The relative congruence of cranial and genetic estimates of hominoid taxon relationships: Implications for the reconstruction of hominin phylogeny Noreen von Cramon-Taubadel a, * , Heather F. Smith b, c a Department of Anthropology, School of Anthropology and Conservation, University of Kent, Marlowe Building, Canterbury CT2 7NR, UK b Department of Anatomy, Arizona College of Osteopathic Medicine, Midwestern University, USA c School of Human Evolution and Social Change, Arizona State University, USA article info Article history: Received 4 November 2011 Accepted 27 February 2012 Available online 17 April 2012 Keywords: Geometric morphometrics Cranial regions Genetic distance Hominoidea abstract Previous analyses of extant catarrhine craniodental morphology have often failed to recover their molecular relationships, casting doubt on the accuracy of hominin phylogenies based on anatomical data. However, on the basis of genetic, morphometric and environmental affinity patterns, a growing body of literature has demonstrated that particular aspects of cranial morphology are remarkably reliable proxies for neutral modern human population history. Hence, it is important to test whether these intra-specific patterns can be extrapolated to a broader primate taxon level such that inference rules for understanding the morphological evolution of the extinct hominins may be devised. Here, we use a matrix of molecular distances between 15 hominoid taxa to test the genetic congruence of 14 craniomandibular regions, defined and morphometrically delineated on the basis of previous modern human analyses. This methodology allowed us to test directly whether the cranial regions found to be reliable indicators of population history were also more reliable proxies for hominoid genetic relationships. Cranial regions were defined on the basis of three criteria: developmental-functional units, individual bones, and regions differentially affected by masticatory stress. The results found that all regions tested were significantly and strongly correlated with the molecular matrix. However, the modern human predictions regarding the relative congruence of particular regions did not hold true, as the face was statistically the most reliable indicator of hominoid genetic distances, as opposed to the vault or basicranium. Moreover, when modern humans were removed from the analysis, all cranial regions improved in their genetic congruence, suggesting that it is the inclusion of morphologically-derived humans that has the largest effect on incongruence between morphological and molecular estimates of hominoid relationships. Therefore, it may be necessary to focus on smaller intra-generic taxonomic levels to more fully under- stand the effects of neutral and selective evolutionary processes in generating morphological diversity patterns. Ó 2012 Elsevier Ltd. All rights reserved. Introduction Reconstructing hominin phylogeny from cranial morphology The reconstruction of hominin phylogeny is necessarily cen- tered on the description and interpretation of fossilized anatomical remains. With the exception of relatively recent Homo paleospecies, where the retrieval of ancient DNA (e.g., Krings et al., 1997 , 2000; Noonan et al., 2006; Green et al., 2006, 2010; Krause et al., 2010) allows for an additional source of information regarding past evolutionary processes, the majority of hominin biological evolu- tion is codified only in the fossilized morphological record. However, numerous cladistic attempts to reconstruct the phylogeny of hominins (e.g., Chamberlain and Wood, 1987; Skelton and McHenry, 1992; Lieberman et al., 1996; Strait et al., 1997; Strait and Grine, 1999) failed to reach unanimous conclusions regarding the evolutionary relationships amongst hominin taxa (e.g., Wood and Collard, 1999). To investigate the possible methodological or evolutionary reasons for this lack of confidence, several studies focused on the phylogenetic reconstruction of catarrhine primate taxa for which the molecular relationships are well understood (e.g., Hartman, 1988; Pilbeam, 1996; Collard and Wood, 2000, 2007; Lycett and Collard, 2005). The results of these analyses found that * Corresponding author. E-mail address: [email protected](N. von Cramon-Taubadel). Contents lists available at SciVerse ScienceDirect Journal of Human Evolution journal homepage: www.elsevier.com/locate/jhevol 0047-2484/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.jhevol.2012.02.007 Journal of Human Evolution 62 (2012) 640e653
The relative congruence of cranial and genetic estimates of hominoid taxonrelationships: Implications for the reconstruction of hominin phylogeny
Noreen von Cramon-Taubadel a,*, Heather F. Smith b,c
aDepartment of Anthropology, School of Anthropology and Conservation, University of Kent, Marlowe Building, Canterbury CT2 7NR, UKbDepartment of Anatomy, Arizona College of Osteopathic Medicine, Midwestern University, USAc School of Human Evolution and Social Change, Arizona State University, USA
a r t i c l e i n f o
Article history:Received 4 November 2011Accepted 27 February 2012Available online 17 April 2012
0047-2484/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.jhevol.2012.02.007
a b s t r a c t
Previous analyses of extant catarrhine craniodental morphology have often failed to recover theirmolecular relationships, casting doubt on the accuracy of hominin phylogenies based on anatomical data.However, on the basis of genetic, morphometric and environmental affinity patterns, a growing body ofliterature has demonstrated that particular aspects of cranial morphology are remarkably reliable proxiesfor neutral modern human population history. Hence, it is important to test whether these intra-specificpatterns can be extrapolated to a broader primate taxon level such that inference rules for understandingthe morphological evolution of the extinct hominins may be devised. Here, we use a matrix of moleculardistances between 15 hominoid taxa to test the genetic congruence of 14 craniomandibular regions,defined and morphometrically delineated on the basis of previous modern human analyses. Thismethodology allowed us to test directly whether the cranial regions found to be reliable indicators ofpopulation history were also more reliable proxies for hominoid genetic relationships. Cranial regionswere defined on the basis of three criteria: developmental-functional units, individual bones, and regionsdifferentially affected by masticatory stress. The results found that all regions tested were significantlyand strongly correlated with the molecular matrix. However, the modern human predictions regardingthe relative congruence of particular regions did not hold true, as the face was statistically the mostreliable indicator of hominoid genetic distances, as opposed to the vault or basicranium. Moreover, whenmodern humans were removed from the analysis, all cranial regions improved in their geneticcongruence, suggesting that it is the inclusion of morphologically-derived humans that has the largesteffect on incongruence between morphological and molecular estimates of hominoid relationships.Therefore, it may be necessary to focus on smaller intra-generic taxonomic levels to more fully under-stand the effects of neutral and selective evolutionary processes in generating morphological diversitypatterns.
� 2012 Elsevier Ltd. All rights reserved.
Introduction
Reconstructing hominin phylogeny from cranial morphology
The reconstruction of hominin phylogeny is necessarily cen-tered on the description and interpretation of fossilized anatomicalremains.With the exception of relatively recentHomo paleospecies,where the retrieval of ancient DNA (e.g., Krings et al., 1997, 2000;Noonan et al., 2006; Green et al., 2006, 2010; Krause et al., 2010)
Cramon-Taubadel).
All rights reserved.
allows for an additional source of information regarding pastevolutionary processes, the majority of hominin biological evolu-tion is codified only in the fossilized morphological record.However, numerous cladistic attempts to reconstruct thephylogeny of hominins (e.g., Chamberlain and Wood, 1987; Skeltonand McHenry, 1992; Lieberman et al., 1996; Strait et al., 1997; Straitand Grine, 1999) failed to reach unanimous conclusions regardingthe evolutionary relationships amongst hominin taxa (e.g., Woodand Collard, 1999). To investigate the possible methodological orevolutionary reasons for this lack of confidence, several studiesfocused on the phylogenetic reconstruction of catarrhine primatetaxa for which the molecular relationships are well understood(e.g., Hartman,1988; Pilbeam,1996; Collard andWood, 2000, 2007;Lycett and Collard, 2005). The results of these analyses found that
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653 641
morphological data could not recover the precise genetic rela-tionships of the primate taxa, suggesting that phylogeneticschemes for the fossil hominins based on craniodental charactersmust also be called into question. Strait and Grine (2004) demon-strated that the inclusion of hominin taxa facilitated the recovery ofthe correct molecular relationship among the extant hominoids,suggesting that phylogenetic reconstruction may be influencedmore by the taxon sample included than the morphometric char-acters employed. Nevertheless, the general lack of congruencebetween molecular and morphological estimates of primatephylogenies has been primarily attributed to the presence ofmorphological homoplasies (i.e., phylogenetically-misleadingsimilarities), caused by evolutionary factors such as parallel/convergent evolution, character reversals, allometric effects (e.g.,Gilbert and Rossie, 2007; Gilbert et al., 2009), and non-heritableenvironmental influences such as osseous remodeling in responseto strain (e.g., Lieberman, 1995; Lockwood and Fleagle, 1999;Collard and Wood, 2000; Collard and O’Higgins, 2001; Lycett andCollard, 2005).
