Abstract
Objectives Extinct hominins can provide key insights into the development of tool use, with the morphological characteristics of the thumb of particular interest due to its fundamental role in enhanced manipulation. This study quantifies the shape of the first metacarpal’s body in the extant Homininae and some fossil hominins to provide insights about the possible anatomical correlates of manipulative capabilities.
Materials and methods The extant sample includes MC1s of modern humans (n=42), gorillas (n=27) and chimpanzees (n=30), whilst the fossil sample included Homo neanderthalensis, Homo naledi and Australopithecus sediba. 3D geometric morphometrics were used to characterize the overall shape of MC1’s body.
Results Humans differ significantly from extant great apes when comparing overall shape. H. neanderthalensis mostly falls within the modern human range of variation although also showing a more robust morphology. H. naledi varies from modern human slightly, whereas A. sediba varies from humans to an even greater extent. When classified using a linear discriminant analysis, the three fossils are categorized within the Homo group.
Discussion The results are in general agreement with previous studies on the morphology of the MC1. This study found that the modern human MC1 is characterized by a distinct suite of traits, not present to the same extent in the great apes, that are consistent with an ability to use forceful precision grip. This morphology was also found to align very closely with that of H. neanderthalensis. H. naledi shows a number of human-like adaptations consistent with an ability to employ enhanced manipulation, whilst A. sediba apparently presents a mix of both derived and more primitive traits.
1 Introduction
There is no doubt that the extremely dexterous human hand is unmatched among animals. As a result, the human hand has been the subject of considerable paleoanthropological research over the years in order to provide a better understanding of the processes that led to its evolution (Almécija, Smaers, & Jungers, 2015a; Key, 2016; Lewis, 1977; Marzke & Marzke, 2000; Susman, 1998). Much of this has been directed at how the hand morphology of higher primates correlates to their manipulative capabilities, and how fossil morphology can be used to infer tool use in extinct hominin species (Almécija & Alba, 2014; Kivell, Kibii, Churchill, Schmid, & Berger, 2011; Marchi, Proctor, Huston, Nicholas, & Fischer, 2017; Napier, 1955). The thumb and its components, most notably the first metacarpal (MC1), plays a fundamental role in object manipulation and the study of its anatomy has therefore been at the center of research in this field (Galletta, Stephens, Bardo, Kivell, & Marchi, 2019; Marchi et al., 2017). However, the constant discovery of new fossils (e.g. Homo Naledi; Berger et al., 2015) and the development of new morpho-functional analysis tools mean that there is still much about the tool use behaviors and manipulative capabilities of extinct hominins that is yet to be uncovered.
Amongst the extant great apes, humans possess the superior manipulative capabilities, with the ability not only to adeptly utilize the objects in their environment, but also to manufacture complex tools in ways that require high levels of dexterity. These advanced skills are facilitated in part by a unique thumb morphology: the human thumb is long relative to the length of the fingers compared to other less dexterous apes, with powerful thenar musculature and robust thumb bones (Almécija, Moyà-Solà, & Alba, 2010; Almécija et al., 2015; Feix, Kivell, Pouydebat, & Dollar, 2015; Tuttle, 1969).
In comparison to human hands, non-human great apes, especially chimpanzees and orangutans, have longer, robust fingers relative to their shorter, more gracile thumbs, which is probably the result of selective pressures associated with locomotor behaviors such as suspension and knuckle-walking (Almécija et al., 2015; Püschel, Marcé-Nogué, Chamberlain, Yoxall, & Sellers, 2020; Richmond & Strait, 2000). Both gorillas and chimpanzees have been observed using tools in the wild, to varying degrees; however, their manipulative capabilities are limited due to the constraints imposed by their hand morphology. Chimpanzees are prolific tool users, known to use tools both in nature and captivity (Boesch & Boesch, 1990). Examples of chimp tool use include termite fishing with specially crafted sticks (Sanz, Call, & Morgan, 2009), hunting bush babies with sharp spears and nut-cracking with stones (Sanz & Morgan, 2007). Gorillas are less reliant on tool use due the fact that they exploit food resources differently from chimpanzees (e.g., crack nuts with their teeth), but they have been observed using sticks to test the depth of water and to support themselves when crossing deep water (Breuer, Ndoundou-Hockemba, & Fishlock, 2005). However, whilst non-human great apes do regularly use their thumbs to manipulate objects, they are not as efficient as humans in using forceful precision grips (Marzke, Marchant, McGrew, & Reece, 2015; Marzke & Wullstein, 1996).
