Supporting Online Material for - Science · 2/25/2009  · the south the surface is heavily poached by numerous superimposed animal prints. A full ichno-taxonomy of the animal prints

www.sciencemag.org/cgi/content/full/323/5918/1197/DC1

Supporting Online Material for

Early Hominin Foot Morphology Based on 1.5-Million-Year-Old Footprints from Ileret, Kenya

Matthew R. Bennett,* John W.K. Harris, Brian G. Richmond, David R. Braun, Emma Mbua, Purity Kiura, Daniel Olago, Mzalendo Kibunjia, Christine Omuombo, Anna K.

Behrensmeyer, David Huddart, Silvia Gonzalez

*To whom correspondence should be addressed. E-mail: [email protected]

Published 27 February 2009, Science 323, 1197 (2009) DOI: 10.1126/science.1168132

This PDF file includes:

Methods SOM Text Figs. S1 to S22 Tables S1 to S6 References and Notes

1

Online Material Early hominin foot morphology based on 1.5 million year-old footprints from Ileret, Kenya Matthew R. Bennett, John W.K. Harris, Brian G. Richmond, David R. Braun, Emma Mbua, Purity Kiura, Daniel Olago, Mzalendo Kibunjia, Christine Omuombo, Anna K. Behrensmeyer, David Huddart, Silvia Gonzalez

1.0 Methods: Footprint Imaging Two sites are discussed in this contribution, FwJj14E in Area 1A 5 km east of Ileret and GaJi10 in

Area 103 just south of Koobi Fora (Fig. S1). The site at FwJj14E has been the focus of the Koobi Fora

Research and Training Program, jointly organised by Rutgers University and the National Museums of

Kenya since 2005. A series of linked excavations, undertaken between 2005 and 2008, have been

made at this site which have yielded a range of finds, including both animal and hominin footprints

(Figs. S2A and S3). In addition we re-examine the prints from GaJi10 first excavated by

Behrensmeyer and Laporte in 1978/79 (S1) and re-excavated as part of the current investigation.

Following excavation all footprints were mapped, photographed, and digitally recorded. Footprints on

all exposed surfaces were scanned using a VI900 Konica-Minolta optical laser scanner used in

horizontal mode and mounted on a custom built rig, approximately 600 mm above the footprint

surface, and enclosed within a canvas cover to control light levels. The scanner simultaneously

digitises the three-dimensional surface below and captures a low resolution digital image. Using a

wide angle lens with a focal length of 8 mm the scanner has a vertical resolving power of ± 0.09 mm

and each scan covers an approximate area of 500 by 250 mm, with approximately 300,000 points per

scan. Scanning was completed during two field seasons, one in July 2007 and a second in July 2008.

Post-processing of the scans was undertaken in RapidformTM and by outputting the scan data as a xyz

point cloud and importing it into ArcGISTM. Contour maps were generated in ArcGISTM using the

complete data set; no data decimation was undertaken. Scans were color rendered for visualisation

using z data and a range of different color schemes from a simple red-blue spectrum to alternating

black and white isopleths (Fig. S4). It was also found that inverting a scan - that is effectively

producing a cast - prior to color rendering often enhanced visualisation of anatomical form.

Scanning was complemented by close-up photography and a vertical mosaic of high resolution digital

images taken with a Nikon D200 from a height of approximately 1.5 m. The photographs were geo-

rectified in ArcGISTM to create a single ortho-rectified photo-mosaic using a grid of 5 mm square

orange cubes laid out in a 1 m x 1 m grid. Each cube was located in three-dimensional space using a

Leica TPS 1100 and cross-referenced to the 1m survey grid established as part of the excavation.

The ortho-rectified photo-mosaics were used to create the footprint maps (Fig. S3).

2

2.0 Methods: Geochronology

The geochemistry and age (40Ar/39Ar) of the tuffs of the Turkana Basin have been well document

(S2,S3), providing a robust tephra chronology for the region. Bulk samples of volcanics were collected

from all the tuffs present at FwJj14E and at GaJi10 and were sent to Professor Frank Brown at the

University of Utah for analysis and comparison with the dated standards (Fig. S5). Each of the

samples contains a range of volcanic shards consistent with them being fluvially re-worked tephra;

although in cases they are geochemically distinct allowing them to be correlated with a known and

dated ash. The trace element data obtained from each of the samples is presented in Table S1.

• Samples JWKH-1. Upper tuff at FwJj14E and with a geochemistry that correlates with the

Northern Ileret Tuff which is dated to between 1.51 and 1.52 Ma.

• Samples JWKH-2. Taken from the middle tuff at FwJj14E. This tuff is most likely the Ileret

Tuff dated to around 1.52 Ma. However, it contains a range of glass shards which can be

correlated with a number of different potential tuffs, including: the Northern Ileret Tuff, a tuff

referred to informally as the Southern Ileret Tuff, one similar to a secondary mode of the Okote

Tuff which outcrops in Area 131, and the Black Pumice Tuff (c. 1.53 Ma).

• Sample JWKH-4. Taken from the lower tuff at FwJj14 East. This is probably the Lower Ileret

Tuff, but it also contains shards similar in composition to the Elomaling’a Tuff, the Orange Tuff,

and two unknown tuffs. On the basis of the age of the Lower Ileret and Elomaling’a tuffs it is

likely to have an age of approximately 1.53 Ma.

• Sample JWKH-5. Taken from the tuff at GaJi10 stratigraphically overlying the footprint site

reported by Behrensmeyer and Laporte (S1). The principal composition corresponds to the

Akait Tuff (c. 1.435 Ma) and the Naiyena Epul Tuff (West of Lake Turkana) which lies below

the Akait Tuff, although one shard closely resembles the Karari Grey Tuff dated to

approximately 1.428 Ma. The most likely date for this tuff is therefore 1.43 ± 0.01 Ma.

Although the samples clearly contain a range of volcanic material consistent with them being

emplaced both by air fall and by fluvial re-working, they provide a robust geochronological framework

for both the footprints at FwJj14E and GaJi10, with the former being approximately 1.5 Ma and the

later slightly younger at 1.4 Ma.

3.0 Additional Text, Data and Illustrations 3.1 FwJj14E At FwJj14E the sedimentary succession consists of a series of fine-grained silt and medium to coarse-

grained sand with a variety of cross-beds consistent with over-bank deposits from a near by stream or

river. There are a variety of fining-upwards sequences at a range of scales, but several large-cycles

can be identified (Fig. 1). Each cycle represents a flood event culminating in a finer-grained silt layer.

