Precision grasping morphology

Extant great-ape PDPs lack all the features related to precision grasping (Figure 2); as such, they display a smooth apical tuft (instead of a developed apical tuberosity with ungual spines), and further lack well-developed basal tubercles, which in humans reflect the presence of collateral intraosseous ligaments for sustaining the nail bed ([3]–[4]; see Figures 1 and 2 and Videos S1, S2, S3, S4, S5 for renderings of the PDPs in Figure 2).

Interestingly, the PDP of Orrorin, being the earliest pollical specimen in the hominin fossil record (ca. 6 Ma; [16]–[17]) displays an overall human-like morphology (Video S3). The latter even includes the most significant features related to precision grasping (Figure 2), although some of them (such as the ridge for insertion of the FPL and the apical tuberosity) are stouter as compared to later hominins and modern humans [17].

The Olduvai Hominid 7 (OH 7) specimen—originally attributed to Homo habilis [17]—differs from that of extant apes and humans (Video S5). This PDP does not display ungual spines [3], and there is no ridge for insertion of the FPL, but a huge palmar fossa that extends until the large apical tuberosity (Figure 2). Furthermore, the lack of a distinctive ungual fossa and spines would be indicative of limited palmar pad compartmentalization and, as such, of a restricted precision-grip capability [3].

PCA

A PCA based on PDP shape variables allow us to discriminate the several extant genera being analyzed between each other (Figure 3, see Materials and Methods and Table 1). Positive values on the PC 1 (68% variance) are related to phalanges with mediolaterally narrow tufts and shafts, and with dorsopalmarly high midshafts and bases, thus having an overall rod-like appearance. Negative values, on the contrary, are related to phalanges with a flat appearance due to high breadths at midshaft and at the distal end (i.e. with apical tuberosities instead of tufts). Positive values on the PC 2 (13% variance) mainly separate phalanges with a relatively large base, in both mediolateral and dorsopalmar diameters (i.e., with a relatively small shaft and apical tuft), from phalanges that are very long relative to other dimensions (see Figures 2 and 3 and Videos S1, S2, S3, S4, S5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Principal components analysis (PCA) based on six shape variables of the pollical distal phalanx.

Blue, Papio; red, Pongo; yellow, H. sapiens; green, Gorilla; grey, Pan. The PC 1 largely reflects the proportions of the tuft and shaft, while the PC 2 is more related to the proportions of the base. The Orrorin PDP overlaps with modern humans in both principal components, and later hominins also resemble modern humans in both components—although to a lesser degree. Paranthropus robustus and OH 7 constitute an exception, because they fall within the human range across the PC 2, but depart from the remaining taxa on the PC 1 by showing exceptionally wide PDPs (Figures 2 and 3). See text for further explanation. Figures at the corners represent the outline of these phalanges in palmar and lateral views.

https://doi.org/10.1371/journal.pone.0011727.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Main results of the principal components analysis (PCA) based on the six shape variables of the pollical distal phalanx, including the variable loadings from the rescaled component matrix for the five principal components.

https://doi.org/10.1371/journal.pone.0011727.t001

To this respect, extant great apes display relatively narrow PDPs with a dorsopalmarly high midshaft, while the apical tuft is not well developed, conferring them a rod-like shape. This is especially true concerning Pan and Pongo, which can be roughly distinguished from each other thanks to the highest loads on PC 2 of orangs, which display very small shafts and apical tufts relative to the base (Figure 3). On this basis, gorillas occupy a central position on the scatter plot, because their PDPs overall resemble a somewhat flatter version of chimps' (see also Figure 2).

Extant humans display PDPs with relatively wide shafts and apical tuberosities (see Figure 2), as indicated by the low loads on the PC1. Moreover, they show low values on the PC2, which are also correlated to a significantly long PDP (Table 2). This combination of relatively long and wide PDP is exclusive of humans among the extant taxa analyzed (Figure 3). It is interesting to point out that, although humans and gorillas overlap in both PC 1 and PC 2, they occupy different regions in the morphospace.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Descriptive statistics for the ratios of distal phalangeal robusticity.

https://doi.org/10.1371/journal.pone.0011727.t002

The position of baboons (Papio) in the scatter plot indicates that they show overall proportions on PDPs more similar to those of humans than to great apes, although being relatively shorter and displaying a larger base (Figure 3).

