The Thumb of Miocene Apes: New Insights From Castell de Barberà (Catalonia, Spain)

Transcription

1 AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 000: (2012) The Thumb of Miocene Apes: New Insights From Castell de Barberà (Catalonia, Spain) Sergio Almécija, 1,2 * David M. Alba, 2 and Salvador Moyà-Solà 3 1 Department of Vertebrate Paleontology, American Museum of Natural History and NYCEP, New York, NY Institut Català de Paleontologia Miquel Crusafont, Universitat Autònoma de Barcelona. Edifici ICP, Campus de la UAB s/n, Cerdanyola del Vallès, Barcelona, Spain 3 ICREA at Institut Català de Paleontologia Miquel Crusafont and Unitat d Antropologia Biològica (Dept. BABVE), Universitat Autònoma de Barcelona. Edifici ICP, Campus de la UAB s/n, Cerdanyola del Vallès, Barcelona, Spain KEY WORDS grasping pollical phalanges; Miocene hominoids; exaptation; power grasping; precision ABSTRACT Primate hands display a major selective compromise between locomotion and manipulation. The thumb may or may not participate in locomotion, but it plays a central role in most manipulative activities. Understanding whether or not the last common ancestor of humans and Pan displayed extant-ape-like hand proportions (i.e., relatively long fingers and a short thumb) can be clarified by the analysis of Miocene ape hand remains. Here we describe new pollical remains a complete proximal phalanx and a partial distal phalanx from the middle/late Miocene site of Castell de Barberà (ca., Ma, Vallès-Penedès Basin), and provide morphometric and qualitative comparisons with other available Miocene specimens as well as extant catarrhines (including humans). Our results show that all available Miocene taxa (Proconsul, Nacholapithecus, Afropithecus, Sivapithecus, Hispanopithecus, Oreopithecus, and the hominoid from Castell de Barberà) share a similar phalangeal thumb morphology: the phalanges are relatively long, and the proximal phalanges have a high degree of curvature, marked insertions for the flexor muscles, a palmarly bent trochlea and a low basal height. All these features suggest that these Miocene apes used their thumb with an emphasis on flexion, most of them to powerfully assist the fingers during above-branch, grasping arboreal locomotion. Moreover, in terms of relative proximal phalangeal length, the thumb of Miocene taxa is intermediate between the longthumbed humans and the short-thumbed extant apes. Together with previous evidence, this suggests that a moderate-length hand with relatively long thumb involved in locomotion is the original hand morphotype for the Hominidae. Am J Phys Anthropol 000: , VC 2012 Wiley Periodicals, Inc. The morphology of the thumb in Miocene apes is important for understanding its evolution in humans, in which it is involved in all precision-grasping activities (Napier, 1960, 1993) 1. The grasping hands of all extant primates display a selective compromise between locomotion and manipulation (Napier, 1960, 1993; Tuttle and Rogers, 1966; Tuttle, 1967, 1969; Etter, 1973; Alba et al., 2003). Thus, extant apes, which display many adaptations to arboreal locomotion, also display long hands with very short and slender pollices. On the contrary, extant humans whose hands are free from locomotor demands possess short fingers with long and robust thumbs. An interesting question is whether the hands of Miocene hominoids displayed similar proportion to those of extant apes or whether they were more primitive, retaining a more generalized catarrhine condition like some monkeys. Previous studies of early Miocene apes such as Proconsul and Afropithecus (Begun et al., 1994; Ward, 1998), middle Miocene taxa such as Nacholapithecus (Nakatsukasa et al., 2003), Sivapithecus (Pilbeam et al., 1980; Madar et al., 2002) and Pierolapithecus (Almécija et al., 2009), and the late Miocene Hispanopithecus (Begun, 1993) and Oreopithecus (Moyà-Solà et al., 1999, 2005a) suggest that all of them displayed relatively long thumbs. For early and middle Miocene taxa, as well as for Hispanopithecus 2, this feature has been related to the important role of the thumb in assisting powerful grasping during arboreal locomotion, whereas in Oreopithecus 1 Throughout the text, the term ape is employed to refer to nonhominin hominoids. 2 There is currently an agreement that the genus Dryopithecus must be restricted to middle Miocene remains, so that late Miocene taxa should be reallocated into one or several different genera. Moyà-Solà et al. (2009) refers to the hominoid from Rudabánya as Hispanopithecus hungaricus (therefore considering Rudapithecus as a junior subjective synonym of Hispanopithecus), whereas Begun (2009) refers to the same taxon as Rudapithecus hungaricus. Grant sponsor: Fulbright Commission; Grant number: 2009 BFUL Grant sponsor: Generalitat de Catalunya; Grant numbers: 2009 BP-A 00226, 2008 BE , 2009 SGR 754 GRC. Grant sponsor: US National Science Foundation; Grant number: NSF Award #BCS Grant sponsor: Spanish Ministerio de Economía y Competitividad; Grant numbers: CGL , CGL , RYC Grant sponsor: European Commission s Research Infrastructure (SYNTHESYS project). *Correspondence to: Sergio Almécija, Department of Vertebrate Paleontology, American Museum of Natural History and NYCEP, New York, NY salmecija@amnh.org Received 21 January 2011; accepted 9 March 2012 DOI /ajpa Published online in Wiley Online Library (wileyonlinelibrary.com). VC 2012 WILEY PERIODICALS, INC.