Animal and hominin prints occur on the bedding plane surfaces of these silt-clay layers and were

3

probably imprinted around contracting water pools following a flood event (Fig. S2B,C). Two footprint

bearing horizons have so far been excavated at FwJj14E (Fig. S3)

Upper Footprint Surface This surface consists of an area approximately 10 m long by 4 m wide at its maximum and is bounded

to the east by vertical faces. The surface consists of a sand-silt horizon which has been traversed by

a variety of animals and hominins to create a palimpsest of footprints (Fig. S2B). A total of 22

potential hominin prints have been recognised on this surface, although only 17 are formally

recognised here as being produced by a hominin (Fig. S3). Three trails account for eleven of the

hominin prints, the rest consisting of either isolated or partial prints. The trails are labelled FUT

(FwJj14E Upper Trail), with the first print of the first trail denoted FUT1-1 and the individual prints

named FUI1, FUI2 etc (FwJj14E Upper Individual print). Additional images of all prints are provided in

Figures S6 to S10; specifically figures S6 to S8 show the same area, but with different data formats, to

illustrate the range of information used in making footprint inferences and interpretations (Prints FUT1-

1 to FUT1-3). Figures S9 and S10 provide scanned illustrations of all the potential hominin prints on

the Upper Footprint Surface. Figure S11 provides a selection of isometric views for some of the

footprints.

Lower Footprint Surfaces The Lower Footprint Surface is divided into two parts, north and south (Fig. S3). No hominin prints are

present on the southern section of this surface and it is not discussed further in this contribution. The

northern surface is approximately 20m long and has a maximum width of 4m, although the southern

part of it forms a narrow strip between 0.5 and 1m wide which is divided into two halves by a runnel of

superimposed and poorly defined prints (Fig. S13A,B). To the north of this runnel the surface is more

compact and uniform in relief with good examples of artiodactyl, perissodactyl and large bird prints, to

the south the surface is heavily poached by numerous superimposed animal prints. A full ichno-

taxonomy of the animal prints present on this surface is the subject of ongoing research, but so far

includes the following (Fig. S13C-F): (1) bovids, (2) carnivores including potential felid prints, (3)

equids, and (4) a. The focus here however is on three hominin prints (Figs. S3 and S13), two of which

form a short trail at the north end of the footprint surface, and an isolated print which occurs just to the

north of the runnel mentioned above.

3.2. GaJi10 Behrensmeyer and Laporte (S1) reported hominin footprints from an excavation undertaken in 1978

and 1979 in Area 103 south of Koobi Fora (GaJi10; Fig. S1). They identified several closely

superimposed footprint horizons, one with hominin tracks. These prints occur in a succession of fine-

grained silts interpreted at the time to be a lake margin environment, a conclusion supported by the

presence of wading birds and hippopotamus prints. The hominin prints are preserved as a latex and

4

fibreglass cast held by the National Museums of Kenya and this cast was examined and scanned both

in July 2007 and March 2008. The original excavation of Behrensmeyer and Laporte (S1) was re-

opened in July 2008 and the hominin footprint trail uncovered; some of the tracks remained in good

condition (Fig. S1). Of the seven hominin prints reported by Behrensmeyer and Laporte (S1) two have

been lost to erosion (R4 & L4), but L1, R1, L3 and R3 remain (Fig. S14E). Of these prints, only L3

and R3 have good anatomical form, but the other prints are too deeply imprinted into the surface to be

of use in this study (Figs. S14 and S15). Print R2 was not relocated on the re-excavated footprint

surface. Consequently only prints L3 and R3 were used in this study; these prints show some

anatomical detail, although much less than that present at FwJj14E, and appear to have been

imprinted by a foot with a relatively adducted hallux, a visible but relatively weak longitudinal arch, and

with some evidence of medial weight transfer prior to toe-off. The GaJi10 prints are limited in number

and topographic detail, but our analysis suggests based on the preserved morphology and size (Table

S3) that Homo ergaster/erectus also made the GaJi10 prints.

The original excavation was extended by 1 m to the west in July 2008 in an attempt to trace the

footprint trail farther. One potential additional hominin print (R0) was uncovered, but it lacks the

anatomical detail present within prints L3 and R3 (Fig. S15F). Scans taken from the actual footprint

surface in July 2008 were used in preference to the casts and moulds made in 1978/79, as there were

noticeable differences between the re-excavated prints and the casts of the 1978/79 excavations

including linear artefacts along the medial boundary of the cast footprint (Fig. S15B) not present in the

re-excavated print (Fig. S15A).

3.3 Laetoli

In order to compare the footprints at FwJj14E and GaJi10 to the Laetoli footprints from 3.75Ma (S4,

S5) a series of casts of both individual Laetoli footprints and selected trail (S5,S6) held at the National

Museums of Kenya were scanned in March 2008. Examples of both G1 and G2/G3 prints were

obtained; however, because the G2/3 prints have the added complication of being superimposed

prints of two individuals, only G1 prints were used in the statistical analysis: G1-23, G1-25, G1-26, G1-

27, G1-31, G1-33, G1-34, G1-35, G1-36, G1-37 and G1-39. A selection of images is presented in

Figure S16.

The interpretation of the Laetoli footprints has been and continues to be a matter of debate over

whether they represent an essentially modern bipedal gait (S5-S9), a primitive gait (e.g., bent-hip,

bent-knee) or unique form of gait (S10-S12), or whether the evidence to date is ambiguous (S13).

Debate also continues over the interpretation of anatomy from the prints, with some researchers

arguing that they point to a primitive foot structure bearing a slightly abducted hallux, relatively long,

possibly curled lateral toes, and lacking both a medial longitudinal arch and evidence of a medial

weight transfer (S10,S11,S14). Others instead argue that the Laetoli prints show evidence of a

relatively modern human-like foot anatomy with evidence of a medial longitudinal arch (S5-S8), and

5

these researchers point to the considerable variation in footprint structure, including variations in the

degree of hallucal abduction and longitudinal arch height in the prints made by habitually shod and

unshod modern humans (S6). Our study supports the hypotheses that the Laetoli prints were made

by a foot whose hallux was much more adducted than in apes, but slightly and significantly more

abducted than that of modern humans (Fig. 4A), with little positive evidence of medial weight transfer

prior to push-off, or a longitudinal arch comparable to that seen in modern humans (Fig. S16). There

is little doubt that compared to the modern great apes (S11, S15) the Laetoli prints and the hard tissue

anatomy of contemporary hominins show evidence of a foot adapted for bipedalism including a more

rigid tarsus but debate continues over whether the Laetoli footprints provide evidence of a modern

human-like longitudinal arch (S6,S11).

3.4 Homo sapiens footprint controls

Control samples of Homo sapiens footprints were obtained from two independent sources:

1. Sefton Coast where Neolithic to Bronze Age (>3649 ± 109 years BP)(S16) footprints are

periodically exposed in coastal sediments of North West England. This population is well

constrained geochronologically, deposited in a similar soft, wet, silt to fine sand as are the

prints from FwJj14E, and a range of prints are available. Prints were optically scanned using

the VI900 Konica-Minolta both in situ in the field in July 2006 and from plaster casts collected

over a number of years by Gordon Roberts and processed as discussed in Section 1.0. A

selection of typical footprints is provided in Figure S17.