Concerning the fossil hominins, Orrorin (BAR 1901′01) most closely resembles modern humans in both components (Figure 3). Neandertals (La Ferrassie I and Kebara 2) and Stw 294 (cf. A. africanus) also fall close to modern humans in both components, although both La Ferrassie I and Stw 294 display slightly highest values on the PC 2. However, both Paranthropus robustus (SKX 5016) and OH 7 depart in PC 1 by having an extremely robust (i.e mediolaterally broad) shaft and apical tuberosity in the PDP.

Ratios

Here we provide additional morphometric evidence regarding the robusticity of the distal phalanges, by comparing the first (DP1) and third (DP3) manual rays of extant taxa, together with OH 7, Paranthropus robustus and Neandertals (Figure 4 and Table 2). In orangutans, the PDP is only slightly more robust than the third distal one, whereas in African apes the reverse condition is found. The 95% confidence intervals (CI) for DP1 and DP3 robusticity between extant great apes do not overlap, suggesting that differences are significant. In modern humans the degree of distal phalanx tuft robusticity for the first and third manual rays is very similar, although it is somewhat higher in the thumbs of Neandertals, with both La Ferrassie I and Kebara II having robusticity values above the 95% CI of extant humans. In monkeys, on the contrary, the distal pollical phalanx is much more robust than the distal third one, with both macaques and baboons showing values well above the 95% CI to that of the third digit. This odd condition is also found in both OH 7 and P. robustus, which the difference between the robusticity of the two digits (DP1 - PD3) falling within the 95% CI displayed by monkeys (see Figure 4 and Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Boxplots of distal phalangeal robusticity in selected extant taxa, Neandertals, OH 7 and Paranthropus robustus.

Robusticity refers to apical tuft width (a) in relation to maximum length (b) of the distal pollical and middle finger phalanges (left and right, respectively; see Materials and for further details). Horizontal lines represent the median values, whereas the boxes represent the 25% and 75% percentiles, the whiskers the maximum-minimum ranges and circles are outliers. OH 7, like Paranthropus robustus, display a robusticity pattern convergent with quadrupedal monkeys (Macaca and Papio), in which the pollical distal phalanx is disproportionally robust relative to that from the middle finger. Note that the pollical and nonpollical distal phalanges attributed to P. robustus may not belong to the same individual.

https://doi.org/10.1371/journal.pone.0011727.g004

Functional analysis

Extant great apes do not significantly use the thumb during locomotion: it does not participate in below-branch suspensory behaviors, and it does not support body weight during knuckle-walking in African apes [18]–[21]. However, it can participate during terrestrial palmigrady and fist-walking in orangutans [18]. In the latter, the hallux is reduced [22], and in some cases the distal phalanx may not be present at all [23]. This is explained by the specialized, four-digit hook grasp of orangs' hands and feet. According to this, the hallux would have been extremely reduced because of locomotor selection pressures, whereas the reduction of the pollex would not have proceeded to the same extent due to contradictory, manipulatory selection pressures favoring instead the possession of a longer thumb [23]. In orangutans, the reduction of the hallux has also affected the long flexor tendon [22]. This condition that can be also found in the thumb of all extant great apes, especially in Pongo and Pan, in which locomotor selection pressures have probably favored the lengthening and increased strength of digits II-V [18].

Thus, although extant great apes display diminished thumbs, especially concerning extrinsic muscle insertions, they do have well-developed intrinsic muscles that enable efficient power and precision grips, the latter being used during food manipulation [1], [2], [5], [18], [24]. Chimps and gorillas do efficiently manipulate small object between their thumb and index finger using different precision grip combinations (e.g. tip-to-tip and pad-to-side), but a human-like, pad-to-pad precision grip is precluded due to the disproportionate length of their digits II-V relative to that of the thumb [1], [2], [5], [24]. Chimps and orangs mostly rely on arboreal foraging by directly putting the foods from branch to mouth, whereas gorillas spend many hours on the ground, where they carefully select, manipulate and hold the food with their hands [18]. Increased terrestriality in such a large ape might have resulted in longer thumbs relative to the rest of the hand [12], [25], as a by-product of their shorter hands relative to body mass [26]. The Gorilla hand therefore displays more balanced proportions between the thumb and index finger (presumably related to somewhat advanced manipulatory capabilities) than chimps and especially orangs [5].