2 2 S. ALMÉCIJA ET AL. Fig. 1. Pollical remains from Castell de Barberà, including the partial distal phalanx IPS 4335 (a d) and the complete right proximal phalanx IPS 4333 (e j). IPS 4335 is shown in palmar (a), dorsal (c) and both lateral views (b,d). IPS 4333 is shown in palmar (e), radial (f), dorsal (g), ulnar (h), distal (i), and proximal (j) views. All Miocene apes for which these bones are known display a thumb with those relatively long and curved phalanges. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] it has been related to enhanced manipulative capabilities in the context of facultative terrestrial bipedalism evolved under insular conditions (Köhler and Moyà-Solà, 1997; Rook et al., 1999; but for alternative views see: Susman, 2004, 2005; Marzke and Shrewsbury, 2006). Here we assess the morphology of new hominoid pollical remains from Castell de Barberà (Vallès-Penedès Basin, Catalonia, Spain) together with the previously available fossil evidence. The remains consist of a complete proximal phalanx (IPS 4333) and a partial distal phalanx (IPS 4335), which come from the same site but were not found associated. With an estimated age of about Ma (either late Aragonian, MN 8, or early Vallesian, MN9), these remains provide additional data on thumb morphology of extinct hominoids from the middle to late Miocene. Morphofunctional considerations based on these and other fossil remains permit us to make functional inferences on thumb use during locomotor behaviors in Miocene apes. In turn, this allows us to make inferences about the hand morphology from which that of early hominins and Pan evolved. CHRONOLOGICAL BACKGROUND The site of Castell de Barberà (Crusafont-Pairó and Golpe, 1972) is located within the Vallès-Penedès Basin, on the NE Iberian Peninsula. This basin is a NNE-SSWoriented half-graben limited by the Littoral and Prelittoral Catalan Coastal Ranges, which is filled by Neogene and Quaternary sediments that span from the early Miocene to the Pleistocene. There are no magnetostratigraphic data available for Castell de Barberà. This locality has been considered to be younger than Sant Quirze (MN8) and Can Llobateres 1 (MN9) (Crusafont-Pairó and Golpe, 1972; Aguilar et al., 1979; Agustí et al., 1985; Casanovas-Vilar, 2007), being traditionally attributed to the late Aragonian (Agustí et al., 1985, 2001; Alba et al., 2006, 2009, 2010a; Casanovas-Vilar, 2007; Casanovas- Vilar et al., 2011), i.e., the MN8 sensu Mein and Ginsburg (2002), albeit with some exceptions (e.g., Andrews et al., 1996, who attributed the site to the MN9). The rodent assemblage from Castell de Barberà certainly corresponds to the Megacricetodon ibericus 1 Democricetodon crusafonti local biozone (Alba et al., 2006; Casanovas-Vilar, 2007; Casanovas-Vilar et al., 2011). However, since late Aragonian and early Vallesian faunas are very similar (e.g., Agustí et al., 2001), an attribution to the MN8 is based merely on the purported absence of Hippotherium at the site. On this basis, Alba et al. (2010a) proposed an age of roughly 11.5 Ma for Castell de Barberà. Nevertheless, Crusafont-Pairó and Golpe-Posse (1974) reported the find of a Hippotherium remain eroded from some level that could be at most situated a few meters above the main fossiliferous layer from Castell de Barberà, and Rotgers and Alba (2011) recently discovered a Hippotherium molar from the site among the collections of the Institut Català de Paleontologia Miquel Crusafont. As such, an uncertain age range of Ma must be assumed for Castell de Barberà (MN8 or MN9), being constrained by the entry of Hippotherium at 11.1 Ma and that of Cricetulodon at 10.4 Ma (see Alba and Moyà-Solà, 2012, for further details). Description The thumb remains from Castell de Barberà (Fig. 1) include a complete proximal phalanx (IPS 4333) and a partial distal phalanx (IPS 4335). The former was ini-

3 THE THUMB OF MIOCENE APES 3 tially identified as a fifth proximal manual phalanx by Moyà-Solà et al. (1990), although more recently Moyà- Solà et al. (2005a) figured this bone and correctly identified it as a proximal pollical phalanx (ibid.: their Fig. 1e). IPS 4335 is a partial pollical distal phalanx lacking most of the basal region. Its preserved length is 15.1 mm while mediolateral and dorsopalmar diameters at the distal end are 2.7 and 4.0 mm, respectively. The shaft tapers distally, especially in mediolateral dimensions, so that at the distal end there is no apical tuft (Fig. 1a d). In palmar view, there are some muscle insertion scars in the most proximal remaining portion of the shaft (Fig. 1a). From what remains of the base and shaft, the proximal phalangeal region appears to have been rather flat. The dorsal surface of the shaft is mediolaterally convex (Fig. 1c). The palmar surface is flat distally, but shows a faint depression proximally, which is also appreciable in lateral and medial views (Fig. 1a,b,d). Interestingly, this phalanx shows a slight degree of curvature (Fig. 1b,d). Because of its peculiar morphology, most similar to that described for other Miocene apes (see Discussion), this specimen is attributed to a pollical distal phalanx. IPS 4333 is a complete, 27-mm long pollical proximal phalanx (Fig. 1e j). It shows an oblique transverse crack in the proximal portion of the shaft and some erosion on the radial side (Fig. 1f). Otherwise, the specimen nicely shows all morphological details. The base displays two bilaterally protruding tubercles; in dorsal view, the right is proximodistally longer, whereas the left one, being shorter, projects more proximally and farther away from the main axis of the phalanx (Fig. 1g). These asymmetries identify this specimen as a right pollical proximal phalanx; the radial expansion corresponds to the insertion of the flexor pollicis brevis and abductor pollicis brevis, whereas the ulnar tubercle corresponds to that of the adductor