Subsequent studies suggested that particular regions of theprimate cranium, such as the temporal bone and the basicranium,may be particularly immune to homoplasy due to their relativeanatomical complexity, thus limiting the direct influence of anysingle selective factors (e.g., Lockwood et al., 2004), and/or theirimmunity to remodeling in response to biomechanical stimuli (e.g.,Lieberman et al., 1996). On this basis, it was suggested that onlyspecific anatomical regions, such as the temporal bone (e.g.,Lockwood et al., 2002, 2004), could yield accurate informationregarding past evolutionary processes. The putative link betweenthe impact of environmental influences on osseous morphologyand the phylogenetic usefulness of osseous characters was formu-lated as the ‘homoiology hypothesis’, where a homoiology isa homoplasy resulting from plasticity or similarities ingenotypeeenvironment interaction (Lieberman, 1995, 1997, 1999,2000; Lieberman et al., 1996; Collard and Wood, 2000, 2001, 2007;Lycett and Collard, 2005; von Cramon-Taubadel, 2009b). However,when the homoiology hypothesis was tested in hominoids (Collardand Wood, 2007) and papionins (Lycett and Collard, 2005), it wasfound that highly strained (and significantly more variable) regionswere not necessarily worse in terms of reconstructing knownphylogenies, especially compared with dental traits that cannotremodel in response to non-genetic stimuli.
It has also been suggested (e.g., Lockwood et al., 2004; Harvatiand Weaver, 2006a, b; Smith et al., 2007) that the quantificationof anatomical regions via 3D geometric morphometrics would yielda more accurate representation of shape variation and thusenhance phylogenetic utility, compared with the traditionalmorphometric methods used in previous cladistic analyses. Arecent study by Bjarnason et al. (2011) highlighted the multitude ofmethodological issues that can plague the accurate reconstructionof hominoid molecular relationships from morphological data (seealso Nadal-Roberts and Collard, 2005). In particular, the choice ofoutgroup can have a considerable effect, as the use of a Hylobatesoutgroup failed to recover the correct Pan-Homo clade (Bjarnasonet al., 2011), in contrast with the results of Lockwood et al. (2004)where the correct Pan-Homo clade was recovered when the treewas rooted using Pongo. In addition, the use of different cranio-metric measurements or characters provided different conclusions,as did the choice of phylogenetic reconstruction technique (i.e.,cladistic versus distance-based phenetic methods). What a reviewof the existing literature suggests, therefore, is that there areseveral potentially confounding factors (e.g., taxa included,methods employed, anatomical regions used etc.) working simul-taneously in creating incongruencies between molecular andmorphological estimates of past evolutionary processes.
Comparing the phylogenetic utility of anatomical regions in modernhumans
In recent years, a number of studies have specifically focused onunderstanding the microevolutionary history of different cranialregions in modern human populations, where detailed data onmorphological, genetic, and environmental variability are available(e.g., Roseman, 2004; Harvati and Weaver, 2006a, b; Betti et al.,2009; Smith, 2009, 2011; von Cramon-Taubadel, 2009a, 2011a).The results of these studies have been remarkably consistent,despite the use of different anatomical datasets, populationsamples and methodologies (Roseman and Weaver, 2007; vonCramon-Taubadel and Weaver, 2009). The comparison of molec-ular and morphometric population affinity patterns (Roseman,2004; Harvati and Weaver, 2006a, b; Smith, 2009; von Cramon-Taubadel, 2009a, 2011a) overwhelmingly suggests that themajority of human craniometric variation can be explained bya neutral model of mutation, genetic drift and gene flow (see alsoRelethford, 1994, 2002, 2004; Roseman and Weaver, 2004; Weaverand Roseman, 2008).
What the results of the modern human analyses suggest,therefore, is that specific aspects of morphology are acting moreneutrally than others, and thus make more reliable indicators ofpast (neutral) population history. Moreover, the modern humanstudies have allowed the teasing apart of neutral and adaptiveevolutionary processes in creating anatomical diversity patterns.However, this level of information is currently only available at theintra-specific level for Homo sapiens, and it is therefore unclear howthe results of modern human population studies relate to patternsof morphological diversity in the hominin fossil record. Hence, it isnecessary to scale analyses of extant molecular and morphologicalcongruence to a larger primate taxonomic level in order to assessthe extent to which the conclusions of modern human studiesmight be extrapolated to understanding the relationships of extincthominin paleospecies.
Cardini and Elton (2008) used 3D geometric morphometrics toinvestigate the extent to which different cranial regions (Func-tional-Developmental Modules) could recover the molecular rela-tionships of the guenons. They found that the chondrocranium(cranial base) was consistently the most reliable for recovering themolecular phylogeny, compared with other regions such as thevault, face, mandible, and masticatory regions, including thezygomatic. However, with the exception of this study on guenons,little research has focused on assessing the phylogenetic utility oflandmark-based morphometric data describing specific cranialregions in nonhuman catarrhine primates. Moreover, no study hastested the relative genetic congruence of all of the cranial regionstested in modern humans in another wider primate group. To buildan accurate inference model for the human fossil record, it isimportant to ascertain whether there are specific cranial regionsthat reflect genetic relationships among primates in general,regardless of the taxonomic scale involved (intra-specific, intra-generic and inter-generic). A system of phylogenetic bracketing(sensu Witmer, 1995) is useful in this regard because if the samegenetic-craniometric congruence patterns were found to bepresent both within humans and among closely related primatespecies, then it would be reasonable to expect these patterns tocharacterize the intervening hominin taxa. This would also furnishinsights into the developmental, functional or anatomical criteriathat best describe phylogenetically-informative anatomicalregions, such that more general assumptions regarding the evolu-tionary processes affecting different aspects of morphology mightbe made. Here, we chose to focus on the analysis of hominoid taxa,as these represent the closest phylogenetic comparison for thehominin lineage. Samples were chosen to represent the inter-
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653642
generic, inter-specific and intra-specific taxonomic diversity of thehominoids, with all major groups and genera represented. Theinclusion of this level of taxonomic diversity provides a moreaccurate representation of the levels of morphometric diversityfound in the hominin fossil record.
In this study, we statistically compare the relative congruenceof different cranial regions against consensus molecular geneticdata for hominoid taxa. There are many ways inwhich the craniummight be sub-divided and the criteria for doing so must beexplicitly stated. Hence, the criteria used here follow those used inprevious modern human analyses (e.g., Smith, 2009, 2011; vonCramon-Taubadel, 2009a, 2011a) to allow for predictions madeon the basis of these analyses to be tested directly using thehominoid taxa. Also the individual landmark configurationsemployed here were designed on the basis of the previous modernhuman analyses, such that not only the same homologous regionscould be compared, but that the manner in which these regionswere captured morphometrically would also be directly compa-rable. Hence, it is possible to make predictions regarding therelative genetic congruence of individual regions of the primatecranium, defined according to several anatomical, functional, anddevelopmental criteria. Our null hypothesis in each case was thathominoid genetic-morphometric congruence patterns wouldmatch the predictions made on the basis of the modern humanstudies.
Criteria for delineating cranial regions and predictions based onmodern human analyses
Functional-developmental: regions defined on the basis ofossification and function ‘Developmental’ is taken here to meanthe individual regions of the primate cranium that have differingphylogenetic origins as characterized by differing modes of ossifi-cation (endochondral versus intramembranous), in contrast withalternative concepts of development such as differing germ layerorigins or modes of growth (e.g., McBratney-Owen et al., 2008;von Cramon-Taubadel, 2011a). This concept separates theendochondrally-ossifying chondrocranium (cranial base) fromthe intramembranously-ossifying cranial vault and face(dermatocranium). Overlaying the developmental concept is theconcept of ‘function’, which is the intuitive notion that differentregions of the skull have differing but not necessarily uniquefunctions (Moss and Young, 1960). Here we use this concept of‘function’ to separate the dermatocranium into the vault and face,which can be assumed to have broadly differing functions ofhousing the brain case, and mastication plus sensory function,respectively (see also von Cramon-Taubadel, 2011a).
