Little Foot and big thoughts—a re-evaluation of the Stw573

R. Kidd,
, C. Oxnard
School of Science Food and Horticulture, University of Western Sydney, Campbelltown, NSW
2560, Australia.
School of Anatomy and Human Biology, University of Western Australia, Stirling Highway,
Crawley, WA 6009, Australia.

Abstract

The part of the fossil assemblage Stw573 consisting of some medial foot bones was initially reported by Clarke & Tobias (Science 269 (2002) 521). They found it to have both ape- and human-like qualities, being human-like proximally and ape-like distally. We have undertaken a re-examination of this pedal assemblage using a multivariate analysis; while we also found ape- and human-like qualities, they are in direct conflict with the original findings of Clarke and Tobias. We report an essentially ape-like morphology proximally and a human-like morphology distally; the talus and navicular were found to be ape-like and the medial cuneiform human-like. We also undertook a morphometric analysis of the medial cuneiform from the fossil assemblage OH8, as this was not included in the original OH8 study of Kidd et
al (J Hum Evol 31 (1996) 269); this cuneiform was found to have a human-like morphology.Thus, the medial column findings from the two assemblages are very similar. This finding,coupled with the re-evaluation of the stratigraphy at Sterkfontein (Am J Phys Anthrop 119(2002) 192), suggests that the two may have been contemporaneous.

We also note that three broad patterns of modification have been identified, equating to proximal–distal lateral–medial (cranio–caudal) and dorsal–plantar(posterior–anterior). It has not escaped our notice that these patterns are each controlled by specific genes or growth factors; we thus see a morphometric expression of our developmental past.


Introduction

That the human foot evolved from some variety of primitive hominoid ancestral stock is not held in doubt. In doing so it has changed in many features, some large scale and obvious, others more subtle. Three large scale modifications stand apart. These may be summarised as follows: (1) a change in the size and proportions of the pedal skeletal segments with the tarsus becoming much greater in humans; (2) the presence of a divergent first ray and opposable first digit in all apes but not in humans; and (3) the presence of the related longitudinal and transverse arch structures in humans but their absence in all other hominoids. These major features, plus a myriad of minor ones distinguish the human foot from that of apes. There is little doubt that the human foot evolved in a mosaic manner, with certain critical features attaining a human form before others. To understand their sequence, a number of questions need to be answered. For instance, did the lateral pedal column, broadly describing the lateral longitudinal arch, become human-like before or after the medial longitudinal arch? Similarly, did the first toe achieve a state of apposition before or after arch formation? And at what stage did the segmental proportions of the evolving foot become human-like? Some clue may be found by scrutiny of the fossil record, though this is frustratingly sparse with no complete prehuman specimens being available. The most complete is that from East Africa, the OH8 foot from Olduvai, but even this is missing certain vital parts, namely portions of the calcaneus, the metatarsal heads and all digits.

In a previous study (Kidd et al 1994, 1996) the four hindmost tarsal elements of this foot were studied morphometrically and unequivocal evidence for mixed functional affinities was found, the medial side being ape-like and the lateral column(with due caveat for the incomplete calcaneus) being human-like. The medial
cuneiform was not reported. Clarke and Tobias (1995) also report on a fossil foot
assemblage, that known as Stw573, consisting of the talus, navicular, medial
cuneiform and first ray fragment from Sterkfontein Cave, Member 2 and attributed
to the species Australopithecus africanus. They also observed mixed affinities, both
human- and ape-like features, though in a manner differing from our findings with OH; they describe the talus to be essentially human-like, the navicular to be of mixed ape and human morphology, and the medial cuneiform to be largely ape-like. Their findings are clearly in conflict with those of the OH8 study, and they do not
seem to be reconcilable with either the OH8 findings of Kidd et al (1996), nor the
model of human pedal evolution predicted in part from the OH8 study (Kidd 1995,1998, 1999). The basis of the findings of Clarke and Tobias is not quantitative; they do not report a metrical analysis. It is appropriate, therefore, to undertake a morphometric analysis of the Stw573 fossil assemblage to establish the functional
affinities of the tarsal elements and to see if the quite different findings of the two assemblages may be reconciled. In addition, in order to present a complete and
unambiguous picture, the OH8 foot is revisited with an analysis of the medial cuneiform to complete the analyses presented in Kidd et al (1996).

Materials and methods

Materials
The materials used in the current study consist of the talus, navicular, medial cuneiform and first metatarsal from Stw573, the medial cuneiform (good quality cast) of OH8, and samples of four extant species, namely human (Homo sapiens), chimpanze(Pan troglodytes),gorilla (Gorilla gorilla), and orang-utan (Pongo pygmaeus).Females and males from each species were treated as groups in their own right, and as far as was possible, were equally represented. The samples used consisted of 20 males and females from each group with the exception of orang-utans in which only 17 females and 10 males were available. Chimpanzee and gorilla specimens were made available courtesy of the Powell-Cotton Museum, England and the orang-utan samples by the Smithsonian Museum, Washington DC. The human sample was made up from two groups in roughly equal proportion, namely British and Zulu. While it is recognised that there are definite, though subtle differences in bony morphology in different human subgroups, these are very small in comparison to those between humans as a whole, and apes. Thus, no attempt was made in this study to sample human variation.Bony shape changes associated with ontogeny, particularly in its early stages are well reported (e.g. Straus 1927; Lisowski 1967); thus specimens used in this study were restricted to nearly adult or adult specimens. Measurements were taken from elements of the left foot wherever a possible, previous statistical studies having shown there to be no significant difference between sides (Kidd 1995).