pollicis. The base is wider than high (due to the mediolaterally protruding tubercles), although the proximal articular surface is not that wide, being slightly ellipsoid (Fig. 1j). Moreover, the proximal articular surface is concave, particularly in radioulnar direction, which can be also appreciated in palmar view (Fig. 1e,j). The dorsal surface of the shaft is mediolaterally convex, like the palmar side, which shows a prominent longitudinal palmar keel. This keel is flanked by two small fossae and associated conspicuous ridges, one on each side, although the ulnar one is more visible because it is not eroded (Figs. 1h and 3e). In dorsal view, the shaft is parallel-sided (Fig. 1g), although in lateral view, the dorsopalmar height decreases distalward (Fig. 1f,h). This phalanx displays a considerable degree of curvature, noticeable in radial and ulnar views, especially at the distal third of the shaft, so that the trochlear region is palmarly bent (Fig. 1f,h). Regarding the trochlear region, the ulnar condyle extends more distally than the radial one, the latter being more palmarly protruding. This is a typical anthropoid feature (Almécija et al., 2009), which has been described in detail in humans by Shrewsbury et al. (2003). The whole trochlear region is radially tilted, so that the pollical distal phalanx would have faced the remaining digits during flexion. The two trochlear condyles flare palmarly, and define a shallow but broad groove between them (Fig. 1e,i). Because of the palmarly bent trochlea, the distal articular surface faces mostly onto the palmar side (Fig. 1e,g,i). Finally, the pits for the insertion of the collateral ligaments are conspicuous, displaying a half-moon morphology that is typical of middle phalanges and proximal pollical phalanges (Fig. 1f,h). All these features clearly separate this specimen from nonpollical proximal phalanges, in which all Miocene apes display a common array of traits: the proximal articular surface extends onto the dorsal shaft; the basal tubercles are extremely developed in the palmar side, to sustain the weight of the body while channeling the long flexor tendons during the hyperextension in palmigrade behaviors; the palmar side of the shaft tends to be flat; the ridges do not extend along all the shaft, but are restricted on the distal half portion; the trochlea is modestly flared, and concomitantly the trochlear groove is very narrow; and finally, the trochlear pits for insertion of the collateral ligaments are not half-moon-shaped as in middle phalanges, but round (Nakatsukasa et al., 2003; Almécija et al., 2007, 2009; Ersoy et al., 2008). Taxonomic attribution Pliopithecid remains have been reported from Castell de Barberà on the basis of dental remains (Crusafont-Pairó, 1975, 1978; Crusafont-Pairó and Golpe-Posse, 1981, 1982; Andrews et al., 1996; Begun, 2002; Alba and Moyà-Solà, 2012). These dental remains, like those from the somewhat older Pliopithecus canmatensis from the local series of Abocador de Can Mata (Alba et al., 2010a), correspond to a small-bodied catarrhine primate (roughly the size of a gibbon). The comparatively large size of the phalangeal remains described here, thus, is incompatible with a taxonomic attribution to the small-bodied pliopithecid recorded from Castell de Barberà, and suggest instead an attribution to a large hominoid, from which a distal humerus has been also recovered (Moyà-Solà et al., 1990; Alba et al., 2011). The appearance of pliopithecoids and hominoids in a single site is rare; currently, both groups are simultaneously recorded only at Rudabánya, Hungary (Begun, 1993; Kordos and Begun, 2001; Begun et al., 2010), and the local stratigraphic series of Abocador de Can Mata, Spain (Alba et al., 2006, 2009; Moyà-Solà et al., 2009; Alba et al., 2012). The attribution of the phalangeal specimens to a hominoid primate cannot be more definitively settled because no pliopithecid pollical proximal phalanges are known, even among the collections of Anapithecus from Rudabánya (Begun, 1988, 1993) or Epipliopithecus from Devínska Nová Ves (Zapfe, 1960). The size criterion previously employed by Begun (1993) to attribute the proximal phalanx RUD 109 from Rudabánya to a hominoid instead of a pliopithecoid, as well as the relatively wide transverse dimensions of the proximal phalanx from Castell de Barberà compared to the manual phalanges of pliopithecids (e.g., Zapfe, 1960), further confirm an attribution of this specimen to the Hominoidea. Previously, the large-bodied hominoid from Castell de Barberà was identified as cf. Dryopithecus by Moyà-Solà et al. (1990), or even as Dryopithecus laietanus (currently Hispanopithecus laietanus; see Moyà-Solà et al., 2009) by Andrews et al. (1996). Unfortunately, no dental material of the hominoid taxon from Castell de Barberà is available, since the upper canine IPS1823, attributed by Golpe Posse (1993, under the label of IPS 26 ) to a female specimen of H. laietanus, doesin fact belong to a larger, male individual of the pliopithecid taxon recorded at the same site (Begun, 2002; Alba and Moyà-Solà, 2012). On the basis of size, the hominoid partial humerus from Castell de Barberà has been attributed to cf. Dryopithecus fontani, but reliable body mass estimates cannot be derived from the phalangeal specimens. As such, and given the presence of other large-bodied hominoids in the Vallès-Penedès basin during the late Arago-

