 Hip bone sexual dimorphism Tab. 2 – P-values from the cross-validated ANOVA of the first three PCs with Bonferroni correction. Significant values (p < 0.05) are in bold. PC1 ADULT FEMALES ADOLESCENT FEMALES ADULT MALES ALL FEMALES ADOLESCENT FEMALES 0.0281 - - - ADULT MALES <0.0001 0.0014 - - ADOLESCENT MALES <0.0001 0.0001 1.0000 - ALL MALES - - - <0.0001 PC2 ADULT FEMALES ADOLESCENT FEMALES ADULT MALES ALL FEMALES ADOLESCENT FEMALES <0.0001 - - - ADULT MALES 0.0544 <0.0001 - - ADOLESCENT MALES <0.0001 0.6470 0.0005 - ALL MALES - - - 0.1200 PC3 ADULT FEMALES ADOLESCENT FEMALES ADULT MALES ALL FEMALES ADOLESCENT FEMALES 0.0132 - - - ADULT MALES <0.0001 <0.0001 - - ADOLESCENT MALES 0.0720 0.0005 0.0768 - ALL MALES - - - <0.0001 After a Bonferroni correction, all groups were the preauricular region, a non-elevated auricular found to differ significantly. Male and female surface, and a thinner post-auricular region. The adult means differed (p < 0.0001) mainly for the acetabulum was also relatively larger in males shape changes described by PC1 and PC3, which than in females. involved sexually dimorphic features of the Typical male and female configurations were greater and lesser pelvis. The male mean could already visible in adolescents, thus their means be distinguished from the female mean by its differed significantly (p = 0.0031). The shape robustness, a more vertical and expanded ilium, changes between adolescent and adult means were a narrower and deeper greater sciatic notch, a significant for both females (p-value < 0.0001) straighter arcuate line, shorter pubic rami, a nar- and males (p-value = 0.0007). Adolescents rower pubic body, and a more inward-oriented showed less robust iliac crests and more vertical ischial tuberosity. The colour maps also showed ilia than adults. The arcuate line was less curved a more pronounced anterior inferior iliac spine in adolescents, resulting in a medio-laterally nar- in the male mean, as well as a flattened groove in rower pelvic inlet. Additionally, the superior rim 124 Hip bone sexual dimorphism Fig. 2 – Surface models of the group means of adult females (a-e), adolescent females (f-j), adolescent males (p-t) and adult males (u-y) showing areas of main deformation (positive deviations in red, nega- tive in blue) from the adolescents mean (k-o, in grey) in internal (a,f,k,p,u), frontal (b,g,l,q,v), lateral (c,h,m,r,w), superior (d,i,n,s,x), and vertical (e,j,o,t,y) views. Positive deformations are mapped in warm (red) gradient colours, while negative deformations are in cold (blue) gradient colours. Given a certain region, a positive deformation signifies a volume increase to the surface reference. Accordingly, a negative deformation signifies a decrease in volume. The group means of adult and adolescent females deviate from those of adolescent and adult males for their wider greater sciatic notch, elon- gated pubic body, depressed preauricular sulcus and elevated auricular surface. Adolescent and adult males possess cranio-caudally higher ilia than females. The adult female mean can be distinguished from the adolescent female mean by features leading to a mediolaterally wider pelvic inlet, a more everted ischial tuberosity, protruding post-auricular area and a posteriorly deflected superior pubic body near the pubic symphysis. Adult males differ from adolescent males by a higher robusticity of the middle and anterior sections of the iliac crest and the anterior inferior iliac spine. Adolescent males and females differ from adults by their more medially oriented iliac blades. Colour bar values in mm. of the symphysis was anteriorly deflected. The Classification accuracy female adults differed from the female adoles- The number of PCs that could be used until cents by their longer superior pubic ramus, while the classification accuracy started to consistently the most visible shape change for males was rep- decrease was 26 (Supplementary Fig. S1). We resented by the expansion of the ilium, which achieved classification accuracies ranging from became taller in adults. 