A Partial Skeleton of the Fossil Great Ape Hispanopithecus laietanus from Can Feu and the Mosaic Evolution of Crown-Hominoid Positional Behaviors

The extinct dryopithecine Hispanopithecus (Primates: Hominidae), from the Late Miocene of Europe, is the oldest fossil great ape displaying an orthograde body plan coupled with unambiguous suspensory adaptations. On the basis of hand morphology, Hispanopithecus laietanus has been considered to primitively retain adaptations to above-branch quadrupedalism–thus displaying a locomotor repertoire unknown among extant or fossil hominoids, which has been considered unlikely by some researchers. Here we describe a partial skeleton of H. laietanus from the Vallesian (MN9) locality of Can Feu 1 (Vallès-Penedès Basin, NE Iberian Peninsula), with an estimated age of 10.0-9.7 Ma. It includes dentognathic and postcranial remains of a single, female adult individual, with an estimated body mass of 22–25 kg. The postcranial remains of the rib cage, shoulder girdle and forelimb show a mixture of monkey-like and modern-hominoid-like features. In turn, the proximal morphology of the ulna–most completely preserved in the Can Feu skeleton than among previously-available remains–indicates the possession of an elbow complex suitable for preserving stability along the full range of flexion/extension and enabling a broad range of pronation/supination. Such features, suitable for suspensory behaviors, are however combined with an olecranon morphology that is functionally related to quadrupedalism. Overall, when all the available postcranial evidence for H. laietanus is considered, it emerges that this taxon displayed a locomotor repertoire currently unknown among other apes (extant or extinct alike), uniquely combining suspensory-related features with primitively-retained adaptations to above-branch palmigrady. Despite phylogenetic uncertainties, Hispanopithecus is invariably considered an extinct member of the great-ape-and-human clade. Therefore, the combination of quadrupedal and suspensory adaptations in this Miocene crown hominoid clearly evidences the mosaic nature of locomotor evolution in the Hominoidea, as well as the impossibility to reconstruct the ancestral locomotor repertoires for crown hominoid subclades on the basis of extant taxa alone.

The postcranial anatomy of H. laietanus is mostly known from the partial skeleton (comprising about 60 elements) from CLL2 [7,8] (see locality and institutional abbreviations in Table 1), associated with the face from a male adult individual from the same locality [5,6]. Several features of the thoracic and lumbar vertebrae indicate the possession of a wide and shallow thorax associated with an orthograde body plan [7]. In turn, inferred limb proportions [7], femoral morphology [7,23,24] and phalangeal features [7,8,25] indicate the possession of adaptations for forelimb-dominated, below-branch suspensory behaviors, including a high intermembral index and long and curved manual phalanges. At the same time, the metacarpal proportions and several morphologic details of the proximal phalanges of H. laietanus have been interpreted as indicating the retention of features functionally-related to above-branch quadrupedalism [7,8,26]. This has led to the contention that, among fossil crown hominids, palmigrady was gradually abandoned as suspensory behavior became progressively more adaptively significant [8,9,25,26]. Most recently, however, it has been argued that the unusual metacarpo-phalangeal morphology of H. laietanus might not reflect the retention of quadrupedal behaviors [22]. Under such view, Hispanopithecus would be simply interpreted to display an essentially modern hominoid-like locomotor repertoire, specialized in vertical climbing and suspensory behaviors, but with no significant quadrupedal component. Here we describe a new partial skeleton of H. laietanus from Can Feu (CF), which reinforces the contention that this taxon displayed a unique locomotor repertoire combining suspensory and palmigrade behaviors. The significant implications of this assessment for the evolution of crown-hominid positional behaviors are further discussed below.

