New hominin dental remains from the ∼2.04–1.95 Ma Drimolen Main Quarry, South Africa

Abstract Background The Drimolen Palaeocave site is situated within the UNESCO Fossil Hominid Sites of South Africa World Heritage Area and has yielded numerous hominin fossils since its discovery in 1992. Most of these fossils are represented by isolated dental elements, which have been attributed to either of two distinct hominin genera, Paranthropus and Homo. Aim This paper provides morphological descriptions for a further 19 specimens that have been recovered from the ∼2.04–1.95 Ma Drimolen Main Quarry (DMQ) deposits since 2008. This paper also discusses the two primary hypotheses used to explain Paranthropus robustus variation: sexual dimorphism, and micro-evolution within a lineage. Subjects and methods These 19 fossils are represented by 47 dental elements and expand the sample of DMQ early Homo from 13 to 15, and the sample of Paranthropus robustus from 69 to 84. Results The evidence presented in this paper was found to be inconsistent with the sexual dimorphism hypothesis. Conclusion Some support was found for the micro-evolution hypothesis.


Introduction
the hominin-bearing palaeocave system of Drimolen is located within the Fossil hominid sites of south africa UNescO World heritage centre in the Gauteng Province of south africa, known locally as the cradle of humankind (coh;Keyser et al. 2000;herries et al. 2020).Discovered in 1992, the locality is approximately 5-6 km north-east of other well-known sites such as sterkfontein and swartkrans (Keyser et al. 2000).excavations conducted since initial discovery have concentrated on the ~2.04-1.95Ma DMQ deposit, which consists of a single large palaeocavern (herries et al. 2020).the site is the third richest site for early Pleistocene hominin fossils in south africa, with specimens representing both Paranthropus robustus and south african early Homo (Keyser 2000;Moggi-cecchi et al. 2010;Martin et al. 2021;leece et al. 2022), as well as one specimen that has been attributed to Homo aff.erectus (herries et al. 2020).the P. robustus sample includes a number of partial crania and skulls including the most complete female (DNh 7) and male (DNh 155) specimens of the species (Keyser 2000;herries et al. 2020;Martin et al. 2021;Rak et al. 2021).the vast majority of hominin fossils, however, are isolated teeth (Keyser et al. 2000;Moggi-cecchi et al. 2010;leece et al. 2022).the specimens presented here significantly expand the DMQ sample and are important for evaluating hypotheses regarding P. robustus variation.two primary hypotheses have been put forward and will be investigated here: (1) variation in P. robustus can be explained by an extreme manifestation of sexual dimorphism (lockwood et al. 2007;Moggi-cecchi et al. 2010) and (2) variation in P. robustus represents micro-evolution in a single lineage (Martin et al. 2021).
as the DMQ and swartkrans assemblages make up the majority of the P. robustus fossil record, and consist largely of isolated dentitions, Moggi-cecchi et al. (2010) performed a statistical comparison of the mesiodistal and buccolingual measurements of the permanent teeth of these two samples (inclusive of all P. robustus-bearing Members at swartkrans).three principal conclusions were drawn from that analysis: (1) the sample from Drimolen represents two species, P. robustus and an "undetermined species of the genus Homo"; (2) P. robustus specimens from Drimolen are generally smaller than specimens from swartkrans while overlapping with the lower end of the size range from the latter site; and (3) this size difference is representative of a highly (sexually) dimorphic species differentially distributed across the two sites as the result of depositional and site-use differences (Moggi-cecchi et al. 2010, p. 404).these results were broadly consistent with those of lockwood et al. (2007) who found marked differences in facial growth/height between DMQ (represented by DNh 7 only) and a combined sample from Kromdraai B (represented by material excavated historically from Members 5 to 7 [KBM4-7] including tM 1517) and swartkrans (represented primarily by swartkrans Member 1 [sM1] with one specimen from Member 2). the authors hypothesised that the observed differences were due to sexual dimorphism as a result of male bimaturism (i.e.secondary maturation) characterising the growth trajectory of P. robustus (lockwood et al. 2007).they further proposed that DNh 7 represented a female P. robustus specimen, while the KBM4-7 and the majority of the swartkrans P. robustus material represented male specimens at various stages of maturation (lockwood et al. 2007).Martin et al. (2021) described a new adult male P. robustus cranium, DNh 155 and compared it to the already-known female specimen DNh 7, the newly described DNh 152 male specimen (herries et al. 2020), and the swartkrans Member 1 hanging Remnant (sM1hR) assemblage.Based on the similarity of DNh 155 to DNh 7, and the consistent differences between these two specimens and the swartkrans sample, Martin et al. (2021) proposed that differences between the DMQ and sM1hR assemblages were not due to sexual dimorphism but rather demonstrated micro-evolution within the P. robustus lineage.
in 2022, leece et al. published an analysis testing between the sexual dimorphism and micro-evolution hypotheses.that study did find statistically significant differences in some dental metrics between the DMQ and swartkrans (all P. robustus-bearing members) assemblages, as in the findings of Moggi-cecchi et al. (2010), but overall, the results were equivocal and did not lend clear support for either the sexual dimorphism or the micro-evolution hypothesis.
this paper utilises a modified sample to assess these two hypotheses, expanding the DMQ sample with newly discovered material and restricting the swartkrans sample to a more comparable time depth based on recent chronological assessments (herries et al. 2020).While both Moggi-cecchi et al. (2010) and leece et al. ( 2022) compared the DMQ assemblage to a complete swartkrans assemblage regardless of Member, it is considered that this is not a valid way to assess these rivalling hypotheses.Firstly, comparing a sample with a time depth of ~90,000 years (DMQ;herries et al. 2020) to a sample with a time depth of ~1.3 million years (swartkrans, all Members; herries 2022) removes any ability to assess micro-evolution in a lineage.second, assessing a deposit with ~1.3 million years of possible accumulation history will also obscure any variability that may represent sexual dimorphism.When discussing the use of coefficient of variation (cV) for identifying sexual dimorphism in primates, Plavcan and cope (2001) warn against using assemblages with a lengthy collection history as well as against the dangers of comparing assemblages with different time depths.Variation due to sexual dimorphism will, in deposits with a deeper time depth, be obscured by temporal variation (Plavcan and cope 2001).For this reason, this study compares the DMQ assemblage to the sM1hR assemblage only in an attempt to restrict comparisons to samples with closer temporal equivalence.
the dental manifestation of sexual dimorphism in large-bodied primates is well understood (see lauer 1975; Plavcan 2001).if the analysis presented here matches this pattern, with significant differences shown in both the buccolingual and mesiodistal dimensions, in the maxillary and mandibular canines, third premolars, second, and third molars to the exclusion of other tooth positions (lauer 1975;Plavcan 2001), this will be taken as support for the sexual dimorphism hypothesis.if the analysis presented here instead shows a pattern of significant differences distinct from this, it will be taken as support for the micro-evolution hypothesis, indicating non-dimorphic differences between two populations of the same species separated by up to ~200,000 years (herries et al. 2020).if the analysis presented here shows differences in only some of the tooth positions found to be critical by lauer (1975) and Plavcan (2001) and in no other tooth positions, it will be taken as equivocal evidence as seen in leece et al. (2022).