70.0% to 100.0% (Tab. 3). In particular, the 125 Hip bone sexual dimorphism Fig. 3 – Surface models of the hip bone in internal (a-d) and lateral (e-h) views with scaled vectors attached to the control points showing the direction and magnitude of deformation (magnified 20 times to allow appreciation of the deformations) from the adolescent mean to the group means of the ado- lescent (a,e) and adult (b,f) females, and adolescent (c,g) and adult (d,h) males. Positive deformations are also mapped in warm (red) gradient colours, while negative deformations are in cold (blue) gradi- ent colours. Given a certain region, a positive deformation signifies a volume increase to the surface reference. Accordingly, a negative deformation signifies a decrease in volume. Colour bar values in mm. highest accuracies were obtained for the clas- Discussion sification of the sexes within the entire sample (females = 97.1%; males = 94.7%) or into female The modern human pelvis has been inten- or male adults only (females = 94.0%; males = sively investigated for its biological relevance in 94.0%). The results for the adults were slightly relation to sex estimation, locomotion, and birth lower than those achieved by Robertson et al. (see references in Haeusler et al. 2021). Thus, (2019) in their geometric morphometric analysis appreciating the nuances of its morphological of a different sample of adult hip bone using a variation is crucial for disentangling the adaptive configuration of 17 landmarks (females = 99.2%; and functional significance of its characteristics. males = 97.8%) or 32 landmarks (females = The hip bone is the most sexually dimorphic 97.6%; males = 97.5%). Classification accuracies bone of the human skeleton because of differ- in the present male and female adolescent groups ential local growth that contributes to shaping were higher for females (100.0%) than for males a more spacious birth canal in females (e.g., (84.0%). Accuracy decreased for the classifi- Huseynov et al. 2016; Robertson et al. 2019). cation based on four groups (female adults = Our deformation-based geometric morphomet- 82.0%; female adolescents = 85.0%; male adults ric investigation of the human hip bone, a large = 78.0%; male adolescents = 84.0%), while and complex anatomical object, successfully accuracies were lowest when classifying males or captured important large-scale, but also subtle females into adolescents and adults (male ado- morphological patterns of variation that previ- lescents = 84.0%; male adults = 78.0%) (female ously required the application of a combination adolescents = 70.0%; female adults = 78.0%). of methods. It allowed us to intuitively visualise 126 Hip bone sexual dimorphism Tab. 3 - Cross-validated classification results based on the first 26 PCs, derived from the Discriminant Function Analysis. The percentages of correctly classified individuals are given for each subgroup. ADULT FEMALES ADOLESCENT ADULT MALES ADOLESCENT FEMALES MALES ADULT FEMALES 82.0 14.0 0.0 4.0 ADOLESCENT FEMALES 10.0 85.0 0.0 5.0 ADULT MALES 4.0 0.0 78.0 18.0 ADOLESCENT MALES 0.0 4.0 12.0 84.0 Overall classification accuracy: 81.38%; Kappa statistic: 0.74 ADULT FEMALES ADULT MALES ADULT FEMALES* ADULT MALES* ADULT FEMALES 94.0 6.0 99.2 0.8 ADULT MALES 6.0 94.0 2.2 97.8 Overall classification accuracy: 94.00%; Kappa statistic: 0.88 ADOLESCENT ADOLESCENT FEMALES MALES ADOLESCENT FEMALES 100.0 0.0 ADOLESCENT MALES 16.0 84.0 Overall classification accuracy: 91.11%; Kappa statistic: 0.82 ALL FEMALES ALL MALES ALL FEMALES 97.1 2.9 ALL MALES 5.3 94.7 Overall classification accuracy: 95.86%; Kappa statistic: 0.92 ADULT FEMALES ADOLESCENT FEMALES ADULT FEMALES 78.0 22.0 ADOLESCENT FEMALES 30.0 70.0 Overall classification accuracy: 75.71%; Kappa statistic: 0.45 ADULT MALES ADOLESCENT MALES ADULT MALES 78.0 22.0 ADOLESCENT MALES 16.0 84.0 Overall classification accuracy: 80.00%; Kappa statistic: 0.58 * In Robertson et al. 2019 127 Hip bone sexual dimorphism Tab. 4 – Significance values (p-values) from the the pelvic canal in females (see Figure 1). This is MANOVA performed using the first 26 PCs with Bonferroni correction. compatible with the fact that the sacrum is not as highly dimorphic as one would expect by con- sidering the total degree of pelvic sexual dimor- ADULT ADULT ADOLESCENT phism (Flander 1978; Zech et al. 2012; Zhan et FEMALES MALES FEMALES al. 2018; Krenn et al. 2021; but see Rusk and ADULT <0.0001 - - Ousley 2016). MALES While most sexually dimorphic traits, such as ADOLESCENT <0.0001 <0.0001 - the width of the greater sciatic notch, the cur- FEMALES vature of the pelvic inlet, and the orientation of ADOLESCENT <0.0001 0.0007 0.0031 the ischiatic