The Hispanopithecus Remains from Can Feu
The partial skeleton of H. laietanus from CF1 (IPS34575; Table 2; Figs. 1, 2) was found in 2001 during the construction of an industrial building at Can Feu [27,28], which is situated in the Industrial Park of Can Feu (Sant Quirze del Vallès, Catalonia, Spain) [UTM 31T 424185, 4598895], about 4 km E from CLL (Sabadell). Both localities correspond to alluvial plain facies of the Castellar fan system ( Fig. 3; Vallès-Penedès Basin) [29,30]. After the initial discovery, associated sediments were carefully excavated and screen-washed, leading to the recovery of additional remains belonging to a single hominoid individual (IPS34575; see Table 2). The primate skeleton was recovered in a greenish lutite layer (CF1), although most associated micromammal remains come from a blackish lutite layer (CF2) situated 1-2 m above the former [28]. The presence of Cricetulodon sabadellensis together with the absence of the murid Progonomys enables to correlate CF to the C. sabadellensis local range zone of the Vallès-Penedès Basin [27,28], which ranges from ca. 10.0 to 9.7 Ma (MN9, early Vallesian, Late Miocene) [21]. CF would be therefore contemporaneous or only slightly older than other Hispanopithecus-bearing localities from the same area, such as CLL1 (ca. 9.7 Ma) [21].

Body Mass Estimates
The values computed for UTML* = 14.9 mm, UTSI* = 17.7 mm and UTDP* = 10.0 mm, yield a value of UTSA   [13]. This suggests that H. laietanus displayed a significant degree of body size dimorphism (males about 50% larger than females), as it is common in Miocene and extant great apes [32], being intermediate between the moderate dimorphism displayed by chimpanzees and bonobos (about one-third larger) and the higher dimorphism displayed by gorillas and orang-utans (more than twice as heavy) [33].

Description of Dental Morphology
Detailed descriptions are reported in the Text S1, so that only comparative descriptions are provided below. The lower central incisor and the lower cheek teeth are preserved ( Fig. 1; Table 2;  see Table 3 for measurements, Fig. 4 for proportions, and Figs. 5, 6, and 7 for comparison with other Hispanopithecus specimens). The i1 (Figs. 1A-D, 5A) is a spatulate and waisted tooth, similar but smaller than the i1 from CLL1 (Fig. 5B) [2,3,34]. Both specimens display a longer and more symmetrical crown than an i2 from CLL1 (Fig. 5C), alternatively interpreted as a di1 [3] or i1 [34].
The lower molars (Figs. 1R-W, 8A-B) are subrectangular and display a Y5 occlusal pattern, with a short mesial fovea, a more extensive talonid basin, and a restricted and lingually-situated distal fovea; there are no cingulids, and the lingual cuspids are more peripheralized than the buccal ones, with the hypoconulid situated buccally but close to crown midline. The  [14] by the less quadrangular occlusal profile and more extensive talonid basin. The longer postmetacristid and longer preentocristid in the only complete CP lower molar (Fig. 8N) is too variable to be a reliable diagnostic criterion [11], like the presence of a distinct metaconulid in the former (since it is also present in some CLL1 specimens; Figs. 7A-B, H). Like other Hispanopithecus specimens, the CF m1 and m2 differ from those of Anoiapithecus in the relatively narrower crown (Figs. 4C-D), the narrower buccal cuspulids, the less centrally-placed hypoconulid, and the lack of cingulids.

Description of Postcranial Remains
Several postcranial bones of the shoulder girdle, rib cage and forelimb are preserved (Table 2; Fig. 2; see Supplementary Information for more detailed descriptions). The former include two scapular fragments (Figs. 2B'-F') and the acromial end of the clavicle (Figs. 2R-U), which were previously unknown for Hispanopithecus-the acromial end is not preserved in the purported clavicular fragment from the CLL2 male individual of H. laietanus [7]. The scapular spine (Fig. 2B') is straighter than in extant hominoids, suggesting a different (more monkey-like) shape of the scapular blade, whereas the acromial fragment (Figs. 2C'-F') indicates a longer and more compressed acromion process than in monkeys (somewhat derived towards the hominoid condition).
The clavicular fragment (Figs. 2R-U) is very straight, differing from extant hominoids (which display a marked sigmoid curvature) and even monkeys (which display a well-defined curvature of the acromial end). Early and Middle Miocene apes (Proconsul, Equatorius, Nacholapithecus and Pierolapithecus) display a robust clavicle with a faint sigmoid curvature [36][37], similar to that of colobines [37], thus being less curved and displaying less marked muscular insertions than in extant apes [39,40]. Among fossil apes, the CF specimen most closely resembles the partial clavicle of Equatorius, although given its incompleteness functional inferences are precluded. From the rib cage, only a first rib proximal portion (Figs. 2M-Q) is preserved. Although no comparisons with fossil apes can be provided, it displays a mix of characters, with a protuberant tubercle as in monkeys, hylobatids and humans, a neck-shaft angle similar to hylobatids and extant hominines (lower than in monkeys and orangutans), and a craniocaudally-compressed shaft (as in extant apes), further lacking the proximal shaft constriction displayed by monkeys.
