A unique Middle Miocene European hominoid and theorigins of the great ape and human cladeSalvador Moya-Solaa,1, David M. Albab,c, Sergio Almecijac, Isaac Casanovas-Vilarc, Meike Kohlera,Soledad De Esteban-Trivignoc, Josep M. Roblesc,d, Jordi Galindoc, and Josep Fortunyc

aInstitucio Catalana de Recerca i Estudis Avancats at Institut Catala de Paleontologia (ICP) and Unitat d’Antropologia Biologica (Dipartimento de BiologiaAnimal, Biologia Vegetal, i Ecologia), Universitat Autonoma de Barcelona, Edifici ICP, Campus de Bellaterra s/n, 08193 Cerdanyola del Valles, Barcelona,Spain; bDipartimento di Scienze della Terra, Universita degli Studi di Firenze, Via G. La Pira 4, 50121 Florence, Italy; cInstitut Catala de Paleontologia,Universitat Autonoma de Barcelona, Edifici ICP, Campus de Bellaterra s/n, 08193 Cerdanyola del Valles, Barcelona, Spain; and dFOSSILIA ServeisPaleontologics i Geologics, S.L. c/ Jaume I num 87, 1er 5a, 08470 Sant Celoni, Barcelona, Spain

Edited by David Pilbeam, Harvard University, Cambridge, MA, and approved March 4, 2009 (received for review November 20, 2008)

The great ape and human clade (Primates: Hominidae) currentlyincludes orangutans, gorillas, chimpanzees, bonobos, and humans.When, where, and from which taxon hominids evolved are amongthe most exciting questions yet to be resolved. Within the Afro-pithecidae, the Kenyapithecinae (Kenyapithecini ! Equatorini)have been proposed as the sister taxon of hominids, but thus farthe fragmentary and scarce Middle Miocene fossil record hashampered testing this hypothesis. Here we describe a male partialface with mandible of a previously undescribed fossil hominid,Anoiapithecus brevirostris gen. et sp. nov., from the Middle Mio-cene (11.9 Ma) of Spain, which enables testing this hypothesis.Morphological and geometric morphometrics analyses of this ma-terial show a unique facial pattern for hominoids. This taxoncombines autapomorphic features—such as a strongly reducedfacial prognathism—with kenyapithecine (more specifically, keny-apithecin) and hominid synapomorphies. This combination sup-ports a sister-group relationship between kenyapithecins (Gripho-pithecus ! Kenyapithecus) and hominids. The presence of bothgroups in Eurasia during the Middle Miocene and the retention inkenyapithecins of a primitive hominoid postcranial body plansupport a Eurasian origin of the Hominidae. Alternatively, the twoextant hominid clades (Homininae and Ponginae) might haveindependently evolved in Africa and Eurasia from an ancestral,Middle Miocene stock, so that the supposed crown-hominid syna-pomorphies might be homoplastic.

Anoiapithecus gen. nov. ! evolution ! Hominidae ! Hominoidea !Paleoprimatology

There is a general consensus on the relevance of MiddleMiocene hominoids for understanding hominid origins (1,

2). However, the question of the initial great-ape/human radi-ation still remains elusive. In this paper we describe a previouslyunderscribed Middle Miocene thick-enameled hominid [seesupporting information (SI) Text and Table S1, regarding thesystematic framework used in this paper], which displays aunique and unusual combination of facial characteristics withsignificant phylogenetic implications. Anoiapithecus brevirostrisgen. et sp. nov. shows the basic great-ape synapomorphies andsome generalized afropithecid and several kenyapithecine-derived features, coupled with a striking reduction of the face.This combination is unknown from any fossil or extant great ape,which has important implications for reconstructing the initialevolutionary history of the great ape and human clade.

Stratigraphic Setting. The description of this taxon is based on ahominoid partial face with mandible (IPS43000) discovered atlocality Abocador de Can Mata (ACM)/C3-Aj, in the area of ElsHostalets de Pierola (Valles-Penedes Basin, Catalonia, Spain).This region is characterized by thick Middle to Late Miocenestratigraphic sequences. The construction of a rubbish dump(ACM) near the country house of Can Mata (Fig. S1) prompteda paleontological intervention to control the removal of Miocene

sediments by the diggers and bulldozers. After 6 years offieldwork, 150 fossiliferous localities have been sampled from the300-m-thick local stratigraphic series of ACM, which spans aninterval of 1 million years (!12.5–11.3 Ma, Late Aragonian,Middle Miocene). To date, 38,000 macrovertebrate remains andthousands of small mammal teeth have been recovered from theabove mentioned localities, some of which have yielded primateremains (3–5). These localities can be accurately dated becauseof detailed litho-, bio-, and magnetostratigraphic control (4). Anage close to 11.9 Ma can be estimated for ACM/C3-Aj, fromwhich IPS43000 was excavated.

Description. The face of IPS43000 (Fig. 1) lacks the nasals and theright maxilla, some parts of the orbits, and parts of bothzygomatics. The palate is nearly complete, lacking only the leftC1 and M3, as well as the incisors; part of the frontal also ispreserved. The mandible preserves the symphysis and a largeportion of the 2 corpora, but lacks the 2 rami; the left I1 andC1-M2 series and the right C1-M1 series are preserved. Completeeruption of the M3 indicates that IPS43000 belongs to an adultindividual, because the slight displacement of this tooth from thealveolar plane merely results from bone distortion at the level ofM2-M3.

Systematic Paleontology. Systematic paleontology is as follows:Order Primates Linnaeus, 1758; suborder Anthropoidea Mivart,1864; infraorder Catarrhini Geoffroy, 1812; superfamily Homi-noidea Gray, 1825; family Hominidae Gray, 1825; subfamily incer-tae sedis; tribe Dryopithecini Gregory and Hellman, 1939; Anoia-pithecus gen. nov. type species A. brevirostris gen. et sp. nov.Etymology is from Anoia (the region where the site is situated) andthe Greek pithekos (ape). Generic diagnosis is as for the typespecies, A. brevirostris gen. et sp. nov. Holotype is IPS43000, a partialface with mandible of an adult male individual, housed at theInstitut Catala de Paleontologia (Fig. 1; see Table 1 for dentalmeasurements). Type locality is ACM/C3-Aj (Abocador de CanMata, Cell 3, locality Aj), in the municipal term of Els Hostalets dePierola (Catalonia, Spain). Age is subchron C5r.3r (Middle Mio-cene, !11.9 Ma), on the basis of the local ACM magnetostrati-graphic series (4), corresponding to the local biozone Megacricet-odon ibericus " Democricetodon larteti (MN 7 Mammal Neogenebiozone), on the basis of biostratigraphic data (3, 4). Etymology isfrom the Latin brevis (short) and rostrum (snout).

Author contributions: S.M.-S. designed research; S.M.-S., D.M.A., and M.K. performedresearch; J.F. contributed new reagents/analytic tools; D.M.A., S.A., I.C.-V., S.D.E.-T., J.M.R.,J.G., and J.F. analyzed data; and S.M.-S., D.M.A., and M.K. wrote the paper.

Conflict of interest: The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: salvador.moya@icp.cat.

This article contains supporting information online at www.pnas.org/cgi/content/full/0811730106/DCSupplemental.

www.pnas.org"cgi"doi"10.1073"pnas.0811730106 PNAS ! June 16, 2009 ! vol. 106 ! no. 24 ! 9601–9606

AN

THRO

POLO

GY

Specific Diagnosis. The face is characterized by reduced nasal andalveolar prognathism with a very short premaxilla. The anteriorborder of the orbit is situated over the P3–P4 limit, the glabellais over the P4, and the anteriormost nasomaxillary contact is over

the posterior part of the C1 (with the alveolar plane horizontal).Nasal aperture edges are vertical. The zygomatic root is mod-erately high and situated over the M1. The anterior surface of thezygomatic root is downwardly inclined. The frontal sinus is welldeveloped, filling the glabellar area and part of the frontalsquama. The maxillary sinus is reduced, situated well above theroots of the molars, occupying a small area below the medial sideof the orbit. The zygomatic root is not pneumatized. Coalescenttemporal lines indicate the presence of a sagittal crest. Thinsuperciliary arches are evident. A large and open incisiveforamen, with the posterior border located at the level of the P3,is shown. The palate is short, wide, and deep. The pyriformaperture is wide, widest close to the base. Dentition is charac-terized by thick enamel (relative enamel thickness, RET # 20)with low dentine penetrance. Low crowns show globulous cusps,blunt crests, and restricted basins with nonperipheralized cusps;there are remnants of cingula in lower teeth. Canines and P3 arelow-crowned whereas upper canines are relatively small and verycompressed. A robust mandible shows highly divergent rami andreduced mandibular length; strong and long inferior torus andweak superior torus that forms a simian shelf.

