Evolutionary developmental origins of the hominin foot and of medial cuneiform bipartition
Elements of the autopodium first appear in Devonian polydactylous tetrapods with six to eight digits (Coates & Clack, 1990). The pattern of basic tarsal units has undergone various evolutionary variations, including fusion or loss of elements and evolution of entirely new elements (Schaeffer, 1941). Because of the lack of evidence for successive fossil anatomy, it is difficult to infer homology within structures (Lewis, 1989). Differential fusion of the tarsal elements produces a wide spectrum of evolutionary developmental variation. In some species, separate cartilaginous precursors, which join during development to form one bone, bear evidence of phylogenetic history, whereas in other species, tarsal elements are highly derived and do not permit phylogenetic inferences (Lewis, 1989). Figure 6 shows the hypothetical basic tarsal units represented in human foot primordia; these include the talus (homologous to os tibiale intermedium and os tibiale centrale proximale), calcaneus (os fibulare and os pisiforme), navicular (tibiale centrale distale and fibulare centrale distale), tuberosity of the navicular (os tibiale externum tarsi), medial cuneiform (os tarsale distale I and distal end of the prehallux primordium), intermediate cuneiform (os tarsale distale II), lateral cuneiform (os tarsale distale III) and cuboid (os tarsale distale IV) (Čihák, 1972; Berman & Henrici, 2003).
Figure 6. Homology of tarsal primordia in human. T, tibia; F, fibula; pi, pisiforme; i, intermedium; t, tibiale; f, fibulare; taph, tarsalia praehallucis; c1–3, centralia; ta I, tarsale distale I; ta II, tarsale distale II; ta III, tarsale distale III; ta IV, tarsale distale IV [redrawn after Čihák (1972)].
Download figure to PowerPoint
Figure 7. (A) Primordia of the bipartite dorsal (cfId) and plantar (cfIp) medial, and intermedial (cfII) and lateral (cfIII) cuneiform in a transverse section of the foot of a 19-mm embryo [from Čihák (1972). With kind permission of Springer Science+Business Media]. (B) Two ossification centers of a medial cuneiform. Copyright © 2010 by the American Orthopaedic Foot and Ankle Society, Inc., originally published in Foot & Ankle International, O'Neal et al. 1995 and reproduced here with permission.
Download figure to PowerPoint
Figure 8. Hypothetical pathways of medial cuneiform development. (A) Mesenchymal pre-chondrogenic condensation. (B) Interzone formation from aggregation of mesenchymal cells: b1, single mesenchymal primordium; b2, divided mesenchymal primordia (plantar and dorsal parts). (C) Separation of cartilaginous anlagen by cavitation and formation of synovial cavities: c1, single anlage; c2, two anlagen. (D) Ossification: d1, from one center; d2, from two centers. (E) Adult stage of bone development: e1, normal medial cuneiform; e2, bipartite medial cuneiform.
Download figure to PowerPoint
Genetic factors underlying variation in medial cuneiform developmental pathways might be identified at two different levels, namely Hox and Sox transcription factors, which specify the patterning and shape of the embryonic skeleton (Hall & Miyake, 2002; Cohen, 2006; Montero & Hurlé, 2007), or local regulators of mesenchymal condensation, which control the shape of the skeletal elements via autocrine/paracrine factors (Garciadiego-Cazares et al. 2004; Hentschel et al. 2004; Pacifici et al. 2005; Cohen, 2006; Newman & Bhat, 2007). In the early stages of development, formation of a divided mesenchymal primordium (Fig. 8b2) could result from altered cell-intrinsic patterning at the gene level and/or variation in activator/inhibitor positional signaling mediators responsible for local control of the mesenchymal primordium. In later stages of development, formation of two centers of ossification inside one single cartilaginous anlage (Fig. 8d2) could result from variation in activator/inhibitor positional signaling mediators responsible for local control of the cartilaginous anlage (Newman & Bhat, 2007). By the end of the adult stage of development, the outcome of both developmental scenarios would be a bipartite bone (Fig. 8e2), unless the separate bone precursors fuse producing a complete bone (Fig. 8e1).
Modern humans exhibit several variations in tarsal bone expression at low frequencies (Marti, 1947). These reflect homologous variants to the examples mentioned earlier: os trigonum (Zeichen et al. 1999; Chao, 2004; Mouhsine et al. 2004), os perineum (cuboideum secundarium, centrale 4) (Bloom, 1991), os sustentaculi (Bloom et al. 1986), os naviculare bipartitum (Cotta, 1961) and os vesalianum (Virchow, 1922; Lepore et al. 1990; Inoue et al. 1999; Boya et al. 2005). A supernumerary ossicle (‘prehallux’) situated at the disto-medial border of the medial cuneiform, or proximo-medial border of the first metatarsal, is common in Ceboidea and Hylobatidae. Such an ossicle also has been observed in pongids and Homo (Lewis, 1972; Wikander et al. 1986). The ‘prehallux’ may be a relict skeletal element of a pre-axial fin ray from the earliest phylogenetic stages of vertebrates (Lewis, 1972).
