First metatarsal trabecular bone structure in extant hominoids and Swartkrans hominins

Abstract

Changes in first metatarsal (MT1) morphology within the hominin clade are crucial for reconstructing the evolution of a forefoot adapted for human-like gait. Studies of the external morphology of the MT1 in humans, non-human apes, and fossil hominins have documented changes in its robusticity, epiphyseal shape and its articulation with the medial cuneiform. Here, we test whether trabecular structure in the MT1 reflects different loading patterns in the forefoot across extant large apes and humans, and within this comparative context, infer locomotor behavior in two fossil hominins from Swartkrans, South Africa. Microtomographic scans were collected from the MT1 of Pongo sp. (n = 6), Gorilla gorilla (n = 10), Pan troglodytes (n = 10), Homo sapiens (n = 11), as well as SKX 5017 (Paranthropus robustus), and SK 1813 (Hominin gen. sp. indet.). Trabecular structure was quantified within the head and base using a ‘whole-epiphysis’ approach with medtool 4.2. We found that modern humans displayed relatively higher bone volume fraction (BV/TV) in the dorsal region of each epiphysis and a higher overall degree of anisotropy (DA), whereas great apes showed higher BV/TV in the plantar regions, reflecting dorsiflexion at the metatarsophalangeal (MTP) joint in the former and plantarflexion in the latter. Both fossils displayed low DA, with SKX 5017 showing a hyper-dorsal concentration of trabecular bone in the head (similar to humans), while SK 1813 showed a more central trabecular distribution not seen in either humans or non-human apes. Additionally, we found differences between non-human apes, modern humans, and the fossil taxa in trabecular spacing (Tb.Sp.), number (Tb.N.), and thickness (Tb.th.). While low DA in both fossils suggests increased mobility of the MT1, differences in their trabecular distributions could indicate variable locomotion in these Pleistocene hominins (recognizing that the juvenile status of SK 1813 is a potential confounding factor). In particular, evidence for consistent loading in hyper-dorsiflexion in SKX 5017 would suggest locomotor behaviors beyond human-like toe off during terrestrial locomotion.

Introduction

One of the central questions within the study of human evolution is how and when obligate bipedalism emerged. The forefoot is of particular importance in addressing this question because it directly reflects the extent to which a species uses its feet for locomotion (either arboreal or terrestrial) and/or manipulation. The first metatarsal (MT1) has undergone a dramatic transformation, from a digit used primarily for grasping, to a digit used mainly for weight-bearing, stabilization, and propulsion in modern humans (Morton, 1922, Elftman and Manter, 1935, Susman, 1983, Harcourt-Smith and Aiello, 2004). Analysis of the partially preserved OH 8 foot formed a critical aspect of the initial diagnosis of bipedalism in Homo habilis (Day and Napier, 1965, Kidd et al., 1996) and a number of studies have incorporated analyses of forefoot bones to argue for committed terrestrial bipedalism in Australopithecus afarensis (Latimer and Lovejoy, 1990; Ward, 2002, Ward et al., 2011) and an opposable hallux in Ardipithecus ramidus (Lovejoy, 2009, White et al., 2015). Equally intriguing is the recent discovery of the Burtele foot, which is similar in age to A. afarensis but displays a number of characteristics that differentiate it morphologically and suggest two different types of bipedal foot loading in the hominin clade at the same time (Haile-Selassie et al., 2012). This variation in hominin foot bone morphology highlights the importance of understanding the form/function relationship of the MT1 in extant and fossil hominoids and, in particular, whether internal bone structure can provide insights into biomechanical loads experienced by the foot during different types of locomotion. Using a comparative sample of modern humans and non-human apes, this study will address whether trabecular structure within the MT1 is reflective of locomotor mode. Furthermore, we will compare them to fossil hominins from Swartkrans (SKX 5017 and SK 1813) to test hypotheses about hominin locomotion in the Plio-Pleistocene of South Africa.

