The method for measuring the implantation angle of the upper canine is shown in a, b. a, The internal alveolar plane was established by best-fitting a plane (outlined in blue) to three landmarks between left P3 and P4, between right P3 and P4 and between left M1 and M2. b, Landmarks were placed at the root tip and at the occlusal-most incursion along the distal enamel line. Upper canine implantation was measured in lateral view as the two-dimensional angle (θ) between a line connecting the canine landmarks and the axis of the alveolar plane. c, d, Upper canine implantation in the fossil specimens. c, A. anamensis paratype maxilla KNM-KP 29283. d, MRD. Of the other two A. anamensis maxillae currently known, one (KNM-KP 58579) is qualitatively similar to KNM-KP 29283 and the other (ARA-VP-14/1) is more inclined. Scale bar, 1 cm. The scale bar applies to images in c and d. e, The magnitude of difference in canine implantation angle between MRD and KNM-KP 29283 (12.5°, red dashed line) is shown in the context of expected conspecific differences. The expectation distributions were constructed using a permutation approach, for which the measurements from two individuals were randomly drawn (without replacement) from a comparative sample and the difference in orientation angle was computed. This procedure was repeated 500 times, separately for comparative species P. troglodytes and G. gorilla (Supplementary Note 6.2).