2. Footprints were collected along with biometric data for a small random sample of modern

people native to Ileret (Dassenach). A total of 31 footprint scans were obtained in this way

using a sand box filled with damp silt from the backfill pile at FwJj14E. Individuals were asked

to walk the length of the sand box (c. 5 m) leaving between 4 and 6 prints at their preferred

walking speed. A right print from each trail was selected from the middle of the sand box and

scanned using the VI900 Konica-Minolta laser scanner. A range of biometric data was also

collected from the individuals. The footprint scans were processed and analysed as described

in Section 1.0.

4.0 Morphometric data & statistical analysis

Berge et al. (S17) recently pioneered the application of geometric morphometrics (S18) to the analysis

of footprints. For simplicity a similar land-mark based approach has been used in this study although

the authors acknowledge that alternative approaches (S19), such as ‘landmark-free’ and ‘sliding-

landmarks’, may also provide equally valid, although more complex, approaches. The underlying

principle was to use traditional pedal landmarks (S20) in order to explore issues of shape with explicit

functional hypotheses. It is important to note that footprint shape is not the same as pedal anatomical

shape. Instead, print shape is a reflection of foot function and has been of considerable interest in

previous research (S20,S21). Print shape varies with foot anatomy, foot function, speed, and

6

substrate properties, and consequently prints made by the same foot will vary to some degree. This

study used a core of 13 anatomical landmarks, the largest number that could be located with

confidence on all the prints. We selected these landmarks based on compatibility with previous

footprint research, and to capture specific measures of functional characteristics (Fig. S16)(S20).

Landmarks were placed independently by several observers and the consensus location adopted

(mean placement error ±3.6 mm). Many of the landmarks were located around the periphery of a print

where maximum variation in depth occurs and consequently inconsistent placement may introduce

small errors in the z co-ordinate for a given landmark. Where 2-dimensional landmark data (x,y co-

ordinates) were used this is of no consequence (Section 4.4), but may impact on the 3-dimensional

analysis (Section 4.5). The potential for error is greatest were the prints are deeply impressed,

however in these cases placement error is unlikely given that the print has a well-defined edge. In

contrast placement errors are more likely were a print is lightly impressed and the print margin poorly

defined. In these cases the increased potential for depth error is not significant. No allowance was

made for post-depositional modification of the prints and the authors accept that this is an issue for

selected prints, but one which is not easily dealt without introducing a subjective correction. For

example, print FLT1-2 appears to be medio-laterally compressed (Fig. 3D, Fig. S13D); however, the

lower prints compare in the same manner as the upper prints with modern human and Laetoli prints

(Fig. S17), suggesting that FLT1-2 was compressed with minimal alteration to its nonaffine shape. It

is also important to note that the material properties and primary moisture content of the Laetoli prints,

made in fine-grained volcanic ash, may be different from those at FwJj14E, GaJi10 or on the Sefton

Coast.

Experimentation by the authors with a range of materials suggests that the overall geometry of a print

does not vary substantially between different fine-grained materials, however the moisture content

does impact on the absolute, although not the relative, depth of a print. Moisture content also impacts

on the immediate preservation and post-depositional modification of a print following foot withdrawal

however this is normally evident in a print via collapsed margins and internal sediment flowage.

These conclusions are supported by other authors working on animal tracks (S22-S24) who find that

while the detailed print definition varies with between different materials and with moisture content the

overall geometry of a print remains remarkable consistent. Prints made in material that is too wet or

dry are simply not preserved and consequently preservation tends to favour comparable

sedimentological conditions, thereby facilitating comparison. We suggest, therefore, that substrate is

unlikely to be a major factor when comparing the print populations used within this study all of which

were made in fine-grained materials.

Landmarks were placed directly on each scan in RapidformTM and x, y, z co-ordinates were extracted

for each, along with other key dimensions such as hallux abduction which is defined in Section 4.2

(Fig. S18). All measurements were made to two decimal places. Subsequent analysis was either

undertaken in PAST, a statistical toolkit developed for palaeontology by Øyvind Hammer of the

University of Oslo (http://folk.uio.no/ohammer/past/), or in SPSSTM. The following prints were used at

7

FwJj14E and GaJi10: FLT1-1,FLT1-2, FUT1-1, FUT1-2, FUI1, FUT1-6, FUT1-7, FUT3-2, FUT3-1, L3

and R3.

4.1 Biometric relationships for the Dassenach footprints

A simple biometric relationship between subject height and footprint length was developed using the

31 scanned right footprints. The height of each subject was measured using a standard medical

height gauge to the nearest millimeter Footprint length was measured digitally on each footprint scan

(right foot) between the heel and second toe landmarks (Fig. S18). Linear regression was performed

on data and the results are summarised in Table S2A. This relationship was used to corroborate

height inferences made from the prints at FwJj14E and presented within the main paper. Linear

regression between footprint length and body mass showed a significant relationship, but one with a

poorer correlation than that between print length and stature (Table S2B). 4.2 Stature and body mass estimates from hominin prints In order to assess the taxonomic implications of the sizes of the footprints, we examined how

estimates of stature and body mass based on the prints would compare with size estimates based on

postcranial remains attributed to relevant east African early Pliocene hominin taxa (Table S3). Mass

and stature estimates of the modern human (non-Dassenach), Pan troglodytes, and fossil hominin

taxa are taken from the literature (S25,S26). The length of the hominin prints from both FwJj14E and

GaJi10 are similar to the mean length of prints of adult Dassenach. Estimates of body mass and

stature based on the fossil print lengths compare well with those of modern humans as well as those

of Homo ergaster/erectus based on postcranial fossil. In contrast, the mass and stature estimates of

Paranthropus boisei and Homo habilis are much smaller.

In each case, the 95% prediction limits (broader and more conservative than the 95% confidence

limits) of the footprint stature estimates include the mean stature estimates of Homo ergaster/erectus

but excludes the mean stature estimate, and the male stature estimate, of Paranthropus boisei and

Homo habilis, indicating that the stature of the individuals who made the fossil prints is significantly

greater than that inferred for the latter taxa. The mass estimates have a broader 95% prediction

interval because of the weaker relationship between print length and mass. However, these too

correspond with the mass estimates of Homo ergaster/erectus and do not include the mean mass

estimates of Paranthropus boisei and Homo habilis. Note that the Homo ergaster/erectus female

mass and stature estimates may be high in light of recent work (S27) showing that more adults of this

taxon are small than had been originally appreciated; however, this does not influence the taxonomic

inferences made here because the prints are larger than size estimates of all taxa other than Homo

ergaster/erectus. This analysis underscores the large size of the footprints at both FwJj14E and

GaJi10, and strongly suggests that the prints were not made by Paranthropus boisei or Homo habilis.

