Developmental simulation of the adult cranial morphology of Australopithecus sediba

AFFILIATIONS: 1Department of Anthropology, Texas A&M University, College Station, Texas, USA 2Evolutionary Studies Institute, School of Geosciences, University of the Witwatersrand, Johannesburg, South Africa 3Department of Wildlife and Fisheries Sciences, Texas A&M University, College Station, Texas, USA 4Department of Anthropology, University of Minnesota, Minneapolis, Minnesota, USA 5Department of Anthropology, Indiana University, Bloomington, Indiana, USA 6European Synchrotron Radiation Facility, Grenoble, France


Introduction
To date, only a single, relatively complete cranium of Australopithecus sediba has been recovered from the Malapa fossil site, belonging to the type specimen MH1. 1 Dating to 1.977±0.002Ma, 2 the Malapa hominins exhibit a unique, mosaic morphology, possessing features that align them with both the genus Homo as well as other species of australopith 1 .Based on this intermediate morphology, Berger et al. 1 suggested a possible ancestordescendant relationship between A. sediba and the genus Homo, with the possibility of A. sediba representing the direct ancestor to H. erectus, or otherwise a close sister group to that ancestor.The cranium itself shows remarkable preservation, possessing a complete facial skeleton and detailed surface anatomy with clearly visible suture lines.The second molars are in occlusion in this specimen, while the third molars remain in the crypt 1 , indicating its sub-adult status.With an estimated cranial capacity of 420 cm 3 , it is estimated that MH1 had achieved approximately 95% of its expected brain growth at age of death. 1,3 response to the announcement, several outside commentators immediately questioned the distinctiveness of A. sediba as a unique species separate from A. africanus. 4,55][6][7][8] Critics of the Berger et al. 1 interpretations argue that the degree of development expected to occur between second and third molar eruption would have been substantial enough to alter our current interpretations of the morphological affinities of A. sediba, especially those features thought to resemble later Homo.For example, Spoor 6 argued the possibility for increased constriction of the MH1 brain case had it reached full adulthood.More recently, Kimbel 8 criticised the use of the juvenile mandible in multivariate statistical comparisons with other species, based on the potential of continued growth and development for impacting linear measurements. 9 address this possibility, we used established geometric morphometric techniques [10][11][12][13][14][15][16] to produce 3D renderings of the inferred adult cranial morphology of A. sediba based on regression of developmental samples in accordance with dental eruption sequence, or dental stage.The goal of this procedure was to generate a series of developmental trajectories for both male and female extant apes and humans (Table 1), and then apply these trajectories to MH1 to create a series of virtual adult crania.This procedure, in turn, allowed us to empirically establish the estimated adult form of A. sediba, and test whether or not current interpretations regarding the cranial morphology of this species should be modified as a result of future developmental changes.