Our results show that extant great apes exhibit rod-like PDPs with barely discernible muscular impressions on the palmar side, further lacking developed tufts and associated palmar pads (Figures 2–4; see also [4], [6], [22]). The lack of dorspalmar flattening in the PDP, especially in Pan and Pongo, would be related to the lack of a developed flexor apparatus (usually without receiving extrinsic muscular component). The overall reduced PDP of orangs, especially referring the shaft and apical tuft, obviously stems from their rudimentary thumbs, which are diminished like in Colobus and Procolobus [22]. In the latter taxa, the PDP, when present, is usually confined to its proximal portion, the base (S.A. personal observation). Gorillas, being the most terrestrial great apes and displaying relatively short and skillful hands with relatively long thumbs, possess a somewhat flatter PDP than the other great apes (Figures 2–3). Furthermore, gorillas also exhibit more barely evidence of muscular insertion scars on the palmar side of the PDP shaft than them (Figure 2). Since the thumbs of gorillas do not participate in locomotion, this morphology should be correlated with an increased use of the thumb for manipulation as compared to Pan and Pongo [18].

Terrestrial monkeys, in its turn, exhibit proportionally short fingers in relation to their thumbs, thereby enabling an efficient precision grip [27]–[28]. This is particularly evident in gelada baboons, which display enhanced manual feeding capabilities [29] thanks to displaying the highest opposability index among extant primates, including humans [28]. Other baboons (Papio spp.) show the same capabilities, although to a lesser degree [28]. These baboons are digitigrade, and only the tip of the thumb contacts the substrate during the touchdown phase [21]. A huge long flexor tendon is inserted onto the distal part of the shaft and tuberosity of their PDP [4], although it does not correspond to the FPL, but to the radial portion of the flexor digitorium profundus [21]. Moreover, during the tip-weight support, their PDPs can be hyperextended, so that, like in humans, their pollical interphalangeal joint displays well-developed sesamoid bones [4]. There are also other similarities between baboon and human PDPs [4], such as a broad distal pad, tuberosity (sometimes with spines) and nail, a large palmar fossa (but not a distinct ridge for insertion of the FPL and ungual fossa), and similar ratios concerning both bone and long flexor tendon dimensions. These similarities can be related to the enhanced manipulative capabilities displayed by these monkeys [27]–[29].

Our results agree with previous findings showing that baboons display PDPs more similar in overall proportions to those of extant humans rather than those of extant great apes, although being shorter and displaying a larger base (Figure 3). This morphology might be an adaptation to frequent weight bearing, during which the tip of the thumb contacts the ground in hyperextended postures. According to this, the enhanced manipulative capabilities displayed by baboons could be merely a by-product of an adaptation of the hand to digitigrady, resulting in a long thumb relative to the rest of the hand that would be suitable for pad-to-pad precision grasping (see Fig. 8a–b in [28]). The precision grasping displayed by these monkeys, in any case, is much less developed than that of humans, which is further reflected by the lack in the former of many PDP traits that are functionally related to human-like precision grasping (Figures 1–2; [3]–[4]).

The morphology displayed by the PDP of modern humans—including a relatively long bone, dorsopalmarly flat and wide at the shaft and apical tuberosity (Figure 3), together with several morphological traits related to a powerful FPL and compartmentalized digital pulp (Figure 2)—is not found among non-human primates, being indicative of a stable and powerful pad-to-pad contact during refined manipulation [3]–[4]. The similar position occupied by the PDPs of modern humans and those of fossil hominins (including Neandertals), together with the possession of morphological features functionally-related to pad-to-pad precision grasping [17], [30]–[31], is highly indicative of shared functional similarities. Apart from Neandertals, this is evident in Stw 294 (cf. A. africanus) and, particularly, in BAR 1901′1 (Orrorin), the latter showing, among early hominins, the greatest resemblance to extant humans (Figures 2–3). On the contrary, SKX 5016 (P. robustus) and OH 7, although more closely resembling extant humans than any other taxa examined, display extremely broad shafts and apical tuberosities (Figures 3–4). Although the phalanx attributed to P. robustus has been described as displaying some traits related to pad-to-pad precision grasping [17], this is not the case of OH 7 [3]–[4], which lacks some of these features (Figure 2 and Video S5; see next section). These results agree with previous ones, according to which both OH 7 and SKX 5016 are extremely broad phalanges, unlike those of humans or apes, thus having a narrow base relative to these dimensions [32].

