The Vertebral Remains of the Late Miocene Great Ape Hispanopithecus Laietanus from Can Llobateres 2 (Valles-Pened� Es� Basin, NE Iberian Peninsula)

Home , Hispanopithecus, Morotopithecus, Proconsul (mammal)

Journal of Human Evolution 73 (2014) 15e34

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jhevol

The vertebral remains of the late Miocene great ape Hispanopithecus laietanus from Can Llobateres 2 (Valles-Pened es Basin, NE Iberian Peninsula)

* Ivette Susanna a, David M. Alba a, b, , Sergio Almecija c, a, d, Salvador Moya-Sol a e a Institut Catala de Paleontologia Miquel Crusafont, Universitat Autonoma de Barcelona, Ediﬁci ICP, Campus de la UAB s/n, 08193 Cerdanyola del Valles, Barcelona, Spain b Dipartimento di Scienze della Terra, Universita degli Studi di Torino, Via Valperga Caluso 35, 10125 Torino, Italy c Department of Anatomical Sciences, Stony Brook University School of Medicine, Stony Brook, NY 11794-8081, USA d NYCEP Morphometrics Group, USA e ICREA at Institut Catala de Paleontologia Miquel Crusafont, Universitat Autonoma de Barcelona, Ediﬁci ICP, Campus de la UAB s/n, 08193 Cerdanyola del Valles, Barcelona, Spain article info abstract

Article history: Here we describe the vertebral fragments from the partial skeleton IPS18800 of the fossil great ape Received 15 November 2012 Hispanopithecus laietanus (Hominidae: Dryopithecinae) from the late Miocene (9.6 Ma) of Can Llobateres Accepted 7 May 2014 2 (Valles-Pened es Basin, Catalonia, Spain). The eight specimens (IPS18800.5eIPS18800.12) include a Available online 18 June 2014 fragment of thoracic vertebral body, three partial bodies and four neural arch fragments of lumbar vertebrae. Despite the retention of primitive features (moderately long lumbar vertebral bodies with Keywords: slightly concave ventrolateral sides), these specimens display a suite of derived, modern hominoid-like Vertebrae features: thoracic vertebrae with dorsally-situated costal foveae; lumbar vertebrae with non-ventrally- Orthogrady Evolution oriented transverse processes originating from a robust pedicle, caudally-long laminae with caudally- Dryopithecinae oriented spinous process, elliptical end-plates, and moderately stout bodies reduced in length and Hominidae with no ventral keel. These features, functionally related to orthograde behaviors, are indicative of a Hominoidea broad and shallow thorax with a moderately short and stiff lumbar region in Hispanopithecus. Despite its large body mass (ca. 39e40 kg), its vertebral morphology is more comparable to that of hylobatids and Ateles than to extant great apes. This is conﬁrmed by our morphometric analyses, also indicating that Hispanopithecus most closely resembles Pierolapithecus and Morotopithecus among Miocene apes, whereas Proconsul and Nacholapithecus resemble pronograde monkeys. Only in a few features (craniocaudally short and transversely wide pedicles, transverse processes situated on the pedicle, and slight ventral wedging), Hispanopithecus is more derived towards the extant great ape condition than other Miocene apes. Overall, the vertebral morphology of Hispanopithecus supports previous inferences of an orthograde body plan with suspensory and climbing adaptations. However, given similarities with Ateles and the retention of a longer and more ﬂexible spine than in extant great apes, the Hispanopithecus morphology is also consistent with some degree of above-branch quadrupedalism, as previously inferred from other anatomical regions. © 2014 Elsevier Ltd. All rights reserved.

Introduction locomotor repertoire of each particular species combines various positional behaviors in different frequencies (Martin, 1990; Hunt The evolution of orthogrady in light of extinct apes et al., 1996). Two main types of positional (postural and locomotor) behaviors are usually distinguished among anthropoids (Keith, Based on limb use and type of substrate, primates are custom- 1903, 1923a; Hunt et al., 1996; Madar et al., 2002; Williams, 2011): arily classified into different locomotor categories, even though the pronogrady, which takes place on (sub)horizontal supports, with the trunk held roughly horizontally and the limbs being under compression; and orthogrady (or antipronogrady; Stern, 1975), in * Corresponding author. which the trunk is held vertically and the limbs may be frequently E-mail address: [email protected] (D.M. Alba). http://dx.doi.org/10.1016/j.jhevol.2014.05.009 0047-2484/© 2014 Elsevier Ltd. All rights reserved. 16 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 under tension. In relation to such positional differences, two diversity of hominoids. This hypothesis can be however tested opposite anatomical body plans (pronograde and orthograde) have based on the information provided by the fossil record. been distinguished among primates in the literature (Benton, 1974; To accept a hypothesis of adaptation, a historical concordance Ward, 1993, 2007; Moya-Sol a and Kohler,€ 1996b; Larson, 1998). The between the origin of a particular trait and the proper selection pronograde body plan is considered to represent the primitive regime to explain its evolution is required (Baum and Larson, 1991; anthropoid condition, being most suitable for quadrupedalism Lauder, 1991; Alba et al., 2003). In this sense, the study of the (with the limbs moving along the parasagittal plane). It is charac- postcranial morphology of Miocene apes, together with the loco- terized by a craniocaudally long, mediolaterally narrow and motor inferences that can be drawn from them, based on mor- dorsoventrally deep thorax; small clavicles, laterally-positioned phofunctional considerations, is of utmost significance for testing scapulae and ventrally-facing glenoid fossae; narrow and laterally whether (or which) orthograde features should be considered an facing iliac blades; arms shorter than legs; and a long and flexible adaptation (Gebo, 1996) instead of an exaptation (Cartmill, 1985) spine. The orthograde body plan, in turn, is derived for crown relative to suspensory behaviors. Central to this discussion is the hominoids and suitable for antipronograde (generally forelimb- postcranial evidence provided by the Miocene great apes Pier- dominated) behaviors. It is characterized by a shorter, wider and olapithecus and Hispanopithecus (Hominidae: Dryopithecinae1), for shallow thorax; longer and stouter clavicles, dorsally-situated which a locomotor repertoire combining orthograde behaviors scapulae and laterally-facing glenoid fossae; expanded and with above-branch, powerful-grasping palmigrady (unknown dorsally-rotated iliac blades; forelimbs generally longer than hin- among extant apes) has been inferred (Moya-Sol a and Kohler,€ dlimbs; a short and stiffened vertebral column (especially the 1996b; Kohler€ et al., 2001; Moya-Sol a et al., 2004; Almecija et al., lumbar spine) without an external tail; and loss of direct ulnocarpal 2007, 2009; Alba et al., 2010, 2012a; Alba, 2012; Pina et al., 2012; articulation. Tallman et al., 2013). On the basis of these and other extinct The vertebral column is one of the anatomical regions in which hominoids, it has been therefore argued that the body plan of the the orthogrady-related features of extant hominids are most clearly last common ancestor of crown hominoids would have been more expressed (Mivart, 1865; Keith, 1903, 1923b; Schultz, 1930, 1961; primitive than inferred from its surviving living counterparts, Rose, 1975; Shapiro, 1990, 1993, 1995; Johnson and Shapiro, 1998; thereby implying a large degree of homoplasy in orthograde fea- Williams, 2011). As a result of the broader and shallower thorax, tures between hylobatids and hominids, if not also between pon- in extant hominoids the spine is more ventrally situated within the gines and hominines (see discussion in Larson, 1998; Alba, 2012). ribcage (approaching the center of mass during upright postures), which is reflected in the position of costal foveae and of the lumbar The IPS18800 skeleton from Can Llobateres transverse processes, among other features. Moreover, the short and stiff lumbar region of crown hominoids is reflected in a lower The dryopithecine Hispanopithecus laietanus is recorded from number and different morphology of the vertebrae (e.g., shorter several early to late Vallesian localities (MN9 and MN10, late and less ventrally-wedged bodies with no ventral keel). These Miocene) from the Valles-Pened es Basin (Catalonia, Spain; features enable making locomotor inferences in extinct apes based Casanovas-Vilar et al., 2011; Alba, 2012; Alba et al., 2012a,b). Many on vertebral morphology (Walker and Rose, 1968; Ward, 1991, of the remains of this taxon come from the site of Can Llobateres in 1993; Sanders and Bodenbender, 1994; Moya-Sol a and Kohler,€ Sabadell, including the early Vallesian locality of Can Llobateres 1 1996b; MacLatchy et al., 2000; MacLatchy, 2004; Moya-Sol a et al., (CLL1) and the slightly younger, late Vallesian locality of Can Llo- 2004; Filler, 2007; Nakatsukasa et al., 2007; Nakatsukasa, 2008; bateres 2 (CLL2) (Agustí et al., 1996; Casanovas-Vilar et al., 2011; Russo and Shapiro, 2013). Alba et al., 2011a). Whereas most of the dental hominoid remains of Although all extant hominoids (hylobatids and hominids) share Can Llobateres come from CLL1 (Alba et al., 2012b, and references a common orthograde body plan, they differ from one another in therein), the most complete cranial and postcranial remains come the frequency and types of positional behaviors included in their from CLL2, including a partial cranium IPS18000 (Moya-Sol a and respective locomotor repertoires (e.g., Hunt, 1991, 2004). However, Kohler,€ 1993, 1995) and associated postcranial skeleton IPS18800 except for the bipedal modern humans, all crown hominoids (Moya-Sol a and Kohler,€ 1996b). practice both vertical climbing and some type of below-branch Based on several features of the CLL2 skeleton, especially from suspensory behaviors (Hunt, 1991, 2004). This has led to various the hand (Moya-Sol a and Kohler,€ 1996b; Almecija et al., 2007; interpretations as to whether orthogrady should be considered an Deane and Begun, 2008), it has been argued that this taxon com- adaptation for forelimb-dominated arboreal behaviors in general bined suspensory adaptations with some plesiomorphic features (e.g., Ward, 1993), for suspensory behaviors only (e.g., Gebo, 1996), functionally-related to above-branch quadrupedalism, although to or for vertical climbing and cautious slow clambering instead a lesser extent than in Pierolapithecus (Moya-Sol a et al., 2004; (Cartmill and Milton, 1977; Cartmill, 1985; Sarmiento, 1995)dsee Almecija et al., 2009; Alba et al., 2010; for a different interpreta- Williams (2011), and references therein, for a more detailed his- tion in this regard, see Deane and Begun, 2008, 2010). The retention torical account. Suspensory behaviors might be selectively advan- of primitive, stem hominoid-like pronograde behaviors in tageous for allowing large-bodied apes to retain access to food H. laietanus is also supported by several features of the more partial items in the periphery of the canopy (Hunt, 1992), although, in fact, skeleton from Can Feu (Alba et al., 2012a), suggesting that the climbing is more frequently employed during feeding activities acquisition of below-branch suspensory adaptations did not imply among extant apes (Gebo, 1996). In this regard, it has been hy- the complete abandonment of quadrupedal behaviors, thus rein- pothesized that, from climbing-adapted ancestors, suspensory be- forcing the mosaic and homoplastic nature of locomotor evolution haviors might have evolved as a less risky or expensive locomotion in the Hominoidea (Moya-Sol a et al., 2004; Almecija et al., 2007; during travel in relation to a progressive increase in body mass Alba et al., 2010, 2012a; Alba, 2012). (Cartmill, 1985). If correct, many orthograde features, originally The purported combination of both suspensory and arboreal evolved as adaptations for climbing, would have been subsequently quadrupedal behaviors in H. laietanus is unknown among extant co-opted to perform a suspensory role (Cartmill, 1985). Given that hominoids and hence, to reconstruct its total morphological current role must not be conflated with historical origin (Gould and Vrba, 1982), it is not possible to discern the original target of selection of orthograde features in light of the currently decimated 1 See Alba (2012) for the systematics employed in this paper. I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 17

Table 1 costal deﬁnition): KNM-MW 13142-I, J and K, corresponding Vertebral specimens of Hispanopithecus laietanus IPS18800 from CLL2. respectively to a ﬁrst, antepenultimate and penultimate lumbar Catalogue no. Brief description vertebrae of Proconsul nyanzae from Mfangano Island (Kenya;

