Footprint evidence of early hominin locomotor diversity at Laetoli, Tanzania Ellison J. McNutt orcid.org/0000-0002-7723-91031,2, Kevin G. Hatala orcid.org/0000-0001-9131-53043, Catherine Miller4,5, James Adams orcid.org/0000-0002-0008-78776,7, Jesse Casana orcid.org/0000-0002-4445-18024,5, Andrew S. Deane8, Nathaniel J. Dominy orcid.org/0000-0001-5916-418X4,5, Kallisti Fabian9, Luke D. Fannin orcid.org/0000-0003-4810-44424,5, Stephen Gaughan7, Simone V. Gill10, Josephat Gurtu9, Ellie Gustafson11, Austin C. Hill orcid.org/0000-0002-8397-81055,12, Camille Johnson5, Said Kallindo9, Benjamin Kilham13, Phoebe Kilham13, Elizabeth Kim2, Cynthia Liutkus-Pierce14, Blaine Maley15, Anjali Prabhat5, John Reader16, Shirley Rubin orcid.org/0000-0002-0520-207X17, Nathan E. Thompson orcid.org/0000-0002-9273-363618, Rebeca Thornburg11, Erin Marie Williams-Hatala3, Brian Zimmer14, Charles M. Musiba orcid.org/0000-0002-4297-517811,19,20 & Jeremy M. DeSilva4,5,19 Nature (2021)Cite this article ArchaeologyBiological anthropology Bipedal trackways discovered in 1978 at Laetoli site G, Tanzania and dated to 3.66 million years ago are widely accepted as the oldest unequivocal evidence of obligate bipedalism in the human lineage1,2,3. Another trackway discovered two years earlier at nearby site A was partially excavated and attributed to a hominin, but curious affinities with bears (ursids) marginalized its importance to the paleoanthropological community, and the location of these footprints fell into obscurity3,4,5. In 2019, we located, excavated and cleaned the site A trackway, producing a digital archive using 3D photogrammetry and laser scanning. Here we compare the footprints at this site with those of American black bears, chimpanzees and humans, and we show that they resemble those of hominins more than ursids. In fact, the narrow step width corroborates the original interpretation of a small, cross-stepping bipedal hominin. However, the inferred foot proportions, gait parameters and 3D morphologies of footprints at site A are readily distinguished from those at site G, indicating that a minimum of two hominin taxa with different feet and gaits coexisted at Laetoli. In 1976, Peter Jones and Philip Leakey discovered five consecutive bipedal footprints at Laetoli site A within locality 7, a 490 m2 area dated to 3.66 million years ago (Ma) and featuring 18,400 animal tracks1,2,3 (Fig. 1). Mary Leakey tentatively suggested that the trackway was made by a hominin1. ‘The footprints,’ she wrote, ‘indicate a rolling and probably slow-moving gait, with the hips swivelling at each step, as opposed to the free-striding gait of modern man [humans].’ Leakey and Hay2 classified the footprints as Hominidae, but with a caveat that ‘the gait was somewhat shambling, with one foot crossing in front of the other.’Fig. 1: Laetoli location and site rediscovery.a, A model of site A generated using photogrammetry showing the five hominin footprints. b, Corresponding contour map of the site generated from a 3D surface scan with scale bar. c, Map of Laetoli localities 7 and 8, indicating the positions of bipedal trackways A, G and S (redrawn from ref. 49). d, e, Topographical maps of the two best preserved A footprints, A2 (d) and A3 (e).Unequivocal hominin footprints were discovered at site G two years later, casting doubt on the hominin status of those at site A3,4,5. Researchers described the footprints at site A as ‘most unusual,’6 ‘curiously shaped,’7 and ‘enigmatic,’8 and yet consensus was uniform: they were produced by a plantigrade mammal moving bipedally.Tuttle4 advanced three hypotheses to account for the morphology of the footprints and cross-stepping gait (that is, when a foot from each side crosses the midline before touchdown): (1) substrate distortion; (2) they were left by a juvenile bear (ursid); or (3) they are evidence of another hominin species. To test the second possibility, Tuttle4,9,10,11,12 collected data from circus bears trained to walk bipedally and found that their short steps and relatively wide feet were a close match to the site A footprints, although bipedal bears take wider steps. Furthermore, the fifth digit is typically the largest in ursids, solving the ‘cross-stepping problem’, although Tuttle4,12 noted that humans do occasionally cross-step. He concluded that ‘until detailed, naturalistic biometric and kinesiological studies are performed on bipedal bears and barefoot humans, we will have to defer choosing among the hominid and ursid hypotheses on Laetoli individual A’4.Complicating matters further, the internal morphology of the site A footprints was never fully cleaned of matrix infill1,2,4,5,12. White and Suwa8 argued that ‘reliable identification of these enigmatic prints at Laetoli site A will be impossible until they are more fully cleaned and followed laterally’. Accordingly, we were motivated to relocate and re-excavate site A and conduct a detailed comparative analysis of the prints as well as the locomotion of bears (Ursus