Among the forelimb remains, the humeral fragments (Figs. 2V-A') do not enable well-founded comparisons (Fig. S1). However, the marked lateral supracondylar crest, the flattened distal shaft and the wide shaft portion lateral to the olecranon fossa suggest a modern hominoid-like distal humeral morphology, more derived than in Proconsul, and more similar to that of kenyapithecines (such as Nacholapithecus), Sivapithecus and, especially, Dryopithecus fontani (Figs. S1B-C) [41][42][43] and H. hungaricus [42,44]. The preserved radial diaphysis ( Fig. 2G-J) is smaller and more slender than the male specimen from CLL2 [7], representing about the same shaft portion. Both display a similar mediolaterally-compressed outline, which differs from the rounder profile displayed by extant hominoids and rather resembles quadrupedal monkeys. The distal fragment of ulnar diaphysis (Figs. 2K-L) is not very informative, unlike the proximal partial ulna (Figs. 2A-F).
The CF specimen most completely preserves the Hispanopithecus proximal morphology of the ulna (Figs. 2A-E, 9), which is very informative for making locomotor inferences. The trochlear notch is short and broader laterally (where it further extends posteriorly onto the shaft), with a moderately-developed median trochlear keel. The coronoid process is large and anteriorly-protruding, with a concave surface facing proximally, like the distolateral portion of the trochlear notch, indicating the presence of a spool-shaped humeral trochlea [45]. The radial notch, situated above a relatively well-developed supinator crest, faces laterally. The quite short olecranon process is somewhat tilted posteromedially. Two distinct ulnar morphotypes can be distinguished amongst Miocene apes (Fig. S2). Proconsulids (Proconsul, Turkanapithecus; Fig. S2C), equatorins (Equatorius, Nacholapithecus; Fig. S2E) and the kenyapithecin Griphopithecus (Fig. S2D) display a colobine-like, primitive morphology (Fig. S2G), characterized by a narrow trochlear notch with a faint medial keel, a proximally-protruding olecranon, a deep shaft and a downward-sloping coronoid process [38,42,[46][47][48]. Turkanapithecus, Nacholapithecus and Griphopithecus also display a flat and laterally-facing radial notch, and Nacholapithecus further combines an overall primitive morphology with a more anteriorlydirected coronoid process [47], like Griphopithecus. Extant hominoids (Figs. S2H-J) differ from the above-mentioned taxa by displaying a more derived morphology, characterized by a wide trochlear notch with a well-developed median keel, a poorlydeveloped olecranon process, and a large and anteriorly-projecting coronoid process (whose medial portion projects proximally, creating an inverted V-shape).
Among Miocene apes, only Oreopithecus (Fig. S2C) and to a large extent Hispanopithecus (Figs. 3, S2B) display this modern hominoidlike ulnar morphology [42,45,[49][50][51], whereas Griphopithecus (Fig.  S2B) displays a more primitive condition (even if incompletely preserved). The CF specimen, however, differs in several respects from Oreopithecus, which most closely resembles extant apes by the extremely reduced olecranon process, the short trochlear notch, and the more marked median keel. Overall, the CF specimen most closely resembles the much larger, male proximal ulna of H. laietanus from CLL2 [7] and the similarly-sized female partial ulna of H. hungaricus from Rudabánya ( Fig. 10) [42,44]. Minor differences with the latter include a more slender proximal shaft and a larger and more anteriorly-protruding coronoid process in the CF specimen, whereas similarities between them include the laterally-facing radial notch, the moderately-developed median keel, and the proximally-facing coronoid process that further defines an inverted V-shape. The two latter features, together with distal humeral morphology, enabled previous authors to infer the presence of a spool-shaped humeral trochlea in H. hungaricus [42,44]. However, unlike the two previously-known specimens, the CF ulna preserves the olecranon process and the proximal portion of the trochlear notch, thus enabling a more complete morphofunctional assessment. Thus, compared to most Miocene apes, Hispanopithecus displays a shorter olecranon process together with a shorter and relatively broader trochlear notch. In contrast, the olecranon process of the CF specimen is still somewhat betterdeveloped than in extant apes and Oreopithecus, further being somewhat posteromedially flexed-as in previous Miocene apes, extant quadrupedal monkeys and the knuckle-walking African apes, but unlike in hylobatids and orang-utans.
