The holotype maxilla (BRT-VP-3/1) has a more anteriorly positioned zygomatic root compared to Au. anamensis and most of the known Au. afarensis maxillae. However, it overlaps with most early hominins in the degree of subnasal prognathism (as determined by the subnasal angle) and palate depth3,4,8 (see Extended Data Table 2). Although the position of prosthion on BRT-VP-3/1 cannot be determined with certainty, the maximum estimated subnasal angle of this specimen is ∼ 39°, falling within the range of most early hominins but considerably less than the angle reported for KNM-WT 40000, the holotype of K. platyops (47°)4.

The incisive foramen in BRT-VP-3/1 is positioned at the distal bi-canine line, whereas in Au. afarensis maxillae of comparable size (for example, A.L. 199-1), and early Homo specimens such as A.L. 666-1, the foramen is positioned posterior to this line, or at the P3 level8. The palate anterior to the incisive foramen is also flexed inferiorly (see Extended Data Fig. 1i, k, l), a feature considered as derived although it is reported to be variable in Au. afarensis9. BRT-VP-3/1 shows very little sagittal overlap between the hard palate and the nasoalveolar clivus (3.5 mm) unlike most Au. afarensis maxillae (see Extended Data Fig. 4).

Further differences between Au. deyiremeda and Au. afarensis relate to the morphology of the upper dentition. The cheek teeth of Au. afarensis tend to exhibit roughly vertical buccal and lingual walls10, whereas the buccal and lingual walls of the Au. deyiremeda premolars and molars converge occlusally, making the occlusal surface narrower than the base of the crown. The P3 and P4 of BRT-VP-3/1 have three roots (Extended Data Fig. 1h, j). This is commonly seen in Paranthopus species, but highly variable in Australopithecus. Kenyanthropus platyops also has three-rooted upper premolars although P3–P4 crown morphology is currently unknown for this species3.

The canine crown of BRT-VP-3/1 is absolutely smaller than almost all known Australopithecus upper canines, specifically in its mesiodistal dimension, which falls close to the mean of P. robustus and P. boisei (Extended Data Table 2). It also appears to be long-rooted relative to its crown height, as in Au. anamensis11,12,13. Furthermore, the canine of BRT-VP-3/1 does not have pronounced lingual relief and entirely lacks the prominent mesial and distal vertical lingual grooves commonly seen in Au. afarensis upper canines10. Its distal tubercle is also small and the mesial crown shoulder relatively short compared to the crown height and length of the mesial crest, thus it is less incisiform than Au. afarensis upper canines. It retains a more primitive, Au. anamensis-like condition in this regard.

The postcanine teeth of BRT-VP-3/1, particularly the P4 and M1, are very small (Fig. 4; see Extended Data Table 2 for summary statistics). The M1 area of BRT-VP-3/1 is the smallest of all known Pliocene hominins (less Ar. ramidus), whereas the P4 size is only approximated by A.L. 199-1, the smallest outlier in the Au. afarensis hypodigm. The M2 also falls at the lower end of the Au. afarensis range and is slightly larger than that of KNM-WT 40000, the holotype of K. platyops. The P4 to M1 area ratio of BRT-VP-3/1 falls within the range of most Pliocene hominin taxa, but outside the range of Paranthropus. In terms of enamel thickness, the two-dimensional linear enamel thickness for the M1 of BRT-VP-3/1 is similar to that of other early hominins such as Au. afarensis, P. robustus and Au. africanus. However, the two-dimensional and three-dimensional enamel thickness values for the M2 of this specimen are high, most similar to P. robustus (see Supplementary Note 4 and Extended Data Tables 3 and 4 for details).

Figure 4: Box-and-whisker diagrams comparing dental dimensions of the holotype BRT-VP-3/1 with other early hominins. Square roots of P4–M2 areas (buccolingual × mesiodistal dimensions) are plotted. The horizontal edges of each box indicate the first quartile (Q 1 ), the median (Q 2 ), and the third quartile (Q 3 ). The superiormost and inferiormost horizontal lines show the maximum and minimum values, respectively, excluding outliers, which are shown above or below these horizontal lines. Metric data for the comparative samples were compiled from refs 9 and 25, 28, 29, 30, 31, 32, 33. Full size image Download PowerPoint slide