IPS18800.5 Partial vertebral body of a mid-lumbar vertebra including Ward, 1991, 1993; Ward et al., 1993); KNM-RU 2036-CY, CD and CZ, the right pedicle root corresponding respectively to three correlative lumbar vertebrae of IPS18800.6 Partial vertebral body of a penultimate or last lumbar indeterminate level of Proconsul heseloni from Rusinga Island vertebra including a small portion of the left pedicle root (Kenya; Walker and Pickford, 1983); KNM-BG 35250R, a lumbar IPS18800.7 Neural arch fragment from a lumbar (?) vertebra vertebra of indeterminate level of Nacholapithecus kerioi from IPS18800.8 Right fragment of a thoracic vertebral body of indeterminate level including the cranial and caudal costal facets, the Nachola (Kenya; Ishida et al., 2004; Nakatsukasa et al., 2007); UMP pedicle root and the prezygapophysis 67-28, a lower midlumbar vertebra of Morotopithecus bishopi from IPS18800.9 Partial vertebral body of a first or second lumbar vertebra Moroto (Uganda; Allbrook and Bishop, 1963; Walker and Rose, including a large portion of the left pedicle and transverse 1968; Sanders and Bodenbender, 1994; Nakatsukasa, 2008); and process root IPS18800.10 Partial neural arch of a lumbar vertebra including the lamina, IPS21350.64 and IPS21350.65, corresponding respectively to a the neural process root and the two postzygapophyses lower midlumbar vertebra and a less complete last lumbar IPS18800.11 Partial neural arch of a lumbar vertebra including the lamina, vertebra2 of Pierolapithecus catalaunicus from Barranc de Can Vila 1 the neural process root and the two postzygapophyses in Abocador de Can Mata (ACM/BCV1, Spain; Moya-Sol a et al., 2004; IPS18800.12 Neural arch fragment from a lumbar (?) vertebra Susanna et al., 2010a,b). For these taxa, high-quality casts housed at the ICP were examined, except in the case of Pierolapithecus, for which the original specimens were studied. The above-mentioned pattern, it is necessary to do further research on all of the taxa are interpreted as possessing a pronograde body plan (e.g., see anatomical regions available from this taxon. After the initial re- review in Ward, 2007, and references therein), except for Moroto- € ports of the IPS18800 skeleton (Moya-Sola and Kohler, 1996b; pithecus and Pierolapithecus, for which orthograde-related features € Kohler et al., 2001), which includes about 60 bones or bone frag- have been documented (Ward, 1993; Sanders and Bodenbender, ments, more detailed descriptions and analyses have been pub- 1994; MacLatchy et al., 2000; Moya-Sol a et al., 2004; Filler, 2007; € lished for the femur (Kohler et al., 2002; Pina et al., 2012; Almecija Ward, 2007). The lumbar vertebrae of Oreopithecus bambolii et al., 2013), the hand (Almecija et al., 2007; Deane and Begun, (IGF11778 and BA72) from Baccinello (Italy) were not included in 2008; Alba et al., 2010) and the tibia (Tallman et al., 2013). The the comparative sample, because their badly crushed condition vertebral remains, however, have been only preliminarily described precludes taking reliable measurements for most of the employed fi (Susanna et al., 2011a,b). Here we gure and provide a detailed variables (Russo and Shapiro, 2013); however, published evidence description of the vertebral specimens from CLL2, together with on this taxon was included in the Discussion. morphometric comparative analyses. On the basis of these comparisons, we discuss the body plan and locomotor repertoire Nomenclature for lumbar vertebral levels inferred for H. laietanus, as well as its implications for the origins of orthogrady in the Hominoidea. Customarily, vertebral segments are denoted by an acronym that corresponds to the vertebral region (‘T’ for thoracic and ‘L’ for Materials and methods lumbar), followed by an Arabic numeral that indicates level within that region (from cranial to caudal). According to this nomencla- Studied sample ture, the first lumbar vertebra should be labeled as ‘L1’. However, this method is problematic for comparing vertebral levels among Fossil sample The available Hispanopithecus laietanus vertebral primates with different number of lumbar vertebrae. To solve this remains, belonging to the partial skeleton IPS18800 from Can Llo- problem, here we follow Sanders and Bodenbender's (1994) € bateres 2 (Moya-Sola and Kohler, 1996b), include up to eight nomenclature, which consists in labeling the lumbar vertebrae specimens (IPS18800.5 to IPS18800.12; Table 1), corresponding to with Roman numerals (based on role similarity in force trans- the thoracic and lumbar regions of the spine. All of them are only mission and the relative branching of lumbar spinal nerves). Ac- partially preserved, in many instances including a portion of the cording to this nomenclature, the last lumbar vertebra is labeled as body and the root of the pedicle and/or the transverse processes, ‘L VII’, the penultimate vertebra as ‘LVI’, and so forth. For example, although several partial neural arches are also included. The the seven lumbar levels of cercopithecoids would be labeled, from specimens are housed at the Institut Catala de Paleontologia last to first, as L VII to L I; in turn, the four lumbar levels of Pan Miquel Crusafont (ICP; Barcelona, Spain). would be labeled as L VII to L IV. As such, generally the first lumbar Comparative sample A total of 438 lumbar vertebrae from 12 vertebra of Gorilla would be equivalent to the second of Pan, the genera of extant anthropoids were measured by one of the authors third of Hylobates, and the fifth of Papio, so that all these vertebrae (I.S.). This comparative sample includes platyrrhines (Ateles, would be labeled as L V. Alouatta) and cercopithecoids (Papio, Nasalis, Colobus), as well as lesser apes (Hylobates, Symphalangus, Hoolock, Nomascus) and great Measurements and statistical comparisons apes (Pongo, Gorilla, Pan)dsee Table 2 for further details on sample composition. Several specimens were measured at the ICP, but most Measurements and shape variables In spite of the fragmentary of the comparative sample comes from the following institutions: nature of the reported vertebral specimens of H. laietanus, American Museum of Natural History (AMNH, New York, USA); morphometric comparisons with other anthropoid primates are Royal Museum for Central Africa (RMCA, Tervuren, Belgium); and possible based on seven metrical variables previously employed by Smithsonian Institution, National Museum of Natural History Sanders and Bodenbender (1994) and other authors (Ward, 1991, (NMNH, Washington D.C., USA). All the specimens were wild-shot 1993; Sanders, 1998; Nakatsukasa et al., 2007; Nakatsukasa, and adult (based on both dental eruption and epiphyseal fusion criteria). The fossil comparative sample includes the following lumbar 2 Also termed ‘ultimate lumbar vertebra’ by some researchers (Nakatsukasa et al., vertebral specimens (lumbar segments identified according to the 2007). 18 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34

Table 2 Composition of the extant comparative sample used in this study.

Taxon Group Individuals Number of vertebrae

Male Female Total I II III IV V VI VII Total

Pongo pygmaeus Pongo 8 4 12 0 0 1 12 12 12 12 49 Gorilla beringei Gorilla 108 18000218181856 Gorilla gorilla Gorilla 2 2 4000144413 Pan troglodytes Pan 7 7 14 0 0 0 13 14 14 14 55 Pan paniscus Pan 5 5 10 0 0 0 10 10 10 10 40 Hylobates lar Gibbons 0 1 1 0 1 1 1 1 1 1 5 Hylobates klossi Gibbons 1 0 1 0 0 1 1 1 1 1 5 Hoolock hoolock Gibbons 0 2 2 0 0 2 2 2 2 2 10 Nomascus leucogenys Gibbons 1 0 1 0 0 1 1 1 1 1 5 Nomascus gabriellae Gibbons 1 0 1 0 0 1 1 1 1 1 5 Symphalangus syndactylus Symphalangus 1 2 3003333315 Ateles sp. Ateles 1 5 6016666630 Nasalis larvatus Nasalis 3 1 4444444420 Alouatta sp. Alouatta 4 4 8088888840 Papio sp. Papio 5 5 10810101010101050 Colobus guereza Colobus 4 4 8888888840

‘Group’ refers to the groups used in boxplots (Figs. 4e5) and labels used in the PCA analysis (Figs. 6e7; Table 5). Roman numerals denote lumbar vertebral segments (for further details, see text and Sanders and Bodenbender, 1994).

2008). These variables (see definition in Fig. 1 and Table 3)were Body mass estimation Body mass (BM, in kg) was estimated based measured to the nearest 0.1 mm in the most complete available on caudal surface area of the vertebral body (ACCA, in mm2) and the specimens of H. laietanus (IPS18800.9 and IPS18800.6; Table 4)as allometric equation derived by Sanders and Bodenbender (1994, well as in the comparative sample. Measurements for specimens their Table 2) for lumbar level L VI in non-human catarrhines: ln of other Miocene hominoids (Pierolapithecus catalaunicus, BM ¼0.29*ln ACCA þ 0.64. ACCA was measured with Adobe Morotopithecus bishopi, Proconsul nyanzae and Nacholapithecus PhotoShop CS4 in IPS18800.6, and compared with ACCA kerioi) were also taken from casts and originals by the authors or measurements for other taxa (Proconsul nyanzae, Morotopithecus from the published literature (Sanders and Bodenbender, 1994; bishopi and Pierolapithecus catalaunicus) taken from the literature. Nakatsukasa et al., 2007; see Table 4 for further details). The geometric mean (GM) of six of the above-mentioned vari- Description ables, including dorsoventral height, mediolateral width, dorsal craniocaudal length and ventral craniocaudal length of the body, as Thoracic region well as craniocaudal length and transverse width of the pedicle (respectively, VH, VB, VLD, VLV, LP and BP), was computed as a IPS18800.8 (Fig. 2aeb) This partial vertebral body is the only proxy of overall vertebral size. To perform comparisons of vertebral available specimen from the thoracic region of the vertebral col- shape, six ‘Mosimann’ shape ratios (Mosimann, 1970; Jungers et al., umn of H. laietanus. It partially preserves the right portion of the 1995) were computed by dividing each of the above-mentioned vertebral body, showing a well-developed, oval cranial costal linear measurements by their GM. The relative position of the articular surface and a subtle but discernible caudal costal facet. It transverse process (RTP) was size-adjusted also by the GM of the also preserves the right prezygapophysis and a portion of the other variables, whereas vertebral wedging (measuring the differ- right transverse process root. The shape and size of the vertebral ence between dorsal and ventral craniocaudal length of the body) body cannot be ascertained due to insufficient preservation. The was measured by means of an additional shape index VLV/VLD. cranial costal facet is dorsally situated at the level of the junction between the body and the pedicle. The caudal costal facet, in Morphometric analyses Shape ratios were compared between the turn, is relatively small compared with the former, being situated fossil and extant specimens by means of boxplots, separately for on the caudal edge of the body at about the same height as the each lumbar vertebral level (as defined above). Specific statistical cranial one. The pedicle, albeit incompletely preserved, seems to comparisons for these ratios between extant taxa and vertebral have been short and robust. The prezygapophysis faces dorsally levels are beyond the scope of this work. Instead, overall vertebral (i.e., the articular surface is horizontally aligned) and slightly shape affinities in our sample were summarized by means of a laterally, and originates close to the body (due to the shortness of Principal Components Analysis (PCA) performed on the covariance the pedicle), resulting in a small cranial notch. The transverse matrix of the seven Mosimann shape variables defined above. The process does not originate very dorsally, but it might have been wedging index (VLV/VLD) was not included in the PCA, because dorsally oriented (although this cannot be unambiguously both VLV and VLD were already included in the analysis as Mosi- determined because it is broken close to its root). Given the mann variables standardized by GM. Although some of the morphology and situation of the costal facets, the orientation of measured variables, and hence the corresponding shape ratios, the prezygapophysis, the size of the cranial notch, and the might be significantly correlated with size, a detailed allometric position of the transverse process, this specimen probably analysis of each of these variables is beyond the scope of this paper. corresponds to some thoracic level between T2 and T5. Instead, to assess size scaling effects on lumbar vertebral morphology (i.e., predictable covariation between vertebral shape Lumbar region and size), we relied on linear regressions of those principal components (PCs) resulting from the PCA providing meaningful IPS18800.9 (Fig. 3aec; see measurements in Table 4) This spec- discrimination among taxa against GM in the extant comparative imen partially preserves the left portion of the body, most of the sample. All analyses were computed by means of SPSS v. 16.0. cranial articular surface, and a small portion of the caudal surface. I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 19

Figure 1. A lumbar vertebra of Pan troglodytes in cranial (left) and left lateral (right) views, showing the measurements taken in this study. See variable abbreviations and deﬁ- nitions in Table 3.

On the lateral surface, a portion of the left pedicle and the root of due to is fragmentary preservation, the particular lumbar level the transverse process are also preserved. It can be ascertained that cannot be determined. the transverse process originates from the robust pedicle (instead e of the body or the pedicle-body junction; Fig. 4f), and that it IPS18800.6 (Fig. 3m o; see measurements in Table 4) This spec- displays a dorsal or coplanar (horizontal) orientation (i.e., not imen preserves most of the vertebral body, including a small ventral). The cranial articular surface displays a roughly rounded portion of the left pedicle root, which originates from the dorsal contour that is only slightly reniform (i.e., kidney-shaped, due to margin of the vertebral body and seems to be robust. The root of the some dorsal concavity) and moderately elliptical (slightly wider transverse process, in contrast, is not preserved, thereby indicating mediolaterally than dorsoventrally high). The body seems that it originated from a more dorsal position (i.e., from the pedicle craniocaudally elongated,3 and displays a somewhat concave instead of the pedicle-body junction). The body is proportionally ventrolateral side and some slight degree of ventral wedging (i.e., shorter craniocaudally than in IPS18800.9. Although only a small the body is craniocaudally somewhat shorter ventrally than portion of the cranial intervertebral surface is preserved, the dorsally; see Table 4). Due to insufficient preservation, it cannot caudal one is almost completely preserved. The latter displays a fi be conclusively ascertained whether a ventral keel was present, suboval pro le that is more elliptical (mediolaterally wider although the preserved morphology suggests that it was relative to dorsoventral height) than in IPS18800.9. Unlike in the notdand that, if present, it would not have been very developed. latter specimen, in IPS18800.6 the dorsal edge is straight, and The incomplete preservation of the specimen also hinders although the ventral edge is slightly more pointed, the overall determining whether this specimen had an accessory process. In morphology of the caudal end-plate is neither reniform nor non-human primates, ventral wedging generally attains its cordiform (heart-shaped). This is related to the lack of a ventral maximum close to the center of the thoracolumbar kyphosis keel, which together with the overall morphology of IPS18800.6 (Ward, 1991), thus being higher in the upper than in the lower compared with the other described lumbar specimens (body less fi portion of the lumbar region. Therefore, the presence of some elongated and more elliptical articular surface) justi es its fi ventral wedging in IPS18800.9, coupled with other morphological identi cation as a penultimate or last lumbar vertebra. details (the small size of the root of the transverse process IPS18800.10 and IPS18800.11 (Fig. 3hej and 3per, respectively) relative to that of the body, and the somewhat elongated body; These specimens are two neural arches of lumbar vertebrae, Schultz, 1961; Ward, 1991), suggest that this specimen which besides the laminae also preserve the root of the spinous corresponds to the upper portion of the lumbar region. processes and the postzygapophyses. The spinous processes, ac- IPS18800.5 (Fig. 3dee) This specimen preserves a right portion of cording to their preserved roots, would have displayed a caudal the vertebral body, including a small part of its cranial articular orientation, which can be most clearly appreciated in IPS18800.11 surface but not of the caudal one. The root of the right pedicle is (Fig. 5b). The lamina is relatively wide and caudally extended until preserved on the dorsal part of the body, showing that the former the origin of the postzygapophyses, being pointed downwards on was robust. The preserved portion of the body displays a moder- its midpoint due to the caudal origin and orientation of the ately concave ventrolateral surface and suggests some degree of spinous process. The mammillary processes and the ventral wedging. Although wedging cannot be measured in this prezygapophyses are not preserved in these specimens. The specimen, due to its incomplete preservation, our qualitative oblique orientation of the articular surfaces of the assessment of the preserved morphology suggests that both the postzygapophyses, together with the caudal orientation of the degree of elongation of the body as well as of ventral wedging spinous processes, indicate that these fragmentary specimens would have been lower than in IPS18800.9, but higher than in correspond to lumbar vertebrae. Although the level within this IPS18800.6. On this basis, we tentatively suggest that IPS18800.5 region cannot be determined, the slightly larger size of would correspond to a lumbar level intermediate between IPS18800.11 suggests a lower level for this specimen compared IPS18800.9 and IPS18800.6 (i.e., to a midlumbar vertebra), although with IPS18800.10.

IPS18800.7 and IPS18800.12 (Fig. 3feg and 3kel) These specimens probably correspond to right (IPS18800.7) and left (IPS18800.12) 3 In this descriptive section, by ‘elongated’ we simply mean qualitatively long in relation to width. See next section for a quantitative numerical assessment of prezygapophyses of lumbar vertebrae, but their fragmentary relative length of the body. preservation precludes providing any additional description. 20 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34

Table 3 Measurements (in mm) taken on the vertebral specimens from CLL2 (see Fig. 1 for further details); deﬁnitions and abbreviations after Sanders and Bodenbender (1994).