8

4.3 Comparison of abduction angles Abduction angle is defined here as the angle between two footprint axes, firstly the longitudinal axis of

the foot (heel to second toe) and secondly the axis of the hallux. The hallux axis was defined on each

footprint by drawing a line running parallel to the long axis of the first toe from the tip of the first toe

through the deepest part of the toe pad to the intersection with the longitudinal axis of the foot.

Abduction angles were compared for each of the print populations using a non-parametric approach

(Krustal-Wallis) which showed a significant difference between the populations (Chi-square =46.75;

df=5; p<0.0001) allowing further examination by a post hoc pair-wise Mann-Whitney U tests. To

mitigate the increased risk of Type 1 error in the multiple Mann-Whitney U tests, an approximate

correction was determined by dividing the usual alpha p=0.05 by the number of comparisons, giving

an approximation to a Bonferroni's adjustment. That gives a new alpha of p=0.003. The results of this analysis are provided in Table S4. 4.4 Two-dimensional landmark analysis

Figure S19 shows a series of simple bivariate plots of key footprint dimensions. Two-dimensional

coordinates for the 13 core landmarks were subject to a Generalised Procrustes Analysis in PAST.

Mean landmark positions for each of the footprints populations were calculated and used as a basis

for comparison as illustrated in Figure 4B via thin plate splines and landmark vector maps (S18).

Figure S20 shows additional thin plate splines and landmark vector maps comparing each of the

footprint layers at FwJj14E and the prints at GaJi10 with the control samples.

These data were also used within a Principal Components Analysis (PCA) to explore the degree of the

difference between the populations (Fig. S21). In this analysis the following populations were

included: FwJj14E, GaJi10, Homo sapiens (Modern), and Laetoli. The first three components which

are plotted explain 64% of the variance. Figure S21 shows the first and second components and

illustrates the differences along each with thin plate splines and vector maps. Much of the difference

along the first component is concerned with the development of the longitudinal arch and the

placement of the ‘Ball’ landmark which provides an indication of the degree of medial weight transfer

during the later part of stance. The second component largely focuses on the relative dimensions of

the proximal to distal portions of the foot. It is important to note that this analysis does not include any

direct measurement of the degree of hallux abduction.

4.5 Three-dimensional landmark analysis

Analysis of the landmark data was also undertaken using three-dimensional landmark coordinates.

These coordinates were subject to a Generalised Procrustes Analysis (GPA) and the GPA coordinates

9

were then used to calculate inter-landmark distances which were then used in the subsequent

analysis. This analysis was undertaken in two stages. The first stage used the inter-landmark

distances which were treated as independent variates, a stepwise Discriminant Function Analysis

(minimising Wilks' λ, F in = 3.84, F out = 2.71) was undertaken with the intention of reducing the initial

79 variates to a smaller number more proportionate to the sample size. This resulted in the selection

of 8 variates (Abduction Angle; B1-Ball1; B1-H2; D1-D2; D1-H1; D1-I1; D2-Hdth; H1-H2) which

significantly discriminated among and perfectly classified the four groups (Homo sapiens [Modern],

Homo sapiens [Holocene], Laetoli, FwJj14E; Fn1: 73.3% of variance, Wilks' λ = 0.006, df=24,

p<0.0001; Fn2: 17.3% of variance, Wilks' λ = 0.085, df=14 , p<0.0001; Fn3: 9.4 % of variance, Wilks' λ

= 0.36, df=6 , p<0.0001). Cross-validation did not reduce the classification success. Individuals and

group centroids are plotted with respect to the first two Discriminant Functions in Figure S22A.

Standardised coefficients of the independent variates in these functions are shown in Table S5A. The

selected distances identify the following anatomical differences within the footprints: hallux abduction,

width of the heel, and the presence and form of the longitudinal arch.

The second stage used the 8 independent variates previously identified and a Discriminant Function

model was fitted to the Laetoli and Homo sapiens (Modern). The function was highly significant

(Wilks' λ = 0.033, df=8, p<0.0001), with a classification success of 100% and cross-validation success

of 100%. The model was used to classify the remaining groups: Homo sapiens (Holocene), FwJj14E

Lower, FwJj14E Upper and GaJi10. Scores on the function were calculated for all individuals, and are

plotted for the four groups in Figure S22B. The standardised coefficients in the function are shown in

Table S5B and the classification of each print against the two end members shown in Table S6.

10

Figure S1. Map showing the collecting areas relevant to this paper in the Koobi Fora Region.

Figure S2. A. General view of FwJj14E in July 2008. B. Oblique photograph of the Upper Footprint Surface showing the

palimpsest of animal and hominin prints. C. Oblique image of part of the lower footprint surface and rear wall showing the

multiple flood cycles and the silting horizons upon which the animal and hominin prints are preserved.

11

Figure S3. Excavation map (July 2008) of FwJj14E showing the two footprint surfaces and indicating the location of each

footprint. Note that the contours are at 1 m elevation relative to an arbitrary field datum.

12

Figure S4. Close-up scans of print FUT1-1 illustrating different types of color render used to visualise the anatomical form of the

print. On the left a simple red-blue color scale is used and on the right an alternating black and white color scheme. In both

cases the isopleths represent 5 mm vertical slices.

Figure S5. Age of key tuffs within the Turkana Basin relevant to this study (S2).

13

Figure S6. A. Vertical photograph showing potential hominin prints FUT1-1 to FUT1-3. B. Vertical photograph of FUT1-2. C.

Vertical photograph of FUT1-3. D. Vertical photograph of FUT1-1.

14

Figure S7. Mosaic of scans viewed vertically showing potential hominin prints FUT1-1 to FUT1-3. The surface has been color

rendered in to 5 mm vertical isopleths.

15

Figure S8. Contour map generated in ArcGISTM from xyz data exported from RapidformTM for part of the Upper Surface showing

the hominin prints FUT1-1 to FUT1-3. The contour interval is 3 mm.

16

Figure S9. Scans of the isolated prints, and Trails 2 and 3 on the Upper Footprint Surface at FwJj14E. See Fig. S3 for location

of each print.

17

Figure S10. Scans of the prints forming Trail 1 on the Upper Footprint Surface at FwJj14E. See Fig. S3 for location of each

print.