tuberosity are already present during MALES adolescence, they become more pronounced in female adults (Figs. 2, 3). The ontogenetic trends we registered in the male and female hip bones gross morphological variation while permitting and their related shape changes add to previous quantitative assessment of the magnitude and observations by Huseynov et al. (2016), who direction of local shape changes (Fig. 3). The found that the female pelvis deviates from males convenient visual depictions produced with the starting from puberty in terms of maturation Deformetrica approach provided insights into and development of sexually dimorphic traits. subtle shape changes that significantly informed Based on a landmark-based geometric morpho- our understanding of hip bone variation. As metric study of the ilium, Bilfeld et al. (2013) such, this method offered all the advantages of concluded that divergent ontogenetic trajecto- geometric morphometrics using dense landmark ries between males and females establish already configurations while avoiding the associated beginning from the age of 9 years. This notion drawbacks related to time-consuming landmark was corroborated by Corron et al. (2021) who collection and susceptibility to placement error. observed the timing of maturation of the pelvic We detected similar morphological signals to epiphyseal sites in relation to the appearance of those previously obtained using other approaches, sexually dimorphic traits. The differential growth from qualitative observations (Coleman 1969; of the hip bone in males and females is reflected Corron et al. 2021), to linear measurements by substantial bone modelling occurring in the (Mestekova et al. 2015), and more sophisti- regions of the ischium, the ischio-pubic ramus, cated landmark-based geometric morphometric the pubis and the auriculum, resulting in larger approaches (Bilfeld et al. 2013, Huseynov et al. morphological differences of the lesser pel- 2016; Robertson et al. 2019). As already noticed vis between adolescent and adult females than by Coleman (1969), we confirmed that the mor- between adolescent and adult males (Figs. 2, 3). phological heterogeneity of the hip bone is accen- Huseynov et al. (2016) found that, during ado- tuated in the anatomical regions formed by more lescence to early adulthood, the female pelvis than one bony element (e.g., the acetabulum, models towards a pelvic canal shape increasingly obturator foramen, ilio-pubic eminence, greater suitable for parturition. Our data confirm that sciatic notch, etc.). Morphological differences female adolescents and adults differ significantly, between males and females also extended to the showing that continued, obstetrically relevant ilium, which is craniocaudally shorter, and thus pelvic directional growth occurs during adoles- relatively broader, in females. In fact, the reduced cence in females. Particularly, the adult female height of the posterior region of the ilium might hip bone is characterized by a more curved arcu- be associated to the widening of the greater sci- ate line than in female adolescents and an elon- atic notch. Further, the elevation of the auricular gated superior pubic ramus, leading to a larger area contributes to the medio-lateral width of pelvic inlet. We also observed that the adolescent 128 Hip bone sexual dimorphism male hip bone goes through significant shape superior pubic ramus and an antero-posteriorly changes, too, before reaching the adult configu- narrower and supero-inferiorly shorter ilium, ration. However, these changes mostly concern regardless of their sex. This finding is compat- the iliac blade, while the lesser pelvis grows in a ible with the existence of obstetrical constraints similar fashion to females but to a lesser degree. in females (see detailed discussion in Fischer and The differential growth of the ilium during Mitteroecker 2015; Fischer et al. 2021; Haeusler adolescence was already described by Coleman et al. 2021) and can possibly reflect genetic sta- (1969), who noted that in males, the anterior bility preserving the structural integrity of pelvic part of the iliac crest corresponding to the ante- morphology in both sexes. rior superior iliac spine is more laterally flaring By adding to these previous findings, we are than females, while the posterior portion of the thereby confirming the utility of