All the evidence reported above suggests that the morphological differences found between the PDPs of extant taxa seem to be also correlated to functional differences in hand function, particularly involving the thumb. It would be of utmost interest to discern whether the subtle morphological differences found amongst the fossil hominins being analyzed—particularly concerning SKX 5016 and OH 7—are also correlated to differences in function, or whether they are alternatively correlated to different overall body types [32]. This cannot be definitively settled until more postcranial remains of these hominins are available. However, given the apparently convergent morphological similarities between these fossil taxa and baboons, both in the middle [33] and distal phalanges [this study], we favor the view that some kind of functional differences as compared to other hominins are likely.

The attribution of the OH 7 hand remains

The original attribution of the partial hand from Olduvai Bed I to the holotype of H. habilis (OH 7, consisting in the parietals and mandible of a subadult; [34]) has been subsequently accepted by most authors [e.g. 2]–[3], [35]–[36], mainly due to its subadult status. However, due to its overall robusticity—especially concerning the PDP—and curved middle phalanges, an alternative attribution to Paranthropus—contemporary to Homo in that region [37]—was also suggested long ago [18], [38]. More recently, this alternative attribution has been further favored on the basis of morphometric and morfofunctional analyses, particularly concerning middle phalangeal morphology [33]. The later study found that the OH 7 remains are more similar to those from South Africa attributed to P. robustus than to earlier Australopithecus and later and contemporary humans [33]. It is also noteworthy that other bones from the OH 7 hand, such as the trapezium, also suggests a taxonomic attribution different than Homo [9].

Be that as it may, the odd morphology of OH 7 PDP led some authors to consider an alternative anatomical identification as a hallucial distal phalanx [35]. This would be supported by the strong muscular attachments and, especially, the slight axial torsion at the apical tuft (functionally related to bipedalism). However, a discriminant analysis by the same authors indicated “that the fossil is closer to human distal phalanges than to those of any other hominoid and is somewhat more like a pollical than a hallucial phalanx” [35, p. 325]. We concur with this anatomical identification, especially given that there are no differences in axial torsion as compared to modern humans or extant apes (Figure 2), in which the apical tuft is slightly twisted, so that it faces the rest of the fingers. It is also noteworthy that the PDP of OH 7 does not display a human-like morphology related to precision grasping ([3]; Figure 2 and Video S5). This includes its odd overall proportions, which like in SKX 5016 (P. robustus) are characterized by a high degree of apical and midshaft robusticity, as it is found here (Figure 3) and in many previous works [10], [32]–[33], [39].

Furthermore, our results regarding relative distal phalangeal robusticity show that, when non-pollical manual rays are also taken into account, the pattern of robusticity of OH 7, like that of Paranthropus, resembles that of quadrupedal monkeys and does not fit either the great-ape or the human pattern (Figure 4). These results agree with a previous study of this partial hand, which interpreted the morphology of the middle phalanges as showing convergent adaptations with gelada baboons [33]. Alternatively, it would be necessary to conclude that early Homo was more similar to Paranthropus than to Australopithecus and Orrorin regarding several aspects of phalangeal morphology, which in evolutionary terms would imply a reversion regarding, among others, the robusticity of the PDP.

The attribution of BAR 1901′01

Hallucial distal phalanges do not display the set of features present in all the human PDPs, which are related to the pad-to-pad contact. Thus, they do show a large proximoplantar fossa, which is further accompanied by plantarly protruding basal tubercles. In proximal view, these structures configure a very wide and shallow channel for the pass and/or insertion of the flexor hallucis longus, which can act as a “supporting muscle” during terrestrial progression in both apes and humans [40].

Thus, besides morphometric similarities, the PDP of the late Miocene Orrorin displays the typically human set of morphological features functionally related to human pad-to-pad precision grasping (Figure 2 and Videos S3–S4; [4]). The presence of these features, among others, indicates that BAR 1901′1 does not belong to the hallux. The other features indicating an anatomical attribution to the pollex include: its degree of elongation, overall flatness, and round dorsal surface [17]; the orientation of the apical tuberosity, which does not face distally (such is the case in distal phalanges that support weight stresses in hyperextended positions; see Fig. 1 in [41] for the case of OH 10); and the lack of axial torsion, which is present in the hallucial phalanges of both humans [41] and apes [42].