Harvati and Weaver (2006a, b) found that vault shape, but notfacial shape was significantly correlated with genetic distance,while facial shape appeared to correlate with climate, especiallywhen cold-adapted populations were included. Smith (2009) foundthat the upper face and the basicranium were correlated withneutral genetic patterns in modern humans but not the vault.However, it is likely that this was the outcome of having fewer,less evenly placed vault landmarks (Smith, 2009). Recently,von Cramon-Taubadel (2011a) tested the three functional-developmental regions of chondrocranium, vault and face withthe aim of statistically assessing the relative neutrality of the threeregions, and found that all three regions were significantly corre-latedwith neutral genetics and statistically indistinguishable in thisregard. It is likely that the ‘face’ results of Harvati and Weaver(2006a, b) are best explained by thermoregulatory adaptation infacial shape of certain cold-adapted populations (Franciscus andLong, 1991; Roseman, 2004; Hubbe et al., 2009; Holló et al., 2010;Noback et al., 2011). However, for the purposes of the current study,
given the lack of cold-adapted hominoid taxa in our sample, itis most conservative to assume that all three functional-developmental regions will be equally valuable for recovering thegenetic relationships of the hominoids.Individual cranial bone regions An additional means of demar-cating regions within the primate cranium is to delineate theexternal form of the individual adult bones. This was used by vonCramon-Taubadel (2009a) as a means of testing whether thetemporal bone might be a particularly reliable indicator ofmodern human population history, as had been suggested byprevious studies (Lockwood et al., 2004; Harvati and Weaver,2006a, b; Smith et al., 2007; Terhune et al., 2007; Smith, 2009).The seven individual bones tested were the temporal, parietal,occipital, sphenoid, frontal, zygomatic and maxilla. The resultsfound that although all seven bones were significantly correlatedwith genetic distance, the temporal, sphenoid, frontal andparietal were all equally genetically congruent while the occipitaland maxilla were less so, and the zygomatic was statistically theleast reliable bone. Hence, on the basis of these results, wepredict that the temporal, parietal, sphenoid and frontal will havethe greatest efficacy for recovering hominoid phylogeny, followedby the occipital and maxilla, and with the zygomatic having thelowest efficacy of the bones tested.‘Masticatory’ regions: regions most affected by biomechanicalstress relating to chewing Several studies have assessed the rela-tive phylogenetic efficacy of regions of the primate craniumthought to be affected by homoiology (i.e., homoplasies caused byphenotypic plasticity) (Collard and Wood, 2001, 2007; Lycett andCollard, 2005). Craniodental datasets were divided according tothe extent to which regions of the primate cranium might beaffected by remodeling due to biomechanical stress associatedwith chewing, with non-remodeling dentition used as a baseline.von Cramon-Taubadel (2009b) used a 3D geometricmorphometric approach to test the predictions of the homoiologyhypothesis in modern human populations. Two ‘masticatory’ (i.e.,high strain) cranial regions were defined: the zygotemporalregions (temporomandibular joint, the temporal fossa, zygomaticarch and attachment sites for the masseter and pterygoidmuscles) and the palatomaxilla (the palate and lower maxillabearing the upper dentition). These were compared against ‘non-masticatory’ regions (the chondrocranium, the upper face and thevault). The results found that the non-masticatory and thezygotemporal region were all equally good at recovering pastpopulation history, while the palatomaxilla was significantly lessreliable than the non-masticatory regions. Smith (2009, 2011)and von Cramon-Taubadel (2011b) also demonstrated that thehuman mandible is a relatively poor predictor of human geneticrelationships, most probably due to phenotypic plasticity inresponse to differing masticatory regimes (von Cramon-Taubadel,2011b). Hence, taking these results together with those of vonCramon-Taubadel (2009b), we predict that the zygotemporal,upper face, vault and chondrocranium should all be equallyreliable indicators of hominoid phylogeny, while thepalatomaxilla will be relatively less reliable and the mandible willbe the least reliable masticatory region.
Materials and methods
Materials: morphological
Morphometric data comprised configurations of 116 cranial and31mandibular 3D landmarks collected by one observer (NVC) usinga Microscribe� digitizer on a sample of 284 males and 275 femalesrepresenting 15 hominoid taxa (Table 1). Landmark configurationswere designed to accurately reflect the morphology of individual
Table 1List of taxon samples collected.
Taxa Museum collections Sample sizes
Males Females Total
Homo sapiens (10 each of Ibo, Italian, Australian,Alaskan Inuit, Japanese)
NHM ¼ Natural History Museum (London), NHMW ¼ Natural History Museum (Vienna), DC ¼ Duckworth Collection (Cambridge), MH ¼ Musée de l’Homme (Paris),AMNH¼ American Museum of Natural History (New York), PC ¼ Powell-Cotton Collection (Kent), MCZ¼Museum of Comparative Zoology (Harvard), RMCA¼ Royal Museumof Central Africa (Tervuren), SNMNH ¼ Smithsonian National Museum of Natural History (Washington D.C), BSC ¼ Bavarian State Collection for Anthropology and Paleo-anatomy (Munich).
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653 643
cranial regions as previously analyzed using modern human pop-ulation samples (Smith, 2009, 2011; von Cramon-Taubadel, 2009a,2011a). The H. sapiens sample (n ¼ 50) was taken from the previ-ously published dataset of von Cramon-Taubadel (2009a, b, 2011a).The sample comprised individuals for whom mandibular data had
Figure 1. Anatomical position of all 116 cranial landmarks digitized. Full anatomical descripthe three main cranial regions defined using developmental and functional criteria; pink ¼tocranium). (For interpretation of the references to color in this figure legend, the reader is
also been collected (von Cramon-Taubadel, 2011b) and specimenswere chosen to represent populations from all major geographicregions. With the exception of Nomascus, all hominoid genera arerepresented in this dataset, and data on as many species andsubspecies as possible were collected.
tions of each numbered landmark can be found in Table 2. Different colors here depictchondrocranium, yellow ¼ face, blue ¼ vault (which together equate to the derma-referred to the web version of this article).
Figure 2. Landmark configurations used to describe the shape of (A) cranial regions defined on the basis of masticatory function; yellow ¼ upper face, blue ¼ zygotemporal, andpink ¼ palatomaxilla, and (B) individual cranial bones; yellow ¼ temporal, blue ¼ zygomatic, light green ¼ frontal, purple ¼ parietal, orange ¼ occipital, green ¼ sphenoid,pink ¼ maxilla. All landmark descriptions can be found in Fig. 1 and Table 2. (For interpretation of the references to color in this figure legend, the reader is referred to the webversion of this article).
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653644
Figs. 1e3 illustrate the anatomical positions of all of the land-marks collected and the cranial regions to which they contribute.Landmarks were chosen on the basis of their homology andrepeatability across all taxa, and to maximize the anatomicalcoverage of the cranial units analyzed. In the case of some speci-mens, it was necessary to transpose the midline landmarks of thecranial vault (bregma and lambda) to one side of the sagittal crestduring the digitizing process. In these cases, the digitized
Figure 3. Landmark configuration used to describe the shape of the mandible.Numbers match the anatomical descriptions given in Table 2. Figure adapted from vonCramon-Taubadel (2011b).
landmarks were subsequently orthogonally projected onto thecentral plane defined by the midline landmarks glabella, prosthionand basion. All specimens sampled were anatomically complete,fully adult with a fused sphenooccipital synchondrosis, andapproximately equal numbers of males and females were sampledfor each taxon. Table 2 provides full anatomical descriptions of allcranial and mandibular landmarks digitized and the individualcranial configurations delineated.
Intra-observer measurement error was tested prior to the onsetof data collection by digitizing the entire cranial and mandibularlandmark configuration five times on two Pan troglodytes and twoGorilla gorilla specimens, with at least one day between measuringrounds. Following the method described by von Cramon-Taubadelet al. (2007), standard deviations for each individual landmarkranged from 0.08 to 0.95 mm for cranial landmarks and between0.12 and 0.98 mm for mandibular landmarks. All landmarks weredeemed repeatable if they had a digitizing error rate of less than1 mm.
Methods
Geometric morphometrics Landmark configurations representingthe entire cranium (n ¼ 116 landmarks), the entire mandible(n ¼ 31 landmarks) and each of the 13 cranial regions depicted inFigs. 1 and 2 were separately subjected to Generalized ProcrustesAnalysis (GPA), followed by orthogonal tangent space projectionin MorphoJ 1.02 (Klingenberg, 2011). All geometric morphometricanalyses were then repeated for males and females separately.GPA removes the effects of rotation, translation and isometricscaling via a least-squares algorithm (Gower, 1975; Chapman,1990), resulting in scaled Procrustes shape variables.
Table 2Anatomical definitions of all cranial and mandibular landmarks digitized and the cranial configurations to which landmarks contribute.