4 4 S. ALMÉCIJA ET AL. nian and Vallesian, the restricted sample of hominoid phalangeal specimens from Castell de Barberà are insufficient to attain a secure taxonomic attribution other than to a large-bodied hominoid. MATERIALS AND METHODS Sample and measurements We collected seven standard variables (measured to the nearest 0.1 mm using digital calipers) on the pollical proximal phalanges of a sample of all extant ape genera and humans, as well as cercopithecines (Macaca) and colobines (Nasalis) and selected fossils: length (L); mediolateral width at the trochlea (MLT), midshaft (MLS), and the base (MLB); and dorsopalmar height at the trochlea (DPT), midshaft (DPS), and the base (DPB). Individual values (for fossil specimens) and descriptive statistics (for extant taxa groups used in our analyses) of phalangeal measurements are reported in Table 1. Among Miocene apes, there are pollical phalanges available for the following genera: Proconsul (complete proximal phalanges and partial distal phalanges; Begun et al., 1994); Afropithecus (both complete proximal and distal phalanges; Leakey et al., 1988); Nacholapithecus (a partial proximal phalanx; Nakatsukasa et al., 2003; and a recently discovered proximal portion joining a previous distal portion, KNM-BG 17813/37295, M. Nakatsukasa, personal communication); Sivapithecus (partial proximal phalanges; Pilbeam et al., 1980; Madar et al., 2002); Pierolapithecus (one distal phalanx; Almécija et al., 2009); Hispanopithecus (a partial proximal phalanx from Rudabánya; Begun et al., 1993); and Oreopithecus (several partial proximal phalanges and two nearly complete distal phalanges; Moyà-Solà et al., 1999, 2005a). Some selected comparisons are reported for pollical distal (Fig. 2) and proximal phalanges (Fig. 3). Regarding the Oreopithecus pollical proximal specimens, due to damage there is no agreement on whether IGF belongs to a pollical proximal or middle phalanx (see complete discussion in Moyà-Solà et al., 1999, 2005a; Susman 2004, 2005; Marzke and Shrewsbury, 2006). Besides, both BA#140 and IGF are too badly crushed to be used in our morphological comparative analysis, and after the examination of all the Oreopithecus isolated phalangeal specimens housed in Basel, we believe that it is more likely that BA#83 belongs to a nonpollical ray. Thus, our description is based only in the small BA#200 specimen (Fig. 3g), which is further associated with a very small partial hand (probably belonging to an adult female). The fossil sample used in our numerical analyses comprises the most complete adult specimens: Castell de Barberà specimen IPS 4333; the recently conjoined Nacholapithecus kerioi specimen KNM-BG 17813/37295 (measurements kindly provided by Masato Nakatsukasa); KNM-WK 18121, attributed to Afropithecus turkanensis (Leakey et al., 1988), for which measurements were taken from a good quality cast; and the Proconsul heseloni specimen PH 100 (from individual KNM-KPS 3), for which measurements were taken from the literature (Begun et al., 1994). Principal components analysis and minimum spanning tree To easily visualize overall morphologic affinities of the IPS 4333 pollical phalanx with other fossils and extant taxa, we performed a principal components analysis (PCA) based on the covariance matrix of seven shape variables using PAST (Hammer et al., 2001). The shape variables for the pollical proximal phalanx were computed on the basis of the original seven measurements reported in Table 1; these variables were transformed into shape variables by dividing each of them by the geometric mean (GM) of all the seven phalangeal measurements (the GM being taken as a variable of overall phalangeal size) and then applying a logarithmic transformation (on the basis of natural logarithms, ln) prior introduction into the analysis, following Jungers et al. (1995). Only specimens with all the seven measurements available were included, thus discarding the Proconsul PH100 proximal phalanx. The PCA was performed using mean sex-specific values for adults of each genus (species in the case of Pan). A minimum spanning tree (MST) was also computed to show nearest-neighbor relationships among fossils and extant means. Similar results were found using separate species/subspecies for extant great apes (not shown), but the overall picture is clearer this way. Relative pollical phalangeal length To compare phalangeal length relative to body size, we relied on the allometric relationship between the length of the pollical proximal phalanx (PP1 Length, in mm) and body mass (BM, in kg) in our sample. The allometric regression line was derived by using natural logarithms (ln) on the basis of mean sex-specific values for adults of each genus/species reported in Table 1. Because of the large differences in BM between the different subspecies of Pan troglodytes (particularly P. t. schweinfurthii), this analysis was performed separately for each of them. Average BM values for extant taxa were taken from Smith and Jungers (1997), whereas in the case of fossil taxa, BM estimates based on postcranial material associated with the phalanges were taken from the literature (see Table 1). Unfortunately, and because the Castell de Barberà specimen IPS 4333 was found in isolation, it cannot be included in the allometric plot of relative phalangeal length. We also decided not to include Oreopithecus, due to the uncertain anatomical attribution of the IGF pollical phalanx, which is the only proximal phalanx of this taxon associated to an individual from which BM can be estimated (see discussion in Morphological comparison section below). This analysis was also precluded for Ardipithecus ramidus, because the available ARA-VP- 7/2I pollical proximal phalanx does not belong to the ARA- VP-6/500 partial skeleton (Lovejoy et al., 2009), for which the BM estimates of this taxon have been derived. Nevertheless, the relative length of other Miocene specimens, as compared to extant taxa, must be taken into account to provide the adequate comparative framework. Relative basal height of pollical proximal phalanges We noticed that Miocene ape pollical proximal phalanges seem to have a particularly low base. Thus, to compare the dorsopalmar height of the base in proximal phalanges (PP1 DBP), we relied on allometric comparisons relative to the geometric mean of this bone (PP1 GM). As such, these variables were logarithmically transformed (using natural logarithms, ln). Again, the GM was taken as a proxy for the overall size of this phalanx (Jungers et al., 1995; Almécija et al., 2009, 2010). Thus, only the complete specimens (where all the seven original variables could be measured to compute