Finally, a PCA based on eight shape variables of the proximal ulna ( Figure 11, Table S1) further confirms that H. laietanus displays a proximal ulna unlike that of extant great apes, and intermediate between them and colobines, being most similar to that of Presbytis and Pan. The PC1, which explains 55.5% of the variance, separates extant great apes from colobine monkeys mainly due to the relatively wider trochlear surfaces and anteroposterior lower proximal shaft of the former, coupled to a lesser degree with the relatively mediolaterally broader proximal shaft and proximodistally shorter radial notches of great apes compared to colobines; along the PC1, the CF proximal ulna falls just in between great apes and colobines. In turn, the PC2, which explains 30.4% of the variance, is basically driven by the anteroposterior diameter of the radial notch, with Pongo, Gorilla, Nasalis and Colobus displaying relatively anteroposteriorly high radial notches, and IPS34575 falling on the opposite side, by displaying an anteroposteriorly very short radial notch. To a lesser extent, this axis also reflects wider proximal articular breadths (positive values), as well as anteroposteriorly higher proximal shafts, broader proximal articular anteroposterior diameters and deeper sigmoid notches (negative values), with Pan and Presbytis displaying intermediate values on this axis, although slightly closer to the CF specimen.

Taxonomic Attribution
Dental comparisons of the CF material with Middle Miocene hominoids from the Vallès-Penedès [9,21] are restricted to Anoiapithecus [35], given the lack of lower teeth for both Pierolapithecus [36] and Dryopithecus [13]. The CF teeth, however, differ from French D. fontani specimens in the same features previously noted to distinguish Hispanopithecus species from Dryopithecus fontani [10,14]. Regarding Anoiapithecus, it differs from the CF and other H. laietanus specimens regarding p4 as well as lower molar morphology and proportions. On the basis of size, proportions and morphology, the CF dental remains fit well into the range of variation of Hispanopithecus laietanus [1][2][3][4]6], in further agreement with its age (10.0-9.7 Ma) [27,28], only slightly older than other H. laietanus remains (9.7-9.5 Ma), but younger than H. crusafonti (10.4-10.0 Ma) [21]. Some authors have favored the distinct species status of H. crusafonti [9,10,13,14,20,22,52], at least for the CP material [20], whereas others have considered that both samples are insufficiently distinct [11,34]. In any case, the CF specimens differ from those of H. crusafonti from CP in several respects: the shorter and relatively wider p3, and the narrower buccal cuspulids and more extensive talonid basins of the lower molars. The CF molars further differ from those of TF-tentatively attributed to H. crusafonti by some authors [9,10,14,21], but assigned to Dryopithecus fontani by others [11,20,34]-in the same features. Therefore, the CF remains are best attributed to H. laietanus.

