Homo naledi was diagnosed as a novel taxon based on the morphology of approximately 1550 fossils, whole and fragmentary, recovered from the Dinaledi Chamber of the Rising Star cave system (Berger et al., 2015; Dirks et al., 2015). Although the Dinaledi material is noteworthy for its abundant hominin fossils, including articulated remains, there is a dearth of recovered faunal material (Berger et al., 2015, 2023; Dirks et al., 2015; Kivell et al., 2015; Dirks et al., 2016; Val, 2016; Bolter et al., 2018; Elliott et al., 2021). This gives the assemblage a taphonomic signature that is distinct from that of nearby hominin-bearing karst sites, like Sterkfontein and Swartkrans, where faunal remains predominate and hominin fossils are proportionally rare (Dirks et al., 2016). Geochronological and radiometric dates from flowstones that bracket the Dinaledi Chamber's fossil-bearing deposits and electron-spin-resonance dates from a small sample of H. naledi teeth suggest that the material entered the chamber between 335 and 241 ka (Dirks et al., 2017; Robbins et al., 2021), but younger dates were noted on some samples (Dirks et al., 2017).

Fossil material attributed to H. naledi comes from several spatially distinct localities within the Rising Star cave system. The largest sample comes from the Dinaledi Chamber, where surface collection and a limited excavation produced remains from a minimum of 15 individuals and more than 190 teeth (Berger et al., 2015). A smaller number of fossils come from the Hill Antechamber and locality U.W. 110, which are also part of the Dinaledi subsystem (Elliott et al., 2021). The Hill Antechamber has been the subject of recent excavation (Berger et al., 2023), but the previously described sample from this setting contains three isolated teeth (Berger et al., 2015; Delezene et al., 2023). The U.W. 110 locality has produced cranial remains and teeth from one immature H. naledi individual (Brophy et al., 2021). The Lesedi Chamber is outside the Dinaledi subsystem and distant from it; this locality has yielded fossils from at least three individuals, including an adult partial skeleton with complete dentition, a subadult with mixed dentition and associated postcrania, and a partial adult mandible retaining M1 and M2 (Hawks et al., 2017; de Ruiter et al., 2019; Feuerriegel et al., 2019; Cofran et al., 2022). Where focused analyses have compared them, the Lesedi and U.W. 110 fossils have been shown to be morphologically indistinguishable from the Dinaledi Chamber and Hill Antechamber homologs (e.g., Hawks et al., 2017; Davies et al., 2019a, 2019b; Feuerriegel et al., 2019; Brophy et al., 2021). Elliott et al. (2021) reported H. naledi fossils from other localities within the Rising Star cave system, and Berger et al. (2023) presented additional material from the Dinaledi Chamber and Hill Antechamber, but these fossils are not yet formally described. Though the tally is sure to grow, the published fossils from the four loci already represent the largest African Middle Pleistocene hominin assemblage recovered to date (e.g., Berger et al., 2017).

Hominin teeth and jaws are numerous in the Rising Star fossil assemblage and represent individuals ranging in age from infant to older adult (Cofran and Walker, 2017; Hawks et al., 2017; Bolter et al., 2018; Bolter and Cameron, 2020; Brophy et al., 2021; Cofran et al., 2022; Demeter et al., 2022). Berger et al. (2015) and Bolter et al. (2018) reported a dental minimum number of individuals (MNI) of 15 for the Dinaledi Chamber and Hill Antechamber material. Adding to that total, the subadult from U.W. 110 and the three individuals from the Lesedi Chamber raise the dental MNI to 19 for the published assemblage.

