Current genetic and fossil evidence point to an African origin of Homo sapiens (e.g., Hublin et al., 2017; Mallick et al., 2016). Fossil and archaeological evidence of the earliest H. sapiens span the African continent, suggesting a pan-African (Gunz et al., 2009a; Hublin et al., 2017; Mallick et al., 2016; Ragsdale et al., 2023; Scerri et al., 2018, 2019) origin of our species. North Africa is a valuable region for understanding the expansion of our species within and out of Africa during the Late Pleistocene. It has yielded hominin fossil remains spanning the last million years, including the oldest currently known members of our lineage from Jebel Irhoud, Morocco, dated to ca. 300 ka (Richter et al., 2017). In addition to Jebel Irhoud, North African Middle Stone Age (MSA) sites yielding human remains have been found exclusively in Morocco. They include Mugharet el’Aliya, Dar es-Soltane II, El Harhoura I, Kébibat (Rabat), and Contrebandiers Cave (Témara). Many of these fossils are associated with the Aterian tool industry and other material culture that documents the emergence of modern behavior in the form of decorated shells, hearths and stone-walled structures, and a bone tool industry (Bouzouggar et al., 2018; Dibble et al., 2013; Hallett et al., 2021; Scerri, 2017).

The site of Contrebandiers Cave is located near the village of Témara and just south of Rabat on the Atlantic coast of Morocco. It was first excavated by Jean Roche in 1955. Subsequent excavations in the mid-1970s yielded human remains from the Aterian layers including an adult mandible and cranial fragments (Bergmann et al., 2022; Debénath, 2000; Ferembach, 1976a; Ferembach, 1998; Hublin, 1992; Roche, 1963; Roche & Texier, 1976). In 2009, new excavations at the site run by a joint Moroccan-American collaboration directed by Mohamed Abdeljalil El Hajraoui and the late Harold Dibble revealed a skull and partial skeleton of a child. The layer where the child was found (5c) was dated by thermoluminescence dating (TL) of burnt stones and single-grain (SG) and single aliquot (SA) optically stimulated luminescence (OSL) dating of quartz grains to around ca. 115 ka, or MIS 5c and d (Dibble et al., 2013; Dibble et al., 2012; Jacobs et al., 2011).

The North African MSA hominins are characterized by extremely robust faces and mandibles (expressed as superstructures), megadonty, as well as considerable size and morphological variability (Bergmann et al., 2022; Harvati & Hublin, 2012; Hublin et al., 2012). Historically, several of these fossils (e.g., Jebel Irhoud 1 and 2, Mugharet el’Aliya) were referred to as African Neanderthals (Ennouchi, 1962; Şenyürek, 1940). Previous studies of the adult North African MSA fossils emphasize a mixture of archaic and modern morphology, often highlighting morphological affinities to older (e.g., Jebel Irhoud) and/or younger (e.g., Iberomaurusian) North African fossils. A study by Ferembach (1976b) on the Dar es-Soltane II H5 facial skeleton described morphological similarities between it and North African Late Stone Age (LSA) humans from Mechta, Afalou, and Taforalt. A more recent geometric morphometric analysis of the Dar es-Soltane II H5 facial skeleton did not identify any strong shape similarities between it and later Iberomaurusians (Harvati & Hublin, 2012). Instead, the Dar es-Soltane II H5 face was most similar to an early H. sapiens from Qafzeh (Israel), a Late Pleistocene North African (Wadi Kubbaniya), and Jebel Irhoud 1. Dentally, the North African MSA hominins are characterized by massive post-canine teeth, mass-additive traits, and thick enamel and resemble early H. sapiens from Skhul and Qafzeh, as well as Peştera cu Oase 2, an Upper Paleolithic human from Romania (Hublin et al., 2012). Interestingly, ancient DNA was retrieved from the Oase 2 fossil which showed that its genome contained 6 to 9% of Neanderthal DNA, suggesting a Neanderthal ancestor four to six generations back (Fu et al., 2015). The robust facial features found in North African MSA hominins may also be a result of interbreeding with archaic populations. A recent geometric morphometric study of North African MSA mandibles (Bergmann et al. (2022) highlighted their remarkable variability in size, shape, and morphology. They also found continuity in mandibular shape compared to both earlier and later Pleistocene North Africans, implying a long-term, accretionary model of evolution.