Traditionally it was believed that extinct hominin species also fell into this category, lacking the manual dexterity of modern humans (Lewis, 1977; Niewoehner, 2001, 2006; Rightmire, 1972). However, there is a growing body of evidence that many hominins as early as O. tugenensis (Gommery & Senut, 2006) show the capacity to efficiently use tools, with the ability to use forceful precision grips as dexterously as humans (Alba, Moyà-Solà, & Köhler, 2003; Feix et al., 2015; Karakostis, Hotz, Scherf, Wahl, & Harvati, 2017; Kivell et al., 2011; Tocheri, Orr, Jacofsky, & Marzke, 2008). Human-like features would have been present along with traits suitable for arboreal locomotion, which lead to the suggestion that the hominin hand evolved in a mosaic fashion showing a manual morphology adapted to these two functional demands (i.e., manipulation and locomotion) (Kivell et al., 2011; Kivell, 2015). This mixed morphology is apparent in the hand of A. sediba and H. naledi (Kivell et al., 2011; 2015) whereas the hand of Neanderthals would be fully derived (Tocheri et al., 2008).
Several studies have focused on different anatomical features of the hand in an effort to understand the extent to which the hand of early hominins is adapted to manipulative abilities (e.g., Almécija et al., 2010; Galletta et al., 2019; Green & Gordon, 2008; Skinner et al., 2015), in particular regarding the joint areas of the MC1, as this bone plays a crucial role in complex manipulative behaviors. However, whilst there is now a greater understanding of the manipulative capabilities of hominins, much of the research on the MC1 dates back to the past century and often contained only qualitative assessments (Aubriot & Tubiana, 1981; Barmakian, 1992; Imaeda, An, & Cooney, 1992; Napier, 1956, 1960; Tuttle, 1969). Even the most recent quantitative research that has been conducted using three-dimensional geometric morphometric (3DGM) techniques have focused only on those certain areas of the MC1 deemed to be most important in controlling manipulation, such as the trapeziometacarpal joint connecting the thumb to the wrist (i.e., proximal articular surface; Marchi et al., 2017) and the first metacarpal distal articular surface (Galletta et al., 2019). As a result, most of MC1’s morphology (i.e., its body) has yet to be fully quantitatively analyzed to assess its possible importance when assessing possible correlates with manipulative abilities.
Consequently, in this study body morphology of the MC1 was quantified using 3DGM in order help in the identification of structures in extant species that may be correlated with human-like manipulative capabilities and determining if similar morphologies are present in fossil hominins. The sample investigated in this study included three extant African ape genera (Homo, Gorilla, Pan) and three fossil hominins (Homo neanderthalensis, Homo naledi and Australopithecus sediba). Based on previous literature about thumb morphology and function, the following hypotheses were tested:
Hypothesis 1: MC1 morphology significantly differs between humans and extant great ape species
Though great apes use their hand for manipulative activities, their specialisation is more a consequence of their locomotion (i.e., knuckle-walking and arborealism) (Almécija, Moyà-Solà, & Alba, 2010). It is therefore expected that the selective pressures associated with locomotor behaviour in chimpanzees and gorillas will result in an MC1 morphology that varies significantly from that of bipedal humans. Furthermore, different use of the human thumb during manipulation and human adaptations to precise and forceful tool use are expected to lead to an MC1 morphology that differs from other extant apes.
Hypothesis 2: All fossil hominin specimens exhibit an MC1 morphology more similar to humans than other great apes
H. neanderthalensis, A. sediba and H. naledi have overall hand morphologies that appear to align with human hands to a greater extent than those of non-human great apes. They possess adaptations, such a long thumb and short fingers, that are associated with advanced manipulative capabilities in modern humans (Holliday et al., 2018; Kivell et al., 2011). Given these morphological characteristics and the inferred tool using abilities of H. naledi, A. sediba and H. neanderthalensis in previous studies, they would be expected to have an MC1 morphology more closely aligned with humans than gorillas or chimpanzees.