5 THE THUMB OF MIOCENE APES 5 TABLE 1. Pollical proximal phalanx measurements (in mm) and body mass estimates (in kg) for the fossil specimens included in the numerical analyses Taxon Sex BM (kg) N PP1L PP1DPT PP1MLT PP1DPMS PP1MLMS PP1DPB PP1MLB Castell de Barberà (IPS 4333) a Nacholapithecus kerioi d (KNM-BG 17813/37295) b Afropithecus e (KNM-WK 18121) a Proconsul heseloni 9.30 f (KNM- KP3 PH100) c P. paniscus females g 11 Mean SD males g 8 Mean SD P. t. troglodytes females g 6 Mean SD males g 2 Mean SD P. t. schweinfurthii females g 2 Mean SD males g 6 Mean SD P. t. verus females g 2 Mean SD males g 3 Mean SD Gorilla females g 27 Mean SD males g 52 Mean SD Pongo females g 18 Mean SD males g 18 Mean SD Homo females g 7 Mean SD males g 7 Mean SD Hylobatidae females 6.66 g 5 Mean SD males 6.92 g 11 Mean SD Macaca females 6.68 g 5 Mean SD males g 11 Mean SD Nasalis females 9.82 g 2 Mean SD males g 7 Mean SD Sex-specific means and standard deviations, as well as average BM for the extant groups included in the analyses are also reported. Abbreviations: PP1 5 pollical proximal phalanx; L 5 maximum length; ML 5 mediolateral width; DP 5 dorsopalmar height; B 5 base; MS 5 midshaft; T 5 trochlea. Sources for longitudinal measurements: a This study. b M. Nakatsukasa (personal communication). c Begun et al. (1994). Sources for BM estimates: d Ishida et al. (2004), maximum BM estimate for KNM-BG skeleton, considered to be a good estimate for the KNM-BG 17813/37295 specimen on the basis of size similarity (M. Nakatsuka, personal communication). e Leakey and Walker (1997). f Rafferty et al. (1995). g Smith and Jungers (1997). Although most analyses were performed at the generic/specific level, the extant comparative sample includes: Gorilla, eastern (G. beringei) and western (G. gorilla) gorillas from Royal Museum for Central Africa and American Museum of Natural History; Pongo, Sumatran and Bornean orangutans (P. p. pygmaeus and P. p. abelii) from Naturhistorisches Museum Basel, American Museum of Natural History and Naturalis ; Hylobatidae, siamangs (H. syndactylus) and two species of gibbons (H. agilis and H. muelleri) from American Museum of Natural History and Naturalis; Homo, modern humans (H. sapiens) from Universitat Autònoma de Barcelona; Macaca, several macaque species (M. fascicularis, M. fuscata, M. nemestrina, M. nigra, M. silenus and M. sylvanus); and finally Nasalis larvatus; both monkeys from American Museum of Natural History and Naturalis. the GM) could be employed in the analysis of relative basal height: IPS 4333 from Castell de Barberà; Nacholapithecus KNM-BG 17813/37250 and Afropithecus KNM-WK Measurements for fossil specimens and descriptive statistics for the extant comparative sample are reported in Table 1.

6 6 S. ALMÉCIJA ET AL. Fig. 2. Pollical distal phalanges of selected extant taxa and Miocene apes, in palmar view: (a) Papio, (b) Gorilla, (c) Pan, (d) IPS 4335 from Castell de Barberà; (e) IPS , Pierolapithecus catalaunicus; (f) IGF 11778, Oreopithecus bambolii; and (g) extant humans. In comparison to Miocene apes and modern humans, extant apes display roughly conical pollical distal phalanges, with barely perceptible muscular impressions. This is especially true for Pan. Note the peculiar distal end of Miocene ape phalanges. All specimens from the right side, with the exception of Gorilla and Oreopithecus phalanges. Original artwork by M. Palmero. Scale bar represents 1 cm. Fig. 3. Morphological comparisons of pollical proximal phalanges in selected extant taxa and Miocene apes: (a) PH 100, Proconsul heseloni; (b) KNM-BG 35250BA, Nacholapithecus kerioi; (c) KNM-BG 17813/37295, Nacholapithecus kerioi; (d) GSP 51344, Sivapithecus parvada; (e) KNM-WK 18121, Afropithecus turkanensis; (f) IPS 4333 from Castell de Barberà; (g) RUD 109, Hispanopithecus hungaricus; (h) BA#200, Oreopithecus bambolii; (i) Papio; (j) Pongo; (k) Gorilla; (l) Pan; and (m) extant humans. Note the presence of well-developed ridges and a median bar in the palmar side of the Miocene phalangeal shafts. They also share a noticeable curvature on the distal shaft and a relatively low base with huge laterally protruding tubercles (especially in relation to the trochlea). The latter feature makes these phalanges to be relatively wide at the base level. All specimens from the right side, with the exception of Sivapithecus, Gorilla and extant human phalanges. Original artwork by M. Palmero. Scale bar represents 1 cm. RESULTS Principal component analysis and minimum spanning tree Our PCA and MST (Fig. 4a,b; Table 2) show that, in spite of some differences, the Miocene pollical specimens analyzed are more similar to each other than to any extant taxon when all the axes in the multivariate space are taken into account. Given the fact that the original variables were standardized by the GM, all the component loadings should be interpreted as relative to the overall size of the bone. On PC 1, the length loads most strongly, and negatively, so the data show that humans