In addition to adult remains, a handful of juvenile MSA North African hominins have been recovered from the sites of Jebel Irhoud, Kébibat, Mugharet el’Aliya, Dar es-Soltane II, and, most recently, Contrebandiers Cave. The Irhoud 3 mandible (ca. 8 years of age (Smith et al., 2007)) is robust with an anatomically modern chin, like the adult Irhoud mandibles (Hublin, 1992). A dental microstructure study on Irhoud 3 found that it had an extended ontogeny like later H. sapiens rather than an accelerated rate of growth as seen in Neanderthals (Smith et al., 2007). The fragmentary maxilla from Mugharet el’Aliya (ca. 9 years of age) was originally classified as a Neanderthal by Şenyürek (1940) because of its posteriorly oriented zygoma, lack of canine fossa, and large permanent dental crowns. However, a reexamination of this fossil by Minugh-Purvis (1993) concluded that although it probably lacked a canine fossa, its maxillary flexion was H. sapiens-like. More recently, using a combination of geometric morphometric and surface registration methods, Röding et al. (2022) found that Mugharet el’Aliya shows shape affinities to H. sapiens, in particular an early H. sapiens subadult individual from Qafzeh, Israel.

The different combinations of ancestral and derived traits found in the earliest H. sapiens fossils indicates that morphological evolution at the root of our species was mosaic. Deep population substructure and possibly hybridization with archaic humans are potential explanations for the morphological, shape, and size variability seen in these fossils (Ackermann et al., 2015; Gunz et al., 2009a; Harvati & Ackermann, 2022; Stringer, 2016). Their mosaic morphology may also be explained by the retention of ancestral features in a regionally evolving North African lineage (Bergmann et al., 2022) combined with structural constraints associated with cranio-mandibular size, shape, and form (Rosas et al., 2019). As evolutionary changes in morphological characters require changes in the developmental process (Gould, 1977), a better understanding of how facial features develop could shed light on the morphological variability present in these hominins. Geometric morphometric shape analysis has become an increasingly popular method to study growth and development in hominins. Comparative studies of ontogenetic allometric trajectories have shown that many aspects of taxon-specific facial morphology develop prenatally or very early postnatally (e.g., Ponce de León & Zollikofer, 2001). It also appears that, while postnatal development contributes to adult facial and mandibular morphology, it is less pronounced than the prenatal contribution (e.g., Bastir et al., 2007). Geometric morphometric approaches have also been useful in identifying and visualizing allometric shape changes in the face and mandible (Bergmann et al., 2021; Freidline et al., 2012a, 2013; Rosas & Bastir, 2004).

In this study, we virtually reconstruct a computed tomographic (CT) scan of the upper face of the undescribed juvenile skeleton from Contrebandiers and quantify its facial shape and growth pattern. We compare its maxillary morphology to fossil and recent H. sapiens, to address continuity in the North African fossil record and explore the shape similarities between North African MSA and early H. sapiens from Skhul and Qafzeh. To better assess the evolution of facial growth within H. sapiens, we also compare its shape and growth pattern to juvenile Neanderthals. Furthermore, as the Contrebandiers fossil is a subadult, we use growth simulations to predict its adult morphology. We assess ontogenetic variability in the African MSA, a period characterized by the origins, emergence, and dispersal of our species, but poorly understood because of the fragmentary and scant human fossil record. The results of this study shed light on the evolution of ontogeny in our species and provide further insight to the role of North Africa in our emergence and dispersal and the tempo and mode of evolution during the Middle to Late Pleistocene.