Methods
Problems in the design of variables

The choice of variable in a study such as this is of the utmost importance. In order to gain as much information as possible regarding functional affinities, it is crucial that the variables chosen represent features that are of biomechanical importance in relation to foot function. Most of the variables chosen may be defined as interlandmark distances, though two were angular and both from the talus.

The problem of homology
The comparison of comparable structures from different species may run into problems associated with analogy and homology. Obvious examples such as the wings of an insect and of a bird are clear-cut and do not cause confusion; compared structures are derived from entirely non-related anatomical elements. However, issues might arise with the comparison of the wings in birds and bats; they are both derived from the pentadactyl forelimb, though not from exactly the same parts. In the case of tarsal elements, achieving exact homology may be neither obvious nor possible. An example from the talus lies in the facet on the head for the plantar calcaneo-navicular ligament, clearly identifiable in humans but not so in apes. Such a feature must be accounted for in the definition of the dimension.

Definition of variables
Measurements obtained from the three bones were defined in the following
manner. Most of the measured dimensions have been defined in the Martin’s technique(Knussmann 1988), but below all are fully described to provide a uniform and comprehensive list.

The talus
The following standard references were defined after Lisowski et al (1974). The standard talar basal plane was defined as the position the talus assumes when it is resting on the tip of the posterior and lateral tubercles and the most inferior aspect of the head. The median sagittal talar plane is the sagittal plane which passes along the midline of the trochlea perpendicular to the standard talar basal plane. The coronal talar plane is the plane which passes through the tip of the lateral tubercle perpendicular to the former two planes. The median trochlear arc is the shape produced by the intersection of the median sagittal talar plane and the superior trochlear surface. The median talar neck plane is the plane which lies in the midline of the neck and is perpendicular to the talar basal plane.The trochlear-head plane is the plane which intersects with the most superior points of the trochlear margins and the talar head. This plane is obtained from the position assumed by the inverted talus. Using these standard planes and definitions the following linear or angular dimensions were obtained:

(1) The maximum medial height is the projected height from the standard basal talar
plane to the highest point on the medial margin of the trochlear facet.
(2) The maximum lateral height is the projected height from the standard talar basal
plane to the highest point on the lateral margin of the trochlear facet.
(3) The maximum median height of the talus is the projected height from the talar
basal plane to highest point on the median trochlear arc.
The above three measurements were obtained by resting the talus upon a glass
plate of known thickness, the measurement being from the underneath surface of the
glass to the required maximum height. The thickness of the glass was included in the
raw data collection to avoid confusion and subsequently subtracted.
(4) The transverse trochlear breadth is the distance between the medial and lateral
margins of the trochlear facet taken in the coronal talar plane.
(5) The anterior trochlear breadth is the maximum distance between the trochlear
margins parallel to the coronal plane.
(6) The posterior trochlear breadth is the minimum breadth of the trochlear margins
taken parallel to the coronal talar plane.
(7) The long dimension of the head is defined as the length of the long dimension of
the talo-navicular articulation of the head and is measured obliquely along its
long axis.
(8) The short dimension of the head is defined as the length of the short dimension of
the talo-navicular articulation of the head and is measured at right-angles to the
long dimension. This dimension includes the facet for the spring ligament where
identifiable.
(9) The maximum functional length is the measured distance length from the groove
for the tendon of muscle flexor hallucis longus posteriorly to the intersection of
the talar neck plane and the articular surface for the navicular.
(10) The maximum breadth is measured from the tip of lateral tubercle to medial
talar margin. The dimension is taken in the coronal talar plane.
(11) The trochlear chord is the length of the chord connecting the intersections of the
median trochlear arc and the anterior and posterior margins of the superior
trochlear facet.
(12) The medial facet length is the maximum distance between the anterior border
and posterior tip of the medial facet.
(13) The lateral facet length is defined as the maximum distance between the anterior
and posterior borders of the lateral facet at their intersection with the superior
trochlear surface.
(14) The maximum neck diameter is the diameter of the talar neck measured
obliquely, coinciding with the long axis of the head.
(15) The minimum neck diameter is the diameter of the talar neck measured at right
angles to above, coinciding with the short axis of the head.
(16) The talar neck-body angle is defined as the angle subtended by the intersection of
the median sagittal talar plane and the median talar neck plane.
(17) The talar head torsion angle is defined as the angle subtended by the long
bisection of the talar head and the trochlear-head plane. The smaller size of the navicular and medial cuneiform compared to the talus means that definition of the position of the bone relative to body planes is not practical. Variables are defined in terms of gross morphology of the bones as an alternative.

The navicular
The horizontal plane coincident with the long axis of the bone, bisecting the proximal and distal articular facets was considered to be analogous to the transverse plane. The corresponding planes perpendicular to this, bisecting the bone into anterior and posterior portions and medial and lateral portions were considered to be analogous to the coronal plane and sagittal plane respectively. The following dimensions were obtained with reference to the above assumed planar references.