Locomotor Inferences
The partial skeleton from CF provides new information on several anatomical regions, such as the first rib, the acromial end of the clavicle and the proximal ulna, which were previously unknown in the partial skeleton from CLL2 [7], thus enabling us to refine previous locomotor inferences for this taxon. The new remains agree well with previous inferences of an orthograde body plan in this taxon [7], as shown among others by the various modern hominoid-like features displayed by the first rib fragment, which represents the first direct evidence of thorax morphology in Hispanopithecus. However, both the rib and the clavicular fragments display a mixture of primitive (monkey-like) and derived (modern hominoid-like) features, suggesting that H. laietanus possessed a locomotor repertoire unlike that of extant hominoids. In this regard, the proximal morphology of the ulna recorded by the CF skeleton is most significant, given the fact that modern hominoids are characterized by a distinctive elbow morphology. The proximal ulnar morphology shared by extant hominoids is functionally related to increased pronation/supination and flexion/extension ranges, by providing substantial stability without compromising mobility at the humeroantebranchial joint [42,51,[53][54][55][56][57][58][59]. In contrast, the ulna of Early and Middle Miocene apes resembles extant non-hominoid anthropoids, reflecting a more restricted range of flexion/extension, and a greater stability only in full pronation [55]. In contrast, the universal stability attained by the elbow of extant apes under a broad range of positions is suitable for extensive forelimb use under both tension and compression during eclectic climbing and below-branch suspensory behaviors [42,56]. The narrow and anteroposteriorly deep proximal ulnar shaft of Early and Middle Miocene hominoids, together with their longer olecranon process-where the principal elbow extensor inserts [60]-and downward-sloping coronoid process, suggest stronger bending stresses along the parasagittal plane with a primarily semiflexed elbow (i.e., a limited range of extension), and are therefore indicative of quadrupedalism [42,46,[61][62][63]. Nevertheless, proconsulids, afropithecids and kenyapithecines already display a mosaic of primitive anthropoids and some derived hominoid features [43,62,64], indicating that the elbow joint was loaded in a variety of flexion/extension and pronation/supination postures, even though higher stability was still attained in full pronation [43,55,65]. In the ulna, the higher degree of forearm rotation of Miocene apes is reflected in their more laterally facing radial notch-an anteriorly-facing radial notch being related to habitually pronated forearms [42,55,63,65]-as well as in their stronger muscular insertionsrelated to enhanced supination capabilities [45]. Together with other anatomical regions, the elbow of these taxa suggests that they were slow-moving, above-branch pronograde quadrupeds with no suspensory adaptations, but already employing more abducted limb postures and more powerful grasping capabilities than other anthropoids [25,43,55,[62][63][64][65][66]. Amongst Middle Miocene African hominoids, Nacholapithecus most clearly shows a humeroulnar complex somewhat more derived towards a higher stability against mediolateral stresses and a somewhat enhanced pronation/supination range, probably indicating a higher reliance on climbing than in previous taxa, in spite of still lacking suspensory adaptations [40,43,47,66,67]. A similar condition is displayed by the proximal ulna of Griphopithecus [10,42,43,68], as shown by the still narrow trochlear notch with no median keel and the long olecranon process.
At the same time, the CF specimen also shows that Hispanopithecus still retained a proximal ulnar morphology unlike that of extant hominoids, suggesting the presence of significant differences in their locomotor repertoires. On the one hand, the PCA reported in this paper indicates that the CF proximal ulna is morphologically distinctive, and intermediate between that of great apes and colobine monkeys in several regards ( Figure 11). Thus, the distinctive anteroposteriorly short radial notch of the CF specimen (as shown by the PC2), coupled with its intermediate proximodistal length (as depicted in the PC1), are reflecting the Ushaped articular surface characteristic of most Miocene apes. The CF specimen is also intermediate regarding anteroposterior shaft and articular diameters at the proximal ulna, with monkeys displaying the highest diameters. This has been related to higher bending stresses on this plane, in relation to predominant parasagittal limb movements [42], and might also be linked to the relatively slender ulnae in comparison to the radius of monkeys compared to apes, further reflecting the higher mediolateral bending stresses of the former, in relation to a predominant quadrupedal posture [75]. Hispanopithecus further retained a somewhat proximally-projecting and posteromedially-tilted olecranon process. Olecranon orientation relative to the forearm determines the elbow position at which the maximum mechanical advantage of the triceps brachii muscle is attained [60]. Therefore, the slightly proximally-protruding olecranon process of Hispanopithecus may be functionally explained by the retention of pronograde behaviors, which require elbow stability also at semiflexed postures [60]. It should be taken into account that the Hispanopithecus olecranon process is also medially protruding, thus more closely resembling the condition displayed by African apes among extant hominoids [76,77]. This condition, termed 'flexor expansion' [77], has been related to the role played by the digital flexors during knuckle-walking [77]. Although such functional relationship remains to be tested, the absence of this feature in orangutans [76] and the presence in monkeys and Miocene apes suggests that it might be related to quadrupedal postures in general. Whereas knuckle-walking adaptations can be discounted in H. laietanus, the proximomedial expansion of its ulna is suggestive of a higher degree of quadrupedalism than in hylobatids and Pongo, and thefore agrees with the presence of palmigrady-related features in the hand of this taxon-the short metacarpals and the morphology of the proximal articulation of the proximal phalanges [8]-although to a lesser extent than in Pierolapithecus and other Middle Miocene taxa [25,26]. Powerful grasping capabilities, suitable for above-branch quadrupedalism, can be also inferred for H. hungaricus on the basis of carpal and phalangeal morphology, suggesting the presence of a large and powerful pollex-as in other Miocene apes [18,72,78]. A significant amount of quadrupedalism is further indicated by the peculiar (Miocene ape-like) configuration of the shoulder girdle and the mediolaterally-compressed shaft of the radius from the CF skeleton. In summary, new evidence provided here confirms that the Late Miocene great ape Hispanopithecus displayed an adaptive compromise between hyperextension capabilities (presumably for suspensory and other orthograde behaviors) and more primitive, pronograde behaviors.