Materials and Methods

Sample

The fossil sample is comprised of non-adult and adult individuals attributed to Eurasian Neanderthals and African and Eurasian Middle-to-Late Pleistocene H. sapiens (Table 1). The recent human sample comprises cross-sectional ontogenetic series, ranging in age from birth to adulthood, from Portugal (Anthropological Collection of the University of Coimbra) and France (Anatomical Institute of Strasbourg). The entire sample was classified into five age groups according to dental development: no teeth erupted (AG 1, or neonates); developing deciduous dentition, until completion (AG 2); first permanent molar (M1) in functional occlusion (AG 3); second permanent molar (M2) in functional occlusion (AG 4); and third permanent molar (M3) in functional occlusion (or adult; AG 5). The age distribution of the recent human sample is as follows: AG 1: n = 9; AG 2: n = 23; AG 3 n = 7; AG 4: n = 6; and AG 5: n = 13.

Table 1 Fossil specimens used in this study; computed tomographic (CT) scan

Virtual Reconstruction of the Contrebandiers Maxilla

Micro-computed tomographic CT scans of the Contrebandiers skeleton were made with a Diondo d3 at the Max Planck Institute for Evolutionary Anthropology Leipzig (Germany) with a scan resolution of 30 µm. While much of the Contrebandiers cranium was found shattered, the maxilla remained largely intact (Figs. 1 and 2). A three-dimensional surface model was generated from the CT scan using Avizo v. 7.1 (FEI Visualization Science Group, Hillsboro) and minor virtual reconstruction was performed on the surface model. This mainly consisted of reflecting the maxilla along its midsagittal plane, smoothing the surfaces, and filling holes. All virtual surface reconstruction was performed using Geomagic Studio 2014 v. 3.0 (Geomagic Inc., Rockhill).

Fig. 1
figure 1

The fragmented Contrebandiers fossil skull (right side) encased in plaster. Photograph courtesy of Contrebandiers excavation team

Fig. 2
figure 2

Contrebandiers maxilla virtual reconstruction. a After virtual segmentation, the maxilla appeared to be well-preserved and nearly complete, only missing small sections around the right and left inferior orbital margins, the right zygomatic process and root, the left nasal margin around alare, and some of the posterior palatal roof. Reconstruction shown in anterior (left), posterior (middle), and inferior (right) views. b To achieve a more complete maxilla, the inferior orbital margin on the better-preserved left side was estimated (arrow) using the hole filling function in Geomagic Studio and the entire left maxilla was mirrored-imaged and aligned to the right side according to the preserved morphology and the midline palatal suture (gray original and blue virtual reconstruction). Shown in anterior (left and middle), and inferior (right) views. c The virtual reconstruction of the Contrebandiers maxilla used in this study shown in anterior (left), posterior (middle), and inferior (right) views

Minor virtual reconstruction was needed for most specimens in the comparative sample and described in detail in previous studies (Freidline et al., 2012a, 2012b, 2013; Hublin et al., 2017). The type of reconstruction varied considerably depending on the specimen, but generally included the filling of cracks or holes, removal of sediments, smoothing abraded areas, and refitting of fragments. For some fossils in which one side was missing or deformed, bilateral symmetry was exploited by mirror-imaging. All reconstructions were done by either SF or PG using Geomagic Studio 2014 v. 3.0 (Geomagic Inc., Rockhill) and Avizo v. 7.1.