(1) The long talar facet dimension is defined as the maximum dimension of the talar
facet.
(2) The short talar facet dimension is defined as the minimum dimension of the talar
facet.
(3) The long cuneiform facet dimension is defined as the maximum dimension of the
cuneiform measured in the transverse plane.
(4) The maximum short cuneiform dimension is defined as the maximum span of the
short cuneiform dimension and is measured in the sagittal plane.
(5) The minimum short cuneiform dimension is defined as the minimum span of the
short cuneiform dimension and is measured in the sagittal plane.
(6) The maximum navicular breadth is measured in the anatomical position.
(7) The maximum height is measured in the anatomical position.
(8) The sagittal plane thickness is the maximum dimension between the talar facet
posteriorly and the cuneiform facet anteriorly measured with respect to the
sagittal plane.
(9) The tuberosity projection is the maximum projection of the navicular tuberosity
medially beyond the margin of the talar facet.

The medial cuneiform
The markedly different morphology of the hominid and pongid bone, largely due to the facet for the first metatarsal, means that a comparable reference position is not possible for all species. In humans, the vertical plane coincident with the long axis of the bone, bisecting the proximal and distal articular facets was considered to be analogous to the sagittal plane. The corresponding planes perpendicular to this,bisecting the bone into superior and inferior portions, and medial and lateral
portions, were considered to be equivalent to the transverse plane and coronal plane
respectively. In the case of apes, the vertical plane was defined about the anterior
facet alone.

(1) The anterior facet height is the maximum height of the anterior articular facet
measured in the assumed sagittal plane.
(2) The anterior facet breadth is measured at right angles to (1) at the maximum
span.
(3) The posterior facet height is the maximum height of the posterior articular facet
measured in the assumed sagittal plane.
(4) The posterior facet breadth is measured at right angles to (3) at the maximum
span.
(5) The posterior height is defined as the maximum bone dimension at its posterior
aspect, measured in the sagittal plane.
(6) The posterior breadth is measured at right angles to (5) at the maximum span.
(7) The anterior plantar breadth is defined as the maximum dimension of the plantar
surface, measured in the transverse plane.
(8) The anterior dorsal breadth is as for (7) but for the dorsal surface.
(9) The anterior height is as for (5) but for the anterior aspect.
(10) The maximum dorsal length is measured dorsally, on the antero-lateral aspect
from the extremes of the posterior facet and the anterior facet.
(11) Maximum plantar length is measured plantarly, from the extremes anteriorly
and posteriorly.

The first metatarsal.

The Stw573 first metatarsal is fragmentary with only the base and a portion,perhaps a third of the shaft, being present. Thus, available dimensions for analysis are very limited. In fact, early exploratory analyses revealed quite clearly that no
discrimination was possible upon the basis of the available part; therefore, this boneis not considered further.

The evaluation of error
In order to test for the presence of intra-observer error, a reproducibility study was undertaken in which all dimensions were measured on six of each tarsal element
on six separate occasions. A two-way analysis of variance without replication was
then undertaken (Sokal & Rohlf 1981) to establish the degree of variation within the replicates for each bone (intra-observer error) compared with the natural variation between the six measured bones. In each of the inter-landmark distance dimensions,the error associated with replicate measurements was found to be significantly less than the real variation between bones ðPo0:05Þ: In the case of the talar angular dimensions, some variation in replicates was noted. However, it was considered valid to include these data for the following reasons. First, there is no significant difference between sexes of the same species and second, there is a significant difference between the apes as a whole and humans. All linear dimensions were obtained using standard osteometric calipers with digital readout. Angular values were obtained by a geometric calculation from digitised photographs. All photographs used for data collection were taken in carefully referenced positions to minimise distortions resulting from camera or object orientation. Inter-observer error was not relevant in this study as all data were collected by one investigator (RSK).

Precision of measurements
Linear data were recorded to 0.1 mm. Angles were recorded in degrees to one decimal place. While it is recognised that this level of accuracy is spurious, it was
considered best to record as such as it was the level of significance produced by the
digitising program and it removed the possibility of human error in rounding.

Problems of measurement of the fragmentary Stw573 talus.
While the fossil assemblage was remarkably complete, there was some erosion of the base of the lateral margin. This landmark is important for the purposes of mensuration of some dimensions, for instance the lateral talar height; thus, some
intelligent reconstruction was required such that estimations could be made. In the
interests of accuracy, several reconstructions were made, but in exploratory analyses,it was found that they made little difference to the actual result with no change in subsequent interpretation of the fossil.

Analytical methods
An initial univariate analysis was undertaken in which the spreads of the individual values for each dimension were compared in each species group. The standard univariate descriptors: the mean, standard deviation, coefficients of variation and distribution shape, were examined. While it may seem superfluous in a multivariate study, this preliminary analysis had three prime objectives. First, it is an essential step in interpretation of subsequent multivariate analyses. Second, it is a most useful way of identifying incorrectly recorded data, for instance a misplaced decimal point. Third, it gives a broad comparison of size differences between the groups. Subsequently, Student ‘‘t’’ tests were undertaken to investigate significance of differences of means between groups. There was no evidence to suggest that any of the variables were not normally distributed.