2 Material and methods
2.1 Sample
The extant sample used in this study includes MC1s of modern humans (Homo sapiens; n=42), chimpanzees (Pan troglodytes; n=30), and gorillas (Gorilla gorilla and Gorilla beringei; n=27) (Table S1). The human MC1s came from a medieval cemetery in Burgos, Spain (Casillas Garcia & Alvarez, 2005) and the surface models were obtained using a Breuckmann SmartSCAN structured light scanner. The non-human sample came from museum collections and they came from different origins (i.e., wild-shot, captivity and unknown origin). Their surface models were collected using photogrammetry as described in Bucchi et al., (2020). Both scanned and photogrammetry models are high resolution, therefore providing a good representation of the original anatomy. The resolution of the models generated using surface scanner and photogrammetry have been previously tested and found to be comparable (Giacomini et al., 2019), thus allowing us to combine these data types in our analyses. Only adult individuals were included in the study and right MC1s were preferred (although left MC1s were reflected when their antimere was not present).
The fossil sample includes the right metacarpal from a Homo neanderthalensis, the right metacarpal from a Homo naledi and the left metacarpal from an Australopithecus sediba. The H. neanderthalensis sample was found in La Ferrassie archaeological site in Savignac-de-Miremont, France. The skeleton was discovered in 1909 and is estimated to be 70–50,000 years old (Guérin et al., 2015). The Homo naledi sample (Morphosource identifier: S2110) was recovered in 2013 from the Rising Star cave system in South Africa and has been dated to around 250,000 years ago (Dirks et al., 2017). The A. sediba sample (Morphosource identifier: S2490) was taken from the near complete wrist and hand of an adult female [Malapa Hominin 2 (MH2)] discovered in Malapa, South Africa (Berger et al., 2010). The latter fossils were downloaded from Morphosource https://www.morphosource.org/, whereas the Neanderthal was obtained from a cast housed at the Catalan Institute of Human Paleoecology and Social Evolution (IPHES).
2.2. 3DGM
3D coordinates were collected using the software Landmark Editor 3.6 (Wiley et al., 2005) to quantify the MC1’s morphology. Eight curves comprising 20 equidistant coordinates each were placed at pre-defined points on the MC1 (Figure 1). These coordinates were chosen to provide a good representation of the overall shape of the shaft of the bone. The first and last coordinates from each one of the eight curves were treated as fixed landmarks, whereas all the rest of the coordinates (i.e., 144 coordinates) were considered as semi-landmarks. A generalized Procrustes superimposition was performed on the coordinate data to remove differences due to scale, translation, and rotation, thus obtaining shape variables (Bookstein, 1991). The semi-landmarks were slid on the MC1’s surface by minimizing bending energy (Bookstein, 1997; Gunz, Mitteroecker, & Bookstein, 2005).
These obtained shape variables were then used in a principal component analysis (PCA) to quantify overall shape variation. The data set of extant hominoids was then grouped by genus and the Procrustes variance of observations in each group (i.e., the mean squared Procrustes distance of each specimen from the mean shape of the respective group) was computed as a simple measure to assess morphological disparity within each one (Klingenberg & McIntyre, 1998; Zelditch, Sheets, & Fink, 2003). Procrustes variance was applied here as way to evaluate intra-genus variation, and absolute differences in Procrustes variances were computed to test differences in morphological disparity among groups (these differences statistically evaluated through permutation). Then, a multi-group linear discriminant analysis (LDA) (also known as canonical variate analysis) was run to maximize separation between groups using the principal components (PCs) that accounted for 90% of the sample variance. Performance was calculated using the confusion matrix from which the overall classification accuracy was computed, as well as the Cohen’s Kappa statistic (Püschel, Marcé-Nogué, Gladman, et al., 2020; Püschel, Marcé-Nogué, Gladman, Bobe, & Sellers, 2018). The complete dataset was resampled using a ‘leave-one-subject-out’ cross-validation, as a way to asses classification performance (Kuhn & Johnson, 2013). Pairwise PERMANOVA tests with Bonferroni corrections for multiple comparisons were performed to assess for shape differences between the three extant genera using the again PCs that accounted for 90% of the sample variance. Euclidean distances were selected as similarity index.
All these analyses were carried out in R 3.5.1 (R Core Team, 2019), using the ‘geomorph’ 3.1.2 (Adams, Collyer, & Kaliontzopoulou, 2019) and ‘MASS’ 7.3-51.5 packages (Venables & Ripley, 2002).
3 Results
3.1 Principal component analysis
The PCA performed using the shape variables returned 102 PCs. The first 22 PCs accounted for ~ 90% of the total variance of the sample, hence offering a reasonable estimate of the total amount of MC1’s shape variation, which were then used in the LDA and pairwise PERMANOVA tests. The first three PCs in the PCA account for ~ 57% of the total variance and display a relatively clear separation between the extant African ape genera (Fig. 2a). PC1 explains 40.8 %, PC2 10.44% and PC3 5.43% of total variance, respectively (Fig. 1a-d).