Geometric Morphometrics and Multivariate Analyses

Geometric morphometric (GM) methods were used to analyze the shape and size of the Contrebandiers maxilla in a comparative context. Three-dimensional surface models were digitally reconstructed from either CT scans of the original fossils, using Avizo v. 7.1 (FEI Visualization Sciences Group, Hillboro), or when CT data was not available, surface scans (Minolta Vivid 910 or a Breuckman optoTOP-He) of high-quality casts (Table 1) (Freidline et al., 2012a, 2012b, 2013). For all individuals, 3D coordinates of landmarks (n = 13), curve-semilandmarks (n = 43), and surface-semilandmarks (n = 82) (Fig. 3 and Table 2) were collected from surface models using Landmark Editor v.3.0.0.6 (Wiley et al., 2005). A template mesh of surface semilandmarks covering the preserved maxilla were digitized on Contrebandiers and a thin-plate-spline (TPS) interpolation was used to warp this template to the surface of every other individual according to their landmark and curve data. Missing bilateral (semi)landmarks were estimated by mirroring the preserved side and missing (semi)landmarks that were lacking a bilateral counterpart were estimated by deforming the sample average onto the deficient configuration TPS interpolation (Freidline et al., 2012a, 2012b, 2013; Gunz et al., 2009b). Curve and surface semilandmarks were slid by minimizing the bending energy of a TPS deformation between each specimen and the sample mean shape (Gunz & Mitteroecker, 2013; Gunz et al., 2005). After sliding, all landmark and semilandmark datasets were symmetrized and converted to shape variables using a generalized Procrustes analysis (Rohlf & Slice, 1990). Centroid size, the square root of the sum of squared distances from each landmark to the specimen’s centroid, was used as a proxy for size (Dryden & Mardia, 1998).

Fig. 3
figure 3

Landmarks and semilandmarks used in the geometric morphometric analyses. (a) anterior maxilla and (b) inferior maxilla. Landmarks are red, curve semilandmarks blue, and surface semilandmarks yellow; landmarks are abbreviated in lower case and curves are capitalized. See Table 2 for landmark and curve and surface semilandmark definitions

Table 2 Landmarks, curves, and surface semilandmarks used in the analyses. Midline landmarks (i.e., non-bilateral) italicized

Principal component analyses (PCA) in shape and form space (including the natural logarithm of centroid size) (Mitteroecker et al., 2004) were performed and ontogenetic trajectories for Neanderthals and recent H. sapiens were visualized by combining the average differences between consecutive age groups (e.g., Bulygina et al., 2006; Coquerelle et al., 2011; Freidline et al., 2015; Neubauer et al., 2009, 2010). This approach accounts for the non-linearity that may be present during ontogeny.

To further explore ontogenetic allometry, multivariate regression analysis was used to predict the adult shape of the Contrebandiers maxilla. A linear ontogenetic allometric trajectory was calculated for early H. sapiens, recent H. sapiens, and Neanderthals by regressing the natural logarithm of centroid size on all Procrustes shape coordinates for each group. The Contrebandiers maxilla was grown along each of these trajectories to the size of Dar es Soltane II H5. We chose this fossil because it is an early H. sapiens of comparable chronological age. The juvenile Contrebandiers maxilla and the predicted adult shapes were then projected into both the shape and form space PCA plots. Shape changes were visualized along PC 1 and PC 2 by warping the sample mean shape along the positive and negative ends of PC 1 and PC 2, plus/minus two standard deviations from the sample mean. To evaluate developmental shape changes, the juvenile Contrebandiers maxilla was compared to the mean shapes of similarly aged Neanderthals and early and recent H. sapiens (i.e., AG 3) and the predicted adult shapes were compared to adult Neanderthals and to early and recent H. sapiens mean shapes.

Nearest neighbors were calculated according to inter-individual Procrustes distances to identify the individuals most similar to Contrebandiers. Maxillary size was evaluated by calculating the natural logarithm of centroid size for each individual and compared across groups. All data processing and analysis was done in RStudio, primarily using the Morpho (Schlager, 2017) and geomorph (Adams et al., 2021; Baken et al., 2021) packages.