Abbreviation Variable Deﬁnition

VH Vertebral body height Dorsoventral diameter of the cranial surface (or caudal, when cranial is unavailable) of the vertebral body along the sagittal plane VB Vertebral body width Maximum transverse (mediolateral) diameter of the cranial surface (or caudal, when cranial is unavailable) of the vertebral body VLD Dorsal vertebral body length Maximum length of body at craniocaudally thickest portion of dorsal annular ring at midline VLV Ventral vertebral body length Maximum length of body at craniocaudally thickest portion of ventral annular ring at midline BP Pedicular width Minimum transverse diameter of pedicle LP Pedicular length Minimum craniocaudal length of pedicle RTP Transverse process height Distance of mid-point of the transverse process root from the dorsal margin of the vertebral body

Table 4 Measurements (in mm) of the two most complete lumbar vertebrae of Hispanopithecus laietanus from CLL2, compared with other extinct hominoids.

Level VH VB VLD VLV BP LP GM RTP VLV/VLD

Hispanopithecus laietanus IPS18800.9a IIIeIV 20.6 (27.8)b 25.8 (24.6) 6.8 13.5 17.5 2.7 0.95 Hispanopithecus laietanus IPS18800.6a VIeVII 18.9c 27.7c 22.2 21.3 (7.1) (10.7) (16.3) e 0.96 Pierolapithecus catalaunicus IPS21350.64a VeVI 21.5 30.0 26.2 22.2 7.2 16.0 18.7 4.4 0.84 Pierolapithecus catalaunicus IPS21350.65a VII ee e21.9 ee e e e Morotopithecus bishopi UMP 67-28d IVeVI 24.4 33.7 27.5f 27.4 8.0 18.1 21.2 4.6f 1.00 Proconsul nyanzae KNM-MW 13142Kd VI 21.0 30.0 31.0f 29.0 5.5 20.2 19.9 5.5f 0.94 Proconsul nyanzae KNM-MW 13142Jd V 22.1 30.0 32.2f 27.3 4.5 22.5 19.7 6.1f 0.85 Nacholapithecus kerioi KNM-BG 35250Re IIIeVI (16.2) (26.6) (21.3) (17.6)f 4.2 16.7 (15.0) 5.5f (0.83)

See abbreviations in Table 3. Estimated values (due to partial preservation or deformation) are reported within parentheses; when, due to incomplete preservation, a variable cannot be estimated, the maximum measurable portion is preceded by ‘>’. a Measurements taken in this study from original specimens. See text for lumbar level identiﬁcation in H. laietanus, and Moya-Sol a et al. (2004) and Susanna et al. (2010a,b) for P. catalaunicus. b Maximum preserved (cranial) VB in IPS18800.9 is >25.3 mm, but the reported value corresponds to an estimate based on a mirror projection of the right (complete) side to the left (incomplete) one. c In IPS18800.6, VH and VB were measured on caudal instead of the cranial surface due to incomplete preservation. d Measurements taken from Sanders and Bodenbender (1994, their Table 2) unless explicitly indicated. e Measurements taken from Nakatsukasa et al. (2007) unless explicitly indicated. f Measurements taken in this study from casts.

Morphometric comparisons and body mass estimates Relative body height (Fig. 6a) Extant great apes (especially Pan and Pongo) and, to a lesser extent, siamangs display taller vertebral Shape variables and functional indices bodies than gibbons and monkeys (including Ateles), only showing slight overlap with Papio and Nasalis. African apes as well as Papio, All of the studied Miocene apes are similar in vertebral size Nasalis and Ateles (but not Pongo, hylobatids, Colobus and Alouatta) (Table 4), with H. laietanus and P. catalaunicus being slightly smaller exhibit a pattern in which the central lumbar levels are higher than than both M. bishopi and Pr. nyanzae, and somewhat larger than their most cranial and caudal counterparts. Hispanopithecus dis- N. kerioi. The lumbar vertebral shape ratios derived in this paper by plays similarly tall bodies as Pierolapithecus and Morotopithecus, standardizing measurements by GM (taken as overall vertebral thus in the low range of Pan, Pongo and Symphalangus, in the size) are reported by means of boxplots in Figs. 6 and 7, and midrange of Gorilla, and in the upper range of Papio and Nasalis. described in greater detail below (GM is further depicted in Fig. 8). Nacholapithecus and Proconsul, in contrast, show proportions more The following shape ratios are meant to assess morphological similar to some monkeys (Colobus and Alouatta exhibit lower patterns in our sample and easily evaluate the afﬁnities of the values) and gibbons, although they also overlap with the low range Hispanopithecus specimens. of Gorilla. Only KNM-MW13142K overlaps with the upper range of Alouatta.

Relative body width (Fig. 6b) This variable shows a similar pattern to relative body height, although with less differences between great apes and cercopithecoid monkeys (in fact, the ranges of all extant genera overlap to some degree). Papio falls in the upper range of extant great apes. Gibbons and siamangs do not differ much, and atelids (both Ateles and Alouatta) exhibit the narrowest vertebral bodies. For this variable, only Papio and Nasalis show a pattern equivalent to that found for relative body breadth, so that the bodies of their midlumbar vertebrae are relatively higher than those of upper or lower vertebrae. Contrarily (and to a lesser degree), Pongo, hylobatids and Alouatta exhibit the opposite pattern. Hispanopithecus is in the low range of extant great apes and Papio, and in the midrange of hylobatids, Colobus and Nasalis; only Figure 2. Partial upper thoracic vertebra (IPS18800.8) of Hispanopithecus laietanus IPS18800.9 overlaps with the upper range of atelids, similarly as from CLL2, in cranial (a) and right lateral (b) views. both Morotopithecus and Pierolapithecus. These taxa show wider I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 21

Figure 3. Lumbar vertebrae of Hispanopithecus laietanus from CLL2. aec: partial ﬁrst or second lumbar vertebra IPS18800.9, in cranial (a), left lateral (b) and ventral (c) views; dee: partial midlumbar vertebra IPS18800.5, in cranial (d) and right lateral (e) views; feg: right prezygapophysis of a partial neural arch of lumbar vertebra IPS18800.7, in dorsal (f) and ventral (g) views; hej: partial neural arch of lumbar vertebra IPS18800.10, in dorsal (h), left lateral (i) and ventral (j) views; kel: left prezygapophysis of a partial neural arch of lumbar vertebra IPS18800.12, in dorsal (k) and ventral (l) views; meo: partial neural arch of lumbar vertebra IPS18800.11, in dorsal (m), left lateral (n) and ventral (B) views; per: partial penultimate or last lumbar vertebra IPS18800.6, in caudal (p), left lateral (q) and ventral (r) views. vertebrae than Proconsul, but not Nacholapithecus, which shows the pattern in which the midlumbar vertebral bodies are relatively wider vertebral bodies among extinct apes. longer (both on the ventral and dorsal sides) than in upper and Relative body length (Fig. 6c, d) All orthograde primates (both lower lumbar vertebrae. Regarding ventral length, hominoids and Ateles) display shorter lumbar bodies than mon- Hispanopithecus shows, like Proconsul and Morotopithecus, keys, both on the ventral (Fig. 6c) and the dorsal (Fig. 6d) side, with relatively long vertebral bodies, largely overlapping with all the only exception of lumbar levels L IeIII and L VII of Papio (which cercopithecoid monkeys. They also overlap with the uppermost resembles orthograde primates). Also, only monkeys (including range of African apes and Ateles, and the lowest range of Alouata. Alouatta, but not Ateles, which follows the ape pattern) display a Pierolapithecus and Nacholapithecus show somewhat shorter 22 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34

Figure 4. The most complete lumbar vertebral specimen of Hispanopithecus laietanus from CLL2 compared with selected extant and extinct hominoids, in cranial view. a: Papio cynocephalus;b:Proconsul nyanzae KNM-MW 13142-J; c: Nacholapithecus kerioi KNM-BG 35250R; d: Pierolapithecus catalaunicus IPS21350.64; e: Morotopithecus bishopi UMP 67.28; f: H. laietanus IPS18800.9; g: Hylobates lar;h:Ateles sp.; i: Pongo pygmaeus;j:Pan troglodytes. The continuous white line denotes the junction between the pedicle and the body, to help ascertain the origin of the transverse processes in each taxon.

Figure 5. Origin and orientation of the spinous process in a lumbar vertebra of Hispanopithecus laietanus from CLL2 compared with selected extant and extinct hominoids, in left lateral view. a: Proconsul nyanzae KNM-MW 13142-J; b: H. laietanus IPS18800.11; c: Pongo pygmaeus;d:Hylobates lar;e:Papio anubis;f:Pan troglodytes;g:Ateles sp. The white line intersects with the origin of the spinous process, thus showing its caudal orientation in H. laietanus and some other taxa (Pan, Ateles and Pongo), contrasting with that in Pr. nyanzae, Papio and Hylobates. vertebral bodies, in the midranges of extant hominoids, Ateles, and great apes and the mid range of hylobatids), whereas the upper and lower lumbar levels of Papio. In contrast, for dorsal Nacholapithecus and Proconsul overlap exclusively with the mid length, most Miocene apes are in general more monkey-like than ranges of all monkeys. extant hominoid-like, although with differences among extinct Transverse process position (Fig. 7a) This feature clearly distin- taxa. Thus, Hispanopithecus, Pierolapithecus and Nacholapithecus guishes great apes and siamangs, which display transverse pro- are just above the great ape range (IPS18800.6 slightly cesses arising from a dorsal position (i.e., from the pedicle or overlapping with the upper range of Pan), but well in the mid-to- lamina; positive values), from all cercopithecoids, which display upper range of Symphalangus and Ateles, as well as in the low transverse processes arising more ventrally (from the vertebral range of Papio and Nasalis. Proconsul clearly displays longer body; negative values). Gibbons and Ateles display values around bodies, exclusively overlapping with monkeys (but not Ateles). 0 for this variable, reflecting the origin of the transverse processes Differences between ventral and dorsal relative length of the close to the junction between the pedicle and the body. Alouatta,in fl vertebral body partly re ect differences in the degree of wedging turn, is intermediate between Ateles and the remaining monkeys. (Fig. 7b; see below). Most of the studied taxa tend to display more dorsally situated Relative pedicle width (Fig. 6e) Although with some overlap, transverse processes in the first and last lumbar vertebrae than in extant hominoids and Ateles are characterized by a higher relative the remaining lumbar segments (this is less marked in Sympha- width of the pedicle than in pronograde monkeys. The former taxa langus, although it might be related to its comparatively smaller (with the exception of Pongo) exhibit a tendency to have wider sample size); Ateles is the exception, since all lumbar levels display pedicles towards the lower lumbar levels (which is particularly comparable values. The only vertebra of Hispanopithecus in which accentuated in Gorilla). In contrast, pronograde monkeys display this feature can be ascertained (IPS18800.9) shows a positive value, narrower pedicles in the midlumbar vertebrae than in upper and, which is most similar to those of hylobatids, overlapping with the especially, lower lumbar levels. Hispanopithecus, like Pierolapithecus upper range of gibbons and the lower range of siamangs, and also and Morotopithecus, resembles extant apes and Ateles (although slightly with the lower-most range of Pan and Pongo. overlapping with the upper range of Papio and Alouatta). In Pierolapithecus and Morotopithecus, with slightly more positive contrast, both Nacholapithecus and Proconsul fall exclusively within values, overlap with the lower ranges of all great apes and the the range of pronograde monkeys. mid range of Symphalangus. In contrast, Nacholapithecus and Relative pedicle length (Fig. 6f) The craniocaudal length of the Proconsul, with more ventrally-situated transverse processes, pedicle does not show the same signal as its width. All taxa share a overlap with cercopithecoids. similar pattern of having shorter pedicles towards the lower lum- Wedging (Fig. 7b) Within each extant taxon, ventral wedging bar vertebra, which is especially accentuated in the last one. This tends to decrease from the upper to the lower lumbar region (a pattern is more accentuated in hylobatids than in extant great apes, value of 1 indicates no wedging, and positive values dorsal instead and even more so in all of the monkey taxa analyzed. In general, of ventral wedging). When differences among taxa are considered, when the last lumbar vertebra (or both the last and penultimate although there is substantial overlap, some trends can be appreci- ones, depending on the taxon) are not considered, it can be ated. Symphalangus, Papio and Ateles exhibit more ventral wedging appreciated that extant great apes exhibit shorter pedicles than all than the remaining taxa. Overall, Gorilla displays a similar degree of monkeys, while hylobatids are intermediate. In this regard, Hispa- wedging in all lumbar vertebrae, which are characterized by less nopithecus falls in the mid-to-low range of extant great apes (and ventral wedging (or even dorsal wedging) compared to the the lowest range of the last lumbar vertebrae of Papio and Ateles). remaining taxa (except for the last lumbar vertebrae of Pan, Pongo Pierolapithecus and Morotopithecus (with virtually the same value) and Colobus). The two specimens of Hispanopithecus show very display somewhat longer pedicles (in the upper range of extant similar values of wedging (almost inexistent), thus fully I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 23 overlapping with great apes but also with the upper ranges of the Gorilla), albeit significant (p < 0.001), only explain a relatively small last lumbar vertebrae of hylobatids and monkeys. Morotopithecus amount of vertebral shape variance (12.5% and 6.2%, respectively); even displays slight dorsal wedging, whereas in contrast Pier- moreover, the regression for hylobatids plus Ateles is not even olapithecus, Nacholapithecus and Proconsul show more clear ventral significant at all (p ¼ 0.879). When predictable covariation is wedging, and hence most closely resemble the condition of both inspected for the whole sample in PC2 and PC3, a significant shape hylobatids (especially Symphalangus), Ateles and Papio. esize correlation is found (p < 0.001 and p < 0.01, respectively), but only accounting for a very small amount of variance (3%). Overall, Principal Components Analysis and size-shape covariation these results indicate that only a very small portion of lumbar vertebral shape, as captured by our PCA, can be accounted for size- The results of the PCA summarizing overall lumbar vertebral scaling effects (i.e., allometry). This is also illustrated by the shape are reported in Table 5 and Fig. 8. The major axis of shape different lumbar vertebral shapes displayed by the various inves- variation (i.e., PC1, accounting for 75% of total variance; Fig. 8a, b) tigated Miocene apes, which in spite of being similar in vertebral clearly shows that great apes differ in lumbar morphology from size to extant great apes (they overlap at least with the range of Pan extant pronograde monkeys, with Ateles displaying an interme- and Pongo, except for Nacholapithecus, which is slightly smaller), diate condition most similar to that of hylobatids (especially they show closer resemblances to either hylobatids-Ateles (Pier- gibbons, since siamangs are slightly more great ape-like in this olapithecus, Morotopithecus and Hispanopithecus) or pronograde regard). Some of the last lumbar vertebrae of Papio occupy also monkeys (Nacholapithecus and Proconsul). The somewhat ‘inter- this intermediate position on the shape space (tending to be more mediate’ vertebral form shared by Morotopithecus, Pierolapithecus ape-likethaninsmallercercopithecoids;Fig. 8aed). Principal and Hispanopithecus is also approximated by a few Papio speci- component1 therefore separates the great apes (towards the mens, corresponding to last lumbar vertebrae of this taxon (Fig. 8). positive end) from the pronograde monkeys (towards the negative end), mostly on the basis of the more dorsally-situated Body mass estimates transverse processes, wider pedicles, and higher but shorter vertebral bodies of the former. Along this axis, the vertebra of The measurement of caudal surface area in IPS18800.6 Hispanopithecus IPS18800.9 (estimated to correspond to level III (ACCA ¼ 504 mm2) yields an estimated BM of 40 kg (Table 6), thus eIV) falls in this intermediate position (mostly overlapping with very similar to that of 39 kg previously derived from femoral head gibbons and Ateles, as well as to some degree with the lower for the same individual of Hispanopithecus laietanus (Moya-Sol a lumbar vertebrae of baboons). Both Pierolapithecus and Moroto- et al., 2009). This suggests that vertebral size in this taxon does pithecus fall very close to one another along PC1, only overlapping not scale differently from extant non-human catarrhines. Note also with hylobatids (both gibbons and siamangs). In contrast, that differences in estimated BM between Hispanopithecus, Pier- Nacholapithecus and especially Proconsul display more negative olapithecus and Morotopithecus, but not Proconsul, are consistent values than the other Miocene apes, thus overlapping with pro- with differences in GM (Table 6). nograde monkeys. The two remaining axes (PC2 and PC3, respectively accounting for 13% and 6% of the variance) essentially Discussion discriminate taxa based on the width of the vertebral body, with wider vertebrae tending to display lower values along PC2 and Morphologic and morphometric comparisons higher values along PC3. When the morphospace defined by PC1 and PC2 is considered (88% total shape variation; Fig. 8a), Pier- The main morphological features of the vertebral specimens of olapithecus and Morotopithecus fall very close to one another, Hispanopithecus laietanus from CLL2 are summarized below. The whereas Hispanopithecus falls not far from them. These three only available partial thoracic vertebral body is too incompletely Miocene apes do not overlap with any extant taxa, and occupy an preserved to warrant quantitative analysis. However, based on its intermediate position between Ateles-gibbons, Symphalangus and preserved morphology, it can be ascertained that it displays a great apes. Nacholapithecus overlaps exclusively with Papio, dorsally situated costal facet, a robust and short pedicle, a moder- whereas Proconsul overlaps with Colobus, Nasalis and only ately dorsal origin and orientation of the transverse process, a marginally with Papio. When the morphospace defined by PC1 nearly dorsally-oriented prezygapophysis and an almost absent and PC3 (81% total shape variation; Fig. 8b) is inspected, Hispa- cranial notch (as it is typical of upper thoracic vertebrae). In the nopithecus overlaps marginally with Papio only. It is worth above-mentioned features, the thoracic vertebra of H. laietanus nothing that Hispanopithecus occupies an intermediate position more closely resembles the condition of extant hominoids. Simi- on the morphospace, only approximated by two Papio last lumbar larly, the various available lumbar specimens of H. laietanus vertebrae (Fig. 8d)dwhich, as we have shown above, exhibit a generally most closely resemble the condition displayed by extant similar morphology to that of Ateles and gibbons (Fig. 8a, c). apes (particularly hylobatids) as well as Ateles. As shown by our We inspected the relationship between overall lumbar vertebral analysis (Fig. 9), the latter display a convergent lumbar vertebral shape and size by means of a bivariate plot of PC1 scores against the form with that of gibbons, in agreement with similarities in fore- GM (Fig. 9). It can be appreciated that cercopithecoid monkeys do limb and overall thorax morphology noted by previous authors not overlap with the great ape shape, even though they fully (Erikson, 1963; Larson, 1998). In some instances, however, Hispa- overlap in vertebral size (especially Papio). Conversely, hylobatids nopithecus more closely resembles the morphology of extant great (especially Symphalangus) overlap with great apes in shape despite apes. Given that, for many details of lumbar vertebral morphology, their much smaller size (on the lower end of the monkey range). It extant great apes differ from hylobatids, a more detailed discussion is also worth noting that gibbons and Ateles fully overlap in both on the lumbar vertebrae of Hispanopithecus is provided below. vertebral shape and size, in spite of their long separate phyletic Proportions of the vertebral body vary to some extent from the histories. However, a trend seems at least visually evident in Papio. first towards the last lumbar vertebrae, although there are also When regressions of shape versus size are computed separately differences in this regard between taxa. The first or second lumbar for the above-mentioned groups with similar vertebral form (i.e., vertebra of Hispanopithecus (IPS18800.9) is very similar in pro- shape plus size), the regression for pronograde monkeys (i.e., portions to the vertebrae of Morotopithecus and Pierolapithecus, and Alouatta, Nasalis, Colobus and Papio) and great apes (Pongo, Pan and hence taller and wider than that of Proconsul, thus most closely 24 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 25