deformation- ilium is more laterally deflected. In accordance based morphometrics to analyse sex- and age- with the greater muscularity, the anterior supe- related shape changes of the hip bone. Using con- rior, and especially the anterior inferior iliac figurations of 32 or fewer landmarks, Robertson spines were more robust in males. The anterior et al. (2019) concluded that a 17-landmark con- inferior iliac spine represents the origin of the figuration was the most appropriate to capture rectus femoris muscle that might be more devel- male and female hip bone morphological differ- oped in males compared to females. Dimorphic ences, allowing them to reach the highest clas- differences in hip bone robusticity might be sification accuracies within their adult sample driven by the same growth patterns observed (see our Table 2). Using a much higher number in other bones. In fact, a general increase in the of variables (~50,000, equivalent to the number robusticity of long bones has been observed after of surface polygons representing each specimen), the adolescent growth spurt (Rantalainen et al. we achieved a marginally lower sex classification 2016). This phenomenon is more pronounced in accuracy for the adult specimens, which means males than in females, and the timing at which that a deformation-based geometric morpho- dimorphic differences in robusticity become evi- metric analysis of the entire hip bone decreases dent varies between different skeletal elements accuracy by only a small percentage. Since the (Iuliano-Burns et al. 2009). current work and the study by Robertson et al. Another geometric morphometric analysis (2019) use different samples, the respective out- of the adult pelvis by Fischer and Mitteroecker comes are not directly comparable. Nevertheless, (2017) revealed that differences in orientation of our evaluation still shows that expected sex-based the iliac blades between males and females were differences and resulting classification accuracies mostly a consequence of differences in overall into male and female categories can be achieved stature. Thus, taller persons, typically males, pre- using both geometric morphometric approaches. sent with a relatively taller and narrower pelvis According to the findings in Huseynov et al. and inwardly rotated iliac blades while shorter (2016), we are aware that the possible inclusion of individuals, typically females, show a relatively hip bones from mature individuals might have con- wider pelvis with shorter and laterally-extending founded the interpretation of sex-related morpho- iliac blades. The direction of the deformation logical variation within the adult sample. Further fields observed for males and females in the pre- studies focussing entirely on samples of known age sent study (Fig. 3) could be interpreted in light of and similar origin might allow for a higher reso- these previous findings, although body height was lution of shape changes occurring during the dif- not known for the individuals considered here. ferent phases of adulthood. Although most of our In addition to sex-based variation, subtle sample (122/147) was from Central Europe, we allometric patterns emerged. As expected, female could speculate that the overlap between the differ- shape changes were associated with smaller sizes. ent age and sex groups might also partially derive Smaller individuals tended to possess a longer from geographical heterogeneity. Based on results 129 Hip bone sexual dimorphism obtained for the sacrum (Krenn et al. 2021), we analysis; CF and AU produced tables and visuali- would expect that using a more geographically sations; all authors discussed the results and their homogeneous sample could have led to better sep- interpretation; CF wrote the manuscript with the aration between the groups and, therefore, higher contribution of all authors. classification accuracies. References Acknowledgements Abitbol MM (1987) Obstetrics and posture in pelvic anatomy. J Hum Evol 16:243–255. htt- We thank the following persons and institutions for ps://doi.org/10.1016/0047-2484(87)90001-7 granting access to or otherwise facilitating data acqui- Bass WM (1995) Human Osteology: a Laboratory sition: Karin Wilschke-Schrotta and Sabine Egg- and Field Manual, Missouri Archaeological ers, Department of Anthropology, Natural History Society, Columbia, Mo. Museum Vienna; Kathia