Results

Morphological Description of the Contrebandiers Maxilla

The right and left sides of the maxilla (Fig. 2a) are nearly complete and preserve the following teeth: left and right deciduous canines, left and right deciduous third and fourth premolars, left and right permanent central incisors, right permanent lateral incisor (not shown in reconstruction), and left and right permanent first molars. The unerupted left and right permanent second molars are crown complete, with no root development. Based on dental development, this juvenile is considered to have been ca. 9 years at the time of death according to Western European standards (AlQahtani et al., 2010).

The right and left maxillary body and alveolar and frontal processes are largely intact. The bone was shattered into tiny fragments just inferior to the inferior orbital margin on both the left and right sides and around the right zygomatic process and root. The right nasal margin appears slightly warped with some damage along the nasal margin around alare. On the anterior maxilla, the anterior nasal spine is present. On the left body, some of the inferior infraorbital plate is present including the infraorbital foramen and the inferior extension of the canine fossa. The lacrimal groove is visible on both sides. The palatine process is largely intact, and the incisive canal is visible. No orbital surface is present and the maxillary tuberosities and posterior walls are missing exposing the maxillary sinuses.

Of taxonomic importance, the Contrebandiers maxilla exhibits some evidence of a canine fossa on the left side, the root of the zygomatic process is located just above the first molar, the inferior border of the zygomatic margin (IZM; also called zygomaticoalveolar crest) is curved and not straight, and when viewed anteriorly, the palate is strongly flexed upon the zygomatic process, presenting an incurvatio inframalaris frontalis along the transverse plane. Each of these features align the juvenile maxilla more to H. sapiens than to Neanderthals (Hublin, 1992; Minugh-Purvis, 1993; Pope, 1991).

Geometric Morphometric Analysis

The first two PCs represent approximately 55% of shape variance (Fig. 4). Subadults plot along the negative end of PC 1 and adults plot along the positive end. Principal component 2 separates Neanderthals and H. sapiens (early, Upper Paleolithic, and recent), suggesting that maxillary shape differences between the two groups are already present at birth. Contrebandiers falls within the recent H. sapiens hull, although it also falls close to the Neanderthal from Pech de l’Azé. Based on Procrustes distances its nearest neighbors are recent H. sapiens (Table 3).

Fig. 4
figure 4

Principal component analysis in shape space. Red triangles: Neanderthals; blue circles: Upper Paleolithic/Late Pleistocene H. sapiens; purple circles: early H. sapiens. Open shapes are non-adults and closed shapes are adults. The blue convex hull outlines the recent H. sapiens range of variation, and the red convex hull outlines the Neanderthal range of variation. Ontogenetic trajectories connect age group means, represented by squares (blue: recent H. sapiens and red: Neanderthals). The Neanderthal trajectory is missing AG 4 (dashed line). The yellow star is Contrebandiers, and black asterisks are its predicted adult shapes grown along early H. sapiens (CEHS), Neandertal (CN) and recent H. sapiens (CRHS) trajectories. Labels for fossils: Qafzeh 6 (Q6), 9 (Q9), 10 (Q10), 11 (Q11), and 15 (Q15); Laetoli Hominin 18 (LH18); Dar es Soltane II H5 (DS5); Jebel Irhoud 1 (Ir1); Skhul 5 (Sk5); Mezmaiskaya 1 (Mz); Le Moustier 2 (Mo2); Roc de Marsal (RM); La Quina 18 (LQ18); Teshik-Tash (TT); and Pech de l’Azé (Pe). Shape changes visualized along PC 1 and PC 2 by warping the sample mean shape along the positive and negative ends of PC 1 and PC 2, plus/minus two standard deviations from the sample mean

Table 3 The top five nearest neighbors according to inter-individual Procrustes distances (in parentheses) of the juvenile maxilla and the predicted adult shape grown along the recent H. sapiens, early H. sapiens, and Neanderthal ontogenetic trajectories

Maxillary shape changes along PC 1 are related to growth and development and mainly include growth in the supero-inferior and antero-posterior direction. Shape changes along the negative end of PC 2 are associated with increased maxillary height (supero-inferiorly), midfacial narrowing, infraorbital inflation, and nasal projection. These are typical Neanderthal features that are already present in the neonate from Mezmaiskaya. Specimens plotting at the positive end of PC 2, e.g., Dar es Soltane II H5, show the opposite pattern with a shorter, wider, and more flexed maxilla.