18

Figure S11. Isometric views of selected prints on the Upper Footprint Surface at FwJj14E. A. Print FUT1-1. B. FUI8. C. FUI8

19

Figure S12. The Lower Footprint surface showing a range of animal prints and three hominin footprints. A. Lower Footprint

Surface North showing the prominent runnel dividing the two areas of the slab with different moisture contents and footprint

densities. B. Runnel on the Lower Footprint Surface, note the bird prints in the fore ground. C. Northern part of the Lower

Footprint Surface, with the scanner and its associated covered frame visible. D. – F. Typical animal prints on the Lower

Footprint Surface

20

Figure S13. Close-up scans of prints on the Lower Footprint Surface at FwJj14E

21

Figure S14. A. Map showing the area of GaJi10 lost to erosion since 1978/79 and the extension to the excavation made by the

authors in July 2008. The two prints highlighted in red were used in the statistical analysis. B. Original site plan for GaJi10

made by Behrensmeyer & Laporte (S1) in 1978/79. C. Stratigraphy as reported in Behrensmeyer & Laporte (S1). Note the

overlying tuff which has been dated to c. 1.435 Ma. D. Conventional photograph of print L1 at GaJi10. E. Conventional

photograph of print L3 at GaJi10. F. Conventional photograph of print R3 at GaJi10.

22

Figure S15. Hominin and animal footprints at GaJi10. A. Color rendered scan of Print R3 made from the original surface in July

2008. B. Color rendered scan of Print R3 made from the latex casts held in the National Museums of Kenya and made in

1978/79 from the original print. Note the linear, fold-like artefacts on the medial side of the print. C. A color rendered scan of

Print L3 made from the original surface in July 2008. D. A color rendered scan of Print R1 made from the original surface in July

2008. E. Color rendered scan of Print L1 made from the original surface in July 2008. F. Color rendered scan of Print R0

excavated by the authors in July 2008. G. Color rendered scan of a bird print at GaJi10 made from the original surface in July

2008.

Figure S16. A selection of the Laetoli prints scanned as part of this work. Note that both G2 and G3 prints have a similar

anatomical morphology to the G1 prints used in the analysis.

23

Figure S17. A selection of the Sefton Coast prints scanned as part of this work.

Figure S18. Anatomical print landmarks used in this work

Figure S19. Bivariate plots of key footprint dimensions. See Figure S18 for a key to landmarks. A. Landmarks B1 and B2 define

the width of the ball area, while the deepest point is recorded by the ball landmark. Prints that plot to the right have a ball which

is deepest in a medial direction; that is towards the first and second metatarsal heads. C. Length to ball width.

24

Figure S20. Thin plate splines and landmark vector maps comparing each of the footprint layers at FwJj14E and those at

GaJi10 with the control samples.

25

Figure S21. PCA on 2-dimensional coordinates using 13 landmarks. The first three components explain 60% of the variance.

Thin plate splines and vector maps show change in the direction of increasing PCA scores.

Figure S22. Plot of first two discriminant functions for Stage One of Analysis One.

26

Table S1. Geochemical profiles for the tuffs sampled within this study and the reference material (19)

JWKH-1 Upper Tuff at FwJj14E Sample SiO2 TiO2 ZrO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O F Cl Sum (-O) Total H2O Total JWKH 1 67.36 0.52 0.09 14.17 5.41 0.28 0.25 0.81 0.88 1.67 0.12 0.1 91.86 0.07 91.78 10.65 102.43 JWKH 1 68.16 0.33 0.15 12.85 4.77 0.24 0.1 0.56 0.58 0.77 0.13 0.14 88.86 0.08 88.77 12.02 100.79 JWKH 1 68.26 0.39 0.12 12.89 4.8 0.23 0.1 0.56 0.82 1.29 0.11 0.14 89.83 0.08 89.75 11.23 100.97 JWKH 1 69.13 0.36 0.1 12.92 4.86 0.24 0.11 0.5 0.79 1.37 0.16 0.16 90.83 0.1 90.73 10.91 101.65 JWKH 1 68.57 0.36 0.12 13.03 4.88 0.25 0.1 0.5 0.94 1.42 0.12 0.15 90.51 0.08 90.43 10.74 101.17 JWKH 1 68.94 0.39 0.14 12.81 4.91 0.26 0.09 0.53 0.62 0.93 0.2 0.14 90.06 0.12 89.94 11.32 101.26 JWKH 1 69.02 0.36 0.12 13.11 4.98 0.26 0.11 0.55 0.67 1.52 0.1 0.17 91.06 0.08 90.98 10.7 101.67 K00-6161 70.48 0.36 0 12.9 5.04 0.21 0.11 0.56 3 2.36 0.1 95.11 0.02 95.09 -0.57 94.51 N. Ileret K04-346 66.05 0.55 0.04 13.89 5.47 0.29 0.25 0.84 4.01 2.71 0.14 0.1 94.48 N. Ileret K04-346 69.5 0.35 0.07 12.84 4.98 0.24 0.1 0.51 4.33 2.99 0.14 0.14 96.23 N. Ileret K83-1566 68.84 0.35 0.1 12.66 4.84 0.24 0.1 0.55 3.21 2.6 0.11 0.14 93.8 0.08 93.73 6.33 100.06 N. Ileret K83-1663 69.03 0.37 0.12 12.9 5.02 0.24 0.11 0.56 2.14 2.39 0.28 0.14 93.34 0.15 93.19 5.38 98.58 N. Ileret Alpha_blue 69.58 0.38 0 12.93 4.91 0.25 0.11 0.55 2.44 2.47 0.33 0.13 94.09 0.17 93.92 7.91 101.83 N. Ileret IL03-064 69.71 0.37 0.11 13.05 4.9 0.23 0.11 0.57 2.78 2.37 0.21 0.14 94.58 0.12 94.46 5.4 99.86 N. Ileret