Differential Diagnosis. Anoiapithecus differs from all known Mio-cene Eurasian hominoids by the very orthognathous face, be-

A B C D

E F G

Fig. 1. Cranium and mandible of Anoiapithecus brevirostris (IPS43000, holotype). (A) Right lateral view. (B) Frontal view. (C) Left lateral view. (D) Superior view.(E) Palatal view. (F) Occlusal view of the mandible. (G) CT scan of the right M2, showing enamel thickness. All photographs were taken with the tooth row orientedhorizontally. For safety reasons, the cranial reconstruction (A–D) is based on casts of the original specimens.

Table 1. Dental measurements (in millimeters) of Anoiapithecusbrevirostris gen. et sp. nov. from ACM/C3-Aj

Upper teeth Lower teeth

MD BL MD BL

L I1 — — 4.8 6.7R C1 14.2 9.6 12.9 8.5L C1 — — 13.2 9.2R P3 — — 12.7 7.3L P3 7.0 11.7 12.3 7.6R P4 — — 7.6 8.8L P4 7.2 10.4 7.8 8.6R M1 9.4 11.2 9.1 8.9L M1 9.4 11.4 9.5 9.1R M2 11.3 11.8 — —L M2 10.7 12.1 11.5 10.0R M3 10.2 10.4 — —

MD, mesiodistal; BL, buccolingual; R, right; L, left.

9602 ! www.pnas.org"cgi"doi"10.1073"pnas.0811730106 Moya-Sola et al.

cause of reduced mid- and lower facial prognathism, with theglabella and orbits situated very anteriorly (close to the premo-lars and C1 on the vertical plane); it further differs from theabove-mentioned taxa except Oreopithecus by the presence of asagittal crest. More specifically, Anoiapithecus further differsfrom proconsulids and afropithecids by the possession of hom-inid facial synapomorphies, including the wide nasal aperturewidest at the base, the high face, the deep palate, and theconfiguration of the nasals. With regard to kenyapithecins, itfurther differs from Kenyapithecus by the reduced heteromorphyof the upper premolars and low-crowned canines and fromGriphopithecus by the strongly reduced cingula and the highzygomatic root. As compared to other Middle Miocene dryo-pithecins, among others it differs from Pierolapithecus by thepresence of a frontal sinus, by the thicker enamel, the lessperipheralized cusps, and the downward inclination of thezygomatic root; and from Dryopithecus by the lower degree ofnasoalveolar prognathism and the more posteriorly situatedanterior zygomatic root. Finally, as compared to the short-faced,Late Miocene Oreopithecus, Anoiapithecus further differs by thehigher face, the shorter muzzle, the reduced maxillary sinus, thevertical nasals, the lack of a nasoalveolar clivus coveringthe palatine fenestra, and the morphology of the dentition.

Morphometric Analyses. Craniofacial angle. The most outstandingcharacteristic of A. brevirostris is the strong reduction of the facialskeleton, because of the combination of an anteriorly positionedglabella with limited nasal and reduced alveolar prognathism.Measurements of the craniofacial angle (CFA) clearly show thispattern (Fig. 2; see also SI Text and Tables S2 and S3). In fossiland living catarrhines, CFA does not surpass 60°, with colobinesand hylobatids displaying the highest values, because of theiranteriorly placed glabella. The value of A. brevirostris is evenhigher and only comparable to that of fossil Homo. The differ-ences in CFA between Anoiapithecus and other fossil taxaincluded in the analysis clearly exceed the normal range ofvariation within extant taxa, as reflected by their 95% confidenceintervals, thus confirming the distinctiveness of the uniquetaxon.Geometric morphometrics. To further evaluate the uniqueness of thepattern of Anoiapithecus, we analyzed the shape of the facialprofile using a geometric morphometrics approach (Fig. 3; seealso SI Text and Tables S4 and S5). Extant great apes arecharacterized by strong alveolar prognathism, because of theposteriorly placed rhinion and nasion in relation to the more

anterior glabella, thus forming the midfacial concavity typical ofthis group (Fig. 3A). Several Miocene great apes such asHispanopithecus, Ouranopithecus, Ankarapithecus, and Sivapithe-cus fit this pattern. More primitive taxa, however, such as thestem hominoids Afropithecus, Turkanapithecus, and Proconsulrather match the pattern of extant cercopithecines, whose facialprofile is characterized by strong alveolar and midfacial progn-athism, with glabella placed posteriorly from nasion, and rhinionbeing much more anteriorly situated. The stem hominid Piero-lapithecus displays a similar or even higher degree of alveolarprognathism, while the degree of midfacial concavity is inter-mediate between stem hominoids and living great apes. Anoia-

Fig. 2. Craniofacial angle (CFA) in living and fossil catarrhines. CFA is theangle formed by the line joining glabella and prosthion with the tooth rowplane (in lateral view). Extinct taxa are represented by individual values,whereas living taxa are represented by the mean and the 95% confidenceinterval for the mean. Note that Anoiapithecus differs from both extantcercopithecoids and nonhuman hominoids by displaying a CFA well above 60°,most closely resembling members of the genus Homo.

Fig. 3. Results of the geometric morphometrics analysis. (A) Scatter diagramshowing the 2 first canonical axes (CA) of a canonical variate analysis (CVA),reflecting the differences in the facial profile of living and fossil catarrhines(see Materials and Methods and SI Text for further information). Visualiza-tions of the shape changes along the CA are also shown; to facilitate theinterpretation, the grids are rotated so that the alveolar plane is horizontal.CA1 reflects the degree of midfacial concavity, while CA2 reflects the degreeof alveolar prognathism. Note that, regarding these axes, Anoiapithecusclosely approaches hylobatids and colobines, far away from stem hominoids,extant great apes, and Pierolapithecus. (B) UPGMA cluster based on Euclideandistances computed from centroids (for extant taxa) and discriminant scores(for extinct taxa). Note that Anoiapithecus clusters with colobines, unlike bothAfrican stem hominoids (which cluster with Macaca and stem catarrhines) andEurasian fossil great apes (which cluster with living great apes).

Moya-Sola et al. PNAS ! June 16, 2009 ! vol. 106 ! no. 24 ! 9603

AN

THRO

POLO

GY

pithecus differs from great apes by the lack of midfacial concav-ity, but unlike Pierolapithecus, it displays a very orthognathousface, with rhinion situated very close to and vertically alignedwith nasion. It, thus, approximates the pattern observed incolobines and, to a lesser extent, gibbons (Fig. 3B).

Given the distinctiveness of the facial morphology of Anoia-pithecus, as compared to both living and fossil hominoids, it islikely that it represents an autapomorphically derived conditionof this taxon. To ensure that great differences found by thecanonical analysis between Anoiapithecus and other extinct taxacannot be accommodated by the normal range of variation thatis customarily found within extant hominoid genera, we followeda randomization approach on the basis of discriminant scores.The results of this analysis (SI Text) allow us to refute thehypothesis that differences between Anoiapithecus and Piero-lapithecus might be attributable to interindividual variationwithin a single genus with P $ 0.001, which confirms the needto erect a previously undescribed taxon.

DiscussionDespite its autapomorphic facial morphology, A. brevirostrisretains primitive stem-hominoid features (1, 6–8), such aslow-crowned teeth (especially the P3 and canines), cheek teethwith flaring labial and lingual walls, short canine roots converg-ing toward the midline, heteromorphic upper premolar cusps,and a frontal sinus that invades the glabella and the frontalsquama. These features, like the autapomorphic facial pattern,are not phylogenetically informative (1, 9). However, A. bre-virostris shares an array of significant features with both MiddleMiocene afropithecids (here included within the Kenyapitheci-nae) and Middle to Late Miocene hominids (see SI Text andTable S1 for further details on the systematics used in thispaper). These features are lacking in Early Miocene proconsu-

lids (1, 2, 6–8, 10) and can be hence interpreted as derivedfeatures that might reflect a phylogenetic link between keny-apithecines and stem Eurasian hominids (Fig. 4).