Landmark Anatomical definition Cranial regionscontributed to
Midline1 Alveolon The intersection of the interpalatal suture and a line tangent to the posterior margins
of the alveolar processesF, Pl
2 Bregma The point where the coronal and sagittal sutures intersect V, Fr, P3 Basion The point where the anterior margin of the foramen magnum intersects the midsagittal plane C, O4 Glabella Most anterior midline point on the frontal bone F, V, Fr, UF5 Hormion The midline point of attachment of the vomer and sphenoid bones C, S6 Inion The midline point where the superior nuchal lines merge in the external occipital protuberance C, V, O7 Incisivon The most posterior, inferior point on the incisive fossa F, M, Pl8 Lambda The midline point where the sagittal and lambdoid sutures intersect V, O, P9 Nasion The point of intersection of the nasofrontal suture and the midsagittal plane F, Fr, UF10 Nasal depth The deepest point of inflection of the nasal profile F, UF11 Opisthion The point where the posterior margin of the foramen magnum intersects the midsagittal plane C, O, C12 Ophryon The midline point of inflection posterior to the brow ridges V, Fr, V13 Palatomaxillare The midline point of intersection of the palatine and the maxillary bones F, M, Pl14 Prosthion The most anterior midline point on the maxillary alveolar process between the two
central incisorsF, M, Pl
15 Sphenobasion The midline point on the sphenooccipital suture C, O, S16 Subspinale The midline point at which the inferior edge of the nasal spine becomes the anterior edge
of the maxillaF, M, Pl, UF
Bilateral17 Alare The most lateral point on the nasal aperture taken perpendicular to the nasal height F, M, UF18 Alveolare The most anterior point on the alveolus of the first molar F, M, Pl19 C/P3 The most inferior point on the external surface of the maxilla between the canine and P3 F, M, Pl20 Posterior M2 The point on the lateral alveolus distal to M2 F, M, Pl21 Asterion The point where the lambdoid, parietomastoid and occipitomastoid sutures meet V, C, O, P, T22 Carotid canal (lat) The most lateral point on the carotid canal C, T23 Carotid canal (med) The most medial point on the carotid canal C, T24 Coronale The most lateral point on the coronal suture V, Fr, P25 Dacryon The point of intersection of the frontolacrimal and lacrimomaxillary suture F, Fr, M, UF26 Ext aud meatus (ant) The most anterior point on the margin of the external auditory meatus V, T27 Ext aud meatus (pos) The most posterior point on the margin of the external auditory meatus V, T28 Ext palate length The point on the inferior surface of the maxilla that denotes the most posterior
point of the alveolar processF, M, Pl
29 Euryon (parietal) The most lateral point on the parietals that defines the greatest cranial breadth on the parietal V, P30 Frontomalare orbitale The point where the zygomaticofrontal suture crosses the orbital margin F, Fr, Z, UF, Zt31 Frontomalare temporale The most lateral point on the zygomaticofrontal suture F, Fr, Z, Zt32 Foramen ovale (ant) The most anterior point on the foramen ovale V, S33 Foramen ovale (pos) The most posterior point on the foramen ovale V, S34 Foramen magnum (lat) The most lateral point on the margin of the foramen magnum and posterior to occipital condyle C, O35 Frontozygomatico-sphenoid (FRED) The point of intersection of the frontozygomatic, zygomaticosphenoid and sphenofrontal
suturesV, F, Fr, S, Z, Zt
36 Infranasion The point of intersection of the nasofrontal, nasomaxillary and maxillofrontal sutures F, Fr, M, UF37 Infratemporale The most medial point on the infratemporal crests V, S38 Jugular (lat) The most inferior, lateral point on the margin of the jugular foramen C, O, T39 Jugular (med) The most inferior, medial point on the margin of the jugular foramen C, O, T40 Jugale The point in the depth of the notch between the temporal and frontal process of
the zygomatic boneF, Z, Zt
41 Krotaphion The most posterior extent of the sphenoparietal suture (pterion) V, P, S, T42 Mandibular fossa (lat) The most lateral point on the mandibular fossa V, T, Zt43 Max maxillary curve The point in the depth of the notch between the zygomaxillary suture and the alveolar process F, M, Pl44 Mastoidale The most inferior, lateral point on the mastoid process C, T45 Nasomaxillare The most inferior point on the nasomaxillary suture F, M, UF46 Occipitocondyle (ant) The most anterior, inferior point on the occipital condyle C, O47 Occipitocondyle (lat) The most lateral, inferior point on the occipital condyle C, O48 Orbitale The most inferior midpoint on the orbital margin F, Z, UF, Zt49 Orbitale (sup) The most superior midpoint of the orbital margin F, Fr, UF50 Palatomaxillare (lat) The most lateral point on the palato-maxillary suture F, M, Pl51 Petrosal The most anterior point of the petrous element of the temporal bone C, T, S52 Porion The most superior point on the margin of the external auditory meatus V, T53 Radiculare The point of maximum inflection of the zygomatic processes V, T, Zt54 Sphenomaxillare (sup) The most superior, lateral point of contact between the maxilla and the lateral
pterygoid plate of the sphenoidV, F, S
55 Sphenobasion (lat) The most lateral, inferior point on the sphenooccipital synchondrosis C, O, S56 Sphenion The most anterior extent of the sphenoparietal suture (pterion) V, Fr, P, S57 Sphenosquamosal The point of intersection of the infratemporal crest and sphenosquamosal suture V, S, T, Zt58 Stenion The most medial point on the sphenosquamosal suture V, C, S, T, Zt59 Styloid foramen The most anterior, inferior point on the styloid foramen C, T60 Sphenozygomatic (pos) The most posterior, inferior point on the sphenozygomatic suture V, F, S, Zt61 Temporal fossa (pos) The most posterior, inferior point on the temporal fossa V, T, Zt62 Zygotemporale (inf) The most inferior point on the zygomaticotemporal suture V, F, T, Z, Zt63 Zygotemporale (sup) The most superior point on the zygomaticotemporal suture V, F, T, Z, Zt
(continued on next page)
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653 645
Table 2 (continued )
Landmark Anatomical definition Cranial regionscontributed to
64 Zygomaxillare The most inferior, anterior point on the zygomaticomaxillary suture F, M, Z, UF, Zt65 Zygoorbitale The point where the zygomaticomaxillary suture intersects with the inferior orbital margin F, M, Z, UF, Zt66 Zygion The most lateral point on the surface of the zygomatic arch V, Z, ZtMandibular1 Condylion (med) The most medial point on the superior surface of the mandibular condyle2 Condylion (lat) The most lateral point on the superior surface of the mandibular condyle3 Mandibular foramen (sup) The most anterior, inferior point on the medial edge of the mandibular foramen4 Alveolus (pos) The most superior, posterior point on the alveolus5 M3 (lat-pos) The most lateral point on the alveolus posterior to M36 M1 (pos) The most lateral point on the alveolus posterior to the M17 Canine/P3 (lat) The most lateral point on the alveolus between the canine and P38 Mental foramen (ant) The most anterior point on the lateral edge of the mental foramen9 Ramus (ant) The most anterior point on the ascending ramus in line with the alveolus10 Gonion The point of maximum curvature on the posterioreinferior border where the posterior
ramus and the corpus intersect11 Ramus (pos) The most posterior point on the ascending ramus in line with the alveolus12 Sigmoid notch The most superior point of maximum inflection in the depth of the sigmoid notch13 Coronion The most superior point on the coronoid process14 Infradentale The most superior midline point on the buccal surface of the alveolus15 Pogonion The most anterior midline point on the mental eminence16 Gnathion The most inferior midline point on the mandibular symphysis17 Mandibular orale The most superior midline point on the lingual surface of the alveolus18 Linguale The most superior-posterior point on the linguale superior transverse torus
Landmark numbers correspond with those shown in Figs. 1 and 2. lat¼ lateral, med¼medial, ant¼ anterior, pos¼ posterior, sup¼ superior, inf¼ inferior, V¼ vault, F¼ face,C ¼ chondrocranium, Fr¼ frontal, O ¼ occipital, P ¼ parietal, T¼ temporal, M ¼maxilla, S ¼ sphenoid, Z ¼ zygomatic, Pl ¼ palatomaxilla, Zt¼ zygotemporal, UF ¼ upper face.
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653646
Morphological distance matrices Pairwise taxon Mahalanobis’distance matrices were computed from the resultant Procrustesvariables for each of the 15 morphological units (entire cranium,mandible, and 13 cranial subunits) for the mixed sex dataset, andfor the males and females separately in MorphoJ 1.02 (Klingenberg,2011), based on a pooled within-taxon covariance matrix. Thisresulted in a total of 45 morphological distance matrices thatcould then be statistically compared to a matrix of moleculardistances for the 15 hominoid taxa.Molecular genetic distance matrix Information regarding themolecular relationships of the 15 hominoid taxa was collated fromthe published literature (e.g., Ruvolo et al., 1994; Garner andRuyder, 1996; Ruvolo, 1996, 1997; Satta et al., 2000; Becquetet al., 2007; Gonder et al., 2011; Israfil et al., 2011; Perelmanet al., 2011) to construct a consensus phylogeny for the 15 taxaconsidered here (Fig. 4). In order to convert these molecular datainto a format that is directly comparable to the morphologicaldata, we followed the method proposed by Cardini and Elton(2008), in generating a pairwise genetic distance matrix for all 15hominoid taxa based on published phylogenetic tree topologiesand branch lengths. We took as our starting point the recentlypublished, high-resolution molecular phylogeny for all livingprimates (including humans) based on nearly 35,000 bp of codingand non-coding DNA sequence (Perelman et al., 2011). Maximumlikelihood (ML) branch lengths for all sections of the hominoidphylogeny were obtained from this analysis (see Fig. 4). Allrelevant nodes had 100% bootstrap support with the exception ofthe node linking Hylobates muelleri and Hylobates agilis, whichhad a bootstrap support of 79%. There were three sections of thephylogeny (marked in bold in Fig. 4) that required additionalinformation to resolve the complete set of molecular distances:
1. Perelman et al. (2011) included only one subspecies ofP. troglodytes (Pan troglodytes troglodytes) in their analyses, so itwas necessary to infer the relative genetic distance betweenP. troglodytes troglodytes, Pan troglodytes schweinfurthii and Pantroglodytes verus using additional published sources. Gonderet al. (2011) provide further evidence that the central andeastern equatorial subspecies of Pan (troglodytes and
schweinfurthii) are the most genetically similar of all the Pantaxa suggested and that, aside from a geographic distinction,these subspecies are effectively a single breeding population.Pairwise genetic distances between Pan taxa provided byGonder et al. (2011) and Becquet et al. (2007) all show the samerelationship between Pan paniscus, P. troglodytes verus and thecentral and eastern subspecies, but vary in terms of whetherP. troglodytes troglodytes or P. troglodytes schweinfurthii are themost distant from the other taxa. Therefore, given their overallsimilarity, we conservatively decided to represent the twosubspecies as equidistant from the node connecting them. Weused the pairwise Rst distances published by Gonder et al.(2011) to estimate the distances between each subspeciesand the node (71a in Fig. 4) and the distance between nodes 71and 71a, by scaling them by the overall distance forP. troglodytes troglodytes provided by Perelman et al. (2011) (seeSOM for further details).