The evolution of refined manipulation

The presence of precision grip features in the PDP of Orrorin indicates that this bone was fully prepared to accommodate objects between the palmar aspect of its pulp and that of the fingers. Some of these precision-grasping features in the Orrorin specimen had been previously reported, although they were interpreted as an adaptation to arboreal locomotion reflecting “the precision grip essential for climbing and balancing, different from that of apes” ([17], p. 372). However, given the fact that no arboreal primate displays this set of features, we favor the hypothesis that functionally relates the striking and detailed similarities between Orrorin and extant human PDPs to refined object manipulation.

Admittedly, although the Orrorin phalanx mostly looks like a human PDP, it also displays some primitive features that are further retained by the later australopiths, such as a small ungual fossa and a proximally protruding median eminence of the articular surface. Moreover, some other features, like the ridge for insertion of the FPL, and the dorsopalmar height of the shaft and apical tuberosity, are stouter than in later hominins and modern humans (Figure 2). The latter display relatively flat PDPs, especially at midshaft and distal tuberosity levels (Figures 2–3). Thus, although the late Miocene Orrorin is somewhat ape-like in dorsopalmar dimensions, it approaches a human-like profile in mediolateral dimensions. Moreover, it is most remarkable that although Plio-Pleistocene australopiths also show the morphological features related to precision grip [17], the 6-million-years-old PDP of Orrorin is more human-like in overall proportions and morphology than these later hominins. To this respect, both OH 7 and SKX 5016 display a degree of mediolateral broadening that highly surpasses the human condition. This is consistent with the femoral morphology of this taxon, which more closely resembles australopiths than later Homo, but among early hominins it is the one that most closely resembles humans [43]. All this evidence suggests that australopiths—especially Paranthropus—are derived by displaying a robusticity pattern on the distal phalanges that is convergent with that of quadrupedal monkeys (Figure 4), as previously suggested for the middle phalanges [33], and also suggested to some degree by the trapezial morphology [9], [36].

Be that as it may, the highest resemblance between the late Miocene Orrorin and modern humans, with the exclusion of australopiths, is an unexpected result that bears important implications for the understanding of the selective pressures originally involved in the evolution of human manual skills. Extant great apes are highly committed to arboreal locomotion (including vertical climbing and suspensory behaviors), their manipulative capabilities being limited by their relative short thumbs [1]–[2], [5], and especially by their absolutely long hands [12]. Because Miocene apes displayed absolutely long thumbs, it has been suggested that their hands were more suitable for manipulative purposes than those of extant hominoids [44]–[47]. The same condition can also be inferred for the stem hominid Pierolapithecus, which displays relatively short manual phalanges—like other early and middle Miocene apes—as well as a relatively long PDP [48]–[49].

From the evidence presented above, it can be suggested that the hand length proportions of humans are plesiomorphic to a large extent. If Orrorin was an early biped hominin with a PDP adapted to refined manipulation, it follows that the whole thumb would be long relative to the hand, thus enabling an efficient pad-to-pad contact. Thus, this leads us to hypothesize that the manual proportions of Orrorin might have more closely resembled those of early and middle Miocene apes than those of extant apes, which would have diverged towards a different direction from the same ancestral morphotype. This idea is similar to that presented by Tuttle [18], who suggested that “By the time early hominids had assumed a bipedal gait, the hand was probably well on its way toward modern human configuration” [18; p. 203]. We therefore favor the hypothesis that human hand proportion enabling a pad-to-pad precision grasping are an exaptation, co-opted by early bipedal hominins for manipulative purposes, but originally evolved within a locomotor selective context as an adaptation to powerful-grasping—assisted by the thumb— within an arboreal setting in Miocene apes. This would explain why early bipedal hominins, such as A. afarensis [12] and A. africanus [13], already displayed human-like hand length proportions prior to the appearance of stone tools in the archeological record. It seem likely that the acquisition of habitual bipedalism—which largely freed the upper extremities from locomotor demands [14]—would have facilitated the refining of the manipulative capabilities displayed by all primates [e.g. 1]–[2], [5], [12], [28], [33]. Furthermore, bipedalism could also affect the hand morphology by means of correlated developmental responses dues to changes in the feet morphology [12], [39]. Thus, most probably “early hominoids [and hominins] […] evidently employed behaviors resulting from the addition of a number of functional morphological innovations to a relatively conservative body plan, including those underlying a more sophisticated manipulative and grasping use of the hand than used before” [50, p. 264].