Among others, Anoiapithecus shares with all afropithecids athick-enameled condition (Fig. 1G), with a RET value of 20,which is in the upper range of Griphopithecus alpani from Pasalar(11). This feature is further combined with other shared den-tognathic features, such as low dentine penetrance, globulousand nonperipheralized cusps with restricted basins, thick androunded crests, a robust mandible, and a procumbent premaxilla.Anoiapithecus also shares derived features with the Keny-apithecinae (Equatorius, Nacholapithecus, Kenyapithecus, andGriphopithecus). These synapomorphies include the anteriorposition of the zygomatic root, indicating a shorter face than inthe Afropithecinae, and a strong mandibular inferior torusentailing a simian shelf (1, 6–8, 10, 12–16). Moreover, Anoia-pithecus shares with Eurasian Middle Miocene Kenyapithecini(Kenyapithecus and Griphopithecus) an extreme reduction of themaxillary sinus, which is situated well above the roots of themolars (Fig. 5A). The extent of the maxillary sinus should beinterpreted with great care, because it can increase throughoutlife. Nevertheless, the holotype of A. brevirostris belongs to a fullyadult individual, as evidenced by the fully erupted M3 and thepresence of some dental wear on both M2 and M3 (albeit with nodentine exposure, given the thick-enameled condition of thistaxon). Moreover, additional CT scans of the skull of Piero-lapithecus (Fig. 5B) have revealed that this stem hominid alsoretains this kenyapithecin trait. Further synapomorphies ofAnoiapithecus and kenyapithecins are a very deep canine fossaand reduced mandibular length with anteriorly placed mandib-ular rami (1, 6–7, 10, 12–17). The extreme shortening of the facein Anoiapithecus denotes a step further in the tendency towardfacial reduction that characterizes kenyapithecins.

Significantly, Anoiapithecus also exhibits the basic facial hom-inid synapomorphies (8), indicating that this taxon is a stem

Fig. 4. Simplified cladogram depicting the phylogenetic hypothesis andbiogeographic scenario favored in this paper. The Afropithecinae include theAfropithecini; the Equatorini include Equatorius and Nacholapithecus; theKenyapithecini include Kenyapithecus and Griphopithecus; and the Dryo-pithecini include Pierolapithecus, Dryopithecus s.s., and Anoiapithecus.Nodes: 0, taillessness and other postcranial and cranial features; 1, thickenamel, dental morphology, robust mandible, procumbent premaxilla; 2,anterior position of the zygomatic root, strong mandibular inferior torus; 3,reduction of maxillary sinus, very deep canine fossa, reduced mandibularlength; 4, high face, high zygomatic root, wide nasal aperture (widest at thebase), flat nasals that project anteriorly beneath the level of the inferiororbital rim, orthograde-related features (as judged from Pierolapithecus).This hypothesis implies a back-to-Africa dispersal of the Homininae and areversal of some features of node 3 in this group, but assumes that features ofnode 4 are homologous between pongines and hominines.

A

B

Fig. 5. CT scans in a parasagittal plane of (A) the left maxilla of Anoiapithe-cus brevirostris gen. et sp. nov. (IPS43000, holotype) and (B) the right maxillaof Pierolapithecus catalaunicus (IPS21350, holotype), to the same scale. EachCT scan is accompanied by a virtual model showing the plane represented bythe scans. Note the lack of sinus over the molar roots in the preserved maxillaof Anoiapithecus. The more completely preserved Pierolapithecus specimensimilarly shows a small and restricted maxillary sinus (the lower and anteriorlimits of the sinus are marked by white points), which does not reach the apicesof the dental roots and anteriorly does not surpass the level of posterior M1.

9604 ! www.pnas.org"cgi"doi"10.1073"pnas.0811730106 Moya-Sola et al.

member of the great ape and human clade: high face, highzygomatic root, pyriform nasal aperture widest at the base, deeppalate, and nasals that project slightly anteriorly beneath thelevel of the lower orbital rims (observed from the nasomaxillarysutures). The same modern facial pattern is also shared byPierolapithecus (5), despite the striking differences in facialprofile as compared to Anoiapithecus.

The retention of highly specialized, derived kenyapithecinefeatures in a stem hominid such as Anoiapithecus has importantimplications for understanding the origin of the Hominidae. Thepresence in Eurasia of kenyapithecin hominoids (Kenyapithecusand Griphopithecus) of putative African origin by !16.5 Ma (18)or 15–14 Ma (19, 20) has led some authors to hypothesize thatlater Middle and Late Miocene Eurasian hominids evolved fromthese taxa (2, 7, 8, 10, 16, 18–24). Hitherto, however, theapparent lack of synapomorphies between both groups pre-cluded testing this hypothesis (7). The unique facial specimen ofAnoiapithecus provides strong support for a sister-group rela-tionship between the Kenyapithecini and Hominidae. Addi-tional support for this hypothesis comes from the association ofkenyapithecine traits with cranial and postcranial great-apesynapomorphies in the stem hominid Pierolapithecus (5).

Kenyapithecus has been considered a good candidate for theancestral form of the Hominidae because it shares severalfeatures with hominids, including the moderately high zygomaticroot, the high-crowned canine, the reduced molar cingula, anddistal humeral morphology (2, 6, 10, 13, 14, 16, 25). Thehominids Anoiapithecus and Pierolapithecus retain plesiomor-phic low-crowned canines and heteromorphic premolar cusps,although they do not exhibit the autapomorphies of Kenyapithe-cus. Furthermore, Pierolapithecus catalaunicus shares with G.alpani the highly distinctive spatulate central incisor with alingual pillar continuous with the lingual cingulum. This is afeature shared with later, more derived, Eurasian hominoids(26–27). This contradictory evidence makes it difficult to de-termine which of the 2 kenyapithecin genera is more closelyrelated to hominids, while it clearly stresses the role of ho-moplasy in hominoid evolution.

When currently available evidence is taken into account, thehypothesis suggesting a Eurasian origin for the Hominidae isfavored, given the following facts: (i) the presence in theEurasian Middle Miocene of both kenyapithecins and hominids,(ii) their likely sister-group relationships, and (iii) their remark-able consistent consecutive time span (kenyapithecins, 15–13Ma; dryopithecins such as Anoiapithecus and Pierolapithecus,11.9 Ma; and Late Miocene hominids, $11.1 Ma). Keny-apithecins retain not only a primitive facial pattern for homi-noids, but also—as far as it can be ascertained—a pronogradepostcranial body plan (21–23, 28, 29). Anoiapithecus and otherdryopithecins (Dryopithecus s.s. and Pierolapithecus) share withLate Miocene Eurasian hominids and extant great apes a derivedfacial morphology (4, 5) and, at least Pierolapithecus, an ortho-grade postcranial body plan (5). This combination of characterssupports the view that crown hominids originated in Eurasiafrom more primitive, kenyapithecin ancestors and radiated inthis continent into pongines and hominines (Fig. 4).

This scenario entails a subsequent ‘‘back to Africa’’ dispersalof the hominine clade (African apes and humans) (9, 18).Alternatively, the basic putative facial and postcranial synapo-morphies of the Hominidae could be homoplastic betweenpongines and hominines, with both groups having independentlyevolved in Eurasia and Africa, respectively, from differentafropithecid ancestors. Independent evolution of suspensorycapabilities has been previously hypothesized (5). However,given the lack of both cranial and postcranial crown-hominidsynapomorphies in afropithecids, this alternative, to the back-to-Africa hypothesis would entail a more pervasive role forhomoplasy than previously suggested. If so, parallelism and

convergence would be far more common during hominoidevolution than the principle of parsimony, customarily applied tocladistic analyses, generally assumes. We expect that futurediscoveries, particularly in the long Middle to Late Miocenestratigraphic sections of Els Hostalets de Pierola section (Cata-lonia, Spain) (3, 4), may help to disentangle the complexquestion of the initial diversification of the great apes.