2. Perelman et al. (2011) included G. gorilla (western lowlandgorilla) but not Gorilla beringei in their analysis. Therefore, weemployed maximum parsimony (MP) branch length differ-ences provided by Ruvolo et al. (1994) to infer the relativedistances between the three Gorilla taxa. We then scaled thevalues given by Ruvolo et al. (1994) by the total ML branchlength for G. gorilla (40.12ML units) provided by Perelman et al.(2011) (see SOM). Garner and Ruyder (1996) point out that thegenetic distance between western lowland gorillas (G. gorilla)and eastern gorillas (G. beringei) is greater than the distancebetween P. paniscus and P. troglodytes. Based on the Ruvolo et al.(1994) MP distances, we calculated scaled distances betweenthe two gorilla species of 27.77 ML units between G. gorilla andG. beringei graueri and 29.32 ML units between G. gorilla andG. beringei beringei, both of which are greater than the range ofdistances between P. paniscus and subspecies of P. troglodytes(16.92e24.37 ML units).
3. Perelman et al. (2011) did not include Hoolock taxa in theiranalysis. Therefore, we utilized the detailed analysis of gibbonand siamang phylogeny by Israfil et al. (2011) to infer both theposition and the relative branch length of Hoolock. This wasmore difficult for Hoolock than for Pan or Gorilla, given
Figure 4. Consensus molecular phylogeny for the hominoid taxa under study based on the current published literature. Nodes are numbered (italics) according to Perelman et al.(2011) and all branch lengths are taken from the maximum likelihood values presented there. Branch lengths in bold are derived from additional published sources on specificmolecular relationships not considered by Perelman et al. (2011). Gonder et al. (2011) and Becquet et al. (2007) in the case of Pan troglodytes; Ruvolo et al. (1994), Ruvolo, 1997 andGarner and Ruyder (1996) in the case of Gorilla species, and Israfil et al. (2011) in the case of Hoolock. See text for further details.
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653 647
uncertainty about the phylogenetic position of Hoolock relativeto Nomascus (not included here but included by Perelman et al.,2011) and Symphalangus genera. Based on the analysis of nearly14,000 bp of DNA from across the genome, Israfil et al. (2011)were able to demonstrate that Hylobates and Symphalangusare the most closely related genera, and that Hoolock andNomascus are consistently basal to the clade containing Hylo-bates and Symphalangus (although there is some doubt as towhich genus is the most basal). Therefore, in the absence ofNomascus taxa in this analysis, we can confidently place Hoo-lock as the most basal taxon in the gibbon/siamang phylogenyprovided here. In order to estimate the length of the branchlinking Hoolock with the basal node linking all gibbon taxa(node 70 in Fig. 4), we utilized the estimation by Israfil et al.(2011) that the length of the Hoolock branch is w75e80% aslong as the distance between Symphalangus and the basal node.According to Perelman et al. (2011), the Symphalangus taxon is31.4 units from the basal node, giving an average adjustedbranch length of 24.35 units for the Hoolock branch.
Fig. 4 shows all the branch lengths inferred from the literature. Apairwise distance matrix was then constructed by simply adding allof the branch lengths connecting each pair of taxa. Neighbor-joining (Saitou and Nei, 1987) and UPGMA (Sneath and Sokal,1973) trees were then estimated from the distance matrix in Phy-lip 3.66 (J. Felsenstein, http://evolution.genetics.washington.edu/phylip) to confirm that the consensus molecular phylogeny couldbe correctly recovered from the resultant genetic distance matrix.Comparing morphometric and molecular genetic distancematrices Each of the 45 morphological Mahalanobis distancematrices was statistically compared against the genetic distancematrix using Mantel tests (Mantel, 1967) with 10,000 matrixpermutations for assigning significance (Smouse and Long, 1992).
The critical alpha level was set at a ¼ 0.05. All Mantel tests wereperformed in the software PASSaGE 1.1 (www.passagesoftware.net). Thereafter, to assess whether any of the morphometricdistance matrices were significantly more strongly correlatedwith the genetic matrix, a series of DoweCheverud tests (Dowand Cheverud, 1985) were performed. DoweCheverud tests areused to examine the null hypothesis (a ¼ 0.05) that any twodependent matrices (i.e., two morphometric matrices) are equallystrongly correlated with a single independent matrix (themolecular genetic matrix). DoweCheverud tests were performedseparately for the mixed sex, female-only and male-onlymatrices, in R using a codewritten by the lab of Charles C. Roseman.
Results
Mantel tests
Table 3 provides the results of all Mantel test comparisonsperformed between morphological and genetic distance matrices.Morphological matrices are listed according to the criteria used todelineate them; i.e., functional-developmental units, individualbones, regions defined according to masticatory stress. Individualmorphological matrices are labeled according to the predictionsmade regarding their genetic congruence based on previous anal-yses of modern human populations, with 1 being statistically morecongruent than 2, which is statistically more congruent than 3, etc.The results for the Mantel tests performed here are remarkablyconsistent irrespective of whether sexes were combined oranalyzed separately. All morphological regions were significantlycorrelated with the genetic matrix, suggesting that, on average,morphological differences are reliable indicators of genetic rela-tionships amongst hominoids. Five regions were more stronglyassociated with the molecular distance matrix than using all of the
All Mantel tests are highly significant (a ¼ 0.05). Cranial regions with an absolute higher r-value than the entire cranium are shown in bold. Func-Dev ¼ Functional-Developmental regions.
a Numbers signify which regions were found to be significantly more genetically congruous in modern humans based on DoweCheverud tests (see Smith, 2009, 2011; vonCramon-Taubadel, 2009a, b 2011a, b). Thus, 1 is significantly better than 2, which is significantly better than 3.
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653648
cranial data; the face, frontal bone, zygomatic bone, upper face andzygotemporal. Given the considerable anatomical overlap betweenthese regions, it is more conservative to say that the face is moregenetically congruent, on average, than data from the entirecranium. The least genetically congruent, on average, were themandible, the palatomaxilla, the temporal and occipital bones, andthe chondrocranium.
DoweCheverud tests
Tables 4e6 document the results of the DoweCheverudcomparisons of individual regions in the mixed sex and single sexanalyses. DoweCheverud comparisons were performed betweencranial regions within particular definition criteria, and all signifi-cantly different results are shown in bold. Comparison of resultsacross Tables 4e6 demonstrates the consistency of the resultsirrespective of whether both sexes are included or not. The face is
Table 4Results of the DoweCheverud test comparisons of all morphological matrices for the mi
Cranial regions are ordered according to the results of the Mantel tests from strongestdiagonals¼ p-values. All significant differences (a¼ 0.05) in bold. Zygotemp¼ Zygotempo
consistently more strongly correlated with the genetic data thanthe other cranial regions or the entire cranium. This differs from thepattern found in modern humans (e.g., von Cramon-Taubadel,2011a) where no significant differences were found between themajor functional-developmental regions in terms of their geneticcongruence. The zygomatic bone is significantly more stronglycorrelated with the genetic matrix than all other bones, while thefrontal, maxilla and sphenoid (in the case of the mixed sex analysis)are all more significantly genetically congruent than the occipitaland the temporal. This also differs from the pattern obtained inmodern humans, where the zygomatic was consistently one of theleast genetically congruent bones, while the temporal was one ofthe most genetically congruent (von Cramon-Taubadel, 2009a). Theupper face and the zygotemporal unit were consistently moregenetically congruent than the other units considered in themasticatory-defined category. Moreover, all morphological regionsconsidered are significantly more congruent that the palatomaxilla
Cranial regions are ordered according to the results of the Mantel tests from strongest (largest r-values) to the weakest correlations. Lower diagonals ¼ p1Z values, upperdiagonals¼ p-values. All significant differences (a¼ 0.05) in bold. Zygotemp¼ Zygotemporal, Zygomat¼ Zygomatic, Chondro¼ Chondrocranium, Palatomax¼ Palatomaxilla.
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653 649
and the mandible. The upper face is a ‘non-masticatory’ region,while the zygotemporal is a ‘masticatory’ region, suggesting thatthe definition of masticatory function per se is not driving theresults of these analyses. Rather, it is the anatomical inclusion infacial morphology (with the exception of the maxillary region) thatis providing the best fit with the genetic data.