Reconstruction
A reconstruction of the MH1 cranium was carried out with the goal of correcting for distortion in the cranium and producing a more complete rendered model of the A. sediba skull (Figure 1a,b).Rapidform ® software (now 3D Systems Geomagic; see http://www.rapidform.com/home) was employed in this reconstruction to refine and process 3D models.The original 3D model of the MH1 cranium employed in this reconstruction was created using synchrotron image data generated on the beamline ID17 at the European Synchrotron Radiation Facility located in Grenoble, France. 17Synchrotron image data were segmented using Avizo 6.3 ® software, resulting in a 3D virtual rendering of the MH1 cranium, removed from the encasing breccia. 17This imaging allowed for the collection of landmark data in areas that were previously obscured by matrix.When examining the juvenile cranium, several preservation issues impacting the integrity of the fossil were apparent.The most prominent among these is a large crack, originating at the left supraorbital torus, which runs posteromedially across the frontal, widening as it continues to bregma to reach a maximum breadth of approximately 7 mm.An additional crack affecting landmark placement extends from the medial margin of the right orbit, inferomedially across the frontal process of the maxilla, breaking across the nasal bridge.Best viewed from frontal perspective, the crack obscures the right frontomaxillary suture and laterally displaces the inferior portion of both nasals.
Entire cranial bones were also displaced.As discussed above, the large crack extending across the frontal has resulted in the displacement of the left portion of the frontal bone.The displacement extended laterally from above the left orbit to the articulation with the zygomatic bone.The zygomatic bone was displaced posteroinferiorly, disarticulating the bone from the zygomatic arch of the temporal and the zygomatic process of the frontal.This distortion can also be noted from the frontal aspect when examining the inferior margin of the left orbit.Reconstruction correcting for the cracks and displacements as just described was a major focus of the present project.An additional goal was to produce a more complete calvaria by reflecting the preserved portions of the left parietal and temporal bones.Fortunately, any plastic deformation was deemed extremely minor, if present at all.The 3D model of the MH1 cranium was first imported into Rapidform ® , and all adjustments were conducted in 'mesh mode'.It was decided that the most efficient and effective way to account for cracks and displacements was by selecting the affected areas that required adjustment, copying and pasting these regions into a separate window, and then re-aligning the selected area with the original model.The selected areas were reoriented using the 'scan tools' property and 'align between scan data'.Using these tools, one selects a reference scan and a moving scan, which are then aligned in accordance with selected analogous points that serve to stitch the scans together.
To correct for the large crack across the frontal bone, we first selected the preserved portions of the frontal, parietal, sphenoid and temporal bones extending to the left of the crack.These portions were then copied into a separate window and selected as the moving scan to align with the remaining portion of the cranium using the 'align between scan data' tool.Once the crack had been corrected, we were then able to move and realign the zygomatic using the same process, in which the zygomatic bone was removed and then rearticulated with the zygomaticotemporal and zygomaticofrontal sutures.The portion of the nasal bones laterally displaced inferior to the break was additionally removed and realigned using the same process used for correcting the zygomatic and frontal bones.
The mirror tool was used to correct for the distortion along the nasal bridge, by reflecting the left side of the superior portion of the nasal bridge to remove the crack across the frontomaxillary suture.After correcting for displacement and distortion in the cranium, the mirror tool was used to reflect the left half of the calvaria posterior to the coronal suture to produce a more complete calvaria.As a final step in the reconstruction, the scans from the corrected model were then merged into a single scan and the resulting model was run through 'global remesh'.This command regenerates the mesh structure with removed defects in accordance with the model's curvature flow.A final product is shown in Figure 1b.

Sample
A summary of the hominoid comparative samples -comprising male and female chimpanzees (Pan troglodytes), gorillas (Gorilla gorilla gorilla) and modern humans (Homo sapiens) -is listed in Table 1.Ape data were collected from wild-shot specimens housed at the Cleveland Museum of Natural History (Cleveland, OH, USA).Records for chimpanzees indicate that specimens were collected in Ebolwa, Cameroon, as well as Abong Mbong, French Cameroons, and Abong Mbong, Djaposten, Cameroons, West A, with the exception of three specimens for which no geographical data are available.The records for the gorilla crania used in the comparative sample indicate that specimens were collected in French West Africa, French Congo, Ebolwa, Cameroon, Abong Mbong, French Cameroons, and Abong Mbong, Djaposten, Cameroons, West A. Both chimpanzee and gorilla crania were sexed using cranial remains.The human sample primarily included cadaver crania obtained from the Raymond A. Dart Collection of Human Skeletons at the School of Anatomical Sciences at the University of the Witwatersrand (Johannesburg, South Africa), although several wellpreserved archaeological crania were included from both the Raymond A. Dart Collection of Human Skeletons and the Cleveland Museum of Natural History.Specimens with obvious pathologies or abnormalities were excluded from the study sample.Archaeological crania were only used if they could be confidently sexed using available records or standard cranial sexing criteria.
Specimens were assigned to developmental categories based on dental eruption sequence.Sub-adults were judged to be of the same developmental age as MH1, and therefore suitable for inclusion within the study, if the second molars were erupted and in occlusion, while the third molars had not yet erupted.Specimens were designated as adults if the third molars were erupted and in occlusion.Both male and female specimens were sampled for parity, although the juvenile sample was pooled for the purpose of developmental simulation.9][20][21] Results of this research indicate that differences in craniofacial morphology of great apes and humans is established early in ontogeny. 20Research further identifies the effects of allometric scaling, as well as disparities in growth rates and growth spurts of male and female hominoids as considerable influences in the onset and resulting degree of sexual dimorphism among species. 18,19,21However, as a result of both the limited number of juveniles, and the fact that sex was often unknown for these specimens, we maintain that pooling of the juvenile sample was appropriate in the present study.We further note that previous studies employing developmental simulation have used pooled juvenile samples for similar reasons. 13e hominin fossil sample used in geometric morphometric comparative analyses included a sample of non-'robust' australopith and early http://www.sajs.co.za