Plots of means against their standard deviations showed a strong positive regression for most dimensions. As a result the data were subsequently transformed to their natural logarithms. A series of similar plots using the transformed data did not show a significant regression in the vast majority of variables. It was therefore
considered prudent to use log-transformed data for the subsequent multivariate
analyses. The multivariate objective of the study was to establish the morphological
affinities between the groups using canonical variates analysis [CVA] (Albrecht 1980,
1992; Reyment et al 1984). We note that in recent years several techniques have been
developed which may provide information in excess of that made available by CVA,
loosely known as ‘‘geometric morphometrics’’. However, it is felt that in order to
have a direct comparison to the OH8 study of Kidd et al (1996) the same techniques
should be used. Computations were undertaken using PC SAS 6.12 (SAS 1988) which produces four standard outputs for subsequent scrutiny: mean values for each group on each canonical variate, eigenvalues of each canonical variate (indicating the proportion of total information contained within the variate), canonical coefficients, and Mahalanobis’s D2 distance matrix.

A series of indices were constructed from the linear inter-landmark distances. The
primary reason for using these indices, however, was not to remove the gross effects
of size (which, incidentally, did swamp the preliminary analyses of log transformed
data). It was to emphasise biomechanically important features of the osseous morphology. The problems associated with the use of indices to deal with size are
well recognised (e.g. Corruccini 1975; Atchley et al 1976; Albrecht 1993). However,
where the intention is primarily to emphasise those features thought to be of
mechanical importance rather than as a deliberate attempt to remove the effects of raw size, their application is not only useful, it is essential. In fact, they are the only way to emphasise those parts of the anatomy which are of mechanical importance.The length of one lever arm is of no importance biomechanically; the ratio of two lever arms is the biomechanically critical variable. Table 1 notes the indices used in the final analysis, the anatomical features that they define and indicates possible biomechanical rationale.

In addition, to allow an assessment of ‘‘overall similarity’’ of the fossil assemblage compared to the extant species, an integrated analysis has been undertaken of the indices from all bones. In all multivariate analyses used in this study, the fossil was entered directly as a part of the original calculation, though with a sample size of one, rather than by subsequent interpolation into the matrix of extant species.

Results
Univariate analysis
The univariate results were difficult to interpret. With respect to some dimensions,
the fossil seemed to have ape-like values, while in other respects it appeared to be
more human-like. For instance, scrutiny of talar index two which describes the
relationship between the medial and lateral heights, would suggest the fossil is most
similar to humans and most distant from gorillas, yet the same scrutiny but of talar index one would suggest the fossil to be most similar to chimpanzees and most distant from orang-utans. While differences between sexes within species, and
between the species and the isolated fossil may be discerned in most cases, the degree of overlap is great and the position of the fossil quite variable. Thus, a multivariate approach is indispensable.

Multivariate analysis
In the analysis of indices from each of the three bones individually, the majority of
discrimination lies within the first two variates, together accounting for at least 91% in all bones and over 95% in the medial cuneiform. The group mean scores along the first, second and third variates are given in Table 2a, and the Mahanalobis’s distance matrix for individual bones given in Table 3a. The same information but for the integrated study is given in Tables 2b and 3b. The first variate contains notably
differing amounts for each bone, 53% in the talus, 69% in the navicular but over 82% in the medial cuneiform. There is a corresponding inverse relationship with that held in the second variate: 38% in the talus, 24% in the navicular and only 12% in
the medial cuneiform. The third variate in each case contains in the order of 4–6%
and later variates only 1% or less. The first two variates between them account for
the vast majority of the discrimination, the between group differences. Thus, though
they were carefully examined and considered, they are not figured.

The talus
The first variate provides a marked separation of humans from apes as a whole, there being about 4 standard deviation units (SDU) between the mean positions of humans and the nearest ape (male chimpanzees). The fossil occupies a position very closely aligned with African apes. The main indices supporting this discrimination are the height index (ind1), the talar neck angle (ind10), and to a lesser extent, the head torsion angle (ind11) and the trochlear groove index (ind2). The second variate demonstrates considerably less separation of groups, with apes occupying both extremes of the variate and humans in between. Notwithstanding this, it does separate quite clearly African and Asian apes, with about 4 SDU between them. The isolated fossil is more-or-less equidistant between African and Asian apes, closely aligned with humans. The most important dimensions affording this discrimination are the trochlear groove index (ind2), the breadth index (ind6),and to a slightly lesser extent, the height index (ind1) and trochlear breadth index(ind4).

A plot of the first two variates clearly separates into three groups:African apes,
Asian apes and humans.The fossil is quite clearly not positioned closely to any of
these groups, but its nearest group is the chimpanzees at 2.5 SDU (Fig. 1),the
human groups being approximately 3.5SDU distant.

The navicular
The first variate separates orang-utans from African apes by nearly 5 SDU.Humans are positioned between them, though much closer to African apes. The isolated fossil lies among the African apes. The major feature supporting this discrimination is the tuberosity index (ind6), though an important contribution is made by the height and thickness indices (inds4&5),and the maximum cuneiform facet index (ind3).

The second variate convincingly discriminates humans from all apes by a minimum of 2 SDU, though it does not provide any discrimination among the ape groups.The isolated fossil navicular lies along the margin of the ape groups, clearly separated from humans. The main dimensions contributing to this discrimination are the tuberosity index (ind6) and the height and thickness indices (inds4&5).This is probably best explained by the somewhat oblique nature of the plot; thus these
dimensions are being expressed in both the first and second variates,though in a
different manner in each.