Figure 7. Boxplots depicting two functionally-relevant indices of the lumbar vertebrae of Hispanopithecus laietanus in comparison to extant and extinct hominoids. a: position of the vertebral transverse processes (RTP), size-adjusted by overall vertebral size, approximated by the geometric mean (GM) of the six remaining linear measurements (Fig. 4); positive values identify specimens with the transverse process arising from the pedicle, while negative values (separated from the former by a vertical thick line) identify specimens with transverse process arising from the vertebral body. b: wedging index, computed as the ratio between ventral and dorsal body lengths (VLV/VLD); values <1 indicate ventral wedging, and values >1 dorsal wedging. Box represents 25th and 75th percentiles, respectively, centerline is median, whiskers are nonoutlier range, dots are outliers, and stars represent extreme outliers. The different lumbar levels are represented for each taxon. The values for the vertebral specimens of Hispanopithecus laietanus IPS18800 are projected onto the remaining taxa to facilitate visual comparisons. See Table 3 for a definition of the variables and Table 4 for the measurements of the fossil specimens. The color code (online version) helps visualizing major taxonomic groups and extant great ape genera. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) approaching the morphology of hylobatids (Fig. 6a, b). The penul- moderately more elongated body (more marked in IPS18800.9) timate or last lumbar vertebra of this taxon (IPS18800.6), in turn, compared with extant hominoids and Ateles, thus being more differs from IPS18800.9 by being slightly wider, thus fitting the similar in craniocaudal length to the lumbar vertebrae of other pattern of the lower-most lumbar levels (Fig. 6b) and more closely Miocene apes except Proconsul, which more closely approaches the approaching the condition of great apes in spite of overlapping more elongated vertebral shape of pronograde monkeys (Fig. 6c, d). more clearly with gibbons. Both specimens also display a In spite of the relatively elongated lumbar vertebral bodies, the two

Figure 6. Boxplots depicting six shape ratios in the lumbar vertebrae of Hispanopithecus laietanus compared with extant and extinct hominoids and selected anthropoid monkeys. Each linear dimension is represented relative to overall vertebral size, approximated by the geometric mean (GM). a: body height (VH); b: body width (VB); c: ventral body length (VLV); d: dorsal body length (VLD); e: pedicular width (BP); f: pedicular length (LP). Box represents 25th and 75th percentiles, respectively, center line is median, whiskers are nonoutlier range, dots are outliers, and stars represent extreme outliers. The different lumbar levels are represented for each taxon. The values for the two vertebral specimens of Hispanopithecus laietanus IPS18800 are projected onto the remaining taxa to facilitate visual comparisons. See Table 3 for a deﬁnition of the variables and Table 4 for the measurements of the fossil specimens. The color code (online version) helps visualizing major taxonomic groups and extant great ape genera. (For interpretation of the references to colour in this ﬁgure legend, the reader is referred to the web version of this article.) 26 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34

Figure 8. Principal Components Analysis (PCA; see Table 5) summarizing overall lumbar vertebral shape in our sample of anthropoid primates. The analysis was performed on the covariance matrix of the seven shape ratios displayed in Fig. 6aeg. a, c: PC1 (75.3%) and PC2 (12.7%); b, d, PC1 and PC3 (6.2%). Panels aeb are taxon-coded (as in Figs. 6e7), whereas panels ced are vertebral level-coded instead. more complete lumbar vertebrae of Hispanopithecus display a very derived than those of other Miocene apes (except Morotopithecus), restricted (almost non-existent) degree of ventral wedging which rather more closely resemble the intermediate condition of (Fig. 7b), thus most closely resembling extant great apes. In this hylobatids and Ateles (characterized by a moderate degree of regard, the lumbar vertebrae of Hispanopithecus appear more ventral wedging, which is higher than in great apes but lower than in pronograde monkeys). Table 5 Qualitatively, the shape of the end-plates or articular surfaces of Results of the Principal Components Analysis (PCA) summarizing overall lumbar Hispanopithecus also resemble those of great apes (as well as vertebral shape in our sample of extant and fossil anthropoid primates. Morotopithecus and Pierolapithecus) in approaching a circular (only PC1 PC2 PC3 slightly reniform) contour, especially in IPS18800.9, thus differing % of Variance 75.368 12.699 6.160 from the more elliptical and markedly reniform morphology of Cumulative % 75.368 88.067 94.227 hylobatids and, especially, the markedly cordiform intervertebral Variable loadings profile of pronograde monkeys (Ward, 1991; Sanders and VH/GM 0.705 0.235 0.473 Bodenbender, 1994, Fig. 4). This is attributable not only to the VB/GM 0.448 ¡0.601 0.569 VLD/GM ¡0.930 0.260 0.184 vertebral body proportions, but also to the lack of a ventral keel in VLV/GM ¡0.819 0.458 0.232 the lumbar vertebrae of Hispanopithecus for which this trait can be BP/GM 0.871 0.132 0.398 ascertained, thereby resembling all extant hominoids in this regard LP/GM ¡0.926 0.110 0.027 (Ward, 1991). In Hispanopithecus, this can be more securely ascer- RTP/GM 0.925 0.370 0.065 tained for the penultimate or last lumbar vertebra (IPS18800.6) Only the three first axes (containing >94% of total vertebral shape variation) pro- than for the first or second one (IPS18800.9). In this regard, His- vided a meaningful discrimination and thus are reported. The analysis was per- panopithecus also resembles the investigated Miocene apes (Pier- formed on the covariance matrix of the seven shape ratios displayed in Fig. 4aef and 5a. Therefore, each original dimension was size-adjusted by the geometric mean olapithecus, Morotopithecus and Nacholapithecus, the latter (GM). Variables with absolute loadings >0.5 are marked in bold. displaying a medial ventral bulge but no well-defined keel; I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 27