Post-hoc analyses: removing humans
The phylogenetic analysis of Creel (1986) revealed the presenceof craniometric homoplasy between humans and hylobatid taxa,which he attributed to convergent similarities in brain size allom-etry (i.e., relatively larger brains in small bodied hominoids andexpanded brain size in modern humans). Similarly, Strait and Grine(2004) found that the most parsimonious tree in a cladistic analysisof hominoid ingroup taxa linked humans with Hylobates asa monophyletic clade. Moreover, when Hylobates were used as an
Table 6Results of the DoweCheverud test comparisons of all morphological matrices for the fem
Cranial regions are ordered according to the results of the Mantel tests from strongest (diagonals¼ p-values. All significant differences (a¼ 0.05) in bold. Zygotemp¼ Zygotempo
outgroup, the correct phylogenetic relationship between humansand the remaining hominoid taxa was established only by addingfossil hominin taxa into the analysis (see also Begun, 1992).Lockwood et al.’s (2004) phenetic analysis revealed exceptionallylong branch lengths for the human lineage relative to otherhominoids, reflective of the relatively high rates of cranial evolutionthat have taken place within the Hominini since the last commonancestor with the genus Pan (Lieberman, 2011). Taken together,what these results suggest is that, of all the taxa considered here, itis the inclusion of H. sapiens that is most likely to confound thereconstruction of phylogeny from craniomandibular morphology.Therefore, to test the extent to which including humans in theanalysis is affecting the results, all Mantel tests were repeated withthe sample ofH. sapiens removed (Table 7). All of the correlations (r-values) improved as a result of removing humans from the analysis.The percentage increase relative to the original r-value was alsocalculated to test whether all cranial regions were affected equally
All Mantel tests were significant (p � 0.001). Percentage differences relate to the relative amount of improvement in the r-values when humans were removed from theanalysis. Cranial regions are ordered according to their original Mantel test results. All improvements of >30% are highlighted in bold. Zygotemp ¼ Zygotemporal,Chondro ¼ Chondrocranium, Palatomax ¼ Palatomaxilla.
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653650
by the removal of humans. Percentage improvements varied from3.7 to 87% in the case of the mixed sex analysis, 5.1 to 82.4% in thecase of the male-only analysis, and 4.2 to 99.4% in the case of thefemale analysis, illustrating that the removal of humans affectedsomemorphological regionsmuchmore than others. Moreover, themorphological units with the absolute lowest original r-values(mandible, palatomaxilla, temporal and occipital) were also thoseto show the greatest improvement in genetic congruence oncehumans were removed. This suggests that the inclusion of humansis a more potent factor in generating correlations betweenmorphometric and genetic data for extant hominoids than theactual anatomical definition of the cranial region itself.
Discussion
Several empirical conclusions can be drawn from the resultspresented here. The criteria employed to delineate cranial regions(functional-developmental, bones, masticatory function) did notappear to impact theanalysis in anysystematicway; that is, no singlecriterion proved to produce particularly informative or non-informative regions. Similarly, the relative performance of regionswas not linkedwith their absolute anatomical size or thenumbers oflandmarks used to delineate them. In addition, sexual dimorphismin the shape of particular regions did not appear to have a strong orsystematic effect on the results of the analyses. In all cases, thesame pattern of results emerged regardless of whether sexes werecombined or treated separately. Interestingly, O’Higgins andDryden(1993) noted that although all hominoid taxa exhibit size dimor-phism between the sexes, only Gorilla and Pongo showed strongshape dimorphism, primarily located in the lower face. Our resultsalso support this finding in the sense that for the face and palato-maxilla regions, the genetic congruencewas lower for themale-onlyanalyses than for the female analyses. Neighbor-joining (Saitou andNei, 1987) phenetic reconstructions (Fig. 5) of the entire cranialdataset and the face found that when the sexes were combined, thetwo dimorphic genera (Gorilla and Pongo) were linked, while in themale-only and female-only topologies (not shown), Gorilla and Panwere linked to the exclusion of Pongo. However, in all cases, humanswere incorrectly positioned between the great apes and the cladecontaining the gibbons and siamangs (Fig. 5). Fig. 5 shows that forthe relativelygenetically congruent face, the intra-generic and intra-specific relationships are always correctly inferred, while relation-ships between genera are generally incorrect. In contrast, theNJ phenogram for the chondrocranium (Fig. 5D) shows that both
intra-generic and inter-generic relationships are incorrectlycaptured by themorphological data, whichwould explain its overalllower genetic congruence when compared with the face (or theentire cranium).
It is important to recognize that, on average, craniomandibularmorphology was significantly and strongly congruent with geneticrelationships amongst hominoid taxa, especially when humanswere removed from the analysis. Indeed the r-values obtained forMantel matrix correlations once humans were excluded, rangedfrom 0.7 to 0.94, which is generally higher than genetic/cranio-metric congruence levels obtained for modern human analyses(e.g., Roseman, 2004; Harvati and Weaver, 2006a, b; Smith, 2009;von Cramon-Taubadel, 2009a, 2011). However, the predictionsregarding the relative efficacy of particular cranial regions made onthe basis of previous modern human population analyses (Harvatiand Weaver, 2006a, b; Smith, 2009, 2011; von Cramon-Taubadel,2009a, 2011a), were not upheld by the results of these analyses.Hence, it does not appear that there are any particular regions orcriteria for delineating regions that are significantly better forreconstructing the genetic relationships of both modern humanpopulations and hominoid taxa.
Overall, facial shape proved to be the most reliable aspect ofhominoid anatomy for reconstructing genetic relationships, bothwith and without humans in the analysis. Moreover, facial regionswere the only cranial regions that were consistently more reliablethan using data from the entire cranium. The relatively poorcongruence of the cranial vault and aspects of the masticatoryapparatus may be due to the presence of numerous muscle attach-ment sites (temporalis,masseter, pterygoid, etc.),which increase thechances of non-genetic shape change due to osseous remodeling.However, although in the initial analyses (humans included)masticatory regions such as the palatomaxilla and the mandibleappeared to be relatively problematic for phylogenetic reconstruc-tion, they were equally reliable once humans were removed. Whatthis demonstrates is that their low genetic congruence is nota function of these regions beingmasticatory-related per se (Collardand Wood, 2001, 2007; Lycett and Collard, 2005; von Cramon-Taubadel, 2009b), but instead suggests homoplasy in masticatorytraits between humans and some other hominoid taxa.
The fact that there do not appear to be any ‘magic bullet’anatomical regions that are effective for reconstructing both homi-noid taxon and modern human population genetic relationshipssuggests thatwemust paycareful attention to the taxonomic level atwhich phylogenetic analyses are conducted. Given that the entire
Figure 5. Unrooted Neighbor-joining phenograms depicting the relationships between all 15 hominoid taxa based on (A) the consensus molecular distance matrix (Fig. 4), andmixed-sex Mahalanobis distance matrices based on morphometric data from (B) the entire cranium, (C) the face, and (D) the chondrocranium. Taxon abbreviations used include thegenus name and capital letters indicative of species (and subspecies, where applicable) according to the taxon names provided in Table 1 and Fig. 4.
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653 651
evolutionary history of the hominin lineage is bracketed by theextant Pan taxa and the single extant species of Homo, futurephylogenetic work might benefit from focusing on smaller taxo-nomic (i.e., intra-generic) scales, if inferences from extant primatephylogenies are to be profitably extrapolated to the hominin fossil
record. Unfortunately, there is a relatively uneven distribution ofextinct to extant morphologies across the primate order, especiallyin the case of the hominins where the considerable morphologicalchanges since the LCA (last common ancestor) are necessarilyinferredwith reference to the verymorphologically distinct Pan and
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653652
Homo taxa. It is for this reason that the inclusion of at least somehominin taxa may be necessary to accurately portray the phyloge-netic relationships of the hominoids based on anatomical data(Begun, 1992; Strait and Grine, 2004).
One of the main impetuses for reconstructing phylogeny frommorphology is to provide a framework against which to documentthe past effects of neutral and non-neutral (adaptive) evolutionaryforces (see also Ackermann and Smith, 2007). The major reason asto why cranial shape covariance patterns are a useful proxy formodern human population history is because intra-specific diver-sification has largely been due to neutral evolutionary forces ofmutation, genetic drift and gene flow. However, if certain homininspeciation events were driven primarily by non-neutral factorssuch as dietary, locomotor or cognitive adaptation then we wouldnot expect fossil specimen covariance patterns to reflect neutralevolution. Ackermann and Cheverud (2004) provided an excellenttemplate for testing the differing effects of neutral and selectiveforces in creating hominin fossil morphological diversity. Theydemonstrated that natural selection most likely played a role increating facial diversification patterns in the early part of thehominin record (gracile and robust australopithecines), but thatdiversification amongst early members of the genus Homo could beexplained by random, neutral processes alone. The same range ofevolutionary processes are likely to have been important inhominoid diversification, whereby some intra-generic and intra-specific taxa may have diversified under random conditions,while generic-level differences reflect the action of directional ordisruptive selection. Therefore, the detection of homoplasy inmorphological datasets could provide important insights into pastinstances of natural or sexual selection (see e.g., Begun, 2007). Theresults presented here are unanimous in demonstrating the overallstrong congruence between molecular and morphological evolu-tionary patterns in the hominoids. Therefore, points of divergencebetween these patterns can now be investigated further to eluci-date specific instances of either non-heritable phenotypic plasticityor parallel/convergent evolution.