The predicted adult shapes resulting from the growth simulations are also projected into shape space. When grown along both an early H. sapiens (CEHS) and Neanderthal (CN) trajectory, Contrebandiers plots within the H. sapiens range of variation. When grown along the recent H. sapiens (CRHS) trajectory, it falls outside of the range of variation for both Neanderthals and H. sapiens along PC 1 and within H. sapiens variation along PC 2. The predicted adult shape of Contrebandiers is always nearest neighbors with early, Upper Paleolithic, and/or recent H. sapiens (Table 3), regardless of which trajectory it is grown along.

Developmental simulations indicate that postnatal shape changes between AG 3 and AG 5 (adulthood) do not appear to be significant, especially in early H. sapiens and Neanderthals (Fig. 5). The shape differences between Contrebandiers (AG 3) and its predicted adult shapes (AG 5) are minimal and the shape differences between them and the Neanderthal (red), early (purple) and recent (blue) H. sapiens mean shapes at AG 3 and AG 5 are consistent. That is, at both AG 3 and AG 5 Neanderthals have a larger and more prognathic nasal aperture, early H. sapiens have slightly more projecting lateral nasal aperture, and recent H. sapiens have a supero-inferiorly taller maxilla.

Fig. 5
figure 5

Developmental simulations. Left column: Contrebandiers (gray) superimposed on the mean Neanderthal (red), early H. sapiens (purple), and recent H. sapiens (blue) shapes at AG 3. Right column: Contrebandiers (gray) grown along the Neanderthal, early H. sapiens, and recent H. sapiens ontogenetic trajectories to adult size and superimposed on the mean Neanderthal (red), early H. sapiens (purple), and recent H. sapiens (blue) adult shapes (AG 5). All predicted and mean adult shapes were scaled to the size of Dar es Soltane II H5. Regions where color is visible or on top of the gray surface are shape differences between the comparison maxilla and Contrebandiers

In form space, PC 1 (87.3% of variance) is correlated with size (r = 0.99) and represents facial growth (Fig. 6). Principal component 2 (3.3% of variance), separates recent H. sapiens, several Upper Paleolithic H. sapiens individuals, and the predicted adult Contrebandiers following a recent H. sapiens trajectory (CRHS) from all other groups, including Contrebandiers. Recent H. sapiens and Neanderthal trajectories begin to diverge at around AG 2. The form changes along the positive end of PC 2 indicate that recent H. sapiens undergo unique growth in maxillary height (supero-inferiorly) especially in the frontal process. Principal component 3 (2.4% of variance) separates Neanderthals from H. sapiens, especially early in ontogeny, with form changes along this axis associated with midfacial projection and facial narrowing. Contrebandiers and its predicted adult shapes, except CRHS, tend to plot more with H. sapiens along PC 3. The size of the Contrebandiers maxilla is comparable to similarly aged (i.e., AG 3) early H. sapiens and Neanderthals, and much larger than recent similarly aged H. sapiens (Fig. 7).