Materials and MethodsThe Primate Sample. The facial profile of A. brevirostris was compared to thatof living and extinct catarrhines by means of univariate and multivariatetechniques (see below). Several fossil specimens and many different individ-uals from several living genera (Table S6) were included in the analyses. Datawere taken from photographs of crania in lateral view, taken with the lensparallel to the midsagittal plane. Regarding the extant comparative sample,photographs were taken by the authors at the Koninklijk Museum voorMidden-Africa (Tervuren, Belgium) and at the Anthropologisches Institut undMuseum (Zurich, Switzerland), were kindly provided by Xavier Jordana (in thecase of living humans from the Coleccoes Osteologicas Identificadas from theMuseu Antropologico of the Universidade de Coimbra), and were taken fromthe ‘‘Mammalian Crania Photographic Archive’’ Second Edition, which isavailable from the Internet (http://1kai.dokkyomed.ac.jp/mammal/en/mammal.html).

CFA. To quantify the degree of orthognathism/prognathism, we measuredCFA as the angle comprised between the plane defined by the nasion–prosthion and that defined by the occlusal plane. It was measured in 256individuals belonging to 11 extant genera (Table S6), including the 5 extantgenera of hominoid primates, 3 cercopithecines (Cercopithecus, Macaca, andPapio) and 3 colobines (the latter being grouped into a single group Colobi-nae). Comparisons were carried out by means of ANOVA and post hoc multiplecomparisons (Bonferroni method).

Geometric Morphometrics. We used a thin-plate spline approach with gener-alized least-squares (GLS) Procrustes superimposition. A canonical variateanalysis (CVA) was performed on the resulting matrix of partial warp scoresand the similarities between the several taxa were depicted by means of acluster analysis based on centroids (extant taxa) and discriminant scores(extinct ones).

Seven landmarks, defined on the lateral facial profile and projected ontothe midsagittal plane, were used (Fig. S2). Most of these landmarks (nos. 2–7)are located at intersections between bones or bone tissues and must betherefore considered type 1 landmarks following Bookstein’s classification(30), while landmark no. 1 (glabella) is of type 2. These landmarks weremeasured in 255 individuals from the same genera included in CFA analysis(Table S6) except for Papio and Homo, which were excluded given theirextreme condition as compared to the remaining taxa; similarly, fossil Homi-nini were also excluded from the analysis. Landmarks were digitized using thetpsDig software (31). Landmark configurations were superimposed by meansof GLS Procrustes superimposition (32). GLS was conducted using the tpsRelwprogram (33). A thin-plate spline approach (30, 34) was adopted for visualiz-ing shape differences and producing a set of variables (the partial warps,including the uniform components of deformation) amenable to statisticalanalysis. The partial warp scores (weight matrix), indicating the position ofeach individual relative to the reference along the partial warps, were calcu-lated using the program tpsRelw (33).

The canonical variate analysis was performed with SPSS 15.0 on the basis ofthe weight matrix for extant taxa alone, to define a morphospace reflectingthe facial differences between living catarrhines. Groups were defined on thebasis of the different genera included in the analysis, except for Colobus,Presbytis, and Procolobus, which were grouped as Colobinae. Extinct generawere left ungrouped and classified on the basis of squared Mahalanobisdistances a posteriori. A cluster analysis was also performed with SPSS 15.0 onthe basis of Euclidean distances computed from centroids (for extant taxa) anddiscriminant scores (for extinct taxa) for all canonical axes, by means of theunweighted pair group method with arithmetic mean (UPGMA) and byrescaling distances to values comprised between 0 and 25 (Tables S4 and S5).

To take into account the range of variation displayed by extant taxa wheninterpreting the differences found between different fossil individuals (par-ticularly between the holotypes of Anoiapithecus and Pierolapithecus) in thegeometric morphometrics analysis, we followed a randomization approach.In particular, on the basis of the discriminant scores of the CVA, we computedthe squared distance for each pair of individuals within several genera (Pan,Gorilla, Pongo, and Macaca separately). The distribution of these distances

Moya-Sola et al. PNAS ! June 16, 2009 ! vol. 106 ! no. 24 ! 9605

AN

THRO

POLO

GY

was then used to test the null hypothesis that the distance between Anoia-pithecus and Pierolapithecus fits interindividual variation within a singleextant genus. This null hypothesis was rejected when the probability offinding such a distance was lower than 5%, on the basis of the distribution ofthe selected extant genera separately.

Enamel Thickness. RET was measured in a coronal plane through the mesialcusps of the right M2 as (area of the enamel cap/length of the dentinoenameljunction)/(area of the dentine)1/2 % 100 (35). The teeth were scanned withhigh-resolution computed tomography (Xylon Compact) at Burgos University(Spain). Images were analyzed with NIH Image software (developed at the U.S.National Institutes of Health and available from the Internet at http://rsb.info.nih.gov/nihimage/).

ACKNOWLEDGMENTS. We thank D. Pilbeam and 3 anonymous reviewers forhelpful comments and suggestions. We are also indebted to Cespa Gestion deResiduos, S.A. for financing the fieldwork and to the Ajuntament dels Hosta-lets de Pierola, the Departament de Cultura i Mitjans de Comunicacio de laGeneralitat de Catalunya, the Mutua de Terrassa, and the staff of the InstitutCatala de Paleontologia M. Crusafont for their collaboration. We thank I.Pellejero and S. Val for the excellent preparation of the specimens, W. vanNeer and Ana Luisa Santos for access to collections, and John Kappelman andXavier Jordana for kindly providing photographic material. This study hasbeen supported by the Comissionat d’Universitats i Recerca [predoctoralfellowship 2006 FI 00065 (to S.A.), postdoctoral grant 2005 BP-B1 10253 (toD.M.A.), the Searching for the Origins of Modern Hominoids Initiative project,and Grup de Recerca Consolidat 2005 00397-GGAC], the National ScienceFoundation (RHOI-Hominid-NSF-BCS-0321893), and the Spanish Ministerio deCiencia e Innovacion (CGL2008–00325/BTE).

1. Ward SC, Brown B, Hill A, Kelley J, Downs W (1999) Equatorius: a new hominoid genusfrom the middle Miocene of Kenya. Science 285:1382–1386.

2. Begun DR (2001) In Hominoid Evolution and Environmental Change in the Neogeneof Europe. Volume 2. Phylogeny of the Neogene Hominoid Primates of Eurasia, edsde Bonis L, Koufos G, Andrews P (Cambridge Univ Press, Cambridge, UK), pp 231–268.

3. Alba DM, et al. (2006) The fossil vertebrates from Abocador de Can Mata (Els Hostaletsde Pierola, l’Anoia, Catalonia), a succession of localities from the late Aragonian (MN6and MN7"8) of the Valles-Penedes Basin. Campaigns 2002–2003, 2004 and 2005. EstGeol 62:295–312. [in Spanish]

4. Moya-Sola S, et al. (2009) First partial face and upper dentition of the Middle Miocenehominoid Dryopithecus fontani from Abocador de Can Mata (Valles-Penedes Basin,Catalonia, NE Spain): taxonomic and phylogenetic implications. Am J Phys Anthropol,in press.

5. Moya-Sola S, Kohler M, Alba DM, Casanovas-Vilar I, Galindo J (2004) Pierolapithecuscatalaunicus, a new Middle Miocene great ape from Spain. Science 306:1339–1344.

6. Andrews P (1992) Evolution and environment in the Hominoidea. Nature 360:641–646.7. Gulec E, Begun DR (2003) Functional morphology and affinities of the hominoid

mandible from Candir. Cour Forsch-Inst Senckenberg 240:89–112.8. Kelley J, Andrews P, Alpagut B (2008) A new hominoid species from the Middle

Miocene site of Pasalar, Turkey. J Hum Evol 54:455–479.9. Rossie JB (2005) Anatomy of the nasal cavity and paranasal sinuses in Aegyptopithecus

and early Miocene African catarhines. Am J Phys Anthropol 126:250–267.10. Begun DR (2005) Sivapithecus is east and Dryopithecus is west, and never the twain

shall meet. Anthropol Sci 113:53–64.11. Smith TM, Martin LB, Leakey MG (2003) Enamel thickness microstructure and devel-

opment in Afropithecus turkanensis. J Hum Evol 44:283–306.12. Kunimatsu Y, et al. (2004) Maxillae and associated gnathodental specimens of Na-

cholapithecus kerioi, a large-bodied hominoid from Nachola, northern Kenya. J HumEvol 46:365–400.