Methods
The ontogenetic samples for extant chimpanzees, gorillas and humans and fossil hominin crania were collected using a NextEngine 3D laser scanner ® .Surface scans were stitched together and fused using Scanstudio HD Pro software ® .These scans were later imported into Geomagic ® where they were smoothed and polished using the 'mesh doctor' tool.'Mesh doctor' provides an automatic polygon mesh improvement tool, which both detects errors in the mesh and corrects them.All holes in the mesh were filled-in using Geomagic ® to prevent any landmarks or semi-landmarks from 'falling through' the mesh during landmark placement.
After fusing and polishing, each scan was imported into Landmark Editor 3.6 ®22 software where 76 traditional landmarks and semi-landmarks were then placed (Table 2).Three semi-landmark curves were placed along the mid-sagittal plane between rhinion and bregma.These landmarks were not slid, but rather served as traditional landmarks.Landmark files for each hominoid species were exported to MorphoJ ®23 software where they were aligned through generalised Procrustes analysis.][26][27] Estimated adult MH1 crania produced from each ontogenetic trajectory (male and female chimpanzee, gorilla and human) were visualised in Landmark Editor 3.6 ® software 22   Note: In addition to the landmarks listed here, three curves of semi-landmarks were added along the mid-sagittal curve between rhinion and bregma, with a density of 10 equidistant spaced semi-landmarks each, making for a total of 76 landmarks.Curve one begins at rhinion and ends at the posterior junction of the glabella and the slope of the frontal bone.Curve two begins at this point and ends at the midpoint along the mid-sagittal curve of the frontal bone.Curve three begins at this midpoint on the frontal bone and ends at bregma.
A subset of 32 landmarks across the calvaria, face and palate of specimens was then used for morphometric comparisons between hominin crania and simulated versions of A. sediba individuals (Table 3).Producing this subset was necessary for comparing extant taxa landmarks to landmarks obtainable on the occasionally incomplete fossil specimens selected for analysis.In order to include specimens Sts 71 and KNM-WT 15000, missing landmarks were estimated through reflection of antimeres.Principal component analysis (PCA) of covariances was conducted on the Procrustes shape coordinates using the subset of 32 landmarks collected from the extant hominoid and fossil sample.The adult gorilla sample was excluded from the PCA because of concerns that, if included, these specimens would largely distort the results as a result of their more divergent morphology compared to other species included in the sample. 18,21,28Analyses were conducted both with and without the extant hominoids to control for the likely dominance of these samples in driving definition of the major principal component axes.In other words, one could ask, to what extent was the position of the fossil specimens affected by the ordination of modern chimpanzees and humans in the same data space?To control for this effect, we also conducted a PCA using only fossil specimens to assess the distribution of crania in the context of fossil hominin variation.
Average Procrustes chord distances both within and between taxa were also calculated using Excel software on the subset of 32 landmarks for the same chimpanzee, human and fossil sample that was used in the PCA.The purpose of this test was to determine if the overall distance within the A. sediba sample (i.e.intragroup variation), including all simulated adults and the reconstructed juvenile cranium, exceeds that observed in extant hominoid species.