Functional interpretations of fossil hominin locomotion largely vary because of a lack of consensus on the functional significance of various external skeletal features. It remains unclear whether ‘primitive’ features represent non-functional evolutionary vestiges, or if they represent functional indicators of locomotor behavior (Stern and Susman, 1983, Clarke and Tobias, 1995, Ward, 2002, Harcourt-Smith and Aiello, 2004, Zipfel et al., 2009). This issue can be partially addressed by studying aspects of bone that are more responsive to external loading. While articular surfaces indicate the joint positions an element is capable of, internal bone is more likely to show the position in which the element was actually loaded (Ruff and Runestad, 1992, Rafferty and Ruff, 1994, Jacobs, 2000, Rubin et al., 2002, Ruff et al., 2006). Diaphyseal cortical bone has been shown to respond to mechanical stress in the shaft and can be indicative of predominant bending forces experienced during loading (Ruff, 1983, Cowin et al., 1985, Doden, 1993, Carlson, 2005, Ruff et al., 2006). However, its function is likely different over joint articular surfaces, where it becomes significantly thinner. It is also covered by cartilage and often contained within a synovial joint. Conversely, the trabecular bone located subchondrally within epiphyses remodels at a faster rate than cortical bone (Eriksen, 2010), and can provide evidence of in vivo loading that may be more useful at reconstructing predominant joint position and associated behaviors (Hodgskinson and Currey, 1990, Rubin et al., 2002, Mittra et al., 2005, Pontzer et al., 2006, Barak et al., 2011; but see; Bertram and Swartz, 1991). However, it should be noted that trabecular bone structure does not always correlate with known locomotor patterns in certain mammals, including mice and several primates (Carlson et al., 2008, Ryan and Walker, 2010, Shaw and Ryan, 2012).

The current study focuses on two main structural properties of trabecular bone: bone volume fraction (BV/TV), which is a measure of trabecular thickness, number, and spacing, and degree of anisotropy (DA), which reflects the degree to which trabecular struts are oriented in the same direction. These parameters account for 87–89% of the variance in the strength of a bone (Young's modulus) (Maquer et al., 2015), have been shown to change in relation to magnitude, frequency, and direction of load in in vivo studies (Lanyon, 1974, Hodgskinson and Currey, 1990, Biewener et al., 1996, Mittra et al., 2005, Pontzer et al., 2006, Barak et al., 2011), and to differ among taxa that employ different modes of locomotion (MacLatchy and Muller, 2002, Ryan and Ketcham, 2002, Ryan and Ketcham, 2005, Ryan and Shaw, 2012, Scherf et al., 2013, Tsegai et al., 2013, Tsegai et al., 2017; but see; Fajardo et al., 2007, Ryan and Walker, 2010). BV/TV and DA are informative parameters because both are less likely to scale allometrically and have been found to respond to loading in predictable ways (Barak et al., 2013). BV/TV is generally higher in areas that experience greater compressive loading, and trabecular orientation adapts to the main axis of joint movement (Biewener et al., 1996, Guldberg et al., 1997, Ryan and Ketcham, 2002, Mittra et al., 2005, Pontzer et al., 2006, Chang et al., 2008, Polk et al., 2008, Harrison et al., 2011, Saparin et al., 2011). Responses in DA and BV/TV to biomechanical stressors have been demonstrated in several classic studies on the mammalian calcaneus (Lanyon, 1973, Lanyon, 1974, Skerry and Lanyon, 1995, Biewener et al., 1996, Skedros et al., 2004, Skedros et al., 2012; Sinclar et al., 2013). In animals in which the calcaneus does not touch the ground during locomotion, trabeculae underlying the Achilles tendon were aligned with the compressive and tensile principal direction of stress (Lanyon, 1974, Biewener et al., 1996). When external loading was removed by detaching the calcaneal tendon, BV/TV reduced as a result of lower trabecular thickness and number (Biewener et al., 1996). Further in vivo studies have supported this. Barak et al. (2011) showed that DA and BV/TV varied predictably in the distal tibiae of sheep that loaded their ankles in different positions. Pontzer et al. (2006) also found strong correlations of DA with changes in external loading at the distal femur of guinea fowl.