Implications for the Evolution of Crown-hominoid Positional Behaviors
Despite phylogenetic uncertainties, Hispanopithecus is considered a crown-hominid by most researchers, being alternatively interpreted as a stem pongine [5,6] (an extinct taxon more closely related to orangutans than to African apes and humans), a stem hominine [10,12,22] (more closely related to the African ape and human clade than to orangutans), or a stem hominid [9,21,35] (a fossil great ape preceding the divergence between pongines and hominines, but postdating the split between hylobatids and the great ape and human clade)-see ref. [9] for further discussion on hominoid systematics and the arguments put forward in favor of each of these phylogenetic alternatives for Hispanopithecus. From a locomotor viewpoint, Hispanopithecus is the oldest ape documenting unquestioned suspensory adaptations, shared by all extant crown hominoids (hylobatids and hominids), thus being of utmost significance for understanding the emergence of modern hominoid positional behaviors. The proximal ulna from CF, being the most complete available for the genus Hispanopithecus, reflects an elbow complex suitable for preserving stability along the full range of flexion/extension and enabling a broad range of pronation/ supination, thus confirming previous inferences of specialized suspensory behaviors [7,8,[23][24][25]74]. However, the rib, clavicular and scapular remains display a mixture of primitive and derived features, suggesting that Hispanopithecus, in spite of orthograde features, possessed a locomotor repertoire currently unknown among extant hominoids. This is further confirmed by the CF ulna, which differs from that of the committed suspensory hylobatids and orang-utans in the slightly more proximally projected olecranon. The latter is functionally interpreted as a compromise between enhanced extension at this joint for suspensory behaviors and for still important weight-bearing postures with a semi-flexed elbow during above-branch arboreal quadrupedalism. Thus, during quadrupedalism Hispanopithecus would not have displayed the fully-extended elbow position most commonly employed by extant hominoids. African apes display a similar morphology (medially but not proximally protruding olecranon) due to adaptation to knuckle-walking, which represents a compromise between terrestrial quadrupedal behaviors-with extended elbow postures [63]-and orthograde arboreal behaviors. However, knuckle-walking can be discounted in Hispanopithecus on the basis of phalangeal and metacarpal morphology [8,18,25]. The CF proximal ulna therefore reinforces the view [8], previously dismissed by other authors [19], that the Hispanopithecus forelimb reflects a different locomotor compromise, combining climbing and suspensory behaviors with powerful-grasping above-branch palmigrady.