7 THE THUMB OF MIOCENE APES 7 Fig. 4. Results of the principal components analysis (PCA) and minimum spanning tree (MST) on pollical proximal phalanges. (a) PC 1 vs. PC 2 (49.9 and 21.9% of the total variance, respectively). (b) PC 1 vs. PC 3 (14.6% of the total variance). The MST shows that Miocene apes are more similar to each other, and subsequently to modern humans, than to extant great apes. See Table 2 and text for further details. TABLE 2. Results of the principal component analysis (PCA) performed on the covariance matrix of seven proximal pollical phalanx measurements (see Table 1) PC 1 PC 2 PC 3 PC 4 PC 5 PC 6 PC 7 Eigenvalue E-20 % Variance E-16 % Cumulative Variance Component loadings PC 1 PC 2 PC 3 PC 4 PC 5 PC 6 PC 7 PP1L PP1DPT PP1MLT PP1DPMS PP1MLMS PP1DPB PP1MLB 2.62E E The variables were size-corrected by the geometric mean and ln-transformed prior to incorporation into the PCA. The most significant component loadings in PC 1, PC 2, and PC 3 are highlighted and discussed in the text. Abbreviations as in Table 1. and Gorilla have relatively short proximal phalanx lengths, likely because of their midshaft mediolateral robusticity relative to other phalangeal variables. On the opposite side, hylobatids show very low values on PC 1 because of their relatively long and slender phalanges. Nasalis, Macaca, Pongo, and Pan occupy a central position in the first axis. Afropithecus is in the right limit of Pan, closely followed by IPS 4333, whereas Nacholapithecus is close to modern humans. Positive values in PC 2 mainly separate Nasalis, Macaca, Pongo, and Gorilla, due to their mediolaterally broad trochleae, whereas negative values segregate both Afropithecus and IPS 4333 due to their dorsopalmarly high midshafts. Nacholapithecus overlaps with Pan along this axis. Macaca and humans have high values in PC 3, which is related, again, to mediolaterally broad midshafts and dorsopalmarly shallow bases, whereas African apes are separated by traits related to dorsopalmarly high bases. Afropithecus and IPS 4333 overlap with Macaca and humans, whereas Nacholapithecus is different on this axis by being exceptionally broad and flat. Interestingly, the MST suggests that Miocene apes resemble modern humans more closely than any other extant taxon analyzed. The same is true for Gorilla, although Pan is different by being less similar to humans in this multivariate morphospace (Fig. 4a,b). The similarities between humans and Miocene apes phalanges could be driven by relative robusticity (depth and/or width) relative to length. Relative phalangeal length Our analysis of relative length of the pollical proximal phalanx (Fig. 5a) shows that in this respect humans display longer phalanges relative to BM than apes and, especially, monkeys, the latter showing very short proximal phalanges relative to BM. The three Miocene apes included in the analysis (Proconsul, Nacholapithecus, and Afropithecus) display somewhat longer proximal phalanges than extant apes, closely approaching the condition of extant humans. Among extant great apes, only the females of the eastern chimpanzee subspecies (P. t. troglodytes) fall within Miocene apes to this regard. Relative basal height of pollical proximal phalanges Our analysis of relative basal height of the pollical proximal phalanx (Fig. 5b) shows that extant apes tend

8 8 S. ALMÉCIJA ET AL. Fig. 5. (a) Allometric bivariate plot of pollical proximal phalanx length (ln PP1 Length) vs. body mass (ln BM), on the basis of the sex-specific mean values reported in Table 1. Because of the uncertainty on the BM estimates for the fossil specimens, we further included a 620% range of uncertainty around the estimate, since the range of mean estimates can provide a first approach to an estimation range, while a closer estimate might be obtained from the 20% range around the grand mean (Delson et al., 2000). (b) Allometric bivariate plot of pollical proximal phalanx dorsopalmar height at the base (ln PP1 DPB) vs. the geometric mean of seven phalangeal measurements (ln PP1 GM, taken as a measure of overall phalangeal size), on the basis of the sex-specific mean values reported in Table 1. A linear regression for extant apes are provided for both plots. In all cases, BM and GM means are higher for males. to display relatively higher phalangeal bases than humans (see also Morphological comparisons and Fig. 3). Extant monkeys appear rather intermediate between extant apes and humans, whereas middle Miocene apes (IPS 4333, Nacholapithecus and Afropithecus) display the lowest values. DISCUSSION Morphologic and morphometric comparisons The main morphologic features of the Miocene pollical phalanges analyzed in this article are reported in Table 3. Basically, the pollical phalanges of these Miocene taxa (Proconsul, Nacholapithecus, Sivapithecus, Afropithecus, Castell de Barberà, Hispanopithecus, and Oreopithecus) are Miocene-hominoid-like, resembling each other more closely than either extant apes or monkeys (Rose, 1983; Ersoy et al., 2008; Almécija et al., 2009). Thus, with the exception of Oreopithecus (which displays subtle differences), the pollical distal phalanges of Miocene apes are wide and dorsopalmarly compressed at the base; they display muscular insertions on the palmar side of the shaft, which markedly tapers distally; they have a bipartite articular surface, which ranges from flat to convex; and they show a slightly developed-to-absent apical tuft (Fig. 2; Table 3). On the contrary, extant great apes display pollical distal phalanges with a relatively narrow base and very faint muscular impressions; slightly concave to flat articular surfaces, which sometimes are not conspicuously divided into two facets by a median eminence; and a round and slightly developed apical tuft (Fig. 2). All the latter features stem from the rudimentary nature of the thumbs of extant great apes (see Straus, 1942; Almécija et al., 2010). Miocene pollical distal phalanges resemble neither hylobatids nor monkeys (Begun et al., 1994), those from the latter being flatter and wider, and with bilaterally protruding apical tufts in the case of the terrestrial cercopithecids (Fig 2a; Almécija et al., 2010). All Miocene apes also share an array of features on their pollical proximal phalanges (see Figs. 3 and 4a,b and Table 3) to the exclusion of extant apes; these features include: a wide base flanked by very bilaterally protruding tubercles; a concave proximal articular surface; the