Violin plots of PC1 (Fig. 2b) show a notable difference between gorillas and humans vs. chimpanzees. Humans and gorillas exhibit the highest PC1 scores, representing a wider distal articular surface, a larger proximal articular surface, a significantly more robust shaft. Chimpanzees show the lowest PC1 scores, representing a narrower proximal articular surface, a smaller distal head, smaller radial and ulnar epicondyles and a more gracile shaft. H. neanderthalensis falls within the human and gorilla distributions and is distinct completely from the chimpanzees. H. naledi falls within the human distribution, whilst A. sediba is characterized by a lower PC1 score and aligns closer to the Pan distribution. None of the analyzed fossils fall within any of interquartile ranges (IQR) (i.e., black bars in Fig. 1b-d) of any of the extant genera.
Violin plots of PC2 (Fig. 2c) shows distinct variation between the extant genera, with a morphological continuum ranging from Gorilla (higher PC2 values), Pan (central PC2 values) and H. sapiens (lower PC2 scores). Interestingly, due to the presence of a couple of outliers, the morphological variation in Gorilla encompasses the whole range of observed morphological variation. The Gorilla sample has the highest PC2 scores, representing an extended palmar lip, a more curved shaft and more rounded ends. The modern human distribution shows the lowest PC2 scores, representing flatter distal and proximal articular ends, as well as larger radial palmar condyles at the distal end. The chimpanzee sample lies in between the gorilla and modern human samples displaying an intermediate morphology. In a similar fashion as chimpanzees, the three fossils are located at intermediate positions in PC2 distribution. H. neanderthalensis and H. naledi display PC2 scores that are within the Pan IQR, whilst A. sediba has lower values.
Violin plots of PC3 (Fig. 2d) show a similar distribution of PC scores for the three extant genera. From a morphological perspective, higher scores are associated with more robust morphologies displaying more marked muscular attachments (for the opponens pollicis, first dorsal interosseous and abductor pollicis longus muscles), while lower values correspond to more gracile MC1s. H. naledi and A. sediba show values which are within the Pan or H. sapiens distribution, but outside their IQR and at opposite extremes of the axis. H. neanderthalensis lies outside the distribution of any of the extant genera, probably due to its particularly robust morphology and associated marked muscular insertion areas.
3.2 Morphological disparity
To compare the amounts of shape variation between the extant genera, we used Procrustes variance as a way to assess intra-genus variation. The obtained results show that three extant genera show a similar magnitude of disparity. Nevertheless, gorillas exhibit a higher Procrustes variance as compared to modern humans and chimpanzees (Table 1a). Gorillas are significantly different to modern humans, and chimpanzees when comparing absolute variance differences, whilst modern human do not significantly differ from chimpanzees (Table 1b).
3.3 Linear discriminant analysis
The LDA model using the first 22 PCs clearly distinguishes between the three extant genera, displaying an outstanding performance with almost perfect classification results after cross-validation (Accuracy: 0.98; Cohen’s Kappa: 0.97; Figure 3). When using the obtained discriminant function to classify the fossils into the extant categories (as way of assessing morphological affinities) (Table 2), the three of them were robustly classified into the Homo category (all posterior probabilities were extremely close to 1), hence indicating that, in spite of their differences, their morphology is closer to that of modern humans. There were significant differences between all extant genera when analyzing 22 PCs from the PCA carried out using the shape variables (Table 3).
4 Discussion
The first hypothesis was that the shape of the human MC1 would differ significantly from that of Pan and Gorilla, due to the variation in their manipulative capabilities and locomotive behaviors. Results from the analyses provide strong support for this hypothesis, confirming that there is indeed significant morphological variation between the extant great apes. Interestingly, we also found clear differences between chimpanzees and gorillas, with gorillas closer (i.e., more similar) to humans than to chimpanzees (PC1). The second hypothesis was that all fossil hominin species would exhibit an MC1 morphology more similar to humans than other great apes. The results also support this hypothesis. However, it is important to notice that even though the three fossils are more similar to the modern humans, they also display some distinct features, different from those which would be typically expected in modern H. sapiens.