A plot of the first two variates reveals a clear separation into the same three groups as found with the talus: humans, African apes and orang-utans. The plot also reveals quite clearly the close affinity of the fossil navicular to those of the African apes (Fig.2). This affinity is confirmed by inspection of the Mahanalobis distance matrix(Table 3a).

The medial cuneiform (Stw573)
The first variate provides a good separation of humans at the positive end of the variate, from chimpanzees at the negative end, of about 9 SDU.Gorillas and orangutans are positioned roughly equidistantly between the two. The fossil bone lies
about 1SDU even more positively than the human group. This separation is largely
afforded by just one dimension, the length index (ind6),though others contribute to
lesser degrees.

The second variate provides a clear discrimination of orang-utans from all other
groups, with the isolated fossil lying roughly equidistant between them.The main
dimensions affording this discrimination are the two anterior breadth indices(inds4&5), though others are also minor contributors.A plot of the first two variates confirms the findings above and serves to emphasise the discriminations (Fig. 3).The fossil cuneiform shows a quite clear affinity with the human groups, a finding confirmed absolutely by the Mahanalobis distance matrix (Table 3a).

The medial cuneiform (OH8)
All findings with the exception of the fossil are practically identical to those above.A plot of the first two variates clearly reveals the position of the OH8 bone to be in the close proximity of humans, in a manner remarkably similar to that of the same bone of Stw573.The same dimensions afford this discrimination with respect to both variates (Fig. 4, Table 3a).

The integrated analysis (Stw573)
An examination of a plot of the first two variates of the integrated analysis is, at
first glance, remarkably similar to that of the medial cuneiform alone, though
horizontally inverted. It thus provides the same separation of orang-utans from all
other groups.However, the degree of discrimination,particularly along the second
variate is very considerably greater being in the order of 10 SDU, clearly indicating
that further information, not found in the cuneiform alone,is being displayed. The
fossil assemblage lies closest to humans, being 2SDU from them, and about 7SDU from the nearest ape group (Fig. 5),borne out by the Mahanalobis distance matrix(Table 3b).

Equally of interest,a plot of the first and third variates displays a separation of
gorillas by about 3SDU from all other groups,the fossil assemblage again being close to humans (not shown). A plot of the first and fourth canonical variates (not shown) provides no separation between extant groups,but does provide separation of the fossil from the extant groups as a whole; this discrimination is statistically
significant ðPo0:01Þ:

The integrated analysis (OH8)
All findings with the exception of the fossil are practically identical to those above. A plot of the first two variates clearly reveals the position of the OH8bone to be broadly equidistant between human, gorillas and orang-utans though marginally closer to humans (Fig. 6). This is confirmed by inspection of the Mahanalobis distance matrix (Table 3b):the fossil assemblage is marginally closer to humans than gorillas or orang-utans, and considerably distant to chimpanzees.

Discussion
The extant species
The four bones, comprising the talus, navicular, medial cuneiform and first
metatarsal form a substantial part of the medial border of the foot.The talus,
together with the calcaneus, represents the posterior portion of the long arch with the midtarsal (transverse tarsal) separating them from the navicular medially,and the cuboid laterally. The midtarsal joint is essentially the apex of the longitudinal arch and as such, is of critical importance. It is not surprising, therefore, that this unit,perhaps more than any other in the foot, has undergone intense evolutionary changes as a component of assuming a bipedal gait. The most obvious large-scale change is that of an overall decrease in motion at this joint, brought about by various modifications, largely bony (Elftman & Manter 1935),but also soft tissue,particularly with respect of the naviculo-cuboid articulation(Lewis 1980; Gomberg 1985).

On the lateral column,the most profound change may be seen in the cuboid with the obvious prolongation of the calcaneus process found in humans but barely noticeable in apes; this structure is of critical functional importance in the manner in which it ‘‘locks’’ into the calcaneus, restricting movement.It is a major factor in
maintaining the lateral longitudinal arch.

On the medial column, an obvious change is that of a ‘‘plantarflexion set’’ to the
talus, resulting in a more steeply positioned talo-navicular joint, affording it a
greater dynamic stability. The other major change in the medial column is to the
talar neck and head torsion angles. The talar neck angle, while being difficult to be
precise, has been correlated with the presence or absence of the divergent first ray
(Volkov 1903, 1904; Duckworth 1904; Straus 1927; Lisowski 1967). That is, a wide talar neck angle is indicative of the presence of a divergent first ray while a narrow angle indicates a non-divergent first ray. In extant apes the angle is wide, while in humans it is narrow (Table 4).

The head torsion angle, while clearly different in humans and apes (Table 4), is perhaps not so easily dismissed, as its explanation is rather more obscure. The
midtarsal joint in humans not only has a decreased range of motion generally, but
also a more variable range of motion. Elftman (1960) describes a restraining mechanism of the midtarsal complex, controlling the range of motion available.
When the subtalar joint complex is everted (pronated), the axes (i.e. bisection lines)of the talonavicular and calcaneocuboid joints are approximately aligned; as a direct consequence, the direction of greatest freedom of movement at the two joints is also aligned. While in this arrangement, the associated ligamentous structures are in a slackened state. Thus, the range of movement of the midtarsal complex is at its greatest. As a consequence, the long arch is able to elongate upon weight bearing; the additional movement allows for shock absorption and occurs during the initial phase of stance, directly after the heel contacts the ground (Wright et al 1964; Inman 1976).

