Reevaluating bipedalism in Danuvius

arising from M. Böhme et al. Nature https://doi.org/10.1038/s41586-019-1731-0 (2019)

Danuvius guggenmosi is a species of Miocene hominoid from the 11.62-million-year-old site of Hammerschmiede. On the basis of interpretations of its vertebrae and limbs, Böhme and colleagues¹ infer that Danuvius exhibited ‘joint positions and loading patterns of both hominin bipedalism that emphasize hindlimb extension and spinal curvatures, and extant great ape suspension’. Although we agree that Danuvius was suspensory, we find the functional interpretation of bipedalism to be unfounded on morphological grounds. We therefore call into question the evolutionary scenario for the origin of hominin bipedalism proposed by Böhme and colleagues.

On the basis of differences in the orientation of the spinous process (41°) between purported ‘first’ and ‘lower’ thoracic vertebrae, Böhme and colleagues infer biped-like cervical lordosis and thoracic kyphosis for the upper spine of Danuvius. However, their comparative data (drawn from ref. ²) are misleading, because they represent a substantially higher thoracic level (T7 in humans and T8 in chimpanzees) than is represented by the Danuvius specimen GPIT/MA/10000-16 (described as the ‘penultimate or ante-penultimate [thoracic] position’, in the supplementary information of Böhme et al.¹). When the relevant comparative data² are used, only minor differences in the inclination of the spinous process are found (T1:ante-penultimate thoracic, 8.4° in humans and −0.1° in chimpanzees; T1:penultimate thoracic, 4.2° in humans and 3.2° in chimpanzees). Danuvius does not resemble humans or chimpanzees in this metric, although it does overlap with some gorillas and orangutans (Fig. 1). Moreover, in the absence of mid-thoracic or lumbar vertebrae, claims regarding the spinal curvature and lumbar lordosis¹ of Danuvius are unsubstantiated.

Fig. 1: Lower thoracic vertebra of Danuvius (GPIT/MA/10000-16) in comparative context.

a, The cranially oriented costal facets and ‘rod-like’ laminapophysis of GPIT/MA/10000-16 (colour) (from ref. ¹) can be found on the last thoracic vertebrae of great-ape specimens (for example, on a gorilla (grey)) (not to scale). Therefore, the contention that GPIT/MA/10000-16 cannot be the ultimate or penultimate thoracic vertebra on the basis of the presence of these morphologies is incorrect. b, Angles of the spinous process of upper and lower (ultimate, penultimate and ante-penultimate) thoracic vertebrae of a range of hominoids: Pongo (n = 7), Gorilla (n = 11), Pan (n = 37) and modern humans (n = 30). Danuvius falls near some of the specimens of Gorilla and Pongo, which indicates that it is neither unique nor humanlike.

The interpretation of a long lower back in Danuvius¹ is based not on a series of vertebrae, but rather on a single lower thoracic vertebra. On the basis of a well-developed costotransverse facet, Böhme and colleagues¹ argue for ‘a non-ultimate thoracic position for the diaphragmatic vertebra and therefore a functionally longer lower back, as in early hominins, stem-hominoids and cercopithecids.’ Although the last thoracic vertebra in humans lacks costotransverse facets, they are in fact commonly present in great apes (in 30 out of 42 specimens examined; personal observations of S.A.W.) and—in some cases—are large and cranioventrally directed (contra ref. ¹) (Fig. 1). Therefore, the presence of a costotransverse facet on a vertebra does not preclude its identification as an ultimate thoracic vertebra (Fig. 1). Moreover, the position of the diaphragmatic vertebra does not directly relate to the length of the lumbar column or to lumbar curvature. All extant hominoid species demonstrate some frequency of cranial displacement of the diaphragmatic vertebra relative to the last thoracic vertebra—approximately 33% of many hominid species, and up to 55% in hylobatids—yet do not have long lumbar columns³. Similarly, atelines, which converge with hominoids on lower back morphology related to suspensory behaviour⁴, exhibit similar frequencies of cranial displacement and possess short lumbar columns (Fig. 2). Stem hominoids possessed six lumbar vertebrae and cranial displacement by one to two elements and are therefore considered long-backed⁵, whereas Oreopithecus bambolii possessed five lumbar vertebrae and demonstrates cranial displacement by at least one element⁶ (Fig. 2). As with Oreopithecus⁶, Danuvius may have had an ‘intermediate’ lower back similar to that of hylobatids rather than a long, monkey-like lower back or a short lower back that recalls those of the extant great apes (Fig. 2). Regardless, neither the morphology of GPIT/MA/10000-16 nor its potential position in the vertebral column indicate the length of the lumbar column or suggest adaptation to bipedal posture or locomotion.