5.1 Humans and great apes MC1 shape
The 3DGM data indicate that modern human MC1 morphology is significantly different from the rest of the extant hominids, therefore allowing us to accept the first hypothesis. The human distal head is characterized by a flatter, larger distal articular surface and larger radial and ulnar epicondyles. The proximal base of the human MC1 is also larger and flatter in both the radioulnar and dorsovolar aspects, with less pronounced curvature than that seen in other hominid species. These are all morphologies that are consistent with previous 3DGM analysis of the proximal (Marchi et al., 2017) and distal (Galletta et al. 2019) surfaces of the human MC1. The shaft, an area that has not previously been analyzed using 3DGM, is characterized by being significantly more robust, with a greater curvature and a larger ridge on its lateral side, corresponding to the insertion of the opponens pollicis muscle.
The flatter and larger distal articular surface in humans has been interpreted as an adaptation that limits dorso-palmar motion whilst preventing radioulnar motion (Barmakian, 1992), thereby stabilizing the MC1 and facilitating forceful power and precision grasping. In apes that distal articular surface has a more pronounced curvature, rendering the metacarpophalangeal joint (MCPJ) less stable and unable to sustain high loads (Galletta et al., 2019). The pronounced radial and ulnar epicondyles found at the distal head of the human MC1 (as described by PC1) serve a similar purpose, reducing the range of motion and stabilizing the MCPJ. These epicondyles act as anchor points for collateral ligaments, which insert at the base of the proximal phalanx. When the thumb is flexed these ligaments tighten and limit the radioulnar motion of the proximal phalanx (Imaeda et al., 1992). Larger epicondyles are therefore thought to act as stronger anchors by providing a greater area for the collateral ligaments to attach to, helping stabilize the MCPJ during the high forces that are experienced by the thumb during manipulation (Galletta et al., 2019). The proximal articular surface in humans is also flatter (as described by PC2), but in contrast this is correlated with a higher range of motion at the trapeziometacarpal joint (TMCJ), rather than a lower one (Marzke et al., 2010). It is this combination in humans of high mobility at the TMCJ and low mobility at the MCPJ that facilitates a high level of manual dexterity, whilst also allowing the thumb to sustain high loads during forceful tool use. High mobility at the TMCJ plays a key role in the pad-to-pad opposition abilities of the human hand, in which the thumb is able to rotate and touch the apical tip of each phalanx. In many human manipulative activities like precision grips the thumb needs to be highly abducted, which means that the load is radially shifted on the joint surface (Lewis, 1977; Marchi et al., 2017). The observed larger radially extended proximal surface is therefore important because, whilst it allows for a greater radial extension, it also helps the joint resist high levels of radial displacement by providing a greater surface area for the abducted MC1 (Hamrick, 1996).
We hypothesize that the morphological characteristics of the human MC1 shaft presented here, such as a significantly more robust build, are likely adaptations that further serve to facilitate forceful tool use. Indeed, a thicker MC1 shaft would be able to better withstand the high levels of stress placed upon the thumb by sustained power and precision grasping (Key & Dunmore, 2015; Marzke, Wullstein, & Viegas, 1992; Rolian, Lieberman, & Zermeno, 2011). It has been also related to a greater development of the thenar musculature attaching into the shaft that is highly active during hard hammer percussion and that would favor thumb opposition (Marzke, 2013; Marzke, Toth, Schick, & Reece, 1998).
5.1 Fossil hominin MC1 shape
The general scientific consensus in recent years is that H. neanderthalensis had a hand morphology and manipulative capabilities that were very similar to those of humans, challenging the previously held beliefs that H. neanderthalensis lacked the derived adaptations for advanced and precise human-like tool use (Karakostis et al., 2017; Karakostis, Hotz, Tourloukis, & Harvati, 2018; Niewoehner, 2001, 2006; Tocheri et al., 2008; Trinkaus & Villemeur, 1991). The results align well with this consensus, with the H. neanderthalensis specimen showing several similarities with the modern humans. The described morphology is one of a flatter (PC2) and larger (PC1) distal articular surface, bigger epicondyles at the distal head (PC1) and a flatter proximal articular surface (PC2). However, H. neanderthalensis also differs in exhibiting a particularly robust MC1 with strongly marked muscular insertions.