Conversely, when the subtalar complex is inverted (supinated), the same two axes
(bisection lines) assume a decidedly oblique relationship. The result is that the
direction of greatest freedom of movement at the two joints is now not coincident;
the range of movement at the midtarsal joint is now restricted. Associated
ligamentous structures are now taut and, in addition, the two joints are stacked
up upon each other and this reduces bending stresses (Close et al 1967). These factorsare collectively responsible for rendering the arch more stable and afford a greater degree of rigidity important in the later propulsive phase of locomotion. This variable range of motion has often been cited as an important adaptation for bipedal gait (e.g. Inman et al 1981; Langdon et al 1991) and was considered in some detail with respect to functional affinities of OH8 in Kidd et al (1996).

The midtarsal joint divides the foot into a fore-part and a hind-part. However, the
foot as a whole may also be considered in terms of two longitudinal units, the medial
and lateral columns. The medial column consists of the talus, navicular, the three
cuneiforms and structures distal to them. The lateral column consists of the calcaneus,cuboid and their distal structures. In the human foot the two columns form discrete components of the longitudinal arch. In many respects, therefore, one may summarise the stability of the lateral arch in terms of calcaneo-cuboid modifications, the medial arch in terms of talo-navicular modifications; for Stw573, we are only able to comment upon those modifications pertinent to the medial long arch
.
Interpolation of the fossil specimens

The Stw573 assemblage, more commonly referred to as ‘‘little foot’’,clearly
represents a major portion of the medial column, though obviously truncated with
much of the first ray missing. Also missing are the other two cuneiforms and their
rays. Thus, any functional interpretation must be based upon a comparison of the
medial column of humans, apes and other fossils alone.Clarke and Tobias find this
assemblage to have ape-like and human-like features, being most human-like
proximally and becoming more ape-like distally. However, the basis for this
judgement is not obvious; they do not present a metrical result. We are thus
presented with a paradox: this ape-human division is clearly fore-hind, while that found in the OH8 study was medial-lateral. In addition, we find the Stw573 talus, to be decidedly ape-like, though perhaps not so much so, as with the OH8 talus. In addition,to compound the paradox, we find the navicular to be decidedly ape-like, in an extremely similar manner to that of OH8. And still further,we find the medial
cuneiform of both OH8 (not examined previously) and Stw573 to be decidedly humanlike.
The differences between our findings and those of Clarke and Tobias are not reconcilable, though it should be emphasised that our findings with respect of Stw573
are reconcilable with those of the pertinent elements of the OH8 study (Kidd et al 1996).The Stw573 talus has decidedly ape-like features,perhaps particularly with respect of the head torsion and neck angles. The former is strongly indicative of a mobile talo-navicular component of the midtarsal joint,and is clearly not indicative of a human-like function; it simply would not have the stability required to maintain a medial longitudinal arch. The talar neck angle of the Stw573 talus is notably ape-like and on its own would suggest quite strongly the presence of a divergent first ray.Certainly this was one of the suggestions made by Kidd et al(1996) with respect of the Olduvai foot, though it must be emphasised that this was based upon findings from the four hindmost bones alone.

The OH8 study provides a clear dichotomy of structure with human-like features
laterally and ape-like features medially.If one makes the assumption that OH8 is a
human ancestor it is possible to speculate constructively, in the light of the OH8
study, as to the chronology of evolutionary modifications that have taken place to
form the modern, human midtarsal joint.And, in so much as the medial and lateral
components of the longitudinal arch of the human foot are inextricably a part of the
medial and lateral columns described earlier, some constructive speculation may be
suggested as to the sequence of events in arch formation.

The well-developed calcaneo-cuboid joint in OH8 is strongly indicative of a stable
lateral column.This in turn suggests the presence of at least a degree of lateral
longitudinal arching in this foot. On the other hand, the low head torsion angle is
highly indicative of an undeveloped midtarsal restraining mechanism.Thus, the
presence of the lateral components of these modifications,but the absence of
equivalent medial components, may be seen as evidence as to the chronology of
evolutionary events.It would appear that formation of the lateral longitudinal arch,
together with increased calcaneo-cuboid stability was an early evolutionary event in
the history of terrestrial bipedalism in hominids.The equivalent modifications to the medial side of the foot seem to have occurred subsequently.

The medial cuneiform study of both OH8 and Stw573 provides findings to suggest
that this model may be expanded further as they provide a more detailed insight into
the medial column. It would appear that within the column, the more distal
structures attain a human-like form before those more proximally sited. Thus,a
highly complex model of development may be proposed in which first the lateral
column achieves a human-like state, followed by the distal component of the medial
column, finally followed by the proximal component.

The human-like nature of the medial cuneiform of both OH8 and Stw573 does not
provide evidence for the presence of a divergent first ray;quite the opposite in fact.