Chaumoitre, UMR7268 Beaudet A, Clarke RJ, Heaton JL, et al (2020) The Anthropologie bioculturelle, Droit, Ethique et Sante atlas of StW 573 and the late emergence of hu- and Assistance Publique des Hôpitaux de Marseille, man-like head mobility and brain metabolism. Marseille; Martin Friess and Véronique Laborde, Sci Rep 10:4285. https://doi.org/10.1038/ Musée de l’Homme, Paris; Marcia Ponce de León, s41598-020-60837-2 Marc Scherrer, Jody Weissmann, Anthropological Beaudet A, Dumoncel J, Thackeray JF, et al Institute and Museum, University of Zurich, Zurich; (2016) Upper third molar internal structural Jocelyne Desideri, Laboratory of Prehistoric Archae- organization and semicircular canal morphol- ology and Anthropology, University of Geneva, ogy in Plio-Pleistocene South African cerco- Geneva; Harald Wilfing, Katrin Schäfer, Katarina pithecoids. J Hum Evol 95:104-20. https://doi. Matiasek, Department of Evolutionary Anthropol- org/10.1016/j.jhevol.2016.04.004 ogy, University of Vienna, Vienna. We thank Stepha- Beaudet A, Holloway R, Benazzi S (2021) A nie Cole, University of Nevada, Reno, and Yaro- comparative study of the endocasts of OH slav Brůžek, Université de Bordeaux, UMR 5199 5 and SK 1585: Implications for the paleo- PACEA, and Charles University, Prague, for the neurology of eastern and southern African in-depth conversations on subadult sex estimation, Paranthropus. J Hum Evol 156:103010. https:// sexual dimorphism, and maturation of the pelvis. doi.org/10.1016/j.jhevol.2021.103010 The Barcelona Supercomputing Center Betti L, Manica A (2018) Human variation in granted computational time to AU (project No the shape of the birth canal is significant and BCV-2020-1-0008). C.F., N.M.W., V.A.K. and geographically structured. Proc Royal Soc M.H. were financially supported by the Swiss B 285:20181807. https://doi.org/10.1098/ National Science Foundation (SNF grant No rspb.2018.1807 31003A_176319/1). AU was funded by the Betti L, Von Cramon-Taubadel N, Manica A, et al Agencia Estatal de Investigación (PID2020- (2013) Global geometric morphometric analy- 117289GB-I00 and PID2020-116908GB-I00, ses of the human pelvis reveal substantial neu- AEI/FEDER, EU) and the Generalitat de tral population history effects, even across sexes. Catalunya (CERCA Programme). PLoS One 8:e55909. https://doi.org/10.1371/ journal.pone.0055909 Bilfeld MF, Dedouit F, Sans N, et al (2013) Author contributions Ontogeny of size and shape sexual dimor- phism in the ilium: a multislice computed CF, NMW, and MH initiated the project; NMW tomography study by geometric morphom- and VK prepared and aligned the surface data; AU etry. J Forensic Sci 58:303–310. https://doi. performed the deformation-based morphometric org/10.1111/1556-4029.12037 130 Hip bone sexual dimorphism Bône A, Louis M, Martin B, et al (2018) Cox SL (2021) A geometric morphometric assess- Deformetrica 4: an open-source software ment of shape variation in adult pelvic mor- for statistical shape analysis. In: Reuter M, phology. Am J Phys Anthropol 176:652–671. Wachinger C, Lombaert H, Paniagua B, Lüthi https://doi.org/10.1002/ajpa.24399 M & Egger B (eds.) Shape in Medical Imaging, Dray S, Dufour A (2007) The ade4 pack- Springer, Cham. age: Implementing the duality diagram for Bookstein FL (1991) Morphometric Tools for ecologists. J Stat Softw 22:1–20. https://doi. Landmark Data: Geometry and Biology, org/0.18637/jss.v022.i04 Cambridge University Press, Cambridge. Dunsworth HM (2020) Expanding the evolu- Bookstein FL (2018) A Course in Morphometrics tionary explanations for sex differences in the for Biologists: Geometry and Statistics for human skeleton. Evol Anthropol 29:108–116. Studies of Organismal Form, Cambridge https://doi.org/10.1002/evan.21834 University Press, Cambridge. Durrleman S, Pennec X, Trouvé A, et al (2012a) Brown WM, Price ME, Kang J, et al (2008) Comparison of the endocranial ontogenies Fluctuating asymmetry and preferences for between chimpanzees and bonobos via tem- sex-typical bodily characteristics. Proc Natl poral regression and spatiotemporal regis- Acad Sci USA 105:12938-12943. https://doi. tration. J Hum Evol 62:74–88. https://doi. org/10.1073/pnas.0710420105 org/10.1016/j.jhevol.2011.10.004 Brůžek J (2002) A method for visual determina- Durrleman S, Prastawa M, Korenberg JR, et al tion of sex, using the human hip bone. Am J (2012b) Topology preserving