Fig. 2: Evolution of vertebral formulae in anthropoids.

a, Regional numbers of thoracic (blue or purple squares, starting at T9 (vertebra 16)) and lumbar (red squares) vertebrae are shown, along with the modal diaphragmatic vertebra (purple squares, with the frequencies listed). Hypothesized ancestral patterns of lower back (lumbar column) length are indicated. LCA, last common ancestor. b, In the IGF11778 Oreopithecus skeleton, two additional lumbar vertebrae (L4 and L5) are entrapped between the iliac blades of the pelvis⁶. Additionally, the last thoracic vertebra is post-diaphragmatic, as evidenced by a cup-shaped, sagittally oriented superior articular facet with a mammillary process lateral to it; this indicates that the specimen is characterized by cranial displacement.

Böhme et al.¹ argue that Danuvius had a valgus knee and hominin-like hip abductor mechanics that were associated with extended-limb arboreal bipedalism. They suggested that ‘the more medial position of the lesser trochanter may result in a more exclusively extension function of iliopsoas, particularly if the ilium were rotated laterally on the hip joint’ (supplementary information of Böhme et al.¹). Yet, given its anatomical position, the iliopsoas functions exclusively as a hip flexor and lateral rotator of the thigh, with no contribution to hip extension⁷. Instead, a more medially positioned lesser trochanter may further diminish the ability of the iliopsoas to contribute to lateral rotation of the thigh, which has no clear connection to bipedalism. Additionally, Böhme et al.¹ infer that the ilium was more ‘inferolaterally’ oriented in Danuvius on the basis of a high femoral neck-shaft angle and a posterosuperior expansion of the articular surface of the femoral head, implying an increased hip abductor function of the lesser gluteal muscles. However, as shown by Böhme et al.¹, neither of these traits is unique to bipeds. Notably, the Danuvius femur appears to lack the elongated femoral neck (figure 1 in ref. ¹) that is characteristic of bipeds, and which increases the internal moment arm of the lesser gluteal muscles to counteract external moments at the hip during the single support phase of the gait cycle⁸.

A tibia with a damaged diaphysis (GPIT/MA/10000-15) is central to arguments for an extended lower limb and bipedalism in Danuvius, as it purportedly displays a hominin-like, relatively large and anteroposteriorly flattened lateral condyle with a ‘buttressing of the tibial metaphysis’¹, combined with a talocrural joint oriented orthogonally to the diaphyseal long axis. However, the analysis of the tibial condyle shape performed by Böhme et al.¹ is preliminary and includes only eight individuals and seven species, which precludes statistical tests of taxon or locomotor group differences. We agree that the morphology of the proximal tibial metaphysis could reflect knee-joint loading regimes associated with various locomotor and postural modes, but Böhme and colleagues¹ do not provide comparative data to support their claim that the tibial metaphysis of Danuvius is expanded relative to those of apes. Moreover, the analysis of the surface area of the tibial plateau relative to tibial length shows Danuvius to be most similar to Pan and Pongo (extended data figure 3 in ref. ¹). The inference that Danuvius habitually loaded its proximal tibia in extended-knee bipedalism on the basis of comparisons of proximal tibia morphology is therefore currently unsubstantiated.

As noted by Böhme and colleagues¹, the anteroposteriorly thin patella of Danuvius resembles those of extant great apes and Miocene hominoids such as Pierolapithecus. The relatively thin patellae of great apes reflect the use of varied knee positions during orthograde climbing and suspension, including extended positions. By contrast, the thicker patellae of cercopithecoids are associated with the generation of higher-magnitude knee extension moments from more-habitually flexed positions as pronograde quadrupeds⁹. However, patellar thickness does not distinguish among living great apes⁹. Although we agree that the Danuvius patella is anteroposteriorly thin and great-ape-like, its morphology cannot therefore support the conclusion that Danuvius used ‘slow and deliberate movements, most similar to Pongo’ (supplementary information of Böhme et al.¹). In addition, the great-ape-like patella of Danuvius reduces the moment arm of the quadriceps at the knee¹⁰, which diminishes the ability of the quadriceps to counteract sagittal-plane moments that flex the knee in the early part of the stance phase of the bipedal gait cycle¹¹. Finally, the damaged tibial diaphysis and distal metaphysis preclude accurate measurement of the frontal-plane angle of the talocrural joint. The intact sections of the tibial diaphysis clearly indicate frontal-plane curvature, particularly along the lateral border of the midshaft and below—potentially resulting in a more obliquely oriented talocrural joint, which is characteristic of African apes¹². The lower limb of Danuvius shares morphometric affinities with great apes that are consistent with a positional repertoire that included orthogrady and suspension, but the evidence for bipedalism is equivocal.

In summary, Danuvius lacks features associated with bipedal posture and locomotion. Its preserved morphology appears to reflect the increased limb mobility and powerful hallucal grasping that are expected to characterize a relatively large-bodied, tailless arboreal ape¹³. The discovery of Danuvius substantially contributes to our understanding of hominoid evolution, but relevant comparative data do not support the hypothesis of Böhme et al.¹ that the last common ancestor of humans and chimpanzees was a long-backed, lordotic and arboreal biped^3,14,15.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.