27

Table S1. Continued

JWKH-2 Middle Tuff at FwJj14E Sample SiO2 TiO2 ZrO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O F Cl Sum (-O) Total H2O Total JWKH 2 72.9 0.17 0.29 9.66 4.69 0.13 0.01 0.13 0.51 1.07 0.32 0.22 90.09 0.19 89.91 10.23 100.14 JWKH 2 68.7 0.4 0.15 13.08 4.96 0.23 0.11 0.54 1.2 2 0.13 0.13 91.72 0.09 91.63 10.32 101.95 JWKH 2 71.39 0.24 0.36 8.94 5.3 0.25 0.06 0.15 0.4 0.47 0.37 0.26 88.18 0.21 87.97 12.41 100.38 JWKH 3 72.21 0.29 0.31 9.19 5.4 0.24 0.05 0.13 0.5 0.54 0.28 0.23 89.4 0.17 89.23 10.91 100.14 JWKH 2 70.85 0.38 0.1 13.81 1.98 0.11 0.25 0.67 1.56 3.06 0.1 0.13 93.07 0.07 92.99 10.14 103.13 JWKH 2 69.53 0.42 0.11 13.9 2.22 0.12 0.28 0.79 1.03 2.11 0.2 0.11 90.89 0.11 90.78 11.03 101.81 JWKH 2 72.75 0.28 0.19 9.58 5.15 0.15 0 0.23 0.63 1.03 0.2 0.08 90.32 0.1 90.22 9.99 100.21 JWKH 2 73.13 0.34 0.16 9.91 4.47 0.18 0 0.19 0.35 1.01 0.14 0.08 89.96 0.08 89.88 11.34 101.22 JWKH 2 73.1 0.15 0.2 10.04 4.25 0.11 0.01 0.15 1.04 1.23 0.34 0.18 90.79 0.18 90.61 10.58 101.19 JWKH 2 69.05 0.5 0.1 14.59 2.7 0.17 0.37 1.09 1.67 3.02 0.08 0.06 93.5 0.05 93.46 9.86 103.32 Black Pumice K83-1637 71.42 0.16 0.31 9.47 4.62 0.13 0.01 0.13 4.72 4.29 0.43 0.2 95.89 0.22 95.67 3.73 99.4 Okote K03-0128 M3 72.06 0.19 0.33 9.61 4.79 0.11 0.01 0.14 1.89 3.83 0.63 0.22 93.79 0.31 93.48 6.76 100.24 K04-346 69.5 0.35 0.07 12.84 4.98 0.24 0.1 0.51 4.33 2.99 0.14 0.14 96.23 N. Ileret K83-1664 72.69 0.25 0.34 9.18 5.37 0.22 0.05 0.15 1.45 1.64 0.65 0.24 92.26 0.33 91.93 5.61 97.54 Ileret K85-2574 73.79 0.26 0.33 9.19 5.46 0.24 0.05 0.16 1.31 1.44 0.57 0.25 93.06 0.3 92.76 4.29 97.05 Ileret K82- 827 72.55 0.28 0.29 8.95 5.47 0.26 0.07 0.15 1.66 2.1 0 0.23 92.02 0.05 91.97 6.38 98.35 Ileret PNG00-ET047 73.02 0.27 0.3 9.29 5.49 0.24 0.06 0.16 1.49 1.45 0.42 0.23 92.44 0.23 92.21 8.59 100.8 Ileret K04-345 Av 72.13 0.26 0.27 9.12 5.51 0.24 0.06 0.16 2.74 1.83 0.26 0.23 92.8 Ileret PNG00-ET045 72.95 0.27 0.32 9.31 5.53 0.25 0.05 0.16 1.41 1.62 0.39 0.23 92.48 0.22 92.27 7.92 100.19 Ileret K00-6158 73.36 0.27 0.35 9.35 5.54 0.24 0.06 0.16 1.41 1.53 0.4 0.24 92.9 0.22 92.68 8.17 100.86 Ileret IL02-056 71.4 0.36 13.66 2.08 0.11 0.25 0.74 2.97 3.14 0.19 0.08 95.01 0.1 94.91 6.78 101.69 Ileret, Mode 3 K85-2565 72.61 0.3 0.17 9.36 4.98 0.2 0.01 0.23 3.06 2.21 0.27 0.08 93.5 0.13 93.36 6.28 99.64 K85-2560 74.28 0.32 0.14 9.67 5.03 0.2 0.01 0.22 2.32 3.08 0.43 0.09 95.8 0.2 95.6 4.35 99.95 K00-6157 74.3 0.33 9.67 5.07 0.17 0 0.24 2.55 2.43 0.08 94.84 0.02 94.82 -0.57 94.25 IL02-057 74.17 0.28 9.47 5.16 0.18 0.01 0.23 2.79 3.46 0.3 0.07 96.13 0.14 95.99 5.44 101.43 K02-0015 73.58 0.32 0.13 9.53 5.27 0.2 0.02 0.23 1.75 1.32 0.42 0.09 92.83 0.2 92.64 7.04 99.68 K02-0007 73.93 0.3 0.15 9.43 5.27 0.2 0 0.23 2.49 1.46 0.46 0.08 94 0.21 93.79 6.73 100.52 K88-3256 73.22 0.33 0.11 9.82 4.49 0.16 0.01 0.25 2.46 3.79 0 0.07 94.7 0.02 94.69 6.02 100.71 K88-3256 IL02-075 71.08 0.31 0.22 9.61 4.46 0.14 0.02 0.21 4.15 2.38 0.47 0.12 93.18 0.22 92.96 6.36 99.32 IL02-075 K83-1662 74.33 0.19 0.2 10.01 4.15 0.11 0.02 0.16 2.29 2.69 0.45 0.15 94.74 0.22 94.52 4.85 99.38 K83-1680 74.24 0.2 0.15 9.88 4.34 0.13 0.01 0.15 2.46 2.04 0.41 0.17 94.18 0.21 93.97 8.98 102.95 S. Ileret K84-1806 71.34 0.47 0.12 14.2 2.71 0.14 0.36 1.08 3.15 3.59 0.18 0.09 97.45 0.1 97.36 5.58 102.94 Black Pumice

28

Table S1. Continued

JWKH-4 Lower Tuff at FwJj14E Sample SiO2 TiO2 ZrO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O F Cl Sum (-O) Total H2O Total JWKH 4 71.75 0.2 0.43 8.91 5.75 0.25 0.04 0.15 0.75 0.69 0.3 0.27 89.47 0.19 89.29 10.12 99.41 Lower Ileret JWKH 4 71.64 0.28 0.36 9 5.83 0.25 0.03 0.12 1.29 0.97 0.39 0.29 90.49 0.23 90.26 10.92 101.18 Lower Ileret JWKH 4 71.97 0.29 0.16 9.65 4.54 0.17 0 0.21 0.62 0.58 0.23 0.08 88.52 0.11 88.4 11.19 99.59 JWKH 4 72.72 0.31 0.18 9.82 4.86 0.17 0 0.19 1.18 2.33 0.16 0.08 92.02 0.09 91.93 10.77 102.7 JWKH 4 73.06 0.3 0.15 9.77 5.02 0.2 0 0.24 0.9 1.52 0.19 0.08 91.49 0.1 91.39 9.5 100.89 JWKH 4 72.52 0.35 0.22 9.8 4.95 0.21 0 0.24 1.04 2.05 0.24 0.08 91.7 0.12 91.58 10.95 102.53 JWKH 4 72.6 0.25 0.16 9.71 4.87 0.22 0 0.2 1 1.63 0.17 0.08 90.93 0.09 90.84 10.2 101.04 JWKH 4 72.46 0.24 0.07 12.48 1.46 0.07 0.14 0.35 1.06 1.89 0.18 0.14 90.55 0.11 90.44 10.94 101.38 New comp. 1 JWKH 4 72.92 0.28 0.08 12.74 1.47 0.06 0.13 0.32 1.37 2.61 0.18 0.15 92.4 0.11 92.28 9.43 101.71 New comp. 1 JWKH 4 72.02 0.25 0.09 12.72 1.48 0.08 0.13 0.36 2.22 3.12 0.17 0.13 92.78 0.1 92.68 8.83 101.51 New comp. 1 JWKH 4 71.71 0.3 0.1 13.32 1.72 0.12 0.18 0.45 1.58 3.11 0.11 0.11 92.91 0.07 92.84 9.56 102.4 New comp. 2 JWKH 4 70.93 0.4 0.05 13.06 1.78 0.1 0.19 0.49 0.95 1.61 0.19 0.11 89.9 0.1 89.79 11.84 101.63 New comp. 2 JWKH 4 72.94 0.1 0.05 12.06 1.96 0.06 0.01 0.33 0.88 1.64 0.12 0.22 90.39 0.1 90.29 10.82 101.11 K77-20 71.04 0.22 0.38 8.87 5.82 0.26 0.04 0.14 1.71 1.6 0.37 0.26 90.72 0.22 90.51 7.93 98.43 Lower Ileret IL02-030 74.57 0.27 0.14 9.8 4.73 0.16 0.01 0.25 3.31 2.93 0.38 0.08 96.62 0.18 96.44 5.69 102.13 Elomaling'a IL03-020 73.09 0.3 0.13 9.72 4.72 0.16 0.01 0.25 2.46 2.59 0.36 0.08 93.87 0.17 93.7 7.76 101.46 Elomaling'a K86-2399 73.34 0.09 11.9 2.05 0.05 0.01 0.35 3.42 4.83 0.23 96.26 101.83 Orange