americanus), chimpanzees (Pan troglodytes) and humans to test whether the footprints at site A were left by a hominin or an ursid.Using detailed maps from Leakey and Harris3, we identified the proboscidean trail adjacent to the bipedal footprints. We cleared the surrounding overburden until one of us (K.F.) found the A3 footprint. The area was then brushed clean to expose A1–A5, which have experienced no discernible erosion since their initial discovery (Fig. 1, Extended Data Fig. 1). Because the footprint tuff is eroded to the north, we excavated south (87 cm) and east (54 cm) from the heel of A1, but no additional footprints were found (Supplementary Information).After brushing sediment from A3, we used a wooden tongue depressor to remove tuff infill left intact during the 1976–1978 field seasons. The hallucial impression is clearly defined and is about 30 mm wide. Crucially, we exposed the impression of the second digit (Extended Data Fig. 2). We removed infill from A2 but could not do so completely without risking damage. Nevertheless, the heel and hallucial impression are clear. Detailed information from the other footprints (A1, A4 and A5) is limited to estimates of length, width and step length.Although preservation quality varies within and between A1–A5, there is no evidence that biologically informative metrics were affected by substrate distortion. Adjacent and comingled tracks of other animals (ranging in size from guinea fowl to elephants) show no evidence of distortion to their perimeters or internal morphologies. Given that track surfaces are likely to represent time scales of hours to days13,14,15, it is parsimonious to infer similar substrate conditions and taphonomic processes during print formation and subsequent epochs.We recorded 50.9 h of video of wild American black bear behaviour. Unsupported bipedal posture and locomotion occurred only 0.09% of the total observation time, of which 59% was postural and 41% was locomotor (Extended Data Fig. 3). In only one instance did a bear take four unassisted bipedal steps. Thus, assuming that our findings are generalizable to other ursids, the probability of observing four consecutive bipedal steps is 0.003%. The low frequency of this behaviour, and the absence of quadrupedal–bipedal transitional footsteps, makes it unlikely, but not impossible, that ursid bipedalism was preserved at site A. Further, Laetoli is devoid of ursid fossils despite the recovery of more than 25,000 fossils attributed to 85 mammalian species16,17. If present at all, ursids were rare on the landscape. Although footprint assemblages can include a surprising number of tracks from taxa whose skeletal fossils are rare (for example, relatively high frequencies of bird tracks at Laetoli3 and at 1.5-Ma sites near Ileret, Kenya14,18), there is no clear taphonomic explanation for why ursid tracks would be present but their fossils absent.In addition, we measured 46 footprints from four bipedally walking wild juvenile black bears specifically chosen because their foot lengths (mean = 145.7 mm) were within 10% of the length of the site A footprints (mean = 161.7 mm). Additionally, we measured the footprints of chimpanzees produced during quadrupedalism (n = 54 from 46 adults; Ngamba Island Chimpanzee Sanctuary, Uganda) and during bipedalism (n = 44 from two subadults; Stony Brook University, USA). We compared these data with human barefoot footprints produced under three conditions: (1) habitually shod (n = 654) walking on a plantar pressure mat19; (2) habitually unshod or minimally shod (n = 41) walking in deformable mud20,21; and (3) Late Pleistocene tracks (n = 113) from Engare Sero, Tanzania, formed in reworked volcanic ash22,23 (summary in Extended Data Table 1).We concur with others4,12 that the ratios of footprint dimensions (for example, heel and forefoot width) to step length observed at site A fall within the ursid range (Extended Data Fig. 4a, b). Yet, for these same measures, site A is also chimpanzee-like and moderately similar to definitive hominin footprints from sites G and S. It follows that the site A individual was taking short steps—as occurs when humans walk slowly or over a slippery substrate24—not that the gait was ursid-like.With additional infill removed from A2 and A3, the perimeter dimensions are decidedly hominin-like with wide heel impressions relative to forefoot width (Extended Data Fig. 4c). By contrast, chimpanzees and bears have relatively narrow heels. Furthermore, with the tracks fully excavated and cleaned, we found no evidence for claw impressions, although they are sometimes absent from ursid footprints12. Here, impressions were absent from 31% of ursid footprints. To test whether A3 was produced by a hominin left foot or an ursid right foot, we compared the width of the hallux to the second digit in human (n = 30) and chimpanzee (n = 50) footprints, and the fifth digit to the fourth in bear (n = 5) tracks. The A3 toe impressions match the distinctive proportions of