atlas construc- Phys Anthropol 117:157–168. https://doi. tion from shape data without correspondence org/10.1002/ajpa.10012 using sparse parameters. Med Image Comput Buikstra JE, Ubelaker DH (1994) Standards for Comput Assist Interv 15:223–230. https://doi. data collection from human skeletal remains: org/10.1007/978-3-642-33454-2_28 Proceedings of a seminar at the Field Museum Ferembach D, Schwidetzky I, Stoukal M (1980) of Natural History, organized by Jonathan Recomendations for age and sex diagnosis of Haas (3rd ed.), Arkansas Archeological Survey, skeletons. J Hum Evol 9:517–549. https://doi. Fayetteville. org/10.1016/j.jchb.2005.07.002 Bytheway JA, Ross AH (2010) A geomet- Fischer B, Grunstra NDS, Zaffarini E, et al (2021) ric morphometric approach to sex de- Sex differences in the pelvis did not evolve de novo termination of the human adult os coxa. in modern humans. Nat Ecol Evol 5:625–630. J Forensic Sci 55:859–864. https://doi. https://doi.org/10.1038/s41559-021-01425-z org/10.1111/j.1556-4029.2010.01374.x Fischer B, Mitteroecker P (2015) Covariation be- Charles BE (2010) A Geometric Morphometric tween human pelvis shape, stature, and head Analysis of the Human Ossa Coxae for Sex size alleviates the obstetric dilemma. Proc Natl Determination, Master of Science, Boston Acad Sci USA 112:5655–5660. https://doi. University, Boston. org/10.1073/pnas.1420325112 Coleman WH (1969) Sex differences in the Fischer B, Mitteroecker P (2017) Allometry and growth of the human bony pelvis. Am J sexual dimorphism in the human pelvis. Anat Phys Anthropol 31:125–151. https://doi. 300:698–705. https://doi.org/10.1002/ar.23549 org/10.1002/ajpa.1330310202 Fischer MCM, Krooß F, Habor J, et al (2019) Corron LK, Santos F, Adalian P, et al (2021) How A robust method for automatic identification low can we go? A skeletal maturity thresh- of landmarks on surface models of the pelvis. old for probabilistic visual sex estimation Sci Rep 9:13322. https://doi.org/10.1038/ from immature human os coxae. Forensic Sci s41598-019-49573-4 Int 325:110854. https://doi.org/10.1016/j. Flander LB (1978) Univariate and multivari- forsciint.2021.110854 ate methods for sexing the sacrum. Am J 131 Hip bone sexual dimorphism Phys Anthropol 49:103–110. https://doi. İşcan MY, Steyn M (2013) The Human Skeleton org/10.1002/ajpa.1330490116 in Forensic Medicine, Springfield, Thomas. Franklin D, Freedman L, Milne N (2005) Sexual Iuliano-Burns S, Hopper J, Seeman E (2009) The dimorphism and discriminant function sexing age of puberty determines sexual dimorphism in indigenous South African crania. HOMO in bone structure: A male/female co-twin con- 55:213–228. https://doi.org/10.1016/j. trol study. J Clin Endocrinol Metab 94:1638– jchb.2004.08.001 1643. https://doi.org/10.1210/jc.2008-1522 Genovés S (1959) L’estimation des différences Karsten JK (2018) A test of the preauricular sulcus as sexuelles dans l’os coxal: différences métriques an indicator of sex. Am J Phys Anthropol 165:604– et différences morphologiques. Bull Mem 608. https://doi.org/10.1002/ajpa.23372 Soc Anthropol Paris X:3–95. https://doi. Klales AR (2020) Sex estimation using pelvis mor- org/10.3406/bmsap.1959.2750 phology. In: Klales AR (ed.) Sex Estimation of Glaunès J, Joshi S. (2006) Template estimation the Human Skeleton: History, Methods, and form unlabeled point set data and surfaces Emerging Techniques, Academic Press. for Computational Anatomy. 1st MICCAI Klales AR, Ousley SD, Vollner JM (2012) A re- Workshop on Mathematical Foundations vised method of sexing the human innominate of Computational Anatomy: Geometrical, using Phenice’s nonmetric traits and statistical Statistical and Registration Methods for methods. Am J Phys Anthropol 149:104–114. Modeling Biological Shape Variability, Oct https://doi.org/10.1002/ajpa.22102 2006, Copenhagen, Denmark. https://hal. Krantz SG, Parks HR, 6.2.4. PT (2013) The im- archives-ouvertes.fr/hal-00263576. plicit function theorem: history, theory, and ap- Grunstra NDS, Zachos FE, Herdina AN, et al plications, Birkhäuser, Basel. (2019) Humans as inverted bats: A comparative Krenn VA, Fornai C, Webb NM, et al (2021) Sex approach to the obstetric conundrum. Am J determination accuracy in a Central European Hum Biol 31:e23227. https://doi.org/10.1002/ sample using the human sacrum. Anthropol Anz ajhb.23227 https://doi.org/10.1127/anthranz/2021/1415 Gunz