Conclusion
The results of our analyses demonstrate that craniomandibularshape differences generally reflect the molecular genetic relation-ships amongst hominoid taxa. Overall, facial anatomy proved to bestatistically more reliable for estimating genetic relationships thanusing data from the entire cranium, although all regions testedwere significantly genetically correlated. However, predictionsmade regarding the relative phylogenetic efficacy of 14 differentregions on the basis of modern human population studies did nothold true for hominoids as a whole, indicating that there are noconsistent anatomical efficacy rules that can be extrapolated to thehominin fossil record. Given both the derived nature of humancraniomandibular morphology and apparent homoplasy betweenmodern humans and the hylobatid taxa, removing humans fromthe analysis further improved the fit between morphological andmolecular estimates of taxon relationships. Further phylogeneticstudies conducted at smaller (i.e., intra-generic) taxonomic scalesand those that specifically investigate the similarities and diver-gences between molecular and morphological estimates of pastevolutionary processes are now required if we are to build moreaccurate inference models for the fossil hominins.
Acknowledgments
We wish to acknowledge the L.S.B. Leakey Foundation forfunding received to undertake this project. We thank CharlesRoseman and Mark Grabowski for sharing their DoweCheverud
test R code and are grateful to the following museum curators forproviding access to the collections in their care: Paula Jenkins andLouise Tomsett (Natural History Museum, London), Malcolm Har-man (Powell-Cotton, Quex Park), Mike Schweissing (Bavarian StateCollection for Anthropology and Paleoanatomy, Munich), Emma-nuel Gilissen and Wim Wendelen (Royal Museum Central Africa,Tervuren), Eileen Westwig (American Museum of Natural History,NY), Linda Gordon (Smithsonian National Museum of NaturalHistory), Judy Chupasko (Harvard Museum of ComparativeZoology). We are grateful to the editor, four anonymous reviewers,and Stephen Lycett for constructive comments on an earlier draft ofthis manuscript.
Appendix A. Supplementary data
Supplementary data related to this article can be found online atdoi:10.1016/j.jhevol.2012.02.007.
References
Ackermann, R.R., Cheverud, J.M., 2004. Detecting genetic drift versus selection inhuman evolution. Proc. Natl. Acad. Sci. 101, 17946e17951.
Ackermann, R.R., Smith, R.J., 2007. The macroevolution of our ancient lineage: whatwe know (or think we know) about early hominin diversity. Evol. Biol. 34,72e85.
Becquet, C., Patterson, N., Stone, A.C., Przeworski, M., Reich, D.E., 2007. Geneticstructure of chimpanzee populations. Plos Genet. 3, 617e626.
Begun, D.R., 1992. Miocene fossil hominoids and the chimp-human clade. Science247, 1929e1933.
Begun, D.R., 2007. How to identify (as opposed to define) a homoplasy: examplesfrom fossil and living great apes. J. Hum. Evol. 52, 559e572.
Betti, L., Balloux, F., Amos, W., Hanihara, T., Manica, A., 2009. Distance from Africa,not climate, explains within-population phenotypic diversity in humans. Proc.R. Soc. B 276, 809e814.
Bjarnason, A., Chamberlain, A.T., Lockwood, C.A., 2011. A methodological investi-gation of hominoid craniodental morphology and phylogenetics. J. Hum. Evol.60, 47e57.
Cardini, A., Elton, S., 2008. Does the skull carry a phylogenetic signal? Evolution andmodularity in the guenons. Biol. J. Linn. Soc. 93, 813e834.
Chamberlain, A.T., Wood, B.A., 1987. Early hominid phylogeny. J. Hum. Evol. 16,119e133.
Chapman, R.E., 1990. Conventional procrustes approaches. In: Rohlf, F.J.,Bookstein, F.L. (Eds.), Proceedings of the Michigan Morphometrics WorkshopSpecial Publication No 2. University of Michigan Museum of Zoology, Ann Arbor,pp. 251e267.
Collard, M., O’Higgins, P., 2001. Ontogeny and homoplasy in the papionin monkeyface. Evol. Dev. 3, 322e331.
Collard, M., Wood, B., 2000. How reliable are human phylogenetic hypotheses?Proc. Natl. Acad. Sci. 97, 5003e5006.
Collard, M., Wood, B., 2001. Homoplasy and the early hominid masticatory system:inferences from analyses of extant hominoids and papionins. J. Hum. Evol. 41,167e194.
Collard, M., Wood, B., 2007. Hominin homoiology: an assessment of the impact ofphenotypic plasticity on phylogenetic analyses of humans and their fossilrelatives. J. Hum. Evol. 52, 573e584.
Creel, N., 1986. Size and phylogeny in hominoid primates. Syst. Zool. 35, 81e99.Dow, M.M., Cheverud, J.M., 1985. Comparison of distance matrices in studies of
population structure and genetic microdifferentiation: quadratic assignment.Am. J. Phys. Anthropol. 68, 367e373.
Franciscus, R.G., Long, J.C., 1991. Variation in human nasal height and breadth. Am. J.Phys. Anthropol. 85, 419e427.
Garner, K.J., Ruyder, O.A., 1996. Mitochondrial DNA diversity in gorillas. Mol. Phy-logenet. Evol. 6, 39e48.
Gilbert, C.C., Rossie, J.B., 2007. Congruence of molecules and morphology usinga narrow allometric approach. Proc. Natl. Acad. Sci. 104, 11910e11914.
Gilbert, C.C., Frost, S.R., Strait, D.S., 2009. Allometry, sexual dimorphism, andphylogeny: a cladistic analysis of extant African papionins using craniodentaldata. J. Hum. Evol. 57, 298e320.
Gonder, M.K., Locatelli, S., Ghobrial, L., Mitchell, M.W., Kujawski, J.T., Lankester, F.J.,Stewart, C.-B., Tishkoff, S.A., 2011. Evidence from Cameroon reveals differencesin the genetic structure and histories of chimpanzee populations. Proc. Natl.Acad. Sci. 108, 4766e4771.
N. von Cramon-Taubadel, H.F. Smith / Journal of Human Evolution 62 (2012) 640e653 653
Good, J.M., Schultz, R., Aximu-Petric, A., Butthof, A., Höber, B., Höffner, B.,Siegemund, M., Weihmann, A., Nusbaum, C., Lander, E.S., Russ, C., Novod, N.,Affourtit, J., Egholm, M., Verna, C., Rudan, P., Brajkovic, D., Kucan, Z., Gusic, I.,Doronichev, V.B., Golovanova, L.V., Lalueza-Fox, C., de la Rasilla, M., Fortea, J.,Rosas, A., Schmitz, R.W., Johnson, P.L.F., Eichler, E.E., Falush, D., Birney, E.,Mullikin, J.C., Slatkin, M., Nielsen, R., Kelso, J., Lachmann, M., Reich, D., Pääbo, S.,2010. A draft sequence of the Neandertal genome. Science 328, 710e722.
Hartman, S.E., 1988. A cladistic analysis of hominoid molars. J. Hum. Evol. 17,489e502.
Harvati, K., Weaver, T.D., 2006a. Reliability of cranial morphology in reconstructingNeanderthal phylogeny. In: Harvati, K., Harrison, T.L. (Eds.), NeanderthalsRevisited: New Approaches and Perspectives. Springer, Dordrecht,pp. 239e254.
Harvati, K., Weaver, T.D., 2006b. Human cranial anatomy and the differentialpreservation of population history and climate signatures. Anat. Rec. 288A,1225e1233.
Holló, G., Szathmáry, L., Marcsik, A., Barta, Z., 2010. Linear measurements of theneurocranium are better indicators of population differences than those of thefacial skeleton: comparative study of 1,961 skulls. Hum. Biol. 82, 29e46.
Hubbe, M., Hanihara, T., Harvati, K., 2009. Climate signatures in the morphologicaldifferentiation of worldwide modern human populations. Anat. Rec. 292,1720e1733.
Israfil, H., Zehr, S.M., Mootnick, A.R., Ruvolo, M., Steiper, M.E., 2011. Unresolvedmolecular phylogenies of gibbons and siamangs (Family Hylobatidae) based onmitochondrial, Y-linked, and X-linked loci indicate a rapid Miocene radiation orsudden vicariance event. Mol. Phylogenet. Evol. 58, 447e455.
Klingenberg, C.P., 2011. MorphoJ: an integrated software package for geometricmorphometrics. Mol. Ecol. Res. 11, 353e357.