Developmental simulations
Qualitative assessment of the simulated adult crania indicated that the majority of morphological changes expected to occur between second and third molar eruption are related to the development of secondary sexual characteristics (Figure 2a,b).The most significant changes were observed along the male gorilla developmental vector, for which we see the glabella becoming more pronounced, the supraorbital torus thickening, and the zygomatic increasing in its superior-inferior dimension.Increased horizontal angulation, or bending across the midface, is also apparent, as is increased lower facial prognathism.Similar transformations were also observed, although to a lesser extent, along the male chimpanzee developmental vector, where one again observes development of moderate lower facial prognathism and a general enhancement of facial robusticity, such as a slight thickening of the supraorbital torus and glabella, as well as a more rugged appearance in the zygomatics.The magnitude of morphological change observed for the female developmental vectors of the gorilla and chimpanzee simulations was comparatively much less.Changes associated with both the male and female human developmental vectors were minimal, producing no notable deviations from the original juvenile form.Thus, while the choice of sex and species did result in observable differences in the estimated adult form of MH1 when applying the chimpanzee and gorilla developmental vectors, applying the human developmental vectors to MH1 resulted in relatively more modest variation in adult simulated form.Given that A. sediba evinces a pattern of canine size dimorphism and facial robusticity similar to other australopiths and to early Homo, 1,9 we consider the gorilla developmental pattern, in particular that of the male gorilla, to be the least suitable model, while chimpanzee or human developmental patterns would likely provide better estimates of the onset of secondary sexual characteristics in MH1.

Morphometric comparisons
Results of the PCA are illustrated in Figure 3 using adult chimpanzee, human and non-robust hominin crania, including MH1 and its simulated adult conformations.Visualisations of principal component shape change for the first three principal component axes are provided in Figure 4.The first axis of variation is dominated by differences in the length and orientation of the frontal bone, in accordance with the high loading for the bregma landmark, whereas the second axis also summarises changes in cheek morphology, with specimens separated based on the overall gracility or robusticity of the zygomatic.This latter interpretation is based on the high loadings for zygomaxillary landmarks as well as the relative distribution of specimens.The third axis is dominated by differences in morphology of the anterior and posterior palate, with the highest loadings being observed for landmarks in this region.These interpretations correspond to the visualisations of shape change for each axis provided in Figure 4.All six simulated adult A. sediba crania -regardless of the species or sex used to estimate them -fall out along both axes with the original, juvenile MH1 cranium in a discrete group relative to other apes or fossil taxa in the analysis (Figure 3a).Early Homo crania tend to cluster together with the exception of KNM-ER 1470, which appears to cluster with Sts 5 on the first two major axes (Figure 3a).However, this proximity vanishes when we examine the third principal component, wherein KNM-ER 1470 plots as an outlier (Figure 3b).In addition, considering component 2 versus component 3 (Figure 3b), KNM-ER 1813, OH 24, and the Dmanisi specimen D2700 are separated from H. erectus sensu lato crania KNM-WT 15000 and KNM-ER 3733, instead plotting near the cluster of simulated A. sediba crania.One outlier to this latter group is the A. sediba cranium simulated using a male gorilla developmental vector, which plots in this space nearest the A. africanus cranium Stw 53.The exceptional location for the MH1 adult version simulated from male gorillas is not entirely surprising, given that one observes the greatest degree of morphological change when applying this vector, although again we consider a male gorilla-like pattern of development to be the least likely for A. sediba.Of perhaps more interest is the fact that a male chimpanzee developmental trajectory makes MH1 appear more similar to early Homo and modern humans.
A second PCA on the fossil sample, excluding chimpanzees and humans, is illustrated in Figure 5, with visualisations of shape change for the first three principal component axes provided in Figure 6.Similar to the PCA including chimpanzees and humans, the first axis of variation is dominated by changes in morphology of the frontal bone, but is also influenced by morphological variation of the posterior palate, likely related to the relative degree of facial prognathism among specimens.
The second axis is also largely dominated by the morphology of the posterior palate, with high loadings for landmarks in this region.While the A. sediba crania generated using chimpanzee and modern human developmental vectors continue to cluster closely to one another, the simulated adult produced from the male gorilla developmental vector shows dissimilarity relative to other A. sediba specimens, plotting nearest A. africanus specimen Sts 71 (Figure 5a).
As the third principal component also accounts for a substantial amount of the total variation (15.2%) in the PCA using only fossils, we examined the distribution of fossil taxa along this axis.This third axis of variation was also influenced primarily by changes in the frontal bone and palate.When comparing principal component 2 versus component 3, all A. sediba estimated adults, along with the original juvenile cranium, clustered together, separated from A. africanus and other hominin taxa along the third axis of variation (Figure 5b).
The results of the above PCA are supported across the full shape space by an analysis of shape differences within and among taxa using average Procrustes distances (Table 4).Examination of the Procrustes chord distances indicates that the within-species variation for A. sediba simulated adults and the juvenile cranium is actually less than that of the modern human and chimpanzee samples used in the study.