presence of a marked longitudinal palmar keel, surrounded by bilateral fossae and associated ridges; and a somewhat curved shaft, especially at the distal end, which makes the trochlear region to be palmarly bent. Furthermore, our PCA and MST show that, although some Miocene apes show pollical proximal phalanges that resemble extant great apes in some traits (such as shaft breadth and in some cases trochlear breadth), the former more closely resemble each other and modern humans when all the axes of our multivariate analysis are taken into account (Fig. 4a,b). This analysis further distinguishes extant great apes by the morphology of their base (Fig. 4b). Regarding pollical length relative to BM, our results show that Miocene apes display relatively long proximal phalanges (Fig. 5a), as previously suggested by other authors (Begun, 1993; Madar et al., 2002; Nakatsukasa et al., 2003). Their length relative to BM is intermediate between short-thumbed extant apes and the longthumbed humans. Cercopithecoid monkeys, on the contrary, display very short proximal pollical phalanges in relative terms (Etter, 1973), being noticeably shorter than apes (Fig. 5a). Only female eastern chimpanzees have a pollical proximal phalanx that, relative to BM, falls within the range of Miocene apes. Moreover, both Miocene apes and humans share a relatively short basal height, significantly lower than that displayed by extant apes (Figs. 4b and 5b). Overall, both morphologic and morphometric data indicate that Miocene ape pollical phalanges more closely resemble each other and even those of modern humans (in terms of relative phalangeal length) than those of extant apes. It is worth noting

9 THE THUMB OF MIOCENE APES 9 TABLE 3. Main morphological features of the pollical phalanges of Miocene apes discussed in this paper as compared to extant great apes and humans Pollical distal phalanx Base Taxon Shape Articular surface Proximally protruding median keel Proconsul Wide and flat Bipartite, flat to convex Yes Afropithecus Wide and flat Yes Pierolapithecus Wide and flat Bipartite, flat to convex Yes Castell de Barberà Wide and flat?? Oreopithecus Wide and flat Bipartite, flat to slightly concave Yes Extant humans Wide and flat Bipartite, concave No Gorilla Wide and flat Barely bipartite, flat to slightly convex Yes Pan Oval, width [ height Barely bipartite, flat to slightly concave Barely perceptible Pongo Oval, width [ height Barely bipartite, flat to slightly concave Barely perceptible Shaft Shape Palmar proximal fossa (muscular attachment) Proconsul Strongly tapering distally, Large and shallow Afropithecus Pierolapithecus Strongly tapering distally, straight Large and shallow Castell de Barberà Strongly tapering distally, curved Large and shallow Oreopithecus Tapering distally, straight Large and shallow Extant humans Slightly tapering distally, straight Very large and deep Extant apes Tapering distally, straight Absent to very faint Distal end Overall shape Tuft development Proconsul Round Afropithecus Round Pierolapithecus Round None Castell de Barberà Round None Oreopithecus Horseshoe-shaped Well developed Extant humans Horseshoe-shaped Highly developed Extant apes Round Moderate

10 10 S. ALMÉCIJA ET AL. TABLE 3. (Continued) Pollical proximal phalanx Base Taxon Shape Tubercles Articular surface Proconsul Width [[ height Highly protruding Laterally restricted, concave Afropithecus Width [[ height Highly protruding Laterally restricted, concave Nacholapithecus Width [[ height Highly protruding Laterally restricted, concave Sivapithecus Width [[ height Highly protruding Laterally restricted, concave Castell de Barberà Width [[ height Highly protruding Laterally restricted, concave Hispanopithecus Oreopithecus Width [[ height?? Extant humans Width [[ height Highly protruding Concave Gorilla Width [ height Moderatelly protruding Concave Pan Width height Slightly protruding Flat to slightly concave Pongo Width height Slightly protruding Flat to slightly concave Shaft Curvature Palmar keel Position and development of sheath ridges and associated fossae Proconsul Moderate Convex First to second third of the shaft, do not protrude palmarly Afropithecus Moderate Convex Around midshaft, protrude moderately Nacholapithecus Moderate Flat Around midshaft, protrude moderately Sivapithecus Moderate Convex Around midshaft, protrude moderately Castell de Barberà Moderate Convex Around midshaft, protrude moderately Hispanopithecus Slight Slightly flat Distal third of the shaft, protrude moderately Oreopithecus Moderate Slightly flat Proximally situated, highly protruding Extant humans None Absent No developed ridges Extant apes Slight to none Absent/very faint No developed ridges Trochlear region Position Condyles Groove Pits for the collateral ligaments Proconsul Palmarly bent Flared, laterally restricted Deep and wide Well developed Afropithecus Palmarly bent Flared, somewhat laterally restricted Shallow and wide Well developed Nacholapithecus Palmarly bent Flared, somewhat laterally restricted Shallow and wide Well developed Sivapithecus Flared, somewhat laterally restricted Shallow and wide Well developed Castell de Barberà Palmarly bent Flared, somewhat laterally restricted Shallow and wide Well developed Hispanopithecus Slightly palmarly bent Slightly flared, bilaterally wide Very shallow and wide Well developed Oreopithecus Palmarly bent Slightly flared, bilaterally wide Faint Well developed Extant humans Aligned with the shaft Slightly flared, bilaterally wide Faint Very faint Extant apes Aligned with the shaft Slightly to not flared, bilaterally wide Faint Very faint The information lacking in Proconsul and Afropithecus pollical distal phalanges is not reported or figured anywhere. In some traits, extant great apes show subtle differences between the different genera. In those cases it is indicated in the table (see text for further details).