morphology, since their pedicles are as wide as but somewhat longer than in great apes and Hispanopithecus (although shorter than in pronograde monkeys). The longer pedicles of hylobatids and Ateles might be related to their also longer bodies relative to great apes, so it is remarkable that Hispanopithecus displays short pedicles in spite of its somewhat elongated lumbar vertebral bodies. The orientation and position of the transverse processes in Hispanopithecus is in agreement with the great-ape morphology of the pedicle. Thus, the roots of the transverse processes in this taxon have a dorsal or coplanar (horizontal) orientation (as in all extant hominoids), instead of a ventral orientation (as in extant pronograde monkeys and stem hominoids such as Proconsul). Further- more, our quantitative assessment indicates that the origin of the transverse processes in Hispanopithecus is comparable to that in Morotopithecus, Pierolapithecus and siamangs, thus being slightly more ventral than in extant great apes, but more dorsal than in gibbons-Ateles and, especially, pronograde monkeys, Proconsul and Nacholapithecus (Fig. 7a). However, Hispanopithecus is somewhat more derived towards the extant great ape condition not only when compared with monkeys and stem hominoids (such as Proconsul and Nacholapithecus), in which the transverse processes originate from the vertebral body (Fig. 4aec), but also when compared with more modern-looking extinct apes (Morotopithecus and Pier- olapithecus), in which they originate from the junction between the body and the pedicle, as in Ateles and hylobatids; (Fig. 4dee, g, h). This is because in Hispanopithecus these processes originate from Figure 9. Overall lumbar vertebral shape and size. Bivariate plot representing the the pedicle, although still closer to the pedicle-body junction, major axis of shape variation (i.e., PC1, accounting for >75% of total variance; see Fig. 8) versus overall vertebral size as approximated by the geometric mean (GM). Color code instead of from the upper portion of the pedicle or even the lamina as in Fig. 8. as in great apes (Fig. 4f, i, j). In other qualitatively inspected morphological features, the lumbar vertebrae of Hispanopithecus also resemble the extant Nakatsukasa, 2008) other than Proconsul, which is more similar to hominoid condition instead of that of pronograde monkeys. They pronograde monkeys (Fig. 4). On the other hand, Hispanopithecus include the obliquely-oriented (ca. 45) rather than ventrally- still displays somewhat ventrolaterally concave lumbar vertebral oriented (ca. 0) articular surfaces of the postzygapophyses bodies, thus being intermediate between the more concave (Shapiro, 1993), the mediolaterally wider and caudally longer morphology of monkeys and the more convex of extant hominoids lamina (authors' own observations), and the clearly caudal orien- (Ward, 1991). tation of the spinous process (Fig. 5; Slijper, 1946; Ward, 1991; With regard to other anatomical structures of the lumbar Shapiro, 1993; Sanders and Bodenbender, 1994). vertebrae, the proportions of the pedicle deserve particular atten- In agreement with the morphological comparisons provided tion. Hispanopithecus most closely resembles extant great apes by above, the results of our PCA (Figs. 8 and 9) clearly show that His- displaying a robust (both wide and short) pedicle (Schultz, 1961; panopithecus, like Morotopithecus and Pierolapithecus, is consider- Ward, 1991; Sanders and Bodenbender, 1994), which contrasts ably derived in lumbar vertebral morphology towards the crown- with the more slender (narrower but longer) pedicle of stem hominoid condition, being characterized among other features by hominoids (Nacholapithecus and Proconsul) as well as pronograde dorsally-situated transverse processes, wider pedicles, and higher monkeys (Fig. 6e, f). Interestingly, like hylobatids and Ateles,both but moderately shorter vertebral bodies. In these regards, all of Morotopithecus and Pierolapithecus display an intermediate these putatively orthograde Miocene apes clearly differ from the more plesiomorphic morphology displayed by extant pronograde monkeys and retained by stem hominoids (Nacholapithecus and Proconsul). A similar morphology is shown by Ateles, as a result of Table 6 Body mass estimates in Hispanopithecus laietanus and other Miocene hominoids evolutionary convergence related to orthograde behaviors (see based on the caudal articular surface of the lumbar vertebrae, compared to the discussion below), as well as by the last lumbar vertebrae of Papio geometric mean of six vertebral measurements (employed in this study as a variable (probably to some extent because of size-scaling effects among of overall vertebral size). pronograde monkeys). Taxon Specimen Level ACCA BM GM Our analysis, in any case, indicates that only a small portion of lumbar vertebral shape variance is associated with differences in Hispanopithecus laietanus IPS18800.6 L VIeVII 504 40.1 (16.3) Pierolapithecus catalaunicus IPS21350.64 L VeVI 570a 43.4 18.7 vertebral size, thus suggesting that the vertebral shape of the Morotopithecus bishopi UMP 67-28 L IVeVI 760b 52.2 21.2 analyzed taxa is likely to mainly reflect locomotor differences (and Proconsul nyanzae KNM-MW 13142K L VI 470b 38.4 19.9 to some degree phyletic constraints) instead of scaling-size effects. Body mass estimates were derived based on Sanders and Bodenbender's (1994) This is evidenced by the fact that Papio overlaps in vertebral size equation (see text for further details). Estimated values (due to partial preserva- with all extant great apes (especially Pan and Pongo) in spite of clear tion or deformation) are reported within parentheses. shape differences (Fig. 9). According to our analysis, despite the Abbreviations: ACCA, caudal surface area of the vertebral body (in mm2); BM, possession of many shared-derived features among extant homi- estimated body mass (in kg); GM, geometric mean (in mm). a Taken from Moya-Sol a et al. (2004). noids, the lumbar vertebrae of hylobatids display a particular b Taken from Sanders and Bodenbender (1994). morphology of their own, which much more closely resembles that 28 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 of spider monkeys, and which is somewhat intermediate (i.e., less a modal number of five lumbar vertebrae (more than 80% in- derived) between that of pronograde monkeys and that of extant dividuals of Hylobates lar), whereas in Symphalangus syndactylus great apes. Interestingly, despite the possession of several great the presence of four and five lumbar vertebrae is equally frequent ape-like features (short pedicle, reduced wedging and transverse (49%). In Ateles geoffroyi, the modal number is four (Pilbeam, 2004), processes situated on the pedicle), Hispanopithecus is, like other although in our sample of Ateles there were also individuals with putatively orthograde Miocene hominoids (Morotopithecus and five lumbar vertebrae and a single individual with as much as six. Pierolapithecus), most similar to the hylobatid-Ateles condition, in The greater morphological similarities of the lumbar vertebrae of spite of overlapping in body size with the range of extant great apes Hispanopithecus with those of hylobatids and Ateles therefore lead (chimpanzees and female orangutans). us to tentatively favor the possession of four (as in siamangs and orangutans) or five (as in gibbons and Ateles) lumbar segments in Morphofunctional interpretation Hispanopithecus. However, it will not be possible to know for certain how many lumbar levels this taxon possessed until a Thoracic vertebra In spite of its fragmentary nature, the single complete lumbar region is eventually found. Therefore, at this available thoracic vertebra of Hispanopithecus displays some point, the presence of as much as six lumbar vertebrae cannot be derived features with extant apes, including the dorsally-situated completely ruled out. costal facet on the vertebral body, the dorsal origin and Shape of lumbar vertebral bodies Besides their higher number of orientation of the transverse process, and the short and robust lumbar elements, cercopithecids also display longer and overall less pedicle. Such features suggest for Hispanopithecus the possession robust lumbar vertebrae than great apes (Keith, 1923b; Benton, of a mediolaterally broad and dorsoventrally shallow thorax, with 1965; Ankel, 1972; Rose, 1975; Ward, 1991), as shown by our a spinal column ventrally situated relative to the ribcage (Schultz, morphometric analyses. The reduced lumbar region of extant 1961; Ward, 1991; Shapiro, 1993; Moya-Sol a and Kohler,€ 1996b). hominoids, especially great apes, is therefore attributable both to However, the incomplete preservation of the transverse processes a lesser number of vertebrae and a reduction in the length of the precludes a secure determination of the height of the articulation vertebral bodies. Such configuration results in a stiffer (less with the rib, so that it is not possible to conclusively ascertain flexible) lumbar spine, but helps to reduce the stresses whether the ribcage proportions of Hispanopithecus were extant experienced by the intervertebral disks due to both bending and great ape-like or instead more similar to those of hylobatids. compressive loads (e.g., Rose, 1975; Ward, 1991). The reduction of Number of lumbar segments The lumbar vertebrae are customarily the lumbar region in extant hominoids has been related to the distinguished from thoracic ones on the basis of either the absence possession of an orthograde body plan ever since Keith (1923b), of ribs or zygapophyseal orientation (e.g., Shapiro, 1993), but such although different functional explanations have been proposed in conflation of costal and zygapophyseal definitions is problematic relation to different orthograde positional behaviors. Ward (1991) (Williams, 2011, 2012). The orientation of zygapophyses should be emphasized the role in minimizing stresses experienced during treated as a separate morphology (Williams, 2011, 2012), but orthograde behaviors, and attributed the less reduced lumbar unfortunately the position of the diaphragmatic vertebra cannot spine of hylobatids (more and longer vertebrae compared with be determined in Hispanopithecus due to incomplete great apes) to the smaller body size of the former. In contrast, preservation. The same applies to the number of lumbar Cartmill and Milton (1977) emphasized the role of a reduced vertebrae according to costal definition. Despite some variation lumbar region in controlling thorax movements during bridging among extant primate species in the number of lumbar vertebrae behaviors, whereas Jungers (1984) argued that this feature would (Schultz and Strauss, 1945; Schultz, 1961), extant hominoids are help counter buckling the vertebral column during vertical characterized by a reduction of the number of lumbar segments climbing at least in African apes. Overall, the main functional (Schultz, 1961; Pilbeam, 2004): generally five in hylobatids and reason explaining the evolution of very short and stiff lumbar humans; four in orangutans; and generally three to four in regions in extant great apes remains unclear (Shapiro, 1993). In chimpanzees and gorillas, although bonobos may have up to five this regard, the lumbar vertebrae of Hispanopithecus are still (Schultz, 1961; Pilbeam, 2004; McCollum et al., 2010). The somewhat elongated in comparison to extant hominoids, derived extant-hominoid condition in this regard contrasts with although to a lesser extent than in pronograde monkeys and the longer lumbar region retained by Old World monkeys (seven Proconsul. Although, as explained above, the exact number of segments), and to a large extent still displayed by the stem lumbar vertebrae in Hispanopithecus is unclear (see above), the hominoid Proconsuldwith an estimated number of six lumbar moderate reduction of vertebral body length displayed by this segments (Ward, 1991, 1993). Such lumbar reduction is related to taxon suggests that it already displayed a somewhat shortened the orthograde body plan of extant hominoids, which is and stiffened lumbar region (as compared at least with Proconsul characterized not only by a broad and shallow thorax, but also by and Nacholapithecus). This is also reinforced by the morphology a stiffened lumbar region (e.g., Filler, 1981). It is noteworthy that of the lumbar vertebral bodies in Hispanopithecus, which are a similar reduction in the number of lumbar segments has been characterized by the apparent lack of a ventral medial keel, convergently acquired by Ateles (Erikson, 1963; Larson, 1998). almost nonexistent ventral wedging, only slightly concave On the basis of the preserved vertebral bodies, the lumbar re- ventrolateral sides, and rather elliptical intervertebral surfaces. gion of Hispanopithecus had at least three different levels. However, Differences in ventral wedging have been related to the short- not all of them seem to be adjacent. In particular, the first or second ening and stiffening of the spine in hominoids, with pronograde lumbar vertebra (IPS18800.9) might be contiguous with the mid- monkeys displaying a more marked ventral wedging, especially lumbar vertebra (IPS18800.5) on the basis of their similar preserved compared with great apes, since hylobatids display an intermediate morphology. In contrast, the penultimate or last lumbar vertebra condition (Schultz,1961; Ward,1991). The reduced ventral wedging (IPS18800.6) is unlikely to be correlative with the latter because of the Hispanopithecus lumbar vertebrae therefore supports the their preserved morphology shows a too abrupt change in view that this taxon would have possessed a relatively short and morphology, suggesting that at least one additional vertebra rigid lumbar region (more rigid than Pierolapithecus in terms of (maybe even two) would have been present between them. These wedging), compatible with an orthograde body plan suitable for considerations lead us to infer the presence of at least four, but suspensory and other orthograde behaviors, in spite of still pos- maybe more lumbar vertebrae for Hispanopithecus. Gibbons display sessing somewhat elongated lumbar vertebrae. The fact that I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 29