humans and chimpanzees rather than those of bears (Extended Data Fig. 5).By establishing that A3 is a left hominin foot, we can now confirm that cross-stepping occurred. Cross-stepping was never observed in our comparative sample, but humans do it occasionally25,26 as a compensatory strategy for re-establishing balance after a perturbation27. In fact, we suggest that cross-stepping supports the hypothesis that the site A footprints were left by a hominin. Cross-stepping is improbable, and perhaps impossible, when bears or chimpanzees walk bipedally. They produce large mediolateral excursions of their centre of mass28 and walk with highly abducted hips29, resulting in a high ratio between stride width and step length. Conversely, human cross-stepping is enabled by their reduced mediolateral centre of mass and body motions, adducted hips and bicondylar angle (that is, valgus knees), resulting in a low corresponding ratio, as expressed in every trackway at Laetoli.The relative step widths of footprints from sites G and S fall squarely within the modern human distribution (Extended Data Figs. 4d, 5). The site A footprints lie outside the distributions of humans, chimpanzees and bears but are most like humans. This result indicates that the maker of the site A footprints had either valgus knees or adducted hips, or both. The presence of either bipedal trait argues for a hominin maker of the site A footprints.It is generally accepted that Australopithecus afarensis produced the footprints at sites G and S8 (but see refs. 12,30,31,32). It is thus tempting to assign the site A tracks to A. afarensis; however, this premise requires an examination of foot ontogeny and intraspecific morphological variation that takes into account the mounting fossil evidence of locomotor (and presumably taxonomic) diversity among Pliocene hominins (for example, in ref. 33).Standing between 101 and 104 cm tall (from equations in Dingwall et al.34), the maker of the site A footprints was smaller in height than other Laetoli trackmakers, which ranged from 111–116 cm (site G1) to 161–168 cm (site S1)35. It is plausible that the site A tracks were made by a juvenile A. afarensis, but this hypothesis is undermined by a distinct footprint morphology from those at sites G or S.The ratio of foot width and length follows a different ontogenetic trajectory in humans and chimpanzees; human feet are consistently narrower than chimpanzee feet (Fig. 2). Footprints made by unshod humans from both modern times and the Pleistocene are slightly wider than those made by shod humans in industrialized populations. The undistorted footprints from sites G and S fall within the human distribution. A3 is more chimpanzee-like in being wide compared with its length (Fig. 2a, Extended Data Fig. 2). In chimpanzees, this wider footprint shape is, in part, driven by the greater divergence of the hallux. We thus measured hallucial divergence as a ratio of the distance between the centre of the impression made by the first and second digits and the length of the footprint. For this metric, humans and chimpanzees are clearly distinct. The best-defined site G footprints overlap the human distribution, whereas the A3 footprint does not—it possesses a slightly more divergent hallux than humans and site G, although not nearly as divergent as those of chimpanzees (Extended Data Fig. 6). This finding alone does not rule out a juvenile A. afarensis, given the foot from Dikika (Afar, Ethiopia) has a slightly more divergent and mobile hallux than its adult counterparts36.Fig. 2: Evidence against Laetoli A belonging to A. afarensis.a, Images of Laetoli A3 (top) length matched to Laetoli G1 (bottom). G1 print is reproduced with permission from Science Photo Library. b, Plot comparing foot length to forefoot width in adult and juvenile humans and bipedal chimpanzees, the Pleistocene human footprints at Engare Sero, and the Laetoli trackways. Lines represent ordinary least-squares regression and shaded bands represent 95% confidence interval of the regression. Data were analysed from the total number of individual footprints indicated in the figure legend. Laetoli footprint markers are slightly enlarged for clarity. Data sources match Extended Data Table 1. c, Histogram of Mahalanobis distances between the mean modern human footprint and the averages of two randomly drawn human footprints (grey; n = 245, resampled 1,000 times) and two randomly drawn bipedal chimpanzee footprints (yellow; n = 45, resampled 1,000 times). The blue and orange lines represent the distances of mean Laetoli G1 (n = 5, with 10 unique two-track combinations) and S1 (n = 2) two-track samples, respectively. All Laetoli G1 and S1 samples fall within the human distribution. The green line indicates the distance between the mean Laetoli A track (n = 2) and the human mean, falling squarely within the chimpanzee di
https://www.nature.com/articles/s41586-021-04187-7
Footprint evidence of early hominin locomotor diversity at Laetoli, Tanzania