However, there are several non-mechanical factors that may affect trabecular structure. It is not clear how genetic, hormonal, and environmental factors constrain its structure (Simkin et al., 1987; Judex et al., 2004, Havill et al., 2010, Devlin, 2011, Devlin and Bouxsein, 2012, Devlin et al., 2013), how its response varies based on frequency versus magnitude of mechanical loading (Skerry and Lanyon, 1995, Lambers et al., 2013), as well as anatomical region (Räth et al., 2013, Wallace et al., 2013). Furthermore, by measuring trabecular bone density throughout 9 anatomical regions in humans, Chirchir (2016) found that most sites have homogenous values, suggesting they are influenced by site-specific genetic factors. Nonetheless, computational models (Odgaard, 1997, Huiskes et al., 2000, Fox and Keaveny, 2001) and in vivo studies (Lanyon, 1974, Biewener et al., 1996, Pontzer et al., 2006, Barak et al., 2011) have demonstrated strong links between trabecular structure and the frequency, magnitude, and direction in which a joint is loaded.

Modern humans are adapted for a bipedal mode of locomotion and possess a forefoot structure in which each metatarsophalangeal (MTP) joint acts as a weight bearing and propulsive structure during the push-off part of the stance phase (Stokes et al., 1979, Christensen and Jennings, 2009, Griffin et al., 2015). During this phase, the MTP joints dorsiflex, moving the proximal phalanges on to the dorsum of their respective metatarsal heads. This causes tightening of the plantar aponeurosis, stabilizing the foot and elevating the longitudinal arch, which changes its conformation to a stiff lever for propulsion and ultimately toe-off (Hicks, 1954, Bøjsen-Moller, 1979, Susman, 1983, Caravaggi et al., 2010, Griffin et al., 2015). As shown by in vivo studies of plantar pressure distribution within the human foot, during dorsiflexion the medial forefoot shows a spike in loading (Hutton and Dhanendran, 1981, Katoh et al., 1983, Soames, 1985; Munro et al., 1987, Lee and Farley, 1998, Hunt et al., 2001, Nester et al., 2007, Griffin et al., 2010a). The MT1 bears a large portion of this load and this is reflected by its large head, which experiences high compressive forces during push-off (Rodgers, 1995, Donahue and Sharkey, 1999, Vereecke et al., 2003, D’Août et al., 2004). Its external shape is also designed to stabilize the MTP joint and facilitate dorsiflexion during push-off. The superior aspect of the articular surface of the head expands to the dorsum of the bone, resulting in a raised appearance in relation to its shaft, which is thought to increase the range of dorsiflexion at the MTP joint (Stokes et al., 1979, Susman and Brain, 1988, Susman and de Ruiter, 2004, Griffin and Richmond, 2005, Griffin et al., 2010a). It is also medio-laterally wider on the dorsal aspect of the head than the plantar aspect, which has been argued to enhance joint stability during push-off and facilitate close-packing of the MTP joint (Susman and Brain, 1988, Hetherington et al., 1989, Susman and de Ruiter, 2004, Pontzer et al., 2010, Fernández et al., 2015).

Analyzing foot kinematics in extant non-human apes is less straight-forward compared to modern humans because they employ a wider range of locomotion, from terrestrial to arboreal quadrupedalism, vertical climbing, suspension, and occasional terrestrial bipedalism. However, in vivo studies of chimpanzee and bonobo footfall patterns show considerable differences from modern humans (Aerts et al., 2000, Vereecke et al., 2003, D’Août et al., 2004, Griffin et al., 2010a, Wunderlich and Ischinger, 2017). During terrestrial quadrupedal locomotion, bonobos show a higher spike in loading on the lateral aspect of the foot during push-off, with relatively little force inflicted upon the MT1 (Vereecke et al., 2003). Additionally, during vertical climbing, chimpanzees show peak loading under the medial metatarsals, when the MTP joints are plantarflexed (Wunderlich and Ischinger, 2017). If these plantar pressure patterns can be broadly applied to non-human apes, they would suggest that the first MTP joint incurs maximum loading when it is used for grasping.