Despite the large number of recovered individuals, the H. naledi dental sample exhibits remarkable morphological homogeneity. Commenting on this feature of the assemblage in the original diagnosis, Berger et al. (2015:24) noted that “almost every aspect of the morphology of the dentition, including the distinctive form of the lower premolars, the distal accessory cuspule of the mandibular canines, and the expression of nonmetric features that normally vary in human populations, is uniform in every specimen from the collection.” Subsequent analyses of crown shape (Davies et al., 2019a, 2019b, 2020), root morphology (Kupczik et al., 2019), molar topography (Berthaume et al., 2018), and nonmetric traits (Irish et al., 2018) have all found low intrasample variation, supporting Berger et al.’s (2015) initial conclusion. Another insight into H. naledi dental variation comes from Garvin et al. (2017), who assessed coefficients of variation (CVs) for the buccolingual widths (BL) of the mandibular and maxillary M1s and M2s and mesiodistal (MD) and labiolingual (LL) dimensions of the mandibular and maxillary canines for the Dinaledi Chamber and Hill Antechamber samples. They reported CVs of 2.0 (M1 BL), 3.2 (M1 BL), and 3.8 (M2 BL and M2 BL) for the molars, and 2.8 (C1 LL), 5.5 (C1 MD), 6.2 (C1 MD), and 7.4 (C1 LL) for the canines. They compared the H. naledi CVs to sex-balanced samples of four extant hominoids and found that the H. naledi values fall within the lower bounds of expectations, or just below them.

Relative to other hominin fossil assemblages, the H. naledi CVs reported in Garvin et al. (2017), especially for molar size, tend to be low. For example, Moggi-Cecchi (2003) analyzed MD, BL, and LL dimensions for all permanent maxillary and mandibular tooth positions in species of Australopithecus and Paranthropus. They found that Australopithecus africanus CVs ranged from 3.9 to 12.5, Paranthropus robustus from 4.2 to 8.9, and Paranthropus boisei from 3.1 to 15.5. For A. africanus and P. robustus, only one dimension for each taxon (I1 LL for A. africanus and M1 MD for P. robustus) had a CV less than 5.0. For P. boisei, only three dimensions (C1 MD, M1 MD, and M1 MD) had CVs less than 5.0, while 15 were greater than 10.0. Lockwood et al. (2000) reported CVs ranging from 4.5 to 13.6 for MD, BL, and LL dimensions of all permanent maxillary and mandibular teeth except the incisors of Australopithecus afarensis; only two of the dimensions they considered (M2 MD and BL) had CVs below 5.0. Thus, the CVs Garvin et al. (2017) reported for H. naledi, especially the molars, fall below most point estimates for tooth size variation in other fossil assemblages.

In this study, we expand upon the analysis of Garvin et al. (2017) by including the entire published hypodigm of H. naledi, by considering all tooth positions, and by setting up hypothesis testing to consider sex-biased sampling as an explanation for low sample variation. Our goals are to 1) determine if the H. naledi sample has unusually low variation in comparison to expectations derived from extant hominoids, 2) generate hypotheses about the Rising Star material that can be tested as the assemblage grows, and 3) facilitate comparisons of variation with other hominin fossil assemblages. These objectives are important because many hominin assemblages, even those small in sample size, exhibit skeletal and dental variation that is perceived to be high in comparison to recent human populations and, in some cases, the most sexually dimorphic extant apes. Sources of high variation in fossil samples have been discussed extensively and include temporal and geographical heterogeneity, phylogenetic polymorphism, population history, sexual dimorphism, ontogeny, pathology, and the pooling of multiple species into a single analytical unit (e.g., Lockwood et al., 2000; Skinner et al., 2006; Harmon, 2009; Royer et al., 2009; Scott et al., 2009; Rightmire et al., 2019; Grine et al., 2021). By contrast, discussions of the meaning of low sample variation are encountered less often owing to the small sizes of most fossil samples. Possible sources of low variation that are informative with regard to a species’ biology include low genetic variation resulting from the action of genetic drift or natural selection, or sexual monomorphism related to aspects of socioecology. However, our ability to make such inferences depends on whether we can confidently exclude sampling error and taphonomy as explanations for low variation.

Taphonomic processes that are not random with regard to sex (e.g., predator bias, size-biased preservation, intersexual differences in home range) can influence the composition of fossil assemblages (e.g., Lockwood et al., 2007; Gower et al., 2019), which could yield a sample with lower variation than the population from which it was drawn. Even where the probability of entering the fossil record does not differ between the sexes, a random process, like the chance of fossil recovery, can produce strong sex bias in fossil samples, especially when the number of individuals in the assemblage is relatively small. Here, we investigate the H. naledi dental sample, asking whether variation at each tooth position is unusually low and, if so, whether this is consistent with sex bias in the fossil assemblage.