29

Table S1. Continued JWKH-5 Tuff above footprint layer at GaJi10 Sample SiO2 TiO2 ZrO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O F Cl Sum (-O) Total H2O Total JWKH 5 73.34 0.16 0.2 10.93 3.06 0.09 0.01 0.21 1.35 1.55 0.23 0.13 91.25 0.13 91.12 9.4 100.52 JWKH 5 73.1 0.17 0.19 10.96 3.07 0.08 0.02 0.2 1.14 1.9 0.27 0.14 91.22 0.14 91.08 9.3 100.38 JWKH 5 73.42 0.18 0.14 10.93 3.07 0.09 0.01 0.21 0.98 1.48 0.19 0.14 90.84 0.11 90.73 9.68 100.41 JWKH 5 72.3 0.3 0.15 8.86 5.75 0.22 0 0.2 0.31 0.5 0.11 0.08 88.78 0.06 88.72 10.91 99.63 JWKH 5 72.48 0.31 0.16 8.86 5.9 0.22 0.01 0.21 0.23 0.48 0.08 0.09 89.08 0.06 89.02 10.18 99.2 JWKH 5 72.05 0.29 0.2 8.99 5.91 0.2 0.01 0.22 0.41 0.64 0.23 0.12 89.25 0.12 89.13 10.82 99.95 JWKH 5 72.64 0.31 0.17 8.95 5.92 0.22 0 0.2 0.47 0.81 0.14 0.1 89.93 0.08 89.85 8.81 98.66 JWKH 5 72.91 0.32 0.14 8.98 5.93 0.2 0.01 0.2 0.27 0.54 0.17 0.12 89.83 0.1 89.73 9.52 99.25 K82- 768 74.83 0.17 0.15 10.69 3.07 0.09 0.01 0.18 4.05 4.27 0.17 0.12 97.82 0.1 97.72 3.96 101.68 Akait K82- 768LoFe-av 74.58 0.18 0.15 10.63 3.07 0.09 0.01 0.19 4.02 4.21 0.17 0.12 97.54 0.1 97.44 4.1 101.54 Akait K88-3263 73.54 0.17 0.14 10.68 3.1 0.09 0.02 0.2 3.11 4 0.18 0.14 95.36 0.11 95.26 4.8 100.06 Akait K88-3267X 73.94 0.17 0.13 10.63 2.96 0.09 0.01 0.21 3.15 4.42 0 0.13 95.85 0.03 95.82 5.07 100.89 Akait K89-3384 75.15 0.18 0.18 10.79 3.11 0.09 0.02 0.2 2.19 4.06 0.45 0.15 96.57 0.22 96.35 2.57 98.92 Akait K93-6013 74.23 0.15 0.14 10.93 3.11 0.08 0.02 0.21 3.65 2.55 0.21 0.14 95.41 0.12 95.29 6.39 101.69 Akait BT- 5 72.09 0.3 0.12 8.93 5.64 0.22 0.01 0.21 1.38 3.13 0.1 0.09 92.24 0.06 92.18 6.55 98.73 Karari Grey BT- 6 73.1 0.29 0.15 9.11 5.76 0.21 0.01 0.21 2.68 4.2 0.14 0.09 95.96 0.08 95.88 4.27 100.16 Karari Grey BT-10 71.97 0.33 0.12 8.95 5.75 0.19 0.01 0.2 1.01 2.99 0.13 0.1 91.75 0.08 91.68 6.91 98.59 Karari Grey K88-3265 72.23 0.28 0.12 9.22 5.71 0.19 0.01 0.2 2.42 3.44 0.21 0.1 94.15 0.11 94.04 7.01 101.04 Karari Grey K89-3395 73.52 0.32 0.17 9.16 5.76 0.2 0.01 0.21 1.82 3.14 0.48 0.1 94.88 0.22 94.66 5.56 100.22 K89-3336 73.74 0.29 0.2 8.95 5.99 0.21 0.01 0.22 0.83 2.07 0.44 0.12 93.07 0.21 92.85 4.55 97.41 Tuff at Naiyena Epul below Akait

30

N R R Square Adjusted R

Square Std. Error of the

Estimate

30 0.804 0.647 0.634 6.21

Unstandardized

Coefficients Standardized Coefficients

B Std. Error Beta t Sig.

Intercept 39.87 17.43 2.287 .03 Print Length (cm) 4.90 0.67 .804 7.28 .000

Table S2A. Summary of linear regression model between F-Length and subject height for a sample of 30 Dassenach. One case was excluded as a statistical outlier.

N R R Square Adjusted R

Square Std. Error of the

Estimate 30 0.696 0.484 0.466 6.10

Unstandardized

Coefficients Standardized Coefficients

B Std. Error Beta t Sig.

Intercept -34.60 17.13 -2.02 0.0527 Print Length (cm) 3.45 0.66 .696 5.21 0.00001

Table S2B. Summary of linear regression model between F-Length and subject mass for a sample of 30 Dassenach. One case was excluded as a statistical outlier.