Most likely, these hand capabilities would have not been co-opted until much later for a regular use in stone tool-making, coinciding with the encephalization increase that took place in hominins with the advent of the genus Homo ca. 2.5 Ma [51]–[53]. Thus, stone tool-making would not have played a significant role until the latter part of human hand evolution, and especially after the advent of the Acheulian culture [54], as already suggested by Tuttle [18]. On the basis of the evidence provided by modern human hands, the fine tuning of manipulative adaptations during the evolution of Homo would have involved an increase of overall robusticity at the several manual joints, an increase in robusticity of the whole thumb, and especially the development of distinct palmar pads on the distal phalanges.

More complete fossil hands from the African Mio-Pliocene transition would be necessary to clarify this issue, especially regarding of the selection pressures underpinning the remarkable human-like features of the Orrorin PDP. For the moment being, the most complete fossil hand around this time corresponds to the 4.4 million-years-old putative hominin Ardipithecus ramidus [55]. Although its hand is not as elongated as in extant apes [55], its thumb seems to be relatively short, and its PDP looks more ape-like than that of Orrorin, in spite of the older chronology of the latter. Thus, although A. ramidus has been claimed to be an early hominin close to the last common ancestor with Pan [56], it could be alternatively interpreted as one of the apes “among the tangled branches that constitute the basal hominine bush” [57, p 533]. Only future comparative studies would help to bring some light onto this question.

Conclusions

The pollical distal phalanx of the early bipedal hominin Orrorin (BAR 1901′01) unequivocally shows precision grasping capabilities in spite of its ancient chronology, most closely resembling modern humans than some later Plio-Pleistocene hominins—which show a derived robusticity pattern. This indicates that refined manipulation is an ancient acquisition already present by the late Miocene. This is consistent with the hypothesis that habitual terrestrial bipedalism and the possession of skillful hands do constitute a single adaptive complex. Both types of behaviors might have been simultaneously selected, by synergistically favoring each other.

From the evidence reported by BAR 1901′01, it is reasonable to assume that human hand length proportions (i.e. short hands and relatively long thumbs) are plesiomorphic to some degree, thus more closely resembling the short hands with relatively long thumbs of Miocene apes, rather than the elongated hands of extant apes, which seem to be secondarily derived. These ancient proportions, suitable for refined manipulation, would not have been co-opted for stone tool-making until much later, coinciding with a significative increase in encephalization in the genus Homo.

The primate sample

The comparative extant sample includes the following taxa: chimpanzees and bonobos (Pan; N = 29), gorillas (Gorilla; N = 13), orangutans (Pongo; N = 19), baboons (Papio; N = 22) and modern humans (Homo sapiens; N = 22). The fossil sample includes the PDPs of Orrorin tugenensis from the Lukeino Formation (BAR 1901′01), Australopithecus africanus from Sterkfontein (Stw 294), the hand from Olduvai (OH 7 A), Paranthropus robustus from Swartkrans (SKX 5016), and H. neanderthalensis from La Ferrassie I and Kebara 2. Apart of the above-mentioned extant taxa, we further employed the following fossil third distal phalanges for computing ratios: OH 7 B, SKX 27504 (attributed to P. robustus), La Ferrassie I and Kebara 2. Measurements for fossil specimens were taken from originals, good quality casts or from the literature [31], [58]–[59].

PCA

A principal components analysis (PCA) based on the covariance matrix was employed to perform morphometric comparisons between the pollical distal phalanges (PDP) of selected fossil specimens and those of other hominins and extant primates, including modern humans. This analysis, which does not assume group membership on a priori grounds, was based on six shape variables of the PDP, in its turn computed on the basis of the following six metrial variables: length (L); mediolateral width at the apical tuft (MLT), midshaft (MLS) and the base (MLB); and dorsopalmar height at midshaft (DPS) and the base (DPB). These measurements were transformed into shape variables by dividing each of them by the geometric mean (GM) of all the six 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), following [43].

Ratios

Distal phalangeal robusticity was analyzed by means of a ratio between apical tuft width and maximum length, separately for both the distal pollical and middle finger phalanges. We further calculated the difference between both ratios, in order to quantify the relative robusticity of the pollical distal phalanx in relation to that of the middle finger. Summary statistics for extant and fossil taxa are reported in Table 2.