Previous analysis on the thumb morphology of Homo naledi fossils has indicated that it has derived characteristics compatible with forceful precision grip and human-like manipulative abilities (Berger et al., 2015; Kivell, 2015). Such characteristics include a well-developed crest for the opponens pollicis insertion and flat distal/proximal articular surfaces (Kivell et al., 2015; Galletta et al. 2019). The results generally agree with these observations and conclusions, whilst also presenting some potentially new insights. The morphology of the H. naledi sample had a human-like robustness of the shaft (PC1), suggesting that the MC1 was adapted to sustain high loads, such as those experienced during forceful tool use. This suggests that H. naledi was potentially capable of a degree of advanced manipulation, such as forceful precision and power grasping. These findings are therefore consistent with previous functional interpretations of H. naledi thumb morphology (Galletta et al. 2019). Whilst the evidence suggests that H. naledi was almost certainly able to perform an certain degree of advanced manipulation, and was likely a tool-user, it also suggests that it had not yet showed the full repertoire of manipulative adaptations exhibited by humans and H. neanderthalensis (Berger et al., 2015; Kivell et al., 2015; Galletta et al., 2019).
Previous analysis of A. sediba hand morphology has found that it possessed a number of advanced Homo-like features, such as a longer thumb relative to shorter fingers, that potentially indicate advanced manipulative capabilities, while retaining primitive traits as a gracile MC1, similar to those of other australopithecines (Kivell et al., 2011). Recent 3DGM studies that have analyzed the MC1 in particular have come to the conclusion that if A. sediba was indeed utilizing tools, as some hand proportion and trabecular evidence suggests (Kivell et al., 2011; Skinner et al., 2015), then it was doing so in a way that differed from that of early Homo and modern humans (Marchi et al., 2017; Galletta et al., 2019). This conclusion was reached due to aspects of their MC1 morphology that were deemed inconsistent with the ability to employ forceful precision grips, namely a gracile MC1 shaft, more curved proximal articular surface and smaller radial and ulnar epicondyles. These morphologies suggest that the range of motion would not have been great enough at the TMCJ to facilitate the necessary abduction-adduction for thumb opposition and pad-pad precision grips. The results agree with this consensus, with the A. sediba sample presenting a gracile shaft (PC1), and smaller epicondyles at the distal head. These morphologies suggest that the A. sediba MC1 did not have the strength or stability to withstand the forces involved with precision grips, nor the range of motion at the TMCJ to facilitate them. Overall the results therefore align with previous research, in the sense that they present A. sediba as having a patchwork of primitive and derived characteristics, a few of which are indicators of an ability to use tools, but most of which suggest that this ability was incipient and certainly not comparable to the forceful precision grip abilities of humans and H. neanderthalensis.
6. Conclusion
The aim of this study was to quantify the 3D morphology of the first metacarpal in extant African hominoids, in order to facilitate a more informed functional interpretation of fossil hominin morphology. The results are in general agreement with previous studies on the morphology of the MC1 in extant and extinct hominids and the inferences made by them. This study found that the human MC1 is characterized by a distinct suite of traits, not present to the same extent in non-human great apes, that are consistent with an ability to use forceful precision and power grips; namely flatter proximal and distal ends, larger epicondyles at the distal head and a more robust shaft. This morphology was also found to align very closely with that of the H. neanderthalensis sample, supporting all the evidence that indicates that Neanderthals were functionally capable of utilizing tools in the same way as modern humans. Analysis of the H. naledi specimen suggested that it had a number of human-like adaptations consistent with an ability to employ advanced manipulation and was therefore likely able to use stone tools in a similar way to humans. The A. sediba fossil presented a number of derived MC1 features that indicate a degree of dexterity, but also several traits which were more similar to the African apes (i.e., probably primitive traits). Overall the results obtained both aligned with and added to past functional interpretations of hominin morphology, thereby reinforcing the validity of 3DGM as a method of quantifying MC1 morphology and providing a deeper insight into the function and structure of the thumb in both extant hominids and fossil hominins.
Author contributions
TAP and AB designed the study. AB and CL generated the 3D models. JM and TAP analyzed the data. TAP carried out the data visualizations. All authors interpreted the data and wrote the manuscript.
Acknowledgments
We are grateful to the following curators and institutions for the access of the ape specimens: Emmanuel Gilissen (AfricaMuseum), Anneke H. van Heteren and Michael Hiermeier (Zoologische Staatssammlung München), Javier Quesada (Museu de Ciències Naturals de Barcelona), José Miguel Carretero (Universidad de Burgos). AB was partially funded by a Becas Chile scholarship, whilst TP was funded by the Leverhulme Trust Early Career Fellowship, ECF-2018-264. This study was funded by the research projects AGAUR 2017 SGR 1040 and MICINN-FEDER PGC2018-093925-B-C32.