P, Mitteroecker P (2013) Semilandmarks: Krishan K, Chatterjee PM, Kanchan T, et al a method for quantifying curves and surfaces. (2016) A review of sex estimation techniques Hystrix 24:103–109. https://doi.org/10.4404/ during examination of skeletal remains in fo- hystrix-24.1-6292 rensic anthropology casework. Forensic Sci Gunz P, Mitteroecker P, Bookstein F (2005) Int 261:165.e1-8. https://doi.org/10.1016/j. Semilandmarks in Three Dimensions. In: Slice forsciint.2016.02.007 DE (ed.) Modern Morphometrics in Physical Krogman WM (1962) The Human Skeleton in Anthropology, Springer, Boston, MA. Forensic Medicine, Thomas, Springfield. Haeusler M, Grunstra NDS, Martin RD, et al Kuchař M, Henyš P, Rejtar P, et al (2021) Shape (2021) The obstetrical dilemma hypothesis: morphing technique can accurately predict pelvic there’s life in the old dog yet. Biol Rev 96:2031– bone landmarks. Int J Legal Med 135:1617–1626. 2057. https://doi.org/10.1111/brv.12744 https://doi.org/10.1007/s00414-021-02501-6 Haeusler MF, Schweitzer W, Braun M, et al Kurki HK (2007) Protection of obstetric dimen- (2004) Evaluation von 3D-Scanner für den sions in a small-bodied human sample. Am J Einsatz in der Rechtsmedizin. Rechtsmedizin Phys Anthropol 133:1152–1165. https://doi. 14:356–357. org/10.1002/ajpa.20636 Huseynov A, Zollikofer CPE, Coudyzer W, et al Legendre P (2018) Lmodel2: Model II Regression. (2016) Developmental evidence for obstetric https://cran.r-project.org/web/packages/lmo- adaptation of the human female pelvis. Proc del2/index.html. Natl Acad Sci USA 113:5227–5232. https:// Mestekova S, Brůžek J, Veleminska J, et al doi.org/10.1073/pnas.1517085113 (2015) A test of the DSP sexing method on 132 Hip bone sexual dimorphism CT images from a modern French sample. Rusk KM, Ousley SD (2016) An evaluation of J Forensic Sci 60:1295–1299. https://doi. sex- and ancestry-specific variation in sacral size org/10.1111/1556-4029.12817 and shape using geometric morphometrics. Am Mitteroecker P, Grunstra NDS, Stansfield E, et al J Phys Anthropol 159:646–654. https://doi. (2021) Did population differences in human org/10.1002/ajpa.22926 pelvic form evolve by drift or selection? Bull Schlager S (2013) Soft-tissue reconstruction of the hu- Mém Soc Anthropol Paris 33:10–25. https:// man nose: population differences and sexual dimor- doi.org/10.4000/bmsap.7460 phism. PhD dissertation, University of Freiburg. Oikonomopoulou EK, Valakos E, Nikita E (2017) Schutkowski H (1993) Sex determination of in- Population-specificity of sexual dimorphism in cra- fant and juvenile skeletons: I. Morphognostic nial and pelvic traits: evaluation of existing and pro- features. Am J Phys Anthropol 90:199–205. posal of new functions for sex assessment in a Greek https://doi.org/10.1002/ajpa.1330900206 assemblage. Int J Legal Med 131:1731–1738. htt- Slice DE (ed.) (2005) Modern morphometrics ps://doi.org/10.1007/s00414-017-1655-x in physical anthropology, Kluwer Academic/ Phenice TW (1969) A newly developed vis- Plenum Publishers, New York. ual method of sexing the os pubis. Am J Stansfield E, Kumar K, Mitteroecker P, et al (2021) Phys Anthropol 30:297–302. https://doi. Biomechanical trade-offs in the pelvic floor con- org/10.1002/ajpa.1330300214 strain the evolution of the human birth canal. R Development Core Team (2021) R: A language Proc Natl Acad Sci USA 118:e2022159118. and environment for statistical computing R https://doi.org/10.1073/pnas.2022159118 Foundation for Statistical Computing, Vienna, Steyn M, Pretorius E, Hutten L (2004) Geometric Austria. ISBN 3-900051-07-0. Available: morphometric analysis of the greater sciatic http://www.R-project.org. notch in South Africans. HOMO 54:197–206. Rantalainen T, Weeks BK, Nogueira RC, et al https://doi.org/10.1078/0018-442X-00076 (2016) Long bone robustness during growth: Tague RG (2003) Pelvic sexual dimorphism in a A cross-sectional pQCT examination of metatherian, Didelphis virginiana: implications children and young adults aged 5–29 years. for eutherians. J Mammal 84:1464–1473. htt- Bone 93:71–78. https://doi.org/10.1016/j. ps://doi.org/10.1644/BME-009 bone.2016.09.015 Tobolsky VA, Kurki HK, Stock JT (2016) Patterns Rmoutilová R, Dupej J, Veleminska J, et al (2017) of directional asymmetry in the pelvis and pel- Geometric morphometric and traditional vic canal. Am J Hum Biol 28:804-810. https:// methods for sex assessment