Although Au. deyiremeda and K. platyops show some similarities in maxillary morphology and molar size, they can be distinguished by a suite of qualitative characters related to the morphology of the nasoalveolar clivus, shape of the dental arcade, and morphology of the anterior maxillary dentition. The nasoalveolar clivus in K. platyops is flat both sagittally and transversely3, whereas in Au. deyiremeda it is anteriorly convex in the transverse plane. In accordance with this difference, Au. deyiremeda and K. platyops differ in the shape of the anterior dental arcade. On the basis of the position of the preserved I1 root, the transversely oblique orientation of the I2 crown, and its placement relative to the canine and postcanine row, it is clear that the canine and incisors of BRT-VP-3/1 formed an arc with the incisor alveoli well anterior to the bi-canine line, as in Au. afarensis specimens such as A.L. 200-1a and A.L. 444-2 (ref. 9; see Extended Data Fig. 1f, g). In K. platyops, the incisors are arranged parallel to the bi-canine line as in Paranthropus and Homo rudolfensis3. Kenyanthropus platyops also has a derived low and curved zygomaticoalveolar crest3, whereas this feature is not sufficiently developed to form a visible crest in BRT-VP-3/1. Instead, the lateral alveolar wall blends smoothly with the base of the zygomatic bone. Furthermore, computed tomography scans show that the incisor roots of BRT-VP-3/1 are straight and inclined posterosuperiorly, indicating procumbent incisors (Extended Data Fig. 1i, k). By contrast, the incisor roots of K. platyops do not show any sign of procumbency3. On the basis of root cross-section approximately 2 mm below the buccal cervicoenamel line of I2, the I1 root of BRT-VP-3/1 is also rounded and more robust than the mesiolaterally compressed I2 root, unlike KNM-WT 40000 in which the I1 and I2 incisor roots are almost equal in size3 (see Extended Data Fig. 1m). In general, K. platyops seems to be more derived than Au. deyiremeda in most of its maxillary and dental morphology.

The overall morphology of the paratype mandibles (BRT-VP-3/14 and WYT-VP-2/10; Fig. 1 and Extended Data Fig. 2) is comparable to that of eastern African robust taxa, although their overall size is within the range of Au. afarensis. The anterior origin of the ascending ramus at the posterior P 4 level is comparable to P. boisei mandibles such as the Peninj mandible and KGA-10-525 (ref. 14), and P. robustus specimens such as SK-12. Australopithecus afarensis mandibles have the origin of the ascending ramus high and more posterior, mostly distal to the first molar9,15. BRT-VP-3/14 and WYT-VP-2/10 have inflated mandibular corpora, especially at the M 2–3 level, and they lack the lateral corpus hollow ubiquitous in Au. afarensis mandibles9. Corpus dimensions of BRT-VP-3/14 and WYT-VP-2/10 are notable because they are broader for their size than specimens assigned to Au. afarensis or early Homo. In this respect, they are more similar to Paranthropus (Fig. 5 and Supplementary Note 5). Mandibular specimens from Lomekwi whose taxonomic affinities have not been established with certainty (KNM-WT 8556 and KNM-WT 16006)3,16, are described as having robust corpora with an ascending ramus take-off at P 4 and M 1 , respectively. However, the robusticity of the Lomekwi specimens is not as pronounced as it is in WYT-VP-2/10 and BRT-VP-3/14. Furthermore, the KNM-WT 8556 premolars are derived: the P 3 has a well-developed metaconid and the P 4 is large and molarized3. By contrast, the P 3 of BRT-VP-3/14 is almost unicuspid, with only a trace of the metaconid, and the P 4 is not molarized.

Figure 5: Bivariate plots of mandibular corpus height and breadth among early hominins. a, Dimensions at the M 1 level. b, Dimensions at the M 2 level. The 95% confidence ellipses are shown for Au. afarensis, P. boisei and early Homo (see Supplementary Note 5 for details). The Au. deyiremeda corpora are relatively broad, comparable to Paranthropus. WYT-VP-2/10 falls outside the Au. afarensis ellipse at the M 1 level. BRT-VP-3/14 (represented by the mean of left and right sides) falls in an area of overlap at the M 1 level but at the edge of the Au. afarensis ellipse at the M 2 level. Mandibular measurements for the comparative taxa were taken from the references listed in Fig. 4 and ref. 14. Full size image Download PowerPoint slide

The symphysis of BRT-VP-3/14 is thinner, superoinferiorly longer, and has a more receding external contour than that of WYT-VP-2/10 (Extended Data Fig. 2h, n). In the latter specimen, the superior half of the external contour is almost vertical but the inferior half is extremely receding similar to L.H.-4, the holotype of Au. afarensis9,17,18. Further similarities between WYT-VP-2/10 and L.H.-4 include the prominence of the superior and inferior transverse tori (hence the deep genioglossal fossa) and the posterior extension of the inferior transverse torus exceeding that of the superior transverse torus. BRT-VP-3/14 has weak transverse tori and a very shallow genioglossal fossa (similar to Au. afarensis specimens such as A.L. 333w-12 and A.L. 315-22) and a pronounced postincisive planum (comparable to A.L. 400-1). These differences are within the range of variation seen in Au. afarensis mandibles9 and further show intra-taxon variability in mandibular symphyseal morphology. Although it was previously argued that two Au. bahrelghazali mandibles (KT 12 and KT 40) have symphyseal morphology that distinguishes them from Au. afarensis mandibles5, they appear to also fall within the range of Au. afarensis variation9,19,20.

In terms of the dentition, BRT-VP-3/14 has small premolars compared to Au. afarensis (see Extended Data Table 5 and Supplementary Note 5 for details). The small mandibular premolars appear to conform well with the small upper premolars and M1 of BRT-VP-3/1. Despite the robust appearance of the mandible, the BRT-VP-3/1 maxilla is only slightly smaller when occluded with the BRT-VP-3/14 mandible (see Extended Data Fig. 5).