This is reflected within the shape of the MT1 head, which in all non-human apes is mediolaterally expanded on the plantar aspect (Susman, 1983, Latimer and Lovejoy, 1990, Marchi, 2005, Marchi, 2010, Griffin and Richmond, 2010, Fernández et al., 2015). The same mechanism that allows for close-packing of the MTP joint during dorsiflexion in humans allows for close-packing during plantarflexion in non-human apes, increasing stability during pedal grasping (Susman, 1983, Griffin et al., 2010a). It should be noted that within non-human apes, there is variation in how the hallux is used for locomotion. Although comparative plantar pressure distribution data for Gorilla and Pongo do not exist, a substantial amount of information can be obtained through observational studies (Tuttle and Beck, 1972, Cant, 1987, Sarmiento, 1994, Remis, 1995, Gebo, 1996, Doran, 1997, Sarringhaus et al., 2014), skeletal (Shea, 1981, Inouye, 1992, Doran, 1993, Marchi, 2005, Richmond, 2007, Congdon, 2012, Drapeau and Harmon, 2013, Jashashvili et al., 2015), and soft tissue morphological analyses (Oishi et al., 2012). Such studies show different habitual positioning of the hallux; orangutans generally do not apply significant force on the hallux during suspension, whereas chimpanzees generally do (Oishi et al., 2012). Gorillas do not use their feet for suspension; when they locomote arboreally, their size restricts them to substrates of larger diameters. Their feet are used for vertical climbing or walking, but because the supports they use for climbing are usually large relative to their foot size, there is little flexion of the metatarsophalangeal and interphalangeal joints (Sarmiento, 1994, Remis, 1995).

The proximal articular surface of the MT1 is equally reflective of locomotor behavior. Within modern humans, it is relatively broad and flat, corresponding to a stable tarsometatarsal (TMT) joint complex that reduces mediolateral mobility of the hallux, and keeps it in line with the other metatarsals (Morton, 1924, Susman and Brain, 1988, Proctor et al., 2008, Proctor, 2010, Gill et al., 2015). Its broad mediolateral width is related to the bending stresses experienced near the base of the MT1, and its increased cross-sectional area is a response to high compressive forces at the joint and high tensile forces inflicted upon the ligaments (Stokes et al., 1979, Griffin and Richmond, 2005). In all other living great apes, the proximal MT1 does not experience such high loading, resulting in a proximal surface that has a smaller overall area. The TMT joint is instead adapted for a wider range of movement associated with grasping and varied locomotion. The proximal articular surface of the MT1 is concave, and the distal articular surface of the medial cuneiform is convex, allowing for multiaxial movement of the hallux that is more effective for climbing and grasping (Latimer and Lovejoy, 1990, McHenry and Jones, 2006, Tocheri et al., 2011).

Here, we focus specifically on Plio-Pleistocene fossil feet from South Africa, which show a diverse range of morphological features. StW 573, attributed to Australopithecus prometheus (but see Berger and Hawks, 2019), shows evidence of a slightly divergent hallux and has been interpreted as adept at tree-climbing (Clarke and Tobias, 1995, Clarke, 2013) but see (Harcourt-Smith et al., 2002). Two isolated MT1s from Sterkfontein (StW 562 and StW 595) suggest there was locomotor diversity in South African hominins. The two metatarsals are of unknown taxonomic status but were both found in Member 4, dated between 2.6 and 2.0 mya (Pickering and Kramers, 2010), and show striking differences in external morphology. StW 562 is described as more human-like based on its distal epiphysis which shows dorsal doming of the head, and because it is relatively robust. StW 595 is relatively gracile and does not show this epiphyseal feature, suggesting this individual had a more ape-like push-off mechanism (Clarke, 2013, DeSilva et al., 2019). Of particular relevance to this study are the postcranial remains of South African robust australopiths attributed to Paranthropus robustus. P. robustus is mainly represented by cranial remains (Grine, 1993, Grine and Daegling, 1993, Wood and Constantino, 2007); postcranial remains are relatively abundant, but cannot be attributed with complete confidence to the taxon. Based on pelvic, femoral, and tarsal morphology, the gait of P. robustus has been described as bipedal, but with a ‘waddling gait’ and a less efficient body weight transfer mechanism (Napier, 1964, Day and Napier, 1965, Robinson, 1972, Gebo and Schwartz, 2006). However, trabecular properties of the talus (Su and Carlson, 2017) and diaphyseal cortical bone morphology of the fifth metatarsal (Dowdeswell et al., 2017) suggest a medial weight transfer of the foot during push-off and loading of the lateral column in a human-like way.