The possession in fossil apes of locomotor repertoires unknown among extant taxa agrees well with the inferred mosaic evolution of the hominoid locomotor apparatus [8,25,26,43,62,64,66,73], but has profound implications for the reconstruction of ancestral locomotor repertoires. The lack of suspensory adaptations in the orthograde, putative stem hominid Pierolapithecus [25,26,36,66,71]see [74] for a different interpretation-otherwise adapted to vertical climbing and powerful-grasping palmigrady, suggests that suspensory behaviors evolved independently at least between hylobatids and hominids [9,25,26,36,66]. Such a contention is reinforced by lack of suspensory adaptations in the pongine Sivapithecus, despite possessing a modern elbow configuration with a spool-shaped trochlea [43,70]. Hispanopithecus, however, stands out as the only Miocene ape in which palmigrady-related features are retained in spite of clear-cut suspensory adaptations. Such a locomotor mosaic is unknown not only among extant, but also among other fossil apes. Given that suspensory features have independently evolved in other primates [43,64,65,68,73], most notably atelines [79], their independent evolution in several crown hominoid lineages, from an orthograde ancestor similar to Pierolapithecus, does not seem unlikely. Atelines display a combination of climbing, quadrupedal and suspensory behaviors, but lack several modernhominoid postcranial adaptations, such as the characteristic hominoid humeroantebrachial complex that provides universal stability at the elbow joint under a variety of positions [43,56]. These features, such as the spool-shaped humeral trochlea, are useful during suspensory behaviors for resisting the mediolateral stresses caused by strong wrist and finger flexor muscles [62]. Nevertheless, they could have originally evolved for stabilizing the humerulnar joint during above-branch quadrupedalism [43,65], i.e. as an adaptation to increase pronation-supination forearm capabilities for maintaining balance above arboreal supports, as required by the tailless hominoid condition [62][63]66,80].
Hispanopithecus differs from other Miocene apes by uniquely showing a transitional stage in which a modern hominoid-like elbow complex appears to be simultaneously an adaptation to keep balance during palmigrady as well as an exaptation for performing suspensory behaviors. The latter eventually replaced above-branch quadrupedalism in all extant ape lineages, ultimately enabling great apes to reach very large body masses that would have been otherwise untenable. Nevertheless, given its quite large body size, the retention of above-branch quadrupedalism in Hispanopithecus suggests that suspensory behaviors did not originally evolve to solve balance problems during horizontal arboreal travel. More specific targets of selection, such as a more efficient feeding on terminal branches in spite of large body size [8,66], could have been involved. If so, the modern-hominoid elbow morphology could have been co-opted several times independently from a  partly quadrupedal ancestor-at least hylobatids and hominids, but perhaps even hominines, pongines and/or dryopithecines-in order to perform these behaviors [25,43,66]. At the very least, the unique locomotor repertoire evidenced by Hispanopithecus should warn us against reconstructing the ancestral positional behaviors of extant hominoid subclades on the basis of the biased evidence provided by their few and very specialized remaining living representatives, without taking the fossil evidence into account.

Morphometric Analysis of the Proximal Ulna
In order to quantify the phenetic affinities of the proximal ulna, we relayed on the published means of the following eight measurements from this anatomical region in extant great apes and selected colobines (the most arboreal catarrhines), extracted from Table 4C in ref. [42]: PAP, proximal shaft height (anteroposterior); PSML, proximal shaft mediolateral diameter; PAB, proximal articular breadth; TAB, trochlear articular breadth; RAP; radial notch anteroposterior diameter; RPD, radial notch proximodistal diameter; PAAD, proximal articular antero-posterior diameter; SND, sigmoid notch depth. Based on these linear measurements, we created eight Mosimann shape variables by dividing each raw measurement by the geometric mean of all the original variables and applying a logarithmic transformation (with natural logarithms, ln) [82,83]. We summarize these logshape data via Principal Components Analysis (PCA) of the covariance matrix and a minimum-spanning tree based on   Table S1. doi:10.1371/journal.pone.0039617.g011 Euclidean distances, using the software Palaeontological Statistics (PAST) [84]. Table S1 Results of the Principal Components Analysis (PCA) of the proximal ulna. This PCA analysis is based on eight Mosimann shape variables, computed from the mean values for the following eight linear measurements [42], by dividing them by their geometric mean (GM) and applying logarithms (ln): PAP, proximal shaft height (anteroposterior); PSML, proximal shaft mediolateral diameter; PAB, proximal articular breadth; TAB, trochlear articular breadth; RAP; radial notch anteroposterior diameter; RPD, radial notch proximodistal diameter; PAAD, proximal articular anteroposterior diameter; SND, sigmoid notch depth. Only those PCs explaining more than 1% of variance have been depicted. The first (PC1) and second (PC2) principal components (see Figure 11) explain more than 85% of the variance. See main text for a morphofunctional interpretation. (DOCX)

Supporting Information
Text S1 Description of dentognathic and postcranial remains of Hispanopithecus laietanus from CF. (PDF)