using the posterior doi.org/10.1002/ajhb.22870 ilium. Leg Med (Tokyo) 26:52–61. https://doi. Torres-Tamayo N, García-Martínez D, Nalla S, et org/10.1016/j.legalmed.2017.03.004 al (2018) The torso integration hypothesis re- Robertson HI, Pokotylo DL, Weston DA (2019) visited in Homo sapiens: Contributions to the Testing landmark redundancy for sex-based understanding of hominin body shape evolu- shape analysis of the adult human os coxa. Am tion. Am J Phys Anthropol 167:777–790. htt- J Phys Anthropol 169:689–703. https://doi. ps://doi.org/10.1002/ajpa.23705 org/10.1002/ajpa.23860 Urciuoli A, Zanolli C, Almécija S, et al (2021) Rogers T, Saunders S (1994) Accuracy of sex de- Reassessment of the phylogenetic relationships termination using morphological traits of the of the late Miocene apes Hispanopithecus and human pelvis. J Forensic Sci 39:1047–1056. Rudapithecus based on vestibular morphology. https://doi.org/10.1520/JFS13683J Proc Natl Acad Sci USA 118:e2015215118. Rosenberg KR (1992) The evolution of mod- https://doi.org/10.1073/pnas.2015215118 ern human childbirth. Yearb Phys Anthropol Urciuoli A, Zanolli C, Almécija S, et al. (2018) 35:89–124. https://doi.org/10.1002/ Analysis of the primate vestibular apparatus: a AJPA.1330350605 comparison of 3D geometric morphometric and 133 Hip bone sexual dimorphism diffeomorphism approaches. Proceedings of the implications for the obstetrical dilemma in humans. European Society for Human Evolution, Faro, Am J Phys Anthropol Supplement 71:111–112. Portugal. Weber GW, Bookstein FL (2011) Virtual Urciuoli A, Zanolli C, Beaudet A, et al (2020) Anthropology: A Guide to a New Interdisciplinary The evolution of the vestibular apparatus in Field, Vienna, Austria, Springer. apes and humans. eLife 9:e51261. https://doi. Wells JC (2017) The New “Obstetrical Dilemma”: org/10.7554/eLife.51261 Stunting, Obesity and the Risk of Obstructed Veneziano A, Landi F, Profico A (2018) Surface Labour. Anat 300:716–731. https://doi. smoothing, decimation, and their effects on org/10.1002/ar.23540 3D biological specimens. Am J Phys Anthropol Wilson LaB, Ives R, Humphrey LT (2016) 166:473–480. https://doi.org/10.1002/ajpa.23431 Quantification of 3D curvature in the iliac crest: Waldeyer HWG (1899) Das Becken. Topographisch- Ontogeny and implications for sex determination anatomisch mit besonderer Berücksichtigung der in juveniles. Am J Phys Anthropol 162:255–266. Chirurgie und Gynäkologie, Cohen, Bonn. https://doi.org/10.1002/ajpa.23114 Waltenberger L, Rebay-Salisbury K, Mitteroecker Zanolli C, Pan L, Dumoncel J, et al (2018) Inner tooth P (2021) Three-dimensional surface scanning morphology of Homo erectus from Zhoukoudian. methods in osteology: A topographical and New evidence from an old collection housed at geometric morphometric comparison. Am J Uppsala University, Sweden. J Hum Evol 116:1-13. Phys Anthropol 174:846–858. https://doi. 10.1016/j.jhevol.2017.11.002 org/10.1002/ajpa.24204 Zech WD, Hatch G, Siegenthaler L, et al (2012) Warrener AG, Lewton KL, Pontzer H, et al (2015) Sex determination from os sacrum by postmor- A wider pelvis does not increase locomotor cost tem CT. Forensic Sci Int 221:39–43. https:// in humans, with implications for the evolution doi.org/10.1016/j.forsciint.2012.03.022 of childbirth. PLoS One 10:e0118903. https:// Zelditch ML, Fink WL, Swiderski DL (1995) doi.org/10.1371/journal.pone.0118903 Morphometrics, homology, and phylogenetics: Washburn SL (1960) Tools and human evolution. Quantified characters as synapomorphies. Syst Biol Sci Am 203:63–75. https://doi.org/10.1038/ 44:179–189. https://doi.org/10.2307/2413705 scientificamerican0960-62schul Zhan MJ, Fan F, Qiu LR, et al (2018) Estimation Weaver DS (1980) Sex differences in the ilia of a of stature and sex from sacrum and coccyx known sex and age sample of fetal and infant measurements by multidetector computed to- skeletons. Am J Phys Anthropol 52:191–195. mography in Chinese. Leg Med 34:21–26. htt- https://doi.org/10.1002/ajpa.1330520205 ps://doi.org/10.1016/j.legalmed.2018.07.003 Webb NM, Fornai C, Krenn VA, et al (2021) Do chimpanzees really have a spacious birth canal? A reanalysis of cephalopelvic proportions and Editor, Giovanni Destro Bisol This work is distributed under the terms of a Creative Commons Attribution-NonCommercial 4.0 Unported License http://creativecom- mons.org/licenses/by-nc/4.0/ 134