Two complete metatarsals from Swartkrans contribute to our understanding of Early Pleistocene hominin locomotion. SKX 5017 is an isolated left MT1 recovered from the Lower Bank deposit of Swartkrans Member 1, dated to approximately 1.5–1.8 mya (Susman and Brain, 1988, Susman and de Ruiter, 2004). Along with other fossils found within this level, SKX 5017 is attributed to P. robustus. The specimen is described as short and similar in length to OH 8 and female bonobos (Susman and Brain, 1988). The base of the metatarsal has a mildly concave and ovoid shape, similar to modern great apes (Susman and Brain, 1988), although the morphology of the base and proximal shaft provide evidence that human-like plantar ligaments (and perhaps an aponeurosis) were present. Based on this basal articular morphology, and on the degree of torsion between the head and the base, there is no indication that the hallux was abducted to an ape-like degree (Susman and Brain, 1988). The head displays a mosaic of primitive and derived features. The superior articular surface of the head extends onto the dorsum of the shaft, which is an indication of increased dorsiflexion at the MTP joint. In contrast, the dorsal medio-lateral breadth of the head is narrower than the plantar breadth, suggesting the joint did not close-pack in dorsiflexion, and thus was less stable during push-off (Susman and Brain, 1988).

SK 1813 is a nearly complete right MT1 found in a backfill hole of Swartkrans and is thought to have come from Member 1 or 2, but attribution to a specific stratigraphic unit or taxon cannot be made with certainty (Susman and de Ruiter, 2004). Presence of an epiphyseal line near the base signals the subadult status of this individual, with an estimated age of approximately 15 years (Susman and de Ruiter, 2004). It bears strong morphological affinities to SKX 5017, albeit the former is smaller. It has the same dorsal mediolateral narrowing on the head, and expansion of the dorsal articular surface onto the dorsum of the shaft. The base is also dorsoplantarly expanded, which is reflective of increased tensile forces from well-developed plantar ligaments in response to a bipedal gait (Susman and de Ruiter, 2004, Proctor et al., 2008, Proctor, 2010). The shape of the proximal articular surface is difficult to discern due to post-mortem damage, but is nonetheless described as concave and ovoid, typical of non-human apes (Susman and de Ruiter, 2004). In this study, we examine trabecular structure within the epiphyses of these two specimens to reconstruct aspects of their biomechanical loading regime and in doing so make inferences about the locomotor behaviors of the individuals they represent.

Predictions about how mechanical loading affects the trabecular bone within the MT1 can be made a priori based on currently-known patterns within the MT1 head of Homo sapiens, Pan troglodytes, Pan paniscus, and Gorilla gorilla (Griffin et al., 2010b). Volume of interest (VOI)-based analysis has shown that modern humans exhibit significantly higher DA values than non-human apes, consistent with a tightly constrained joint with a relatively uniaxial range of movement. The same study has shown BV/TV to be less effective at differentiating locomotor behavior between species, but this may be caused by the methodological limitations of using VOIs in analyzing trabecular structure. Overall, these results suggest that among trabecular bone properties, a high degree of anisotropy is the most indicative factor of a forefoot habitually used for propulsion during bipedal gait (Griffin et al., 2010b).

Based on known loading patterns within the forefoot of modern humans and great apes, and the mechanical adaptations of trabecular bone, we test the following hypotheses:

• Modern humans will have a higher BV/TV within the dorsal aspect of the MT1 head and base, and non-human apes will display the opposite pattern. This corresponds to the position in which the joints close-pack and incur the highest compressive load.

• Modern humans will show higher overall DA and non-human apes will show lower overall DA within the entire element, corresponding to the range of motion at the TMT joint and the MTP joint. We also predict that modern humans will show relatively higher DA in the dorsal regions of the epiphyses compared to the non-human apes.

• Based on their external morphology, we hypothesize that SKX 5017 and SK 1813 will show similar trabecular distribution to modern humans. However, because of a more concave and rounded proximal articular facet in SKX 5017 indicative of a relatively mobile joint, its base will have a lower DA than modern humans.

To ensure no confounding factors relating to body size, we also test for interspecific allometry in trabecular bone variables. Following the results of previous studies on the talus and tibia of modern humans and chimpanzees (Tsegai et al., 2017), and on the lower limbs of modern humans (Saers et al., 2016), we predict that there will be no significant allometric scaling between any trabecular parameter and bone size. Furthermore, because Tsegai et al. (2017) found that the tibia and talus do not show the same patterns of scaling in regards to Tb.Th., BV/TV, or DA, we expect that the MT1 will also display allometric patterns not displayed in either element.