13. Pickford M (1986) Hominoids from the Miocene of East Africa and the phyletic positionof Kenyapithecus. Z Morphol Anthropol 76:177–130.

14. Harrison TA (1992) Reassessment of the taxonomic and phylogenetic affinities of thefossil catarrhines of Fort Ternan, Kenya. Primates 33:501–522.

15. Alpagut B, Andrews P, Martin L (1990) New hominoid specimens from the MiddleMiocene site at Pasalar, Turkey. J Hum Evol 19:397–422.

16. McCrossin ML, Benefit BR (1997) In Function, Phylogeny and Fossils: Miocene HominoidEvolution and Adaptation, eds Begun DR, Ward CV, Rose MD (Plenum, New York), pp327–362.

17. Cameron DW (2004) Hominid Adaptations and Extinctions (University of New SouthWales Press, Sydney), 235 pp.

18. Begun DR, Gulec E, Geraads D (2003) Dispersal patterns of Eurasian hominoids:Implications from Turkey. Deinsea 10:23–39.

19. Agustí J, Cabrera L, Garces M (2001) In Hominoid Evolution and Environmental Changein the Neogene of Europe. Volume 2. Phylogeny of the Neogene Hominoid Primatesof Eurasia, eds Bonis L de, Koufos GD, Andrews P (Cambridge Univ Press, Cambridge,UK), pp 2–18.

20. Van der Made J (1999) In The Miocene Land Mammals of Europe, eds Rossner GE,Heissig K (Friedrich Pfeil Verlag, Munchen), pp 457–472.

21. Andrews P, Martin L (1987) Cladistic relationships of extant and fossil hominoids. J HumEvol 16:101–118.

22. Begun DR, Ward CV, Rose MD (1997) In Function, Phylogeny and Fossils: MioceneHominoid Evolution and Adaptation, eds Begun DR, Ward CV, Rose MD (Plenum, NewYork), pp 389–415.

23. McCrossin ML, Benefit BR (1993) Recently recovered Kenyapithecus mandible andits implications for great ape and human origins. Proc Natl Acad Sci USA 90:1962–1966.

24. Begun DR, Nargolwalla MC, Hutchinson MP (2006) Primate diversity in the PanonianBasin: In situ evolution, dispersal, or both? Beitr Palaontol 30:43–56.

25. Begun DR (2007) In Handbook of Paleoanthropology, eds Henke W, Tattersall I(Springer, Heidelberg), pp. 921–977.

26. Pilbrow V (2006) Lingual incisor traits in modern hominoids and an assessment of theutility for fossil hominoid taxonomy. Am J Phys Anthropol 129:323–338.

27. Kelley J, Ward S, Brown B, Hill A, Duren DL (2002) Dental remains of Equatoriusafricanus from Kipsaramon, Tugen Hills, Baringo District, Kenya. J Hum Evol 42:39–62.

28. Begun DR (1992) Phyletic diversity and locomotion in primitive European hominids.Am J Phys Anthropol 87:311–340.

29. Ishida H, Kunimatsu Y, Takano T, Nakano Y, Nakatsukasa M (2004) Nacholapithecusskeleton from the Middle Miocene of Kenya. J Hum Evol 46:69–103.

30. Bookstein FL (1991) Morphometric Tools for Landmark Data: Geometry and Biology(Cambridge Univ Press, Cambridge, UK), 456 pp.

31. Rohlf FJ (2005) TpsDig, Digitize Landmarks and Outlines, Version 2.05 (Department ofEcology and Evolution, State University of New York, Stony Brook, NY).

32. Rohlf FJ, Slice D (1990) Extensions of the Procrustes method for optimal superimposi-tion of landmarks. Syst Zool 39:40–59.

33. Rohlf FJ (2003) TpsRelw, Relative Warps Analysis, Version 1.36 (Department of Ecologyand Evolution, State University of New York, Stony Brook, NY).

34. Bookstein FL (1996) In Advances in Morphometrics, eds Marcus LF, Corti M, Loy A,Naylor GJP, Slice DE (Plenum, New York), pp 153–168.

35. Martin L (1985) Significance of enamel thickness in hominoid evolution. Nature314:260–263.

9606 ! www.pnas.org"cgi"doi"10.1073"pnas.0811730106 Moya-Sola et al.

Supporting InformationMoya-Sola et al. 10.1073/pnas.0811730106SI TextSystematic Framework. In Table S1 we provide a systematicclassification of living and fossil Hominoidea to the tribe level,by further including extant taxa and extinct genera discussed inthis paper. Hominoidea are defined as the group constituted byHylobatidae and Hominidae, plus all extinct taxa more closelyrelated to them than to Cercopithecoidea. Hominidae, in turn,are defined as the group containing Ponginae and Homininae,plus all extinct forms more closely related to them than toHylobatidae. While this broad concept of Hominidae is currentlyused by many paleoprimatologists (e.g., refs. 1–2), the systematicposition of primitive (or archaic) putative hominoids is far fromclear (see below). Begun (3–5) employs the terms ‘‘Eohomi-noidea’’ and ‘‘Euhominoidea’’ to informally refer to hominoidsof primitive and modern aspect, respectively. These terms,however, are roughly equivalent to stem-lineage and crown-group hominoids, respectively, and are not further used here.

The systematic status of many Late Oligocene and Early-Middle Miocene fossil catarrhines, lacking both cercopithecoidsynapomorphies and crown hominoid synapomorphies, has beensubject to different interpretations (6–8). Harrison (8) arguesthat most fossil ‘‘apes’’ from the Late Oligocene and EarlyMiocene of Africa have not crossed the hominoid ‘‘cladisticthreshold’’ and classifies them into 2 distinct superfamilies(Dendropithecoidea and Proconsuloidea), which he considers tobe successive sister taxa of stem catarrhines, more derived thanpropliopithecoids and pliopithecoids, but presumably precedingthe cercopithecoid–hominoid split (8–12). On the other ex-treme, a few authors have considered that proconsuloids mightbe stem hominids (13–15). Although the latter view is apparentlyabandoned (16), most authors consider that Early Mioceneforms, especially Proconsul, are more closely related to extanthominoids than to cercopithecoids (3, 6–7, 16–20). The system-atic scheme used here (Table S1) follows Harrison (8) inrecognizing that proconsulids and dendropithecids are distinctclades, but considers that at least the former are stem hominoids.These putative stem hominoids lack crown hominoid synapo-morphies, in particular, features functionally related to ortho-grady that are presumably homologous between hylobatids andhominids. Yet stem hominoids already share with both hylo-batids and hominids some facial (16) and several postcranial(17–18, 21) synapomorphies. Prominently, the lack of externaltail in Proconsul (17, 19, 21), albeit disputed by some authors(22), is now firmly established (23) and has been further ascer-tained in the putative stem hominoid Nacholapithecus (24).

Among putative stem hominoids, the position of Afropithecusand other related taxa is the most problematic. Harrison (8)distinguishes a single family Proconsulidae with 3 distinct sub-families (Proconsulinae, Nyanzapithecinae, and Afropitheci-nae), while other authors (20) take an opposite approach byclassifying their Afropithecinae (including Nacholapithecus andEquatorius) within the Hominidae, and others (2) consider theformer to be a distinct family. Especially problematic is theplacement of Kenyapithecus and Griphopithecus: in some sys-tematic schemes (20, 25), the latter taxa are classified into thesubfamily Kenyapithecinae within the Hominidae; others (2)distinguish a subfamily Griphopithecinae (for Griphopithecus)within the Afropithecidae; and Begun reunites both Keny-apithecinae and Griphopithecinae into a distinct family Gripho-pithecidae (2) or combines these taxa into an informal grouping(‘‘griphopiths’’) of stem hominoids (5). The systematic schemefollowed in this paper (Table S1) recognizes the close phyloge-

netic relationships between all these taxa by classifying them allinto a single family Afropithecidae with 2 subfamilies (Keny-apithecinae and Afropithecinae).