Hispanopithecus most closely resembles the extant great ape con- accessory processes would mainly restrict lateral flexion and dition, rather than the intermediate degree of ventral wedging rotation of the spine, but this is at odds with the lack of such pro- displayed by hylobatids-Ateles, might be attributed to the larger cesses in great apes (Ward, 1991). In fact, the accessory processes in (great ape-like) body size of Hispanopithecus, although such an monkeys serve as insertion sites for various muscles of the lower explanation does not hold based on Pierolapithecus. back, which are involved not only in tail extension, but also in Besides wedging, Hispanopithecus also resembles extant homi- lateral flexion and extension of the lower portion of the spine, noids in the overall morphology of the intervertebral articular probably having also a largely proprioceptive role (Ward, 1991). areas, which reflects the size and shape of the intervertebral disks Among these muscles inserting in the accessory processes of and, hence, their ability to absorb loads and distribute them monkeys, only the longissimus is retained in great apes, in which it effectively (Ward, 1991). To a large extent, differences in interver- similarly originates from the iliac crest, but inserts instead onto the tebral surface profile between extant hominoids and monkeys are transverse processes and the ribs (Ward, 1991). The reduction attributable to the possession of a dorsoventrally taller (especially, (hylobatids) and loss (great apes) of the accessory processes is thus dorsally expanded) lumbar vertebral bodies in the former (espe- likely related to the rearrangement of the back musculature in cially great apes), coupled with the lack of a ventral keel as well as relation to orthogrady, which would have implied a loss of their concave ventrolateral sides (Ward, 1991). The more cylindrical original function as muscular insertion sites (Ward, 1991). The vertebral bodies of hominoids, and especially great apes, imply a retention of accessory processes (even if broken) in Proconsul, being larger cross-sectional area, which has been functionally related to better developed on the upper lumbar vertebrae and more rudi- the need to resist the greater loads experienced by the lumbar spine mentary on the lower ones, most closely resembles the hylobatid during climbing and suspensory behaviors (Cartmill and Milton, condition and suggests that a great-ape like reorganization of the 1977; Jungers, 1984). This might be particularly related to the lower back musculature had not yet taken place in this taxon compressive loads caused by the contraction of the latissimus dorsi, (Ward, 1991, 1993). Unfortunately, the absence or presence of although in fact a greater cross-sectional area would also provide accessory processes, due to incomplete preservation, cannot be the vertebral bodies with increased bending strength against other ascertained in Hispanopithecus, thus hindering the determination loads imposed to the spine during climbing and other orthograde of whether the lower back of this taxon was still more hylobatid- behaviors (Ward, 1991). like or already more great ape-like on this basis. The moderately elliptical (only slightly reniform) and dorso- In contrast, both the position of the transverse processes and the ventrally tall intervertebral surfaces of Hispanopithecus, coupled orientation of the spinous processes can be ascertained in Hispa- with its only slightly ventrolaterally concave bodies with no ventral nopithecus. Extant great apes display more caudally inclined lum- keel, most closely resemble those of Morotopithecus, Pierolapithecus bar spinous processes than monkeys (Ward, 1991; Shapiro, 1993). and hylobatids, rather than the more circular articular surfaces and The morphology of the former was related by Slijper (1946) to the cylindrical vertebral bodies of great apes. All of the taxa, however, possession of a better-developed multifidus muscle, which would clearly differ from pronograde monkeys and stem hominoids such contribute to trunk stabilization during orthograde postures. This as Proconsul, which display more markedly cordiform articular remains a valid possibility, since the fascicles of the multifidus contours and ventrolaterally concave vertebral bodies with a originate from the spinous process (inserting on the lumbar ventral keel (Ward, 1991, 1993). The extant hominoid-like config- mammillary processes of lower lumbar vertebrae, the iliac crest uration, which would imply a decrease in flexion capabilities along and the sacrum) and act as extensors of the lower back that the sagittal plane, has been inferred to minimize stresses caused by contribute to trunk stabilization (Shapiro, 1993). Alternatively, tensile loads on the end-plates and intervertebral disks (Ward, however, the orientation of the spinous processes might be more 1991). The Hispanopithecus morphology in this regard would be directly related to the rearrangement of the erector spinae group in thus in agreement with other features discussed above, indicating hominoids, and especially great apes (Ward, 1991). Detailed quan- the possession of a somewhat reduced and stiffened lumbar spine. titative analyses would be required to better determine the rela- Although differences in intervertebral surface profile between tionship between more caudally-oriented spinous processes and hylobatids (more elliptical) and great apes (more circular) have orthogrady in hominoids and other primates (Shapiro, 1993). been previously attributed to differences in size (Ward, 1991), the Nonetheless, the extant great ape-like, caudal orientation of the great ape-like body size of Hispanopithecus suggest that this is spinous processes inferred for Hispanopithecus (based on their unlikely to be the main functional explanation of these differences. preserved roots) suggests at least that, unlike in Proconsul, a sig- Lumbar vertebral apophyses, processes and pedicle Compared nificant rearrangement of the lower back muscles had already with monkeys, great apes have rearranged the musculature of the taken place, presumably in relation to a somewhat shortened and lower back (the erector spinae group, which extends the lumbar more rigid lumbar spine as compared with pronograde monkeys. column), so that these muscles mainly oppose movements of the Such a rearrangement is more clearly reflected in the position of lower spine in several directions, instead of emphasizing flexion- the lumbar transverse processes, which in pronograde monkeys extension as in monkeys (Ward, 1991). Both the spinous and arise from the broadest part of the vertebral body and are ventrally transverse processes of the lumbar vertebrae act as bony levers oriented. In contrast, in extant great apes the transverse processes for erector spinae muscles (Ward, 1991; Shapiro, 1993), so that are situated very dorsally (about the junction between the pedicle their position and orientation, together with the presence/ and the lamina) and are dorsally oriented, whereas in hylobatids absence of accessory processes, have important functional (especially gibbons) and Ateles they arise from or near the junction implications. between the pedicle and the body (Fig. 7a) and display a coplanar As a result of the rearrangement of lumbar back musculature, in (horizontal) orientation (Benton, 1965; Ankel, 1972; Filler, 1986; great apes the intrinsic lumbar muscles are less fasciculated than in Ward, 1991, 1993; Shapiro, 1993). These processes provide attach- monkeys; this fact implies lesser fine control of this region and has ment for several muscles that originate from the pelvis and the been related to the lack of accessory processes of the lumbar sacrum, including muscles of the erector spinae group (portions of vertebrae in these taxa (Ward, 1991). In this regard, hylobatids the longissimus and iliocostalis, which insert on the dorsal portion display a somewhat intermediate condition, by possessing well- of the processes), as well as portions of the psoas major and developed accessory processes only on the upper portion of the quadratus lumborum, which insert on their ventral aspects lumbar spines (Ward, 1991). Ankel (1972) concluded that the (Shapiro, 1993, 1995, and references therein). The shape and 30 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 position of the lumbar transverse processes therefore determines condition. Hispanopithecus most closely resembles the condition of the leverage of these muscles (Ward, 1991; Shapiro, 1993, 1995). extant great apes, which are characterized by short and robust Among others, the dorsal position of the transverse processes in (transversely wide) pedicles (Schultz, 1961; Ankel, 1972; Ward, great apes would enable to maintain a consistent line of action of 1991), thereby differing from the monkey-like condition of both the erector spinae muscles, as well as to increase their moment arm Nacholapithecus and Proconsul in this regard (both long and narrow to counteract flexion of the trunk (Ward, 1991, 1993; Shapiro, 1993, pedicles). The wider pedicles of great apes have been related to the 1995; Sanders and Bodenbender, 1994; Gommery, 2006). Overall, need to maintain their cross-sectional area due to the concomitant the great-ape configuration of the transverse processes seems to be reduction in pedicle length (Ward, 1991), which in turn would be functionally related to the reorganization of the lower back related to the reduction in the length of the vertebra as a whole. musculature, in relation to concomitant changes in the pelvis However, hylobatids and Ateles (as well as Morotopithecus and (rotated and expanded iliac blades) and in torso morphology (wider Pierolapithecus) also display wide pedicles in spite of their moder- ribcage with a more ventrally situated spine) (Waterman, 1929; ately long pedicles, which are intermediate in length between Schultz, 1961; Benton, 1965; Cartmill and Milton, 1977; Ward, those of great apes and those of monkeys. This suggests that pedicle 1991, 1993; Shapiro, 1993). robusticity is also related to the position of the transverse pro- In Hispanopithecus, the dorsal origin of the transverse processes cesses, which given their role as a site of muscular attachment from the pedicle, slightly above its junction with the body, is thus would have an impact on the loading regime of the pedicle itself completely extant hominoid-like, and differs from the more (Ward, 1991). Accordingly, those taxa in which the transverse monkey-like configuration displayed by Proconsul. In particular, the processes do not arise from the body (i.e., all extant hominoids and transverse processes in Hispanopithecus are slightly more dorsally Ateles) would also display wider pedicles, irrespective of cranio- positioned than in Morotopithecus, Pierolapithecus, Ateles and caudal length of the vertebra. The great-ape-like pedicle hylobatids, the former being therefore somewhat derived towards morphology of Hispanopithecus thus entirely agrees with the the extant great-ape condition (although the latter display an even possession of considerably dorsal transverse processes, but does more dorsal origin of the transverse processes, from the junction not reflect the moderately elongated lumbar region inferred from between the pedicle and the lamina). This feature most clearly the vertebral bodies, thus strengthening the view that this taxon supports that some significant degree of reorganization of the was already somewhat derived towards the great-ape condition in lower back musculature was already under way in Hispanopithecus, this regard (more so than either Morotopithecus or Pierolapithecus). in agreement with other orthograde features of this taxon and the possession of a somewhat stiffened lumbar region inferred from Implications for hominoid locomotor evolution other vertebral features, although probably to a lesser extent than in extant great apes. It needs to be stressed that a dorsal origin and Among fossil hominoids, orthograde features are first docu- not very marked ventral orientation of the transverse processes, mented in the putative stem hominoid Morotopithecus from the together with shorter lumbar bodies, are associated with ortho- early Miocene of Africa, mainly on the basis of its vertebral remains grade positional behaviors not only in non-hominoid primates (Ward, 1993, 2007; Sanders and Bodenbender, 1994; MacLatchy (such as indriids, lorisids and sloth lemurs; Mivart, 1865; Schultz, et al., 2000; MacLatchy, 2004; Nakatsukasa, 2008), although this 1961; Cartmill and Milton, 1977; Shapiro, 1995; Shapiro et al., taxon apparently retained a higher number of lumbar segments 2005), but also in more distantly related taxa such as giant ant- than extant hominoids (Nakatsukasa, 2008). Other anatomical re- eaters and giant pandas (Davis, 1964; Williams, 2011). The striking gions (scapular glenoid and proximal femur) attributed to Moro- convergence in lumbar vertebral morphology between giant topithecus (Gebo et al., 1997; MacLatchy et al., 2000; MacLatchy, pandas and extant hominoids (including an origin of the transverse 2004; Ward, 2007) similarly show some orthograde-related fea- processes from the pedicle; Davis, 1964, his Figs. 39e40) may be tures, although the morphology of the femur is apparently inter- attributed to the frequent upright trunk postures of the former mediate between monkeys and extant apes. An unambiguous while feeding, so as to free its forelimbs to manipulate bamboo interpretation of the evolutionary implications of the orthograde stalks (Davis, 1964; Chorn and Hoffmann, 1978; Williams, 2011). features of Morotopithecus (i.e., whether it is homologous or ho- Although more research is necessary to understand the functional moplastic with the crown hominoid condition) is complicated not role of the lumbar region in the giant panda, given its pronograde only by the lack of additional remains of this taxon, but also the lack body plan and positional behaviors otherwise comparable to other of comparable postcranial elements in Afropithecus (MacLatchy, ursids (Chorn and Hoffmann, 1978), its hominoid-like lumbar 2004)dapparently closely-related to the former on the basis of vertebral morphology suggests that this region might be very prone cranial morphology (e.g., Harrison, 2010), but generally thought to to reorganization as an adaptation to more frequent orthograde display a pronograde body plan similar to that of Proconsul (Ward, positional behaviors, maybe in relation to freeing the hands during 2007). On the basis of this evidence, an independent development feeding behaviors (Williams, 2011). of some degree of orthogrady in Morotopithecus does not seem To some extent, the possession of a relatively rigid lumbar re- unlikely (Ward, 2007; Harrison, 2010; Alba, 2012), and it would not gion in Hispanopithecus is also supported by the oblique orientation be surprising in light of the apparent high degree of homoplasy of the postzygapophyses (the prezygapophyses are not preserved). regarding these features between hylobatids and crown hominids The possession of less ventrally-oriented postzygapophyses has (Moya-Sol a et al., 2004, 2005; Almecija et al., 2009; Alba, 2012). been functionally related to providing resistance against both Moreover, the convergent similarities in lumbar vertebral ventral displacement and rotation in hominoids (especially great morphology between extant hominoids and pronograde taxa (such apes) compared with cercopithecids (Shapiro, 1993). The orienta- as indriids and giant pandas; Shapiro, 1995; Williams, 2011) should tion of the postzygapophyses in Hispanopithecus would therefore prevent us from inferring an overall orthograde body plan for indicate a decrease in lower back flexibility compared with mon- Morotopithecus until additional remains enabling further evalua- keys, although being more comparable with hylobatids than to tion of thorax and pelvic morphology are eventually found. great apes in this regard. The possession of an orthograde body plan can be more securely The proportions of the pedicle also enable some functional in- inferred for the putative stem hominoid Pierolapithecus from the ferences, since they are highly divergent between great apes and European middle Miocene (11.9 Ma [millions of years ago]), on the monkeys, with hylobatids displaying a somewhat intermediate basis of a lumbar vertebral morphology most similar to that of I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 31 hylobatids and Morotopithecus, marked rib curvature and a long Hispanopithecus, but not in either Morotopithecus and Pier- and stout clavicle (indicating a broad and shallow thorax with olapithecus, suggests a greater emphasis on orthogrady in the dorsally situated scapulae), and a wrist (triquetrum) morphology former, in agreement with the suspensory adaptations that have indicating the lack of direct ulnocarpal articulation (Moya-Sol a been inferred for Hispanopithecus from the morphology of its et al., 2004, 2005; Susanna et al., 2010a,b; Alba, 2012). At the forelimb and axial skeleton (Almecija et al., 2007; Deane and same time, however, Pierolapithecus retains adaptations for Begun, 2008; Alba et al., 2010, 2012a; Alba, 2012). powerful grasping, above-branch palmigrady in the phalanges and On the other hand, the overall hylobatid-like lumbar vertebral metacarpals (Moya-Sol a et al., 2004, 2005; Almecija et al., 2009; morphology of Hispanopithecus would indicate the possession of a Alba et al., 2010; Alba, 2012). Such a mosaic of features has led to less shortened and somewhat more flexible lumbar spine than in conflicting views as to whether this taxon possessed (Begun and extant great apes. The biomechanical demands of orthograde be- Ward, 2005; Deane and Begun, 2008, 2010) or lacked (Moya-Sol a haviors become more pronounced at increasing body sizes, and et al., 2005; Almecija et al., 2009; Alba et al., 2010; Alba, 2012) therefore the differences in vertebral morphology between hylo- suspensory adaptations. Irrespective of the latter, the phalangeal batids and great apes have been attributed to the lower loads morphology of Pierolapithecus shows that this taxon displayed a experienced at the lumbar region by the former, as a result of their locomotor repertoire unlike modern hominoids, combining much smaller body weight (Ward, 1991). Such an explanation, orthograde behaviors with above-branch quadrupedalism (palmi- however, does not hold for the skeleton of H. laietanus from CLL2, grady). An unambiguous interpretation of such a mosaic of primi- for which a body mass of ca. 39e40 kg has been estimated from the tive and derived features in several areas of the locomotor skeleton femora (Moya-Sol a et al., 2009) and lumbar vertebra (this study), in Pierolapithecus is precluded by phylogenetic uncertainties, this well within the range of Pan and female Pongo (Smith and Jungers, taxon being alternatively interpreted as a stem hominid (Moya-Sol a 1997) and in further agreement with their similar overall lumbar et al., 2004), a stem hominine (Begun, 2007, 2009, 2010) or maybe a vertebral size. This relatively large body size, similar to that of stem pongine (Perez de los Ríos et al., 2012; see discussion in Alba, Morotopithecus and Pierolapithecus, much closely approaches the 2012). However, a similar mosaic condition is even more clearly range of extant great apes than that of hylobatids. Therefore, the displayed by Sivapithecus from Asia (Madar et al., 2002), interpreted hylobatid-like lumbar vertebral morphology of Hispanopithecus as a pongine by most authors (e.g., Kelley, 2002; Begun, 2010; Alba, might alternatively reflect some differences in the locomotor 2012), thus reinforcing the contention that the last common repertoire of this taxon as compared with extant great apes, ancestor of crown hominids must have been more primitive than thereby supporting the view, based on other anatomical regions, inferred on the basis of extant taxa alone (Moya-Sol a and Kohler,€ that it has no close extant positional analogues (Almecija et al., 1996b; Moya-Sol a et al., 2004; Alba et al., 2010, 2012a; Alba, 2007; Alba et al., 2012a). In this regard, it should be taken into 2012). The latter contention is strengthened by Hispanopithecus, account that the similarities in lumbar vertebral morphology be- in spite of the phylogenetic controversies that persist regarding this tween Hispanopithecus and hylobatids can be also made extensive taxon, which like Pierolapithecus is interpreted either as a stem to spider monkeys, which besides vertical climbing and suspension, pongine (Moya-Sol a and Kohler,€ 1993, 1995; Kohler€ et al., 2001), a also practice a considerable amount of arboreal quadrupedalism stem hominine (Begun, 2007, 2009, 2010) or a stem hominid (Alba, (Cant, 1986; Youlatos, 2008). Thus, although the vertebral 2012). morphology of Hispanopithecus lacks clear adaptations to pronog- The postcranial evidence of Hispanopithecus suggests a mosaic rade behaviors, its more elongated and flexible lumbar region of derived, modern ape-like features related to orthograde behav- compared with extant great apes is fully compatible with the iors, coupled with more primitive, stem hominoid-like adaptations retention of a significant degree of above-branch, palmigrade to quadrupedalism, albeit to a lesser extent than in Pierolapithecus, quadrupedalism in this taxon, as inferred from various features of and also differing from both Pierolapithecus and Sivapithecus by its forelimb and hindlimb (Moya-Sol a and Kohler,€ 1996b; Almecija displaying undoubted suspensory adaptations (Moya-Sol a and et al., 2007; Alba et al., 2012a; Tallman et al., 2013). Kohler,€ 1996b; Kohler€ et al., 2001, 2002; Almecija et al., 2007; Final mention needs to be made of the late Miocene ape Oreo- Deane and Begun, 2008; Alba et al., 2010, 2012a; Alba, 2012; Pina pithecus bambolii from Tusco-Sardinia (Straus, 1957, 1963; Hürzeler, et al., 2012; Tallman et al., 2013). The morphology displayed by 1958; Harrison, 1986; Sarmiento, 1987; Harrison and Rook, 1997; the vertebral specimens of Hispanopithecus laietanus from CLL2 Moya-Sol a and Kohler,€ 1996a), which represents the last surviv- described in this paper, in spite of their fragmentary preservation, ing member of the European radiation of apes (Casanovas-Vilar are consistent with such interpretation, for the reasons discussed et al., 2011; Rook et al., 2011). Oreopithecus displays, like Hispano- below. First, both the thoracic and lumbar vertebral remains of pithecus, clear extant hominoid-like orthograde features suitable Hispanopithecus display numerous derived, extant hominoid-like for forelimb-dominated locomotor behaviors (Schultz, 1960; Straus, features that are highly suggestive of an orthograde body plan 1963; Harrison, 1986, 1991; Jungers, 1987; Sarmiento, 1987; and indicate the possession of a mediolaterally broad and dorso- Harrsion and Rook, 1997; Kohler€ and Moya-Sol a, 1997; Ward, ventrally shallow thorax, coupled with a somewhat shortened and 2007; Alba et al., 2011b), including: high intermembral index, stiffened lumbar region. Second, in many regards, the condition wide and shallow thorax (curved ribs, robust clavicle and broad displayed by Hispanopithecus, like in Morotopithecus and Pier- pelvis), modern great ape-like elbow joint (with a very short olapithecus, most closely resembles that of hylobatids and Ateles olecranon process), highly mobile humeral and femoral heads, rather than that of extant great apes (e.g., obliquely-oriented taillessness, and short lumbar region. Although most of the verte- postzygapophyses, somewhat elongated vertebral bodies with bral column of this taxon is preserved, comparisons with Hispa- slightly concave ventrolateral sides, high but less circular end- nopithecus are hindered by the badly crushed condition of the plates than in extant great apes), although with some exceptions. Oreopithecus specimens, which precludes a complete assessment of In particular, the pedicle morphology, the orientation of the spinous their original morphology (Harrison, 1986; Russo and Shapiro, process and the reduced ventral wedging of Hispanopithecus are 2013) as well as its inclusion in the numerical analyses performed extant great ape-like features, whereas the position of the trans- in this paper. The lumbar region of Oreopithecus is composed of five verse processes is also slightly more derived towards the extant lumbar vertebrae, which are short and relatively broad, with rela- great ape condition than in gibbons and Ateles. The possession of a tively small articular and accessory processes, transverse processes few more derived (extant great ape-like) features in originating from the base of the pedicle, and a moderately- 32 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 developed ventral keel (Schultz, 1960; Straus, 1963; Harrison, 1986, regard than it can be inferred based on the few surviving great apes 1991; Sarmiento, 1987; Harrison and Rook, 1997; Kohler€ and Moya- and humans (especially if Hispanopithecus is considered a crown, Sola, 1997; Lovejoy and McCollum, 2010; Russo and Shapiro, 2013). instead of a stem, hominid; e.g., Moya-Sol a and Kohler,€ 1996b; In these features, Oreopithecus is essentially comparable with His- Begun, 2007). The results of this paper therefore reinforce the panopithecus (except for the more developed keel in the former), view that homoplasy and mosaic evolution played a very important thus indicating the possession of an orthograde body plan with a role in the emergence of the locomotor apparatus of the various somewhat shortened and stiffened lumbar region. In Oreopithecus, surviving hominoid lineages (e.g., Larson, 1998; Moya-Sol a et al., the reduction of the lumbar spine is also confirmed by the likely 2004; Ward, 2007; Alba, 2012; Alba et al., 2012a; Almecija et al., sacralized morphology of the last lumbar vertebra (Lovejoy and 2013), not only between hylobatids and hominids, but probably MacCollum, 2010), as well as by the possession of six sacral verte- also between pongines and hominines. brae (Harrison, 1986; Russo and Shapiro, 2013), which have been functionally related to increased stability of the lower spine in slow climbers (Ankel, 1972; Cartmill and Milton, 1977; Harrison, 1986). Acknowledgments The evolutionary significance of orthograde features in Oreopi- thecus is however difficult to ascertain, given its uncertain phylo- This work has been funded by the Spanish Ministerio de genetic affinities among hominoids (Harrison, 1986; Sarmiento, Economía y Competitividad (CGL2011-28681, CGL2011-27343, and 1987), although several features suggest that, like Hispanopithe- RYC-2009-04533 to D.M.A.); the Generalitat de Catalunya (2009 cus, it might be a member of the great ape and human clade SGR 754 GRC, and 2009 BP-A 00226 to S.A.); the AAPA Professional (Harrison and Rook, 1997; Moya-Sol a and Kohler,€ 1996a). Based on Development Grant (to S.A.); and the SYNTHESYS Project (BE-TAF- its postcranial morphology, most authors have concluded that 2068 to I.S.; http://www.synthesys.info/), which is financed by the Oreopithecus was adapted not only to climbing (Schultz, 1960; European Community Research Infrastructure Action under the FP6 Sarmiento, 1987), but also to below-branch suspensory behaviors ‘Structuring the European Research Area’ Program. We thank (Harrison, 1986, 1991; Jungers, 1987; Susman, 2004; Ward, 2007; Massimo Delfino for sending us some relevant literature, as well as Lovejoy and McCollum, 2010), or even habitual bipedalism the following curators or collection managers for access to extant (Kohler€ and Moya-Sol a, 1997; Rook et al., 1999). However, an comparative material under their care: E. Gilissen and W. Wende- adaptation to suspensory behaviors in Oreopithecus is at odds with len (RMCA), Esther Langan (NMNH) and Eileen Westwig (AMNH). the relatively short metacarpals and phalanges (Moya-Sol a et al., We also acknowledge the comments by two anonymous reviewers, 1999; see discussion in Almecija et al., 2012, 2014). Both in its which significantly helped to improve a previous version of this moderate phalangeal relative length (Moya-Sol a et al., 1999) and paper. This is NYCEP morphometrics contribution number 78. curvature (Susman, 2004; Deane and Begun, 2008; Almecija et al., 2014), Oreopithecus seems more comparable with Pierolapithecus than to either Hispanopithecus or extant highly suspensory apes References (i.e., hylobatids and orangutans; Almecija et al., 2007, 2009; Alba € et al., 2010), which might be explained as a symplesiomorphy Agustí, J., Kohler, M., Moya-Sola, S., Cabrera, L., Garces, M., Pares, J.M., 1996. Can Llobateres: the pattern and timing of the Vallesian hominoid radiation recon- (Almecija et al., 2014). In any case, the possession of a less short- sidered. J. Hum. Evol. 31, 143e155. ened lumbar region in both Oreopithecus and Hispanopithecus Alba, D.M., 2012. Fossil apes from the Valles-Pened es Basin. Evol. Anthropol. 21, 254 strongly suggests that the morphology shared by extant great apes e269. Alba, D.M., Moya-Sol a, S., Kohler,€ M., 2003. Morphological affinities of the Austral- (i.e., African apes and pongines) might have been to some extent opithecus afarensis hand on the basis of manual proportions and relative thumb independently derived from a more generalized condition, as length. J. Hum. Evol. 44, 225e254. similarly suggested recently for the proximal morphology of the Alba, D.M., Almecija, S., Moya-Sol a, S., 2010. Locomotor inferences in Pierolapithecus femur (Almecija et al., 2013). and Hispanopithecus: Reply to Deane and Begun (2008). J. Hum. Evol. 59, 143 e149. Alba, D.M., Casanovas-Vilar, I., Robles, J.M., Marmi, J., Moya-Sol a, S., 2011a. New Conclusions paleontological excavations at the Late Miocene site of Can Llobateres 1 (Valles- Penedes Basin, NE Iberian Peninsula): preliminary results of the 2010 campaign. In: Perez-Garc ía, A., Gasco, F., Gasulla, J.M., Escaso, F. (Eds.), Viajando a Mundos The morphology of the vertebral remains of H. laietanus most Preteritos. Ayuntamiento de Morella, Morella, pp. 35e43. closely resembles that of hylobatids and Ateles, although unlike in Alba, D.M., Moya-Sol a, S., Almecija, S., 2011b. A partial hominoid humerus from the preceding putatively orthograde Miocene apes (Morotopithecus and middle Miocene of Castell de Barbera (Valles-Penedes Basin, Catalonia, Spain). Am. J. Phys. Anthropol. 144, 365e381. Pierolapithecus) it already displays some more derived (extant great Alba, D.M., Almecija, S., Casanovas-Vilar, I., Mendez, J.M., Moya-Sol a, S., 2012a. ape-like) features. The vertebral morphology of Hispanopithecus A partial skeleton of the fossil great ape Hispanopithecus laietanus from Can Feu thus agrees well with previous inferences of an orthograde body and the mosaic evolution of crown-hominoid positional behaviors. PLoS ONE 7, e39617. plan for this taxon, suggesting the possession of a broad and Alba, D.M., Casanovas-Vilar, I., Almecija, S., Robles, J.M., Arias-Martorell, J., Moya- shallow thorax with a somewhat shortened and stiffened lumbar Sola, S., 2012b. New dental remains of Hispanopithecus laietanus (Primates: region, although to a lesser degree than in extant great apes. The Hominidae) from Can Llobateres 1 and the taxonomy of Late Miocene hominoids from the Valles-Pened es Basin (NE Iberian Peninsula). J. Hum. Evol. 63, hylobatid-like morphology of Hispanopithecus, coupled with its 231e246. great ape-like body size (ca. 39e40 kg), suggest that there is no Allbrook, D., Bishop, W.W., 1963. New fossil hominoid material from Uganda. Nature extant locomotor analogue for this taxon among crown hominoids. 197, 1187e1190. € Our results are thus compatible with previous inferences, based on Almecija, S., Alba, D.M., Moya-Sola, S., Kohler, M., 2007. Orang-like manual adaptations in the fossil hominoid Hispanopithecus laietanus: First steps towards other anatomical regions (Almecija et al., 2007; Alba et al., 2012a; great ape suspensory behaviours. Proc. R. Soc. B 274, 2375e2384. Tallman et al., 2013), indicating the possession of a locomotor Almecija, S., Alba, D.M., Moya-Sol a, S., 2009. Pierolapithecus and the functional repertoire combining modern hominoid-like orthograde behaviors morphology of Miocene ape hand phalanges: Paleobiological and evolutionary implications. J. Hum. Evol. 57, 284e297. (such as climbing, clambering and suspension) with powerful- Almecija, S., Alba, D.M., Moya-Sol a, S., 2012. The thumb of Miocene apes: New in- grasping, and above-branch quadrupedalism. Given the generally- sights from Castell de Barbera (Catalonia, Spain). Am. J. Phys. Anthropol. 148, accepted great ape status of Hispanopithecus (e.g., Alba, 2012), its 436e450. Almecija, S., Tallman, M., Alba, D.M., Pina, M., Moya-Sol a, S., Jungers, W.L., 2013. The lumbar vertebral morphology suggests that the last common femur of Orrorin tugenensis exhibits morphometric affinities with both Miocene ancestor of crown hominids might have been more primitive in this apes and later hominins. Nat. Comm. 4, 2888. I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34 33