11 THE THUMB OF MIOCENE APES 11 that, among extant great apes, Gorilla most closely approaches the Miocene condition, while Pan and Pongo display more rudimentary pollical phalanges (see also discussion in Almécija et al., 2010). These results on the pollical distal phalanges agree with previous findings on pollical metacarpal resemblances between humans and Gorilla (Hamrick and Inouye, 1995; Ohman et al., 1995). Functional interpretations In their preliminary description of the Proconsul forelimb, Napier and Davis (1959) observed that the thumb was unlike that of any extant ape. In particular, the latter authors noted that the Proconsul thumb was longer than in extant apes, thus indicating a lesser emphasis on suspensory behaviors. The morphology of the proximal articular surface of the thumb proximal phalanx, being slightly dorsally oriented, mediolaterally wide, and dorsopalmarly short (further shown in this article for other Miocene taxa; Figs. 3, 4b, and 5b), suggests more abducted thumb postures suitable for powerful grasping (Begun et al., 1994). This contention would be also supported by the long pollical distal phalanx, which in Proconsul is nearly as long as the hallucial one (Begun et al., 1994). Powerful grasping is consistent with other hand features of Miocene apes, notably the bases of their nonpollical proximal phalanges, which are wide and dorsopalmarly short, with dorsally-oriented joint surfaces, indicating more emphasis on flexion-hyperextension at the metacarpophalangeal joint during above-branch palmigrady (Rose, 1986; Madar et al., 2002; Almécija et al., 2007, 2009; Ersoy et al., 2008; Alba et al., 2010b). Thus, the slightly dorsally oriented articular surface on the pollical proximal phalanges of Miocene apes could suggest as slight degree of hyperextension at the pollical metacarpophalangeal joint, which would result in increased abduction of the thumb. On the other hand, the lack of palmarly protruding tubercles would indicate that weight-support would have relied more on the nonpollical rays, as already suggested (e.g., Begun et al., 1994; Almécija et al., 2007, 2009). However, the pollical proximal phalanges of Miocene apes do display laterally protruding tubercles at the base, which partially accounts for their relative dorsopalmar shortness. These tubercles serve as muscle tendon insertions (e.g., flexor pollicis brevis and abductor pollicis brevis on the radial side), also indicating a well-developed flexor apparatus and palmar plate, as previously suggested for Sivapithecus (Madar et al., 2003), therefore further confirming a higher emphasis on flexion strength than in extant great apes. A strong thumb has been also inferred for Afropithecus on the basis of the saddle-shaped joint between the trapezium and first metacarpal, indicating enhanced grasping capabilities of the thumb in this taxon (Rose, 1992; Ward, 1998), as well as for Nacholapithecus (Nakatsukasa et al., 2003), Sivapithecus (Pilbeam et al., 1980; Madar et al., 2002), Hispanopithecus (Begun, 1993), and Oreopithecus (Moyà-Solà et al., 1999, 2005a), on the basis of their pollical proximal phalanges. In this respect, their palmarly bent trochleae, together with the morphology of the trochlear articular surfaces highly restricted to the palmar side would be indicative of a greater emphasis on flexed positions of the pollical interphalangeal joint in all these taxa as compared to extant apes, the thumbs of these Miocene apes being actively used for powerful grasping (Begun et al., 1994; Madar et al., 2002; Nakatsukasa et al., 2003; Fig. 3). In the case of the late Miocene Hispanopithecus, suspensory behaviors have been inferred for the specimens from Rudabánya and Can Llobateres (Catalonia, Spain) on the basis of elbow morphology (Morbeck, 1983; Begun, 1992), intermembral proportions (Moyà-Solà and Köhler, 1996) and manual anatomy (Almécija et al., 2007; Kivell and Begun, 2009). Moreover, the combination of specific adaptations for suspensory behaviors plus palmigrady, unknown amongst extant suspensory apes (Begun, 1993), has been inferred on the basis of manual remains for the late Miocene Hispanopithecus from Spain (Almécija et al., 2007). The morphology of RUD 109 with a more restricted trochlear surface on the palmar side departs slightly from earlier specimens, suggesting that this taxon may have had less emphasis on the use of the thumb in locomotion relative to other Miocene taxa (Fig. 3g). However, its relative dimensions indicate that it might have still combined suspensory behaviors with some degree of thumb-assisted, powerful-grasping capabilities during above-branch palmigrady. The peculiar morphology of the pollical proximal phalanges of Miocene apes characterized by a longitudinal middle palmar bar or keel, associated with the lateral fossae, and somewhat developed ridges could be analogous to the morphology (with the same association of palmar keel and associated lateral fossae) described by Marzke et al. (2007) for the middle phalanges of extant primates. Thus, these features could also be similarly used in Miocene apes as a predictor of the force potential of the flexor muscle in their pollices, which would be used in a similar way to other digits although in abducted positions to powerfully assist grasping during arboreal locomotion. This would be also indicated by the higher degree of pollical phalangeal curvature of Miocene apes in comparison to extant great apes (Fig. 3). Because dorsopalmar curvature in phalanges is an adaptation to reduce bending moments when grasping the locomotor substrate with flexed joint postures, an increased curvature is favored in arboreal taxa (e.g., Preuschoft, 1973; Nakatsukasa et al., 2003; Richmond, 2007). With regard to the pollical distal phalanges of Miocene apes, they differ from both extant monkeys and apes by being relatively longer (Begun et al., 1994; Almécija et al., 2009) and displaying conspicuous muscular insertions on the palmar side (Fig. 2; Begun et al., 1994). These features indicate a strong development of the flexor apparatus in the thumb. Furthermore, the pollical distal phalanges of Miocene apes would have displayed a higher degree of mobility, as indicated by their flat to slightly convex articular surface (Begun et al., 1994; Almécija et al., 2009). The peculiar shape of the strongly tapered distal end, without any discernible apical tuft, indicate that these apes possessed very poorly developed palmar pads and narrow nails (Shrewsbury et al., 2003; Mittra et al., 2007), which would be even less developed than in extant forms. Probably, the main reason why extant great apes especially Pongo and Pan still preserve their thumbs is because of selective pressures driven by manipulation (e.g., Napier, 1960, 1993; Tuttle and Rogers, 1966; Tuttle, 1967, 1969; Alba et al., 2003). Thus, the somewhat developed apical tufts in their pollical distal phalanges could have evolved to support a bigger digital pulp than in Miocene apes, allowing these short thumbs to secure their specific precision grips (e.g., Napier, 1960, 1993; Tuttle, 1967, 1969). On the other hand, the morphology of the nonpollical distal phalanges of extant great apes, with a tough