The human-like nature of this bone in the two fossils would suggest that the first ray and digit were likely to be at least broadly human-like.In Kidd et al (1996) we
speculated that OH8 may have possessed a divergent first ray and opposing first digit
but in view of the medial cuneiform findings for both OH8 and Stw573,it may be
that this speculative suggestion needs rethinking. Day and Wood (1968) in their
analysis of the OH8 talus, make the pertinent comment that the OH8 talar neck
angle is wide, and in isolation would indicate a divergent hallux. They continue to
state that ‘yyequally true that the hallux is not divergent in this form compensatory adjustments have been made more distally’(p. 454).The new evidence
presented in the form of the medial cuneiform analyses provides tentative evidence
for this notion, and also provides us with further information upon which to build an
evolutionary model.

Some criticism of this line of thinking may be levelled upon the basis of the
supposed ages of the two fossil assemblages and their difference.Isotope dating
suggests that OH8 is of the order of 1.7Ma (Day & Wood 1968), while the Stw573
may be as old as 3.5 Ma.However, the accuracy of the stratigraphy of Sterkfontein
has been questioned, most recently by Berger et al (2002)who suggest an upper age
of no more than 3 Ma, and possibly as young as 1.5 Ma.It is quite possible,therefore, that they are in fact contemporaneous.

It is worth pausing for thought at some further implications of what we have
described here. First, in the foot generally we can visualise a lateral column (that
leads to the fourth and fifth digits) and a medial column (that leads to the hallux and the second and third digits).Though described as lateral and medial columns in
terms of adult functional anatomy, these components are embryologically caudal
and cranial, respectively.Embryologically, therefore, these are elements produced by
a caudo-cranial process in the terminal portion of the limb (cheiridium) and this
process is controlled by the sonic hedgehog series of genetic factors (Larson 1997). In the OH8 assemblage there exists a human-like lateral (caudal) column and an apelike medial (cranial) column.It is thus possible that this difference from an ape-like situation to a human-like situation is brought about through modifications of the sonic hedgehog system.

In a similar way, second, we can visualise differences leading from posterior
elements of the foot (calcanea, tali) through cuneiforms and cuboid, to the anterior
foot elements, metatarsals (digits being absent in these fossils). These components
are embryologically proximal to distal, respectively and are produced by a proximodistal process that is controlled by the Hox gene system (Larson 1997).In the Stw573 assemblage there exist ape-like proximal components, and human-like distal
components in the medial column—likewise a human-to-ape difference. Again,
therefore, it is possible that these differences are brought about through modifications of the Hox gene system.

Finally, we can visualise anatomical arrangements that are dorsal and plantar in the foot (for example not only the longitudinal arches, but also the related transverse arches).These elements are embryologically produced by a dorso-ventral gradient within the limb bud that is effected by the Wnt system of factors (Larson 1997).In the OH8 assemblage there exists an ape-to-human dorso-ventral difference along the lateral column.It is thus possible that this is brought about through modifications of the Wnt system.

Thus the adult anatomical differences we describe, though undoubtedly adaptive in terms of foot function, may also be understood as having been brought about by appropriate changes of genetic developmental factors, Hox, sonic hedgehog and Wnt
genes/growth factors and of course their upstream and downstream qualifiers. That
is, morphometric examination that reveals obvious features of functionally adaptive
significance also reveals factors that may relate to their developmental expression. In other words, the adult comparisons achieved by morphometrics show not only the
expected direct functional connotations, but also unexpected ‘‘ghosts’’ of their
developmental pasts. Though unexpected, this is not really surprising. It is not only
entirely logical to expect adult morphological comparisons to mirror underlying
developmental bases but also that they may indicate how, in developmental terms,
such evolutionary adaptations have been achieved.

If this were the only example where developmental factors had been revealed
through morphometric analyses it might not be overly convincing. However, similar
findings have been made in studies of overall body proportions and dental
dimensions (e.g. Oxnard 1992, 2000). In each of these cases, too, the morphometric
groupings of the variables defining differences among adult forms seem to be related
to underlying developmental phenomena. For body proportions these included
proximo-distal and, separately, caudo-cranial arrays of variables (in that study
dorso-ventral variables were not examined). For the dental apparatus, these included
groups of variables mirroring the different populations of neural crest cells from
which the different components of the mandible arise.

Integrated analyses can be useful summaries of a ‘‘total morphological pattern’’,
advocated so long ago by the late Professor Sir Wilfred Le Gros Clark. Thus, a firm
indication of morphology may be gained not based upon bones as if they were
independent structures, but as if they were parts of functional groups such as the
medial column. The integrated analysis of the OH8 talus, navicular and medial
cuneiform suggest a unique overall morphology, neither decidedly ape-like nor
human-like. The same cannot be said for Stw573, in which the fossil assemblage enjoys
a decidedly human-like position. This is confirmed by inspection of the Mahanalobis
distance matrices. Thus, upon the basis of analyses of independent bones, there is a
case for suggesting that in fact the two feet, at least as judged by those parts of the medial column available, are remarkably similar. However, when seen from the
viewpoint of total morphological pattern, it is clear that there are fundamental
differences, the Stw573 assemblage being considerably ‘‘more human-like’’.

Acknowledgements
We wish to acknowledge the contribution and advice given by Professor Paul
O’Higgins, Hull-York Medical School, and for his comments on the manuscript. We
also wish to acknowledge the comments of two reviewers, Professor Robert
Eckhardt and Professor Carstens Niemitz.