The systematic scheme used here requires several nomencla-tural decisions, which deserve further explanation. The nominaKenyapithecini and Kenyapithecinae are adopted instead ofGriphopithecinae and Griphopithecini (see also ref. 25) merelybecause the former have priority. It is unclear why neither Begun(1) nor Kelley (2) specify the authorship of Griphopithecinae (orGriphopithecidae), but, to our knowledge, the authorship of thelatter nomina must be attributed to Begun (ref. 4, p. 232: Table10.1), which therefore do not have priority over KenyapithecinaeAndrews, 1992. Griphopithecinae thus remains potentially validonly if Kenyapithecus (and Afropithecus, see below) are excludedfrom it. This notwithstanding, there has been some confusionregarding the authorship of the nomen Kenyapithecinae. BothBegun (1) and Ward and Duren (20) attribute its authorship toLeakey (26). However, the earliest usage of a supragenericnomen with Kenyapithecus as the type genus is attributable toAndrews (6), who erected it as a previously undescribed tribe,Kenyapithecini, in the same paper that he erected Afropithecini(27). To our knowledge, Kenyapithecus and Afropithecus have notbeen previously included into a single family with the exclusionof Hominidae (as in ref. 20), but we have chosen the nomenAfropithecidae (instead of Kenyapithecidae) because the formerhas been already used at the family level—albeit with a differentmeaning—by some previous workers (2, 5). If, as suggested inthis paper, there is a close phylogenetic relationship betweenKenyapithecinae (in particular, the Kenyapithecini) and Hom-inidae, Afropithecidae as conceived here would be paraphyletic.Transferring the Kenyapithecinae into the Hominidae (25),however, would not solve this problem, since the remainingAfropithecidae (including only the Afropithecinae) would re-main paraphyletic in excluding the Kenyapithecinae. Paraphylycan be transferred from one group or rank to another, but cannotbe completely eliminated unless Linnean ranks are aban-doned—a view that is not advocated here.

The Late Miocene genera Hispanopithecus and Ouranopithe-cus are not classified here at the tribe level. Both genera havebeen previously suggested to be either pongines (28) or homi-nines (4), but this issue is not definitively resolved and liesoutside the scope of this paper. Nevertheless, it is worth men-tioning that Hispanopithecus was recently resurrected (29) forLate Miocene species previously included in Dryopithecus, sothat the latter genus is restricted to its type species, Dryopithecusfontani. As a result, the tribe Dryopithecini is here used with adifferent meaning from previous usages, to refer to MiddleMiocene stem hominids that apparently do not belong to any ofthe 2 crown-hominid subfamilies; this tribe includes Pierolapithe-cus, Dryopithecus s.s., and Anoiapithecus gen. nov., but mostlikely excludes Hispanopithecus. Given the uncertain phyloge-netic relationships between the several dryopithecin genera, it iscurrently unclear whether this tribe is paraphyletic or representsa clade of stem European hominids.

ResultsMorphometric Analyses. ANOVA comparisons show that thereare significant differences (P ! 0.001; F " 178.6) among severalextant catarrhines regarding the CFA (Table S2 and S3), withgorillas, colobines, and hylobatids displaying the highest values,which nevertheless generally do not surpass 60°. Living humans,on the contrary, differ from all of the remaining taxa (P ! 0.001)

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 1 of 10

by the much higher CFA of the former, while extinct homininsdisplay intermediate values. Most extinct catarrhines clearly fallwithin the ranges of extant nonhuman cercopithecoids andhominoids, with the exception of Anoiapithecus. The latter taxadisplay values of CFA beyond the maximum value recorded inextant nonhuman catarrhines, and #20° higher than other fossilapes, most closely resembling the values displayed by extinctmembers of the genus Homo.

The canonical variate analysis (CVA) indicates that Anoia-pithecus displays a unique morphology, previously unknownamong living and fossil hominoids, which confirms the need toerect a previously undescribed genus. This analysis (Table S4 andTable S5) correctly classifies 93% of the original cases, only withminor confusion between some chimpanzees and gorillas andbetween very few colobines, cercopithecines, and hylobatids.Stem catarrhines, stem hominoids, and even the stem great apePierolapithecus most closely resemble cercopithecines, whereasall previously known putative crown hominids show a morederived condition that closely resembles one of the several extantgreat ape genera (Table S5). Anoiapithecus differs from crownhominids not only by the lack of facial concavity (as reflected byCA1, which explains 66% of variance), but also by the highlyverticalized (orthognathous) facial profile (as reflected by CA2,which explains 23% of variance). Anoiapithecus, in particular,displays a vertical alignment of glabella, nasion, rhinion, andnasospinale in relation to the alveolar plane. Glabella andrhinion are more anteriorly situated, whereas rhinion and na-sospinale are slightly more posteriorly placed. Thus, in Anoia-pithecus, the rhinion is situated very close and only slightly moreanterior than the nasion. In regard to the latter, Anoiapithecusshows the opposite condition from Pierolapithecus, most closelyresembling hylobatids and colobines. In fact, Anoiapithecusclusters with colobines when the several canonical axes are takeninto account simultaneously. It is thus the only known fossil great

ape to have ever displayed a colobine-like facial profile, to whichit may have autapomorphically converged from an ancestralcondition more similar to that displayed by stem hominoids andliving cercopithecines.

Squared distances based on the CVA discriminant scores, andcomputed for pairs of fossil individuals included (Table S5),indicate that Anoiapithecus is particularly far from the 2 otherMiocene hominoids from Spain included in the analysis (Piero-lapithecus and Hispanopithecus); D. fontani could not be includedin the analysis because of incomplete preservation, although aprevious analysis of facial morphology based on new remainsfrom Abocador de Can Mata indicates a gorilla-like morphology(29), which is thus quite different from the Anoiapithecuscondition. The results of the randomization approach furtherconfirm that differences between Anoiapithecus and other fossilindividuals cannot be accommodated within the range of vari-ation of a single genus. When the chimpanzee, the orangutan, orthe gorilla distributions of intrageneric individual pair differ-ences are used, the null hypothesis can be rejected with at leastP ! 0.05 with the single exception of Ankarapithecus. The latteris the fossil individual closest to Anoiapithecus when all canonicalaxes are taken into account (Table S5), although both taxadisplay a different facial profile—as shown by the highly diver-gent discriminant scores for CA1 and, especially, CA2 (Fig.3A)—and cluster very far from one another (Fig. 3B). In anycase, when the macaque distribution is used, the null hypothesiscan be rejected with P ! 0.05 in all instances, including An-karapithecus. With regard to Pierolapithecus and Hispanopithe-cus, the null hypothesis can be rejected with P ! 0.001 in allinstances, i.e., irrespective of the distribution used. All thisevidence clearly indicates that differences between Anoiapithe-cus and other fossil individuals cannot be interpreted as repre-senting the extremes of the range of variation within a singletaxon, so that the erection of a genus is fully justified.

1. Begun DR (2002) In The Primate Fossil Record, ed Hartwig WC (Cambridge Univ Press,Cambridge, UK), pp 339–368.

2. Kelley J (2002) In The Primate Fossil Record, ed Hartwig WC (Cambridge Univ Press,Cambridge, UK), pp 369–384.

3. Begun DR, Ward CV, Rose MD (1997) In Function, Phylogeny and Fossils: MioceneHominoid Evolution and Adaptation, eds Begun DR, Ward CV, Rose MD (Plenum, NewYork), pp 389–415.

4. Begun DR (2001) In Hominoid Evolution and Environmental Change in the Neogeneof Europe. Volume 2. Phylogeny of the Neogene Hominoid Primates of Eurasia, edsde Bonis L, Koufos G, Andrews P (Cambridge Univ Press, Cambridge, UK), pp 231–268.

5. Begun DR (2005) Sivapithecus is east and Dryopithecus is west, and never the twainshall meet. Anthropol Sci 113:53–64.