Krause, J., Fu, Q., Good, J.M., Viola, B., Shunkov, M.V., Derevianko, A.P., Paabo, S.,2010. The complete mitochondrial DNA genome of an unknown hominin fromsouthern Siberia. Nature 464, 894e897.
Krings, M., Stone, A., Schmitz, R.W., Kranitzki, H., Paabo, S., 1997. Neanderthal DNAsequences and the origin of modern humans. Cell 90, 19e30.
Krings, M., Capelli, C., Tschentscher, F., Geisert, H., Meyer, S., von Haeseler, A.,Grossschmidt, K., Possnert, G., Paunovic, M., Paabo, S., 2000. A view of Nean-dertal genetic diversity. Nature 26, 144e146.
Lieberman, D.E., 1995. Testing hypotheses about recent human evolution fromskulls: integrating morphology, function, development, and phylogeny. Curr.Anthropol. 36, 159e197.
Lieberman, D.E., 1997. Making behavioural and phylogenetic inferences fromhominid fossils: considering the developmental influences of mechanicalforces. Annu. Rev. Anthropol. 26, 185e210.
Lieberman, D.E., 1999. Homology and hominid phylogeny: problems and potentialsolutions. Evol. Anthropol. 7, 142e151.
Lieberman, D.E., 2000. Ontogeny, homology and phylogeny in the hominidcraniofacial skeleton: the problem of the browridge. In: O’Higgins, P., Cohn, M.J.(Eds.), Development, Growth and Evolution: Implications for the Study of theHominid Skeleton. Academic Press, London, pp. 85e122.
Lieberman, D.E., 2011. The Evolution of the Human Head. Harvard University Press,Cambridge.
Lieberman, D.E., Wood, B.A., Pilbeam, D.R., 1996. Homoplasy and early Homo: ananalysis of the evolutionary relationships of H. habilis sensu stricto andH. rudolfensis. J. Hum. Evol. 30, 97e120.
Lockwood, C.A., Fleagle, J.G., 1999. The recognition and evaluation of homoplasy inprimate and human evolution. Yearbk Phys. Anthropol. 42, 182e232.
Lockwood, C.A., Lynch, J.M., Kimbel, W.H., 2002. Quantifying temporal bonemorphology of great apes and humans; an approach using geometricmorphometrics. J. Anat. 201, 447e464.
Lockwood, C.A., Kimbel, W.H., Lynch, J.M., 2004. Morphometrics and hominoidphylogeny: support for the chimpanzee-human clade and differentiationamong great ape subspecies. Proc. Natl. Acad. Sci. 101, 4356e4360.
Lycett, S.J., Collard, M., 2005. Do homoiologies impede phylogenetic analyses of thefossil hominids? An assessment based on extant papionin craniodentalmorphology. J. Hum. Evol. 49, 618e642.
Mantel, N.A., 1967. The detection of disease clustering and a generalized regressionapproach. Cancer Res. 27, 209e220.
McBratney-Owen, B., Iseki, S., Bamforth, S.D., Olsen, B.R., Morriss-Kay, G.M., 2008.Development and tissue origins of the mammalian cranial base. Dev. Biol. 322,121e132.
Moss, M.L., Young, R.W., 1960. A functional approach to craniology. Am. J. Phys.Anthropol. 18, 281e291.
Nadal-Roberts, M., Collard, M., 2005. Impact of methodological choices on assess-ments of the reliability of fossil primate phylogenetic hypotheses. Folia Pri-matol. 76, 207e221.
Noback, M.L., Harvati, K., Spoor, F., 2011. Climate-related variation in the humannasal cavity. Am. J. Phys. Anthropol. 145, 599e614.
O’Higgins, P., Dryden, I.L., 1993. Sexual dimorphism in hominoids: further studies ofcraniofacial shape differences in Pan,Gorilla and Pongo. J. Hum. Evol. 24,183e205.
Perelman, P., Johnson, W.E., Roos, C., Seuánez, H.N., Horvath, J.E., Moreira, M.A.M.,Kessing, B., Pontius, J., Roelke, M., Rumpler, Y., Schneider, M.P.C., Silva, A.,O’Brien, S.J., Pecon-Slattery, J., 2011. A molecular phylogeny of living primates.Plos Genet. 7, e1001342.
Pilbeam, D.R., 1996. Genetic and morphological records of the Hominoidea andhominid origins. Mol. Phylogenet. Evol. 5, 155e168.
Relethford, J.H., 1994. Craniometric variation among modern human populations.Am. J. Phys. Anthropol. 95, 53e62.
Relethford, J.H., 2002. Apportionment of global human genetic diversity based oncraniometrics and skin color. Am. J. Phys. Anthropol. 118, 393e398.
Relethford, J.H., 2004. Global patterns of isolation by distance based on genetic andmorphological data. Hum. Biol. 76, 499e513.
Roseman, C.C., 2004. Detecting interregionally diversifying natural selection onmodern human cranial form by using matched molecular and morphometricdata. Proc. Natl. Acad. Sci. 101, 12824e12829.
Roseman, C.C., Weaver, T.D., 2004. Multivariate apportionment of global humancraniometric diversity. Am. J. Phys. Anthropol. 125, 257e263.
Roseman, C.C., Weaver, T.D., 2007. Molecules versus morphology? Not for thehuman cranium. Bioessays 29, 1185e1188.
Ruvolo, M., 1996. A new approach to studying modern human origins: hypoth-esis testing with coalescence time distributions. Mol. Phylogenet. Evol. 5,202e219.
Ruvolo, M., 1997. Molecular phylogeny of the hominoids: inferences from multipleindependent DNA sequence data sets. Mol. Biol. Evol. 14, 248e265.
Ruvolo, M., Pan, D., Zehr, S.M., Goldberg, T., Disotell, T.R., von Dornum, M., 1994.Gene trees and hominoid phylogeny. Proc. Natl. Acad. Sci. 91, 8900e8904.
Saitou, N., Nei, M., 1987. The neighbor-joining method: a new method for recon-structing phylogenetic trees. Mol. Biol. Evol. 4, 406e425.
Satta, Y., Klein, J., Takahata, N., 2000. DNA archives and our nearest relative: thetrichotomy problem revisited. Mol. Phylogenet. Evol. 14, 259e275.
Skelton, R.R., McHenry, H.M., 1992. Evolutionary relationships among early homi-nids. J. Hum. Evol. 23, 309e349.
Smith, H.F., 2009. Which cranial regions reflect molecular distances reliably inhumans? Evidence from three-dimensional morphology. Am. J. Hum. Biol. 21,36e47.
Smith, H.F., 2011. The role of genetic drift in shaping modern human cranialevolution: a test using microevolutionary modeling. Int. J. Evol. Biol.. Article ID145262, 1e11.
Strait, D.S., Grine, F.E., 1999. Cladistics and early hominid phylogeny. Science 285,1209.
Strait, D.S., Grine, F.E., 2004. Inferring hominoid and early hominid phylogeny usingcraniodental characters: the role of fossil taxa. J. Hum. Evol. 47, 399e452.
Strait, D.S., Grine, F.E., Moniz, M.A., 1997. A reappraisal of early hominid phylogeny.J. Hum. Evol. 32, 17e82.
Terhune, C.E., Kimbel, W.H., Lockwood, C.A., 2007. Variation and diversity in Homoerectus: a 3D geometric morphometric analysis of the temporal bone. J. Hum.Evol. 53, 41e60.
von Cramon-Taubadel, N., 2009a. Congruence of individual cranial bonemorphology and neutral molecular affinity patterns in modern humans. Am. J.Phys. Anthropol. 140, 205e215.
von Cramon-Taubadel, N., 2009b. Revisiting the homoiology hypothesis: the impactof phenotypic plasticity on the reconstruction of human population historyfrom craniometric data. J. Hum. Evol. 57, 179e190.
von Cramon-Taubadel, N., 2011a. The relative efficacy of functional and develop-mental cranial modules for reconstructing global human population history.Am. J. Phys. Anthropol. 146, 83e93.
von Cramon-Taubadel, N., 2011b. Global human mandibular variation reflectsdifferences in agricultural and hunteregatherer subsistence strategies. Proc.Natl. Acad. Sci. 108, 19546e19551.
von Cramon-Taubadel, N., Weaver, T.D., 2009. Insights from a quantitative geneticapproach to human morphological evolution. Evol. Anthropol. 18, 237e240.
von Cramon-Taubadel, N., Frazier, B.C., Lahr, M.M., 2007. The problem of assessinglandmark error in geometric morphometrics: theory, methods and modifica-tions. Am. J. Phys. Anthropol. 134, 24e35.
Weaver, T.D., Roseman, C.C., 2008. New developments in the genetic evidence formodern human origins. Evol. Anthropol. 17, 69e80.
Witmer, L.M., 1995. The extant phylogenetic bracket and the importance ofreconstructing soft tissues in fossils. In: Thomason, J.J. (Ed.), FunctionalMorphology in Vertebrate Paleontology. Cambridge University Press, Cam-bridge, pp. 19e33.
Wood, B., Collard, M., 1999. The human genus. Science 284, 65e71.