Fig. 6
figure 6

Principal component analysis in form space. Red triangles: Neanderthals; blue circles: Upper Paleolithic/Late Pleistocene H. sapiens; purple circles: early H. sapiens. Open shapes are non-adults and closed shapes are adults. The blue convex hull outlines the recent H. sapiens range of variation, and the red convex hull outlines the Neanderthal range of variation. Ontogenetic trajectories connect age group means, represented by squares (blue: recent H. sapiens and red: Neanderthals). The Neanderthal trajectory is missing AG 4 (dashed line). The yellow star is Contrebandiers, and the black asterisks are its predicted adult shapes grown along early H. sapiens (CEHS), Neandertal (CN) and recent H. sapiens (CRHS) trajectories. Labels for fossils: Qafzeh 6 (Q6), 9 (Q9), 10 (Q10), 11 (Q11), and 15 (Q15); Laetoli Hominin 18 (LH18); Dar es Soltane II H5 (DS5); Jebel Irhoud 1 (Ir1); Skhul 5 (Sk5); Mezmaiskaya 1 (Mz); Le Moustier 2 (Mo2); Roc de Marsal (RM); La Quina 18 (LQ18); Teshik-Tash (TT); and Pech de l’Azé (Pe). Shape changes were visualized along PC 1, PC 2, and PC 3 by warping the sample mean shape along the positive and negative ends of each PC, plus/minus two standard deviations from the sample mean

Fig. 7
figure 7

Box plot of log maxillary centroid size for Neanderthals (N), recent H. sapiens (rHs), early H. sapiens (eHs), Upper Paleolithic/Late Pleistocene H. sapiens (UPHs), and Contrebandiers (C) according to age groups (AG). Labels for fossils: Qafzeh 10 (Q10), 11 (Q11), and 15 (Q15); Dar es Soltane II H5 (DS5); La Quina 18 (LQ18); and Teshik-Tash (T-T)

Discussion

Geometric morphometric analyses on the virtually reconstructed Contrebandiers juvenile maxilla reveal clear shape and size similarities to North African and Levantine early H. sapiens. The results of the GM analyses support previous studies linking North African MSA and Aterian craniomandibular and dental remains to Late Pleistocene Levantine fossils (Bergmann et al., 2022; Harvati & Hublin, 2012; Hublin et al., 2012; Röding et al., 2022). Results of our ontogenetic analyses support previous studies (e.g., Ackermann & Krovitz, 2002; Bastir & Rosas, 2004; Cobb & O'Higgins, 2004; Freidline et al., 2013) showing that the distinct maxillary shapes of Neanderthals and H. sapiens are present at birth and do not change substantially during development. Developmental simulations used to predict the adult maxillary shape of Contrebandiers suggest that its morphology would not change substantially into adulthood and even when grown along a Neanderthal trajectory its adult shape would be most similar to early (e.g., Jebel Irhoud 1 and Qafzeh 6) and later (Liujiang 1, Oberkassel 1, and Holocene) H. sapiens. Visualization of the developmental simulations demonstrate that similarly aged Neanderthals have a more projecting midface and nasal aperture compared to Contrebandiers and that these shape differences are maintained into adulthood. Interestingly, results also show that the maxillary development in our sample of recent H. sapiens undergoes greater ontogenetic changes in the superior-inferior dimensions compared to the Contrebandiers fossil, early H. sapiens, and Neanderthals.