6. Andrews P (1992) Evolution and environment in the Hominoidea. Nature 360:641–646.7. Fleagle JG (1999) Primate Adaptation and Evolution (Academic, San Diego), 2nd Ed,

608 pp.8. Harrison T (2002) In The Primate Fossil Record, ed Hartwig WC (Cambridge Univ Press,

Cambridge, UK), pp 311–338.9. Harrison T (1987) The phylogenetic relationships of the early catarrhine primates: a

review of the current evidence. J Hum Evol 16:41–80.10. Harrison T (1988) A taxonomic revision of the small catarrhine primates from the early

Miocene of East Africa. Folia Primatol 50:59–108.11. Harrison T, Rook L (1997) In Function, Phylogeny and Fossils: Miocene Hominoid

Evolution and Adaptation, eds Begun DR, Ward CV, Rose MD (Plenum, New York), pp327–362.

12. Rossie JB, Simons EL, Gauld SC, Rasmussen DT (2002) Paranasal sinus anatomy ofAegyptopithecus: Implications for hominoid origins. Proc Natl Acad Sci USA 99:8454–8456.

13. Walker A, Teaford M (1989) The hunt for Proconsul. Sci Am 260:76–82.14. Walker A (1997) In Function, Phylogeny and Fossils: Miocene Hominoid Evolution and

Adaptation, eds Begun DR, Ward CV, Rose MD (Plenum, New York), pp 209–224.15. Rae T (1997) In Function, Phylogeny and Fossils: Miocene Hominoid Evolution and

Adaptation, eds Begun DR, Ward CV, Rose MD (Plenum, New York), pp 59–77.

16. Rae TC (1999) Mosaic evolution in the origin of the Hominoidea. Folia Primatol70:125–135.

17. Kelley J (1997) In Function, Phylogeny and Fossils: Miocene Hominoid Evolution andAdaptation, eds Begun DR, Ward CV, Rose MD (Plenum, New York), pp 173–208.

18. Rose MD (1997) In Function, Phylogeny and Fossils: Miocene Hominoid Evolution andAdaptation, eds Begun DR, Ward CV, Rose MD (Plenum, New York), pp 79–100.

19. Ward CV (1997) In Function, Phylogeny and Fossils: Miocene Hominoid Evolution andAdaptation, eds Begun DR, Ward CV, Rose MD (Plenum, New York), pp 101–130.

20. Ward SC, Duren DL (2002) In The Primate Fossil Record, ed Hartwig WC (CambridgeUniv Press, Cambridge, UK), pp 385–397.

21. Ward CV, Walker A, Teaford MF (1991) Proconsul did not have a tail. J Hum Evol21:215–220.

22. Harrison T (1998) Evidence for a tail in Proconsul heseloni. Am J Phys Anthropol26(Suppl):93–94.

23. Nakatsukasa M, et al. (2004) Tail loss in Proconsul heseloni. J Hum Evol 46:777–784.24. Nakatsukasa M, et al. (2003) Definitive evidence for tail loss in Nacholapithecus, an East

African Miocene hominoid. J Hum Evol 45:179–186.25. Cameron DW (2004) Hominid Adaptations and Extinctions (University of New South

Wales Press, Sydney), 235 pp.26. Leakey LSB (1962) A new lower Pliocene fossil primate from Kenya. Ann Mag Nat Hist

4(Ser 13):689–697.27. Pickford M, Kunimatsu Y (2005) Catarrhines from the Middle Miocene (ca. 14.5 Ma) of

Kipsaraman, Tugen Hills, Kenya. Anthropol Sci 133:189–224.28. Kohler M, Moya-Sola S, Alba DM (2001) In Hominoid Evolution and Environmental

Change in the Neogene of Europe. Volume 2. Phylogeny of the Neogene HominoidPrimates of Eurasia, eds de Bonis L, Koufos G, Andrews P (Cambridge Univ Press,Cambridge, UK), pp 192–212.

29. Moya-Sola S, et al. (2009) First partial face and upper dentition of the Middle Miocenehominoid Dryopithecus fontani from Abocador de Can Mata (Valles-Penedes Basin,Catalonia, NE Spain): Taxonomic and phylogenetic implications. Am J Phys Anthropol,in press.

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 2 of 10

Fig. S1. Situation of Abocador de Can Mata (ACM). The old rubbish dump is indicated in yellow, whereas the area under exploitation and/or excavation isindicated in orange. The location of the classical sites of this area is indicated by circles, whereas the location of the published hominoid-bearing localities isindicated by a cranium. Three hominoid-bearing sites are geographically close (albeit at different stratigraphic horizons) at C3: C3-Aj and C3-Az, which haveyielded remains of Dryopithecus fontani, and C3-Aj, which is the type locality of Anoiapithecus brevirostris gen. et sp. nov.; BCV1 is the type locality ofPierolapithecus catalaunicus. Abbreviations: BCV, Barranc de Can Vila; BDA, Bassa de Decantacio d’Aigues Pluvials; BDL, Bassa de Lixiviats; CCV, Camí de Can Vila;VIE, Vial Intern d’Explotacio; C, Cel!la.

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 3 of 10

Fig. S2. Composite polarity stratigraphy of the ACM local series and correlation with the astronomically tuned geomagnetic polarity timescale ATNTS2004.The stratigraphic situation of the classical sites of Can Mata I and III, together with that of the published hominoid-bearing localities, is indicated. Stratigraphyis modified from ref. 5.

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 4 of 10

Table S1. Systematic classification of living and fossil Hominoidea at the tribe level, includingall extant genera and extinct taxa included in this paper

Order Primates, Linnaeus, 1758Semiorder Haplorrhini, Pocock, 1918

Suborder Anthropoidea, Mivart, 1864 (" Simiiformes, Hoffstetter, 1974)Infraorder Catarrhini, É. Geoffroy Saint-Hilaire, 1812

Superfamily incertae sedisFamily Dendropithecidae*, Harrison, 2002

Superfamily Hominoidea, Gray, 1825Family Proconsulidae*, L.S.B. Leakey, 1963

Subfamily Proconsulinae*, L.S.B. Leakey, 1963Genus Proconsul*, Hopwood, 1933

Subfamily Nyanzapithecinae*, Harrison, 2002Genus Turkanapithecus*, R. E. Leakey and M. G. Leakey, 1986

Family Afropithecidae*, Andrews, 1992Subfamily Afropithecinae*, Andrews, 1992

Tribe Afropithecini*, Andrews, 1992Genus Afropithecus*, R. E. Leakey and M. G. Leakey, 1986Genus Heliopithecus*, Andrews and Martin, 1987Genus Morotopithecus*, Gebo et al., 1997

Subfamily Kenyapithecinae*, Andrews, 1992Tribe Kenyapithecini*, Andrews, 1992

Genus Kenyapithecus*, L. S. B. Leakey, 1962Genus Griphopithecus*, Abel, 1902

Tribe Equatorini*, Cameron, 2004Genus Equatorius*, Ward et al., 1999Genus Nacholapithecus*, Ishida et al., 1999

Family Hylobatidae, Gray, 1870Genus Hylobates, Illiger, 1811

Family Hominidae, Gray, 1825Subfamily incertae sedis

Tribe Dryopithecini*, Gregory and Hellman, 1939Genus Dryopithecus*, Lartet, 1856Genus Pierolapithecus*, Moyà-Solà et al., 2004Genus Anoiapithecus*, gen. nov.

Tribe incertae sedisGenus Hispanopithecus*, Villalta and Crusafont, 1944Genus Ouranopithecus*, de Bonis and Melentis, 1977

Subfamily Ponginae, Elliot, 1913Tribe Pongini, Elliot, 1913

Genus Pongo, Lacépède, 1799Genus Sivapithecus*, Pilgrim, 1910Genus Ankarapithecus*, Ozansoy, 1957

Subfamily Homininae, Gray, 1825Tribe Gorillini, Frechkop, 1943

Genus Gorilla, I. Geoffroy Saint-Hilaire, 1853Genus Pan, Oken, 1816

Tribe Hominini, Gray, 1825Genus Homo, Linnaeus, 1758Genus Australopithecus*, Dart, 1925Genus Paranthropus*, Broom, 1938

*Extinct taxa.