Conclusions
The results of this study support the hypothesis that the expected degree and nature of development that would have occurred between second and third molar eruption in MH1 would not have been substantial enough relative to other hominin taxa to alter our current craniofacial understanding of the phylogenetic relationships of A. sediba.In other words, had MH1 lived to adulthood, its craniofacial morphology would look similar enough to its present, juvenile cranial morphology that we can reliably differentiate the taxon using the cranium of this particular specimen.Any future changes would likely be related to the onset of puberty, and the concomitant development of secondary sexual characteristics.As a result, this study further reinforces the importance of separating developmental vectors by sex to avoid obscuring this type of variation in ontogenetic projections.
The fact that the juvenile MH1 cranium continues to cluster with simulated A. sediba adults in PCAs conducted in this study reinforces the above position.While we cannot ultimately rule out the scenario that qualitative apomorphies shared with other fossil taxa (e.g. A. africanus) might have developed and become expressed only during this later stage of ontogeny, the expression of such features in developmentally much younger fossil hominin specimens 29,30 argues against the likelihood of this possibility.For example, anterior pillars are already present and easily observable in the Taung child, for which only the first molars are in occlusion. 30Therefore, the late juvenile status of the type specimen, MH1, would not likely influence our diagnosis of this species.These traits on the reconstructed cranium of MH1 and its ontogenetic simulated extrapolations concur with those for the mandibles of MH1 and MH2. 9 Both studies indicate the verity of A. sediba as a unique taxon.Based upon our current understanding of the postcranial anatomy of the Malapa hominins, which indicates an Australopithecus-level adaptive grade, the Homo-like features observed in the simulated adults generated when applying the H. sapiens and P. troglodytes trajectories reinforce the transitional nature of this species.

Figure 2 :
Figure 2: Visualisations of MH1 and its simulated adult forms from hominoid developmental trajectories in (a) frontal and (b) lateral aspects.

Table 1 :
Summary of hominoid sample used to create developmental vectors http://www.sajs.co.za Volume 112 | Number 7/8 July/August 2016 Sts 71, Stw 53), H. habilis (OH 24, KNM-ER 1813), H. rudolfensis (KNM-ER 1470) and H. erectus (D2700, KNM-WT 15000, KNM-ER 3733).Laser surface scans of A. africanus fossils were collected from original fossil material at the Ditsong Museum of Natural History in Pretoria, South Africa.The original OH 24 fossil, referred to as H. habilis, was scanned at the National Museum of Tanzania in Dar es Salaam.Scans from the Dmanisi and Kenyan fossil material were obtained from casts, as the original fossil material was not available.

Table 2 :
Landmark definitions for landmarks used in developmental simulation

Table 3 :
Landmark definitions for the subset of landmarks used in morpho metric comparison

Table 4 :
Average Procrustes chord distances within taxa (on diagonal) and among taxa (off diagonal)Bolded numbers indicate average within species variation.The range of variation within P. troglodytes is 0.0597-0.14093.The range of variation within H. sapiens is 0.06654-0.12589.Distances were calculated using Excel software for the subset of 32 landmarks collected from the adult P. troglodytes, H. sapiens and fossil sample.