Almecija, S., Shrewsbury, M., Rook, L., Moya-Sol a, S., 2014. The morphology of Ishida, H., Kunimatsu, Y., Takano, T., Nakano, Y., Nakatsukasa, M., 2004. Nacholapi- Oreopithecus bambolii pollical distal phalanx. Am. J. Phys. Anthropol. 153, 582 thecus skeleton from the Middle Miocene of Kenya. J. Hum. Evol. 46, 69e103. e597. Johnson, S.E., Shapiro, L.J., 1998. Positional behavior and vertebral morphology in Ankel, F., 1972. Vertebral morphology of fossil and extant primates. In: Tuttle, R. atelines and cebines. Am. J. Phys. Anthropol. 105, 333e354. (Ed.), Functional and Evolutionary Biology of Primates. Aldine Press, Chicago, Jungers, W.L., 1984. Scaling of the hominoid locomotor skeleton with special pp. 223e240. reference to lesser apes. In: Preuschoft, H., Chivers, D., Brockelman, W., Creel, N. Baum, D.A., Larson, A., 1991. Adaptation reviewed: a phylogenetic methodology for (Eds.), The Lesser Apes: Evolutionary and Behavioral Biology. Edinburgh Uni- studying character macroevolution. Syst. Zool. 40, 1e18. versity Press, Edinburgh, pp. 146e169. Begun, D.R., 2007. Fossil record of Miocene hominoids. In: Henke, W., Tattersall, I. Jungers, W.L., 1987. Body size and morphometric affinities of the appendicular (Eds.), Handbook of Paleoanthropology. Springer Verlag, Heidelberg, pp. 921 skeleton in Oreopithecus bambolii (IGF 11778). J. Hum. Evol. 16, 445e456. e977. Jungers, W.L., Falsetti, A.B., Wall, C.E., 1995. Shape, relative size, and size- Begun, D.R., 2009. Dryopithecins, Darwin, de Bonis, and the European origin of the adjustments in morphometrics. Yearb. Phys. Anthropol. 38, 137e161. African apes and human clade. Geodiversitas 31, 789e816. Keith, A., 1903. The extent to which the posterior segments of the body have been Begun, D.R., 2010. Miocene hominids and the origins of the African apes and transmuted and suppressed in the evolution of man and allied primates. J. Anat. humans. A. Rev. Anthropol. 39, 67e84. Physiol. 37, 18e40. Begun, D.R., Ward, C.V., 2005. Comment on “Pierolapithecus catalaunicus, a new Keith, A., 1923a. Hunterian lectures on Man’s posture: Its evolution and disorders. Middle Miocene great ape from Spain”. Science 308, 203c. Lecture I. Theories concerning the evolution of Man’s posture. Brit. Med. J. 1, 451 Benton, R.S., 1965. Morphological evidence for adaptations within the epaxial re- e454. gion of the primates. In: Vagtborg, H. (Ed.), The Baboon in Medical Research. Keith, A., 1923b. Hunterian lectures on Man’s posture: Its evolution and disorders. University of Texas Press, Houston, pp. 10e20. Lecture II. The evolution of the orthograde spine. Brit. Med. J. 1, 499e502. Benton, R.S., 1974. Structural patterns in the Pongidae and Cercopithecidae. Yearb. Kelley, J., 2002. The hominoid radiation in Asia. In: Hartwig, W.C. (Ed.), The Primate Phys. Anthropol. 18, 65e88. Fossil Record. Cambridge University Press, Cambridge, pp. 369e384. Cant, J.G.H., 1986. Locomotion and feeding postures of spider and howling mon- Kohler,€ M., Moya-Sol a, S., 1997. Ape-like or hominid-like? The positional behavior of keys: Field study and evolutionary interpretation. Folia Primatol. 46, 1e14. Oreopithecus bambolii reconsidered. Proc. Natl. Acad. Sci. USA 94, 11747e11750. Cartmill, M., 1985. Climbing. In: Hildebrand, M., Bramble, D., Liem, K., Wake, D. Kohler,€ M., Moya-Sol a, S., Alba, D.M., 2001. Eurasian hominoid evolution in the light (Eds.), Functional Vertebrate Morphology. Belknap Press, Cambridge, pp. 73 of recent Dryopithecus findings. In: de Bonis, L., Koufos, G.D., Andrews, P. (Eds.), e88. Hominoid Evolution and Climatic Change in Europe, vol. 2. Cambridge Uni- Cartmill, M., Milton, K., 1977. The lorisiform wrist joint and the evolution of versity Press, Cambridge, pp. 192e212. ‘brachiating’ adaptations in the Hominoidea. Am. J. Phys. Anthropol. 47, 249 Kohler,€ M., Alba, D.M., Moya-Sol a, S., MacLatchy, L., 2002. Taxonomic affinities of the e272. Eppelsheim femur. Am. J. Phys. Anthropol. 119, 297e304. Casanovas-Vilar, I., Alba, D.M., Garces, M., Robles, J.M., Moya-Sol a, S., 2011. Updated Larson, S.G., 1998. Parallel evolution in the hominoid trunk and forelimb. Evol. chronology for the Miocene hominoid radiation in Western Eurasia. Proc. Natl. Anthropol. 6, 87e99. Acad. Sci. USA 108, 5554e5559. Lauder, G.V., 1991. Biomechanics and evolution: integrating physical and historical Chorn, J., Hoffmann, R.S., 1978. Ailuropoda melanoleuca. Mamm. Spec. 110, 1e6. biology in the study of complex systems. In: Rayner, M.V., Wooton, R.J. (Eds.), Davis, D.D., 1964. The giant panda. A morphological study of evolutionary mecha- Biomechanics in Evolution. Cambridge University Press, Cambridge, pp. 1e19. nisms. Fieldiana Zool. Mem. 3, 1e339. Lovejoy, C.O., McCollum, M.A., 2010. Spinopelvic pathways to bipedality: why no Deane, A.S., Begun, D.R., 2008. Broken fingers: retesting locomotor hypotheses for hominids ever relied on a bent-hip-bent-knee gait. Phil. Trans. R. Soc. B 365, fossil hominoids using fragmentary proximal phalanges and high-resolution 3289e3299. polynomial curve fitting (HR-PCF). J. Hum. Evol. 55, 691e701. MacLatchy, L., 2004. The oldest ape. Evol. Anthropol. 13, 90e103. Deane, A.S., Begun, D.R., 2010. Pierolapithecus locomotor adaptations: a reply to Alba MacLatchy, L.M., Gebo, D., Kityo, R., Pilbeam, D., 2000. Postcranial functional et al.'s comment on Deane and Begun (2008). J. Hum. Evol. 59, 150e154. morphology of Morotopithecus bishopi, with implications for the evolution of Erikson, G.E., 1963. Brachiation in New World monkeys and in anthropoid apes. modern ape locomotion. J. Hum. Evol. 39, 159e183. Symp. Zool. Soc. Lond. 10, 135e163. Madar, S.I., Rose, M.D., Kelley, J., MacLatchy, L., Pilbeam, D., 2002. New Sivapithecus Filler, A.G., 1981. Anatomical specializations in the hominoid lumbar region. Am. J. postcranial specimens from the Siwaliks of Pakistan. J. Hum. Evol. 42, 705e752. Phys. Anthropol. 54, 218. Martin, R.D., 1990. Primate Origins and Evolution. A Phylogenetic Reconstruction. Filler, A.G., 1986. Axial Character Seriation in Mammals. An Historical and Princeton University Press, Princeton. Morphological Exploration of the Origin, Development, Use, and Current McCollum, M.A., Rosenman, B.A., Suwa, G., Meindl, R.S., Lovejoy, C.O., 2010. The Collapse of the Homology Paradigm. BrownWalker Press, Boca Raton. vertebral formula of the last common ancestor of African apes and humans. Filler, A.G., 2007. Homeotic evolution in the Mammalia: Diversification of therian J. Exp. Zool. 314B, 123e134. axial seriation and the morphogenetic basis of human origins. PLoS ONE 2, Mivart, G., 1865. Contributions toward a more complete knowledge of the axial e1910. skeleton in the primates. Proc. Zool. Soc. Lond. 1865, 545e592. Gebo, D.L., 1996. Climbing, brachiation, and terrestrial quadrupedalism: historical Mosimann, J.E., 1970. Size allometry: Size and shape variables with characteriza- precursors of hominid bipedalism. Am. J. Phys. Anthropol. 101, 55e92. tions of the lognormal and generalized gamma distributions. J. Am. Stat. Ass. 65, Gebo, D.L., MacLatchy, L., Kityo, R., Deino, A., Kingston, J., Pilbeam, D., 1997. 930e945. A hominoid genus from the early Miocene of Uganda. Science 276, 401e404. Moya-Sol a, S., Kohler,€ M., 1993. Recent discoveries of Dryopithecus shed new light Gommery, D., 2006. Evolution of the vertebral column in Miocene hominoids and on evolution of great apes. Nature 365, 543e545. Plio-Pleistocene hominids. In: Ishida, H., Tuttle, R.H., Pickford, M., Ogihara, N., Moya-Sol a, S., Kohler,€ M., 1995. New partial cranium of Dryopithecus Lartet, 1863 Nakatsukasa, M. (Eds.), Human Origins and Environmental Background. (Hominoidea, Primates) from the Upper Miocene of Can Llobateres, Barcelona. Springer, New York, pp. 31e44. Spain. J. Hum. Evol. 29, 101e139. Gould, S.J., Vrba, E.S., 1982. Exaptationda missing term in the science of form. Moya-Sol a, S., Kohler,€ M., 1996a. The phylogenetic relationships of Oreopithecus Paleobiology 8, 4e15. bambolii Gervais, 1872. C. R. Acad. Sci. Paris 324, 141e148. Harrison, T., 1986. A reassessment of the phylogenetic relationships of Oreopithecus Moya-Sol a, S., Kohler,€ M., 1996b. A Dryopithecus skeleton and the origins of great- bambolii Gervais. J. Hum. Evol. 15, 541e583. ape locomotion. Nature 379, 156e159. Harrison, T., 1991. The implications of Oreopithecus bambolii for the origins of Moya-Sol a, S., Kohler,€ M., Rook, L., 1999. Evidence of hominid-like precision grip bipedalism. In: Senut, B., Coppens, Y. (Eds.), Origine(s) de la Bipedie Chez les capability in the hand of the Miocene ape Oreopithecus. Proc. Natl. Acad. Sci. Hominides. Editions du CNRS, Paris, pp. 235e244. USA 96, 313e317. Harrison, T., 2010. Dendropithecoidea, Proconsuloidea, and Hominoidea (Catarrhini, Moya-Sol a, S., Kohler, M., Alba, D.M., Casanovas-Vilar, I., Galindo, J., 2004. Pier- Primates). In: Werdelin, L., Sanders, W.J. (Eds.), Cenozoic Mammals of Africa. olapithecus catalaunicus, a new Middle Miocene great ape from Spain. Science University of California Press, Berkeley, pp. 429e469. 306, 1339e1344. Harrison, T., Rook, L., 1997. Enigmatic anthropoid or misunderstood ape? The Moya-Sol a, S., Kohler,€ M., Alba, D.M., Casanovas-Vilar, I., Galindo, J., 2005. Response phylogenetic status of Oreopithecus bambolii reconsidered. In: Begun, D.R., to comment on “Pierolapithecus catalaunicus, a new Middle Miocene great ape Ward, C.V., Rose, M.D. (Eds.), Function, Phylogeny and Fossils: Miocene Homi- from Spain”. Science 308, 203d. noid Evolution and Adaptation. Plenum Press, New York, pp. 327e362. Moya-Sol a, S., Kohler,€ M., Alba, D.M., Casanovas-Vilar, I., Galindo, J., Robles, J.M., Hunt, K.D., 1991. Positional behavior in the Hominoidea. Int. J. Primatol. 12, 95e118. Cabrera, L., Garces, M., Almecija, S., Beamud, E., 2009. First partial face and Hunt, K.D., 1992. Positional behavior of Pan troglodytes in the Mahale mountains upper dentition of the Middle Miocene hominoid Dryopithecus fontani from and Gombe stream national parks, Tanzania. Am. J. Phys. Anthropol. 87, 83e105. Abocador de Can Mata (Valles-Pened es Basin, Catalonia, NE Spain): taxonomic Hunt, K.D., 2004. The special demands of great ape locomotion and posture. In: and phylogenetic implications. Am. J. Phys. Anthropol. 139, 126e145. Russon, A.E., Begun, D.R. (Eds.), The Evolution of Thought. Evolutionary Origins Nakatsukasa, M., 2008. Comparative study of Moroto vertebral specimens. J. Hum. of Great Ape Intelligence. Cambridge University Press, Cambridge, pp. 172e189. Evol. 55, 581e588. Hunt, K.D., Cant, J.G.H., Gebo, D.L., Rose, M.D., Walker, S.E., Youlatos, D., 1996. Nakatsukasa, M., Kunimatsu, Y., Nakano, Y., Ishida, H., 2007. Vertebral morphology Standardized descriptions of primate locomotor and postural modes. Primates of Nacholapithecus kerioi based on KNM-BG 35250. J. Hum. Evol. 52, 347e369. 37, 363e387. Perez de los Ríos, M., Moya-Sol a, S., Alba, D.M., 2012. The nasal and paranasal ar- Hürzeler, J., 1958. Oreopithecus bambolii Gervais. A preliminary report. Verh. Naturf. chitecture of the Middle Miocene ape Pierolapithecus catalaunicus (Primates: Ges. Basel 69, 1e48. Hominidae): Phylogenetic implications. J. Hum. Evol. 63, 497e506. 34 I. Susanna et al. / Journal of Human Evolution 73 (2014) 15e34