The distinction between early H. sapiens and Neanderthal maxillary shape is less clear when size and shape are considered together (cf. Figure 6). Like the Qafzeh and Neanderthal juveniles, the Contrebandiers maxilla is large compared to the similarly aged recent H. sapiens in this study. This finding is consistent with studies on other juvenile early H. sapiens faces from Mugharet el’Aliya (Röding et al., 2022) and Herto BOU-VP-16/5 from Ethiopia (Zollikofer et al., 2022). Comparison of growth trajectories between Neanderthals and the recent H. sapiens in this study indicates a truncated growth in the latter beginning around adolescence (AG 4; cf. Figure 6). That is, the amount of maxillary growth in recent H. sapiens between adolescence and adulthood (AG 3 to 5) is less compared to Neanderthals, explaining their smaller facial size. In this respect, Contrebandiers and the Qafzeh juveniles differ from the growth pattern observed in recent humans. Given the similarly large facial sizes of the Mugharet el’Aliya and Herto maxillae, the growth pattern observed in Contrebandiers may be an ancestral one that is shared among early H. sapiens and Neanderthals. However, it is important to keep in mind that only one recent H. sapiens population was evaluated in this study. A study incorporating a larger recent human ontogenetic sample appears to support these results (Schuh et al., 2023) even though subtle population level differences in maxillary growth are known to exist (Freidline et al., 2012a, 2015; Schuh et al., 2020). When and where the human growth pattern evolved remains to be determined, but it may have been relatively recent (Williams, 2013). Like the Aterians, the Late Stone Age (LSA) Iberomaurusians from Taforalt, Morocco, and Afalou, Algeria, are characterized by megadonty, robust facial morphology, and large facial sizes. The morphological and shape similarities between the two groups suggest long-term regional continuity in North Africa (Bergmann et al., 2022; Ferembach, 1976a, 1976b). Future studies on facial growth and development incorporating the North African MSA mandibular remains (e.g., from Jebel Irhoud and Contrebandiers) and LSA Iberomaurusians, as well as expanding the Upper Paleolithic juvenile sample, will help elucidate the evolution of the recent human growth pattern. The results of this study and previous research on facial growth and development in early H. sapiens (e.g., Röding et al., 2022; Zollikofer et al., 2022) suggest a dissociation between the evolution of maxillary size and shape in later human evolution, by at least 300 ka in Jebel Irhoud (Richter et al., 2017). The Jebel Irhoud faces are large yet indistinguishable from recent H. sapiens using geometric morphometric analyses (Hublin et al., 2017). Large facial size in the MSA North Africans may be an ancestral retention or it may be the result of hybridization with archaic hominins (Ackermann et al., 2015; Harvati & Ackermann, 2022; Harvati & Roksandic, 2016; Stringer, 2016).The latter hypothesis is difficult to test without genetic evidence.

The overall morphological pattern in MSA North African faces suggests there was both regional continuity from the Early to Late Stone Age, as well as gene flow between H. sapiens groups living in other parts of Africa and the southwest Asia (Bergmann et al., 2022; Ferembach, 1976a, 1976b; Harvati & Hublin, 2012; Röding et al., 2022). These findings are consistent with the view that our species originated and diversified within subdivided populations that were connected by intermittent gene flow (Scerri et al., 2018; Stringer, 2016). The shared maxillary shape in MSA humans across Africa and the southwest Asia is likely a result of such gene flow. Positioned at the interface between the Mediterranean Basin and the Sahara, climatic shifts between hyper-arid and warm, humid (i.e., “green Sahara”) conditions during the Middle to Late Pleistocene likely facilitated population contractions and expansions back and forth between North Africa and the Levantine corridor, as well as into sub-Saharan Africa (Breeze et al., 2016; Drake et al., 2011; Larrasoana et al., 2013; Rosenberg et al., 2013).

Conclusion

This study addresses ontogenetic variability in the North African MSA, a period characterized by the origins, emergence, and dispersal of our species, but poorly understood because of the fragmentary and scant fossil record. The results of this study help shed light on when and where the evolution of our unique ontogenetic pattern evolved and provides further insight to the role of North Africa in the emergence and dispersal of our species within, and out of, Africa.

Geometric morphometric analyses on a virtual reconstruction of the juvenile Contrebandiers maxilla and simulations of its hypothesized adult form and ontogenetic pattern confirm its H. sapiens-like morphology and overall affinity with other North African MSA fossils and Late Pleistocene humans from Qafzeh. The growth pattern observed in the Contrebandiers and Qafzeh juveniles differs from recent humans and resembles that of Neanderthals. This may be an ancestral feature shared among all early H. sapiens. If so, it suggests that the changes to facial growth patterns we observe in Homo sapiens may have post-dated the MSA. Future studies on the juvenile North African MSA mandibles, including one from Contrebandiers, will shed additional light on the evolution of ontogeny in H. sapiens and help tease apart the development of the mosaic features characteristic of early H. sapiens facial remains (Bergmann et al., 2022).