These non-random developmental variants provide the raw material on which selection has the potential to act during evolution (Alberch, 1983; Erlebacher et al. 1995). Morphological variants of the medial cuneiform also seem to result from a non-random pattern of developmental variation, i.e. different degrees of bipartition. Regarding the evolutionary developmental origins of bipartition, it may be speculated that the medial cuneiform is homologous to os tarsale distale I and the distal part of the prehallux primordium (Čihák, 1972; Berman & Henrici, 2003).
Structure and function
In our modern human comparative sample, there was no indication that the bipartite medial cuneiforms of the Sigtuna sample significantly exceeded normal patterns of cuneiform shape variation. Morphological differences between the Dmanisi bipartite medial cuneiform and modern human cuneiforms thus most likely reflect species-specific differences, irrespective of whether these bones display bipartition. The articulation between the two segments of the Dmanisi medial cuneiform has a smooth structure, similar to the structure of other articular surfaces on these bones. This pattern is also apparent in articular surfaces of modern human bipartite medial cuneiforms. It is thus probable that the intracuneiform joint of Dmanisi, as in modern humans, was covered by cartilage and a synovial membrane.
Overall, it appears that the bipartite condition in the medial cuneiform represents developmental variation that does not cause significant overall morphological differences. Presumably the lack of morphological differences also implies a lack of functional differences. This provides an interesting perspective on the relationship of developmental and adaptive/functional constraints. The fact that the taxon-specific (most likely functionally relevant) morphology of the medial cuneiform can be reached by different developmental pathways, some of which imply bipartition to various degrees, points toward higher-order, epigenetic constraints that canalize the development of midfoot morphology as a whole. This indicates morphogenetic homeostasis in the sense that foot ontogeny could be buffered against environmental noise as well as against developmental noise. Accordingly, it appears that the network of developmental pathways graphed in Fig. 8 not only gives rise to patterns of medial cuneiform variation but also provides the required developmental homeostasis, as developmental disturbance at any node or link in the network can be compensated by alternative pathways.
Fossil hominin medial cuneiforms and first metatarsals
In hominoids, the distal articular surface of the medial cuneiform is convex, wide and medially-oriented, such that the articulating hallux is medially divergent. In humans this surface is flat, narrow and anteriorly-facing, such that the hallux does not diverge medially (Lewis, 1980, 1989). The latter condition exists in all described fossil hominin medial cuneiforms, which indicates that none of them are likely to have had a divergent hallux (McHenry & Jones, 2006). The Dmanisi bipartite medial cuneiform parallels this pattern.
In modern humans, morphology of the proximal joint surface of the first metatarsal (i.e. the surface articulating with the cuneiform) typically reflects the distal joint surface of the medial cuneiform. It is flat in the case of a single, non-bipartite medial cuneiform and it displays two facets separated by a transverse ridge (‘8-shaped’ circumference) in the case of a bipartite medial cuneiform. The latter association is observed in the relatively convex/concave first tarsometatarsal joint of the Dmanisi D3442/D4111a/b bones.
The situation is more complex in other Plio-Pleistocene hominins (Harcourt-Smith & Aiello, 2004; DeSilva, 2008; Proctor et al. 2008). The first tarsometatarsal joint surface is flat and undivided in OH8 and in SKX5017/SKX31117 (both attributed to Homo). In the Stw573 foot (attributed to Australopithecus), the proximal metatarsal surface has two facets. There is evidence that the same is true for the distal surface of the medial cuneiform. In the A. afarensis first metatarsal (AL333-54), the proximal articular surface exhibits two facets, whereas the only partially preserved medial cuneiform from the same stratigraphic location (AL333-28) bears no evidence of bipartition. However, not all first metatarsals attributed to Australopithecus reportedly exhibit a double facet (Day & Napier, 1964; Susman, 1989).
If we assume that a double-faceted proximal metatarsal joint surface is indicative of bipartition of the medial cuneiform in any of the three categories, it appears that this condition was relatively frequent in Plio-Pleistocene hominins compared with modern human populations. The significance of a potentially higher incidence remains to be clarified. It could reflect increased developmental variation during evolutionary diversification of the hominin foot but it could also represent a sampling artifact.