Pilbeam, D., 2004. The anthropoid postcranial axial skeleton: Comments on Smith, R.J., Jungers, W.L., 1997. Body mass in comparative primatology. J. Hum. Evol. development, variation, and evolution. J. Exp. Zool. 302B, 241e267. 32, 523e559. Pina, M., Alba, D.M., Almecija, S., Fortuny, J., Moya-Sol a, S., 2012. Brief communi- Stern Jr., J.T., 1975. Before bipedality. Yearb. Phys. Anthropol. 9, 59e68. cation: Paleobiological inferences on the locomotor repertoire of extinct hom- Straus Jr., W.L., 1957. Oreopithecus bambolii. Science 126, 345e346. inoids based on femoral neck cortical thickness: The fossil great ape Straus Jr., W.L., 1963. The classification of Oreopithecus. In: Washburn, S.L. (Ed.), Hispanopithecus laietanus as a test-case study. Am. J. Phys. Anthropol. 149, 142 Classification and Human Evolution. Aldine, Chicago, pp. 146e177. e148. Susanna, I., Alba, D.M., Almecija, S., Moya-Sol a, S., 2010a. Las vertebras lumbares del Rook, L., Bondioli, L., Kohler,€ M., Moya-Sol a, S., Macchiarelli, R., 1999. Oreopithecus gran simio antropomorfo basal del Mioceno medio Pierolapithecus catalaunicus was a bipedal ape after all: evidence from the iliac cancellous architecture. Proc. (Primates: Hominidae). Cidaris 30, 311e316. Natl. Acad. Sci. USA 96, 8795e8799. Susanna, I., Alba, D.M., Almecija, S., Moya-Sol a, S., 2010b. The lumbar vertebrae of Rook, L., Oms, O., Benvenuti, M., Papini, M., 2011. Magnetostratigraphy of the Late the Middle Miocene stem great ape Pierolapithecus catalaunicus (Primates: Miocene BaccinelloeCinigiano basin (Tuscany, Italy) and the age of Oreopithecus Hominidae). Am. J. Phys. Anthropol. 141 (Suppl. 50), 227. bambolii faunal assemblages. Palaeogeogr. Palaeoclimatol. Palaeoecol. 305, 286 Susanna, I., Alba, D.M., Almecija, S., Moya-Sol a, S., 2011a. Vertebral remains of the e294. Late Miocene ape Hispanopithecus laietanus (Primates: Hominidae): Functional Rose, M.D., 1975. Functional proportions of primate lumbar vertebral bodies. J. Hum. morphology and paleobiological implications. J. Vert. Paleontol. 31 (Suppl. 2), Evol. 4, 21e38. 202. Russo, G.A., Shapiro, L.J., 2013. Reevaluation of the lumbosacral region of Oreopi- Susanna, I., Alba, D.M., Almecija, S., Moya-Sol a, S., 2011b. Vertebral remains of thecus bambolii. J. Hum. Evol. 65, 253e265. Hispanopithecus laietanus (Primates: Hominidae) from the Late Miocene of Can Sanders, W.J., 1998. Comparative morphometric study of the australopithecine Llobateres 2 (Catalonia, NE Iberian Peninsula). Paleontol. Evol. Mem. Esp. 5, 383 vertebral series Stw-H8/H41. J. Hum. Evol. 34, 249e302. e386. Sanders, W.J., Bodenbender, B.E., 1994. Morphometric analysis of lumbar vertebra Susman, R.L., 2004. Oreopithecus bambolii: an unlikely case of hominidlike grip UMP 67-28: Implications for spinal function and phylogeny of the Miocene capability in a Miocene ape. J. Hum. Evol. 46, 105e117. Moroto hominoid. J. Hum. Evol. 26, 203e237. Tallman, M., Almecija, S., Reber, S.L., Alba, D.M., Moya-Sol a, S., 2013. The distal tibia Sarmiento, E.E., 1987. The phylogenetic position of Oreopithecus and its significance of Hispanopithecus laietanus: more evidence for mosaic evolution in Miocene in the origin of the Hominoidea. Am. Mus. Nov. 2881, 1e44. apes. J. Hum. Evol. 64, 319e327. Sarmiento, E.E., 1995. Cautious climbing and folivory: a model of hominoid differ- Walker, A.C., Pickford, M., 1983. New postcranial fossils of Proconsul africanus and entiation. Hum. Evol. 10, 289e321. Proconsul nyanzae. In: Jungers, W.L. (Ed.), New Interpretations of Ape and Hu- Schultz, A.H., 1930. The skeleton of the trunk and limbs of higher primates. Hum. man Ancestry. Plenum Press, New York, pp. 325e351. Biol. 2, 303e438. Walker, A.C., Rose, M., 1968. Fossil hominoid vertebra from the Miocene of Uganda. Schultz, A.H., 1960. Einige Beobachtungen und Maße am Skelett von Oreopithecus in Nature 217, 980e981. Vergleich mit anderen catarrhinen Primaten. Z. Morphol. Anthropol. 50, 136 Ward, C.V., 1991. Functional anatomy of the lower back and pelvis of the Miocene e149. hominoid Proconsul nyanzae of Mfangano Island, Kenya. Ph.D. Dissertation, The Schultz, A.H., 1961. Vertebral column and thorax. In: Hofer, H., Schultz, A.H., Johns Hopkins University. Stark, D. (Eds.), Primatologia, Handbuch der Primatenkunde, vol. IV/4. S. Karger, Ward, C.V., 1993. Torso morphology and locomotion in Proconsul nyanzae.Am.J. Basel, pp. 1e66. Phys. Anthropol. 92, 291e328. Schultz, A.H., Straus Jr., W.L., 1945. The numbers of vertebrae in Primates. Proc. Am. Ward, C.V., 2007. Postcranial and locomotor adaptations of hominoids. In: Phil. Soc. 89, 601e626. Henke, W., Tattersall, I. (Eds.), Handbook of Paleoanthropology. Springer Verlag, Shapiro, L.J., 1990. Vertebral morphology in the Hominoidea. Am. J. Phys. Anthropol. Heidelberg, pp. 1011e1030. 81, 294. Ward, C.V., Walker, A., Teaford, M.F., Odhiambo, L., 1993. Partial skeleton of Proconsul Shapiro, L., 1993. Functional morphology of the vertebral column in primates. In: nyanzae from Mfangano Island, Kenya. Am. J. Phys. Anthropol. 90, 77e111. Gebo, D.L. (Ed.), Postcranial Adaptation in Nonhuman Primates. Northern Illi- Waterman, H.C., 1929. Studies on the evolution of the pelves of man and other nois University Press, DeKalb, pp. 121e149. primates. Bull. Am. Mus. Nat. Hist. 58, 585e642. Shapiro, L., 1995. Functional morphology of indrid lumbar vertebrae. Am. J. Phys. Williams, S.A., 2011. Evolution of the hominoid vertebral column. Ph.D. Dissertation, Anthropol. 98, 323e342. University of Illinois. Shapiro, L.J., Seiffert, C.V.M., Godfrey, L.R., Jungers, W.L., Simons, E.L., Randria, G.F.N., Williams, S.A., 2012. Placement of diaphragmatic vertebra in catarrhines: Implica- 2005. Morphometric analysis of lumbar vertebrae in extinct Malagasy lemurs. tions for the evolution of dorsostability in hominoids and bipedalism in hom- Am. J. Phys. Anthropol. 128, 823e839. inins. Am. J. Phys. Anthropol. 148, 111e122. Slijper, E.J., 1946. Comparative biologic-anatomical investigations on the vertebral Youlatos, D., 2008. Locomotion and positional behavior of spider monkeys. In: column and spinal musculature of mammals. Verh. Kon. Ned. Akad. Wet. 42, 1 Campbell, C.J. (Ed.), Spider Monkeys. Behavior, Ecology and Evolution of the e128. Genus Ateles. Cambridge University Press, Cambridge, pp. 185e219.