Mass (kg) Stature (cm) Print L (cm)

Taxon Dates (Ma) Avg Female Male Avg Female Male -

Pan troglodytes Extant 45 41 49 - - - -

Paranthropus boisei 2.3-1.4 41.5 34 49 130.5 124 137 -

Homo habilis 1.9-1.4 34.5 32 37 115.5 100 131 -

Homo ergaster/erectus 1.9-1.5 61 56 66 170 160 180 -

Homo sapiens Extant 53.5 49 58 168 161 175 - Homo sapiens (Dassenach) Extant 55 50.2 59.8 167.9 160.0 175.8 25.9

Mass (kg) Stature (cm)

Print Estimate -95% PL +95%

PL Estimate -95% PL +95%

PL

FUT1 Average 1.52 56.5 43.8 69.2 169.3 156.4 182.2 26.4

FUT3 Average 1.52 56.1 43.4 68.8 168.8 155.9 181.7 26.3

FUI1 1.52 65.5 52.1 78.9 182.2 168.5 195.8 29.0

GaJi10 Average 1.43 53.7 41.0 66.4 165.4 152.5 178.3 25.6 Table S3. Mass and stature estimates of relevant early Pliocene hominin taxa based on postcranial fossil remains, and mass

and stature estimates of individuals who made the FwJj14E and GaJi10 footprints based on modern Dassenach sample.

Laetoli FwJj14E L FwJj14E U GaJi10 Homo sapiens (Formby)

Homo sapiens

(Modern) Laetoli √ √ x √ √

FwJj14E L 0.003 x x √ √ FwJj14E U 0.0001 0.921 x √ √

GaJi10 0.026 0.8 0.533 x x Homo sapiens

(Formby) 0.0001 0.003 0.0001 0.007 x

Homo sapiens (Modern)

0.0001 0.003 0.0001 0.014 0.83

Table S4. Post hoc pair-wise Mann-Whitney U tests for abduction angle (p<=0.003 Significant difference between medians)

31

A. Stage 1 B. Stage 2 Function Function 1 2 3 1 Abduction angle -1.129 .048 .110 1.381 H1-H2 -.359 .882 .280 .297 D1-I1 .530 .067 -.277 -.724 B1-H2 .508 .646 .443 -.450 D1-D2 .498 .378 .364 .462 B1-Ball1 -.256 .152 -.810 -.664 D1-H1 -.167 .530 -.421 -.924 D2-Hdth .659 .236 .733 .646

Table S5. A. Standardised canonical discriminant function coefficients for Stage One. B. Standardised canonical discriminant function coefficients for Stage Two

Locality Print Analysis One FwJj14E FLI1 Laetoli FwJj14E FLT1-1 H. sapiens (Modern) FwJj14E FLT1-2 H. sapiens (Modern) FwJj14E FUT1-1 H. sapiens (Modern) FwJj14E FUT1-2 Laetoli FwJj14E FUI1 H. sapiens (Modern) FwJj14E FUT1-6 H. sapiens (Modern) FwJj14E FUT1-7 H. sapiens (Modern) FwJj14E FUT3-2 H. sapiens (Modern) FwJj14E FUT3-1 H. sapiens (Modern) GaJi10 L3 Laetoli GaJi10 R3 Laetoli

Table S6. Results from Stage Two showing how the prints at FwJj14E and GaJi10 classify against the two end members Laetoli versus H. sapiens (Modern)

Supplementary References and Notes S1. A.K. Behrensmeyer, L.F. Laporte, Nature 289, 167-169 (1981).

S2. F.H. Brown, B. Haileab, I. McDougall, J. Geol. Soc. Lond. 163, 185-204 (2006).

S3. McDougall, F.H. Brown, J. Geol. Soc. Lond. 163, 205-220 (2006).

S4. M.D. Leakey, R.L. Hay, Nature 278, 317-323 (1979)

S5. M.D. Leakey, in Laetoli: A Pliocene Site in Northern Tanzania, M.D. Leakey, J.M. Harris, Eds.

(Clarendon Press, Oxford, 1987), pp. 490-523.

S6. T.D. White, G. Suwa, G. Am. J. Phys. Anthropol. 72, 485-514 (1987).

S7. M.H. Day, E.H. Wickens, Nature 286, 385-387 (1980).

S8. R.H. Tuttle, D. Webb, E. Weidle, M. Baksh, J. Arch. Sci. 17, 347-362 (1990).

S9. J. Charteris, J.J. Wall, J.W. Nottrodt, Am. J. Phys. Anthropol. 58, 133-144 (1982).

S10. D.J. Meldrum, in From Biped to Strider: The Emergence of Modern Human Walking, Running,

and Resource Transport. D.J. Meldrum, C.E. Hilton, Eds. (Kluwer Academic/Plenum Publishers,

New York 2004), pp. 63-83.

S11. J.T Stern, R.L. Susman, Am J. Phys. Anthropol. 60, 279-317 (1983).

S12. Y. Rak, J. hum. Evol. 20: 283-291 (1991).

S13. D. A. Raichlen, H. Pontzer, M.D. Sockol, J. hum. Evol. 54: 112-117 (2008).

S14. C.B. Ruff, A. Walker, in The Nariokotome Homo erectus Skeleton. A.C. Walker, R.E.F. Leakey,

(Harvard Univ Press, Cambridge 1993), pp. 234-265.

S15. N.L. Griffin, B.Wood, in The Human Foot: A companion to clinical studies, Eds. L. Klenerman,

B. Wood, Eds. (Springer-Verlag, London, 2006), pp. 27-79.

S16. G. Roberts, S. Gonzalez, D. Huddart, Antiquity 70, 647-651 (1996).

32

S17. C. Berge, X. Penin, E. Pelle, Comptes. Rendus. Palevol. 5, 561-569 (2006).

S18. M.L. Zelditch, D.L. Swiderski, H.D. Sheets, W.L. Fink, Geometric Morphometrics for Biologists:

a Primer. (Elsevier Academic Press, London 2004).

S19. J. Niu, Z. Li, G. Salvendy, in Digital Human Modeling, HCII 2007, LNCS 4561. V.G. Duffy,

(Springer-Verlag Berlin Heidelberg, 2007), pp. 161-170.

S20. L.M. Robbins, Footprints. Collection, analysis and interpretation. (Charles C. Thomas

Publisher, Springfield 1985).

S21. See a collection of papers and reviews within a special issue on footprints in Ichnos 15 (2008).

S22. L.R. Brand, J. Paleon. 70, 1004-1010 (1996).

S23. J. Milàn, Palaeontology 49, 405-420.

S24. J.R.L. Allen, Phil. Trans. R. Soc. Lond. B 352, 481-518 (1997).

S25. C.B. Ruff, A. Walker, in The Nariokotome Homo erectus Skeleton. A.C. Walker, R.E.F. Leakey,

(Harvard Univ Press, Cambridge 1993), pp. 234-265.

S26. H.M. McHenry, K. Coffing, K. Annu. Rev. Anthropol. 29, 125-146 (2000).

S27. F. Spoor et al., Nature 448, 688-691 (2007).