References
Albrecht G (1980) Multivariate analysis and the study of form, with special reference to canonical variates
analysis. Am Zool 20: 679–693.
Albrecht G (1992) Assessing the affinities of fossils using canonical variates and generalised distances.
Hum Evol 7: 49–69.
Albrecht G (1993) Ratios as a size adjustment in morphometrics. Am J Phys Anthrop 91: 441–468.
Atchley WR, Gaskins CT, Anderson D (1976) Statistical properties of ratios. 1 Empirical results. Syst
Zool 25: 137–148.
Berger LR, Lacruz R, de Ruiter DJ (2002) Brief communication: revised age estimates of
Australopithecus-bearing deposits at Sterkfonein, South Africa. Am J Phys Anthrop 119: 192–197.
Clarke RJ, Tobias PV (1995) Sterkfontein member 2 foot bones of the oldest South African Hominid.
Science 269: 521–524.
Close JR, Inman VT, Poor PM, Todd FN (1967) The function of the subtalar joint. Clin Orthopaed 50:
159–179.
Corruccini RS (1975) Multivariate analyses in biological anthropology: some considerations. J Hum Evol
4: 1–19.
Day MH, Wood BA (1968) Functional affinities of the Olduvai hominid 8 talus. Man 3: 440–445.
Duckworth WLH (1904) Morphology and Anthropology. Cambridge University Press, Cambridge.
Elftman H (1960) The transverse tarsal joint and its control. Clin Orthopaed 16: 41–45.
Elftman H, Manter J (1935) The evolution of the human foot, with special reference to the joints. J Anat
70: 56–67.
Gomberg D (1985) Functional differences of three ligaments of the transverse tarsal joint in hominoids.
J Hum Evol 14: 553–562.
Inman VT (1976) The Joints of the Foot and Ankle. Williams and Wilkins, Baltimore.
Inman VT, Ralston HJ, Todd F (1981) Human Walking. Williams and Wilkins, Baltimore.
Kidd R(1995) An investigation into the patterns of morphological variation in the Hindtarsus of selected
human subgroups, hominoids and fossils: a morphometric analysis. PhD thesis, The University of
Western Australia.
Kidd R(1998) The past is the key to the present: thoughts on the origins of human foot structure, function
and dysfunction as seen from the fossil record. Foot 8: 75–84.
Kidd R(1999) Evolution of the hindfoot: a model of adaptation with evidence from the fossil record. J Am
Podiatric Med Assoc 89: 2–17.
Kidd RS, O’Higgins P, Oxnard CE (1994) The OH8 foot, a reappraisal of the functional morphology of
the talus, calcaneus, navicular and cuboid. Am J Phys Anthrop 17 (Suppl): 123.
Kidd RS, O’Higgins P, Oxnard CE (1996) The OH8 foot: a reappraisal of the functional morphology of
the hindfoot utilizing a multivariate analysis. J Hum Evol 31: 269–291.
Knussmann R(1988) Anthropologie: Handbuch der vergleichenden Biologie des Menschen, Fisher,
Stuttgart, New York.
Langdon JH, Bruckner J, Baker HH (1991) Pedal mechanics and bipedalism in early hominids. In
Origine(s) de la Bipe´ die chez les Hominide´ s (Cahiers de Pale´ anthropologie), Editions du CNRS, Paris.
Larson WJ (1997) Essentials of Human Embryology. Churchill Livingstone, New York.
Lewis OJ (1980) The joints of the evolving foot-2: the intrinsic joints. J Anat 130: 833–857.
Lisowski FP (1967) Angular growth changes and comparisons in the primate talus. Folia Primat 7: 81–97.
Lisowski FP, Albrecht GH, Oxnard CE (1974) The form of the talus in some higher primates: a
multivariate study. Am J Phys Anthrop 41: 191–216.
Oxnard CE (1992) Developmental processes and evolutionary diversity: some factors underlying form in
primates. Arch Ocean 27: 95–104.
Oxnard CE (2000) Morphometrics of the primate skeleton and the functional and developmental
underpinnings of species diversity. InO’Higgins P (ed) Development, Growth and Evolution:
Implications for the Study of the Hominid Skeleton. Academic Press, London, 235–263.
Reyment RA, Blackwith RE, Campbell NA (1984) Multivariate Morphometrics. Academic Press,
London.
ARTICLE IN PRESS
R. Kidd, C. Oxnard / HOMO—Journal of Comparative Human Biology 55 (2005) 189–212 211
SAS Institute Inc (1988) SAS/STAT User’s Guide. Release 6.03 Edition. SAS Institute Inc, Cary, NC
174–177.
Sokal RR, Rohlf FJ (1981) Biometry, The Principles and Practice of Statistics in Biological Research. 2nd
ed. WH Freeman, New York.
Straus WL (1927) The growth of the human foot and its significance. Contrib Embryol 101: 93–134.
Volkov T (1903) Variations sque´ lettiques du pied chez les primates et dans les races humaines. Bull Soc
Anthropol (Paris) 4: 632–708.
Volkov T (1904) Variations sque´ lettiques du pied chez les primates et dans les races humaines. Bull Soc
Anthropol (Paris) 5 (1–50): 201–331.
Wright D, Desai S, Henderson W (1964) Action of the subtalar and ankle-joint complex during the stance
phase of walking. J Bone Joint Surg 46A 361–382, 464.

Source: Reproduced with the kind permission of Dr.Robert Kidd.

Tables unable to be re-produced in this blog.