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 5 of 10

Table S2. Descriptive statistics of the craniofacial angle (CFA) in extant genera and values for extinct taxa included in the analysis

Taxon N Mean SD 95% CI Range

Macaca 55 46.8 4.3 45.6 47.9 35 56Cercopithecus 16 52.7 5.1 50.0 55.4 37 61Colobinae 26 55.7 5.1 53.7 57.8 48 65Pongo 17 44.4 5.6 41.5 47.2 34 52Hylobates s.l. 21 57.8 4.7 55.7 60.0 50 67Pan 63 51.1 4.3 50.0 52.2 40 59Gorilla 14 54.5 3.8 52.3 56.7 47 58Homo 31 78.5 3.9 77.0 79.9 68 85Papio 13 35.8 2.9 34.0 37.5 30 40Anoiapithecus 1 72Ankarapithecus 1 45Proconsul 1 46Pierolapithecus 1 43Sivapithecus 1 52Aegyptopithecus 2 43 0.0 43.0 43.0 43 43Ouranopithecus 1 47Afropithecus 1 36Hispanopithecus 1 52Turkanapithecus 1 46Victoriapithecus 1 49Paranthropus 3 62.0 7.5 43.2 80.8 54 69Australopithecus 2 66.0 0.0 66.0 66.0 66 66Fossil Homo 4 71.5 6.4 61.3 81.7 65 77

N, sample size; SD, standard deviation; CI, confidence interval.

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 6 of 10

Table S3. ANOVA and Bonferroni results for comparisons of the craniofacial angle (CFA) among extant taxa

Taxon Macaca Cercopithecus Colobinae Pongo Hylobates Pan Gorilla Homo

Cercopithecus 0.000Colobinae 0.000 0.000Pongo 0.000 1.000 0.000Hylobates s.l. 1.000 0.000 0.000 0.000Pan 0.000 0.026 0.000 1.000 0.000Gorilla 0.000 1.000 0.000 0.001 0.000 0.000Homo 0.000 1.000 0.000 1.000 0.000 1.000 0.424Papio 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000Cercopithecus ***Colobinae *** NSPongo NS *** ***Hylobates s.l. *** * NS ***Pan *** NS *** *** ***Gorilla *** NS NS *** NS NSHomo *** *** *** *** *** *** ***Papio *** *** *** *** *** *** *** ***

***, significant at P ! 0.001; *, significant at P ! 0.05; NS, nonsignificant.

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 7 of 10

Table S4. Main results of the canonical variate analysis (CVA) performed on the matrix of partial warp scores (PW), including theuniform component (UNI)

Discriminant functions (canonical axes)

CA1 CA2 CA3 CA4 CA5 CA6

Eigenvalues 9.512 3.355 0.870 0.436 0.143 0.067% of variance 66.1 23.3 6.0 3.0 1.0 0.5Cumulative % 66.1 89.5 95.5 98.5 99.5 100.0Canonical correlation 0.951 0.878 0.682 0.551 0.354 0.251

Standardized coefficients of the canonical discriminant functionsCA1 CA2 CA3 CA4 CA5 CA6

X1-PW1 $0.162 0.391 $0.417 0.253 0.088 0.522Y1-PW2 0.503 $0.185 0.046 0.405 $0.230 $0.106X2-PW3 $0.098 $0.186 0.616 0.790 0.187 $0.123Y2-PW4 0.442 $0.215 0.521 0.157 0.738 0.363X3-PW5 0.136 $0.106 $0.125 $0.453 $0.336 0.423Y3-PW6 $0.17 $0.099 $0.327 $0.146 $0.108 0.242X4-PW7 $0.842 $0.232 0.502 0.056 0.162 0.348Y4-PW8 $0.029 0.401 0.069 0.425 0.517 0.007X5-UNI1 0.617 0.059 0.37 0.273 $0.288 0.707Y5-UNI2 $0.214 0.573 0.841 $0.134 $0.353 $0.058

Functions at group centroidsCA1 CA2 CA3 CA4 CA5 CA6

Cercopithecus 4.513 $1.819 0.261 1.059 0.300 $0.747Colobinae 3.025 $3.100 0.228 $1.366 $0.185 0.041Gorilla $2.898 0.554 $1.596 $0.770 1.160 $0.057Hylobates s.l. 1.698 $2.631 $1.656 1.239 $0.055 0.505Pan $4.017 $0.171 $0.189 0.041 $0.316 $0.131Pongo $4.035 $2.131 3.175 0.592 0.659 0.359Macaca 1.819 1.848 0.207 $0.004 $0.036 0.073

Discriminant scores for fossil taxaCA1 CA2 CA3 CA4 CA5 CA6

Afropithecus turkanensis 3.157 2.870 2.584 $0.566 $1.345 $0.574Anoiapithecus brevirostris 0.942 $4.093 1.834 $1.603 0.750 $0.457Hispanopithecus laietanus $5.002 0.657 $1.852 $0.981 $2.609 $0.077Ouranopithecus macedoniensis $4.465 0.749 $1.860 $0.514 0.349 0.045Sivapithecus indicus $7.581 $3.490 1.281 1.675 0.255 $2.144Turkanapithecus kalakolensis 2.351 0.948 $1.142 $1.004 $0.450 $1.899Victoriapithecus macinnesi 1.573 $1.636 3.830 $2.987 $3.589 $2.354Proconsul heseloni 3.400 0.757 0.690 $0.558 $2.790 $0.130Pierolapithecus catalaunicus $0.947 3.489 2.773 $0.810 $0.079 0.023Aegyptopithecus zeuxis 4.279 $0.764 0.868 $0.673 $2.931 $1.547Ae. zeuxis 3.741 1.020 1.497 0.525 $2.845 $2.297Ankarapithecus meteai $2.343 $0.334 1.222 $1.732 $0.289 $0.259

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 8 of 10

Table S5. Classification results of the canonical variate analysis and squared Mahalanobis distance of Anoiapithecus to extantcentroids and fossil taxa

Classification of fossil taxa

Predicted group (first)Squared Mahalanobisdistance to centroid

Predicted group(second)

SquaredMahalanobis

distance to centroid

Afropithecus turkanensis Macaca 10.932 Cercopithecus 34.6Anoiapithecus brevirostris Colobinae 9.082 Hylobates s.l. 24.539Hispanopithecus laietanus Pan 10.726 Gorilla 18.75Ouranopithecus macedoniensis Gorilla 3.296 Pan 4.618Sivapithecus indicus Pongo 25.611 Pan 32.928Turkanapithecus kalakolensis Macaca 7.971 Cercopithecus 20.44Victoriapithecus macinnesi Colobinae 37.176 Macaca 52.733Proconsul heseloni Macaca 11.859 Cercopithecus 20.601Pierolapithecus catalaunicus Macaca 17.577 Pan 32.397Aegyptopithecus zeuxis Cercopithecus 15.613 Colobinae 17.98Ae. zeuxis Macaca 19.838 Cercopithecus 22.766Ankarapithecus meteai Pan 7.985 Gorilla 12.108

Classification of original cases (extant taxa)Cercopithecus Colobinae Gorilla Hylobates s.l. Pan Pongo Macaca

Cercopithecus 14 (93.4%) 1 (6.7%)Colobinae 1 (4.0%) 24 (96.0%)Gorilla 15 (100%)Hylobates s.l. 1 (5.9%) 16 (94.1%)Pan 10 (15.6%) 54 (84.4%)Pongo 11 (100%)Macaca 1 (1.1%) 1 (1.1%) 1 (1.1%) 92 (96.8%)

Squared Mahalanobis distance of Anoiapithecus to extant centroids and fossil taxa

Cercopithecus Colobinae Gorilla Hylobates s.l.27.77 9.08 49.13 24.54Pan Pongo Macaca Afropithecus48.01 35.91 42.17 59.43Hispanopithecus Ouranopithecus Sivapithecus Turkanapithecus83.29 67.92 87.15 40.13Victoriapithecus Proconsul Pierolapithecus Aegyptoptihecus34.76 44.60 63.48 38.75Aegyptoptihecus Ankarapithecus55.93 26.43

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 9 of 10

Table S6. Composition of the extant comparative sample employed in the morphometric analyses

Genus N (craniofacial angle) N (geometric morphometrics)

Cercopithecus 16 15Colobus 10 9Gorilla 14 15Homo 31Hylobates s.l. 21 17Macaca 55 95Pan 63 64Papio 13Pongo 17 11Presbytis 12 12Procolobus 4 4

N, sample size.

Moya-Sola et al. www.pnas.org/cgi/content/short/0811730106 10 of 10