Site, stratigraphy, and context
Unlike other sites in the region, such as sterkfontein, swartkrans, and Kromdraai located in the Bloubank stream Valley, the geological sequence at DMQ is relatively simple (herries et al. 2020).this appears to be because the cave is located in an upland area outside the Bloubank stream Valley and was a stream sink for water from an upland restricted watershed (herries et al. 2020).the excavated sediments at DMQ also infilled the site over a relatively short period of time (~90,000 years) as compared with swartkrans where Paranthropus material accumulated over a much longer period of time (~1.6-1.2 million years) (herries et al. 2020; herries 2022).the age of the entire Kromdraai sequence is not known with certainty, but given the complexity of the deposits, it likely formed over a longer period of time much like swartkrans (herries 2022).after the formation of a basal speleothem prior to ~2.6 Ma, the central part of the DMQ palaeocavern filled with a talus cone formed from a vertical entrance.this talus cone became cemented into a breccia and is the source for all the hominin fossils.however, lime mining of the basal speleothem in the late 19th and early 20th century led to the undermining of this breccia and its partial collapse.today, the breccia (consisting of Jangi Buttress, Western Wall Breccia, Marcel Pinnacle, southern italian Job Pinnacle) and sandstone and siltstone deposits (Warthog cave section, Walls of Jericho Pinnacle, northern italian Job Pinnacle; Figure 1a) that occur in the western portion of the palaeocavern (west of the central excavation area; hereafter cea; Figure 1B) are still in situ.however, the area occupied by the western part of the cea consists of breccia that has collapsed away from these deposits, while the eastern side of the cea consists of lime mining rubble that is a mixture of speleothem and breccia debris.None of the hominin teeth described here were recovered from this lime mining debris.
the age of the youngest fossil bearing deposits is constrained by the occurrence of a magnetic field reversal from reversed to normal polarity in the uppermost units (above −1.6 m) of the italian Job Pinnacle, Walls of Jericho Pinnacle, and West Wall Breccia (herries et al. 2020; Figure 1).this magnetic reversal has been correlated to the base of the Olduvai subchron at ~1.95 Ma based on a uranium-lead age of 1.96 ± 0.11 Ma for an intermediate polarity flowstone occurring within the magnetic reversal itself as well as a uranium series-electron spin resonance (Us-esR age) of 1.97 ± 0.15 Ma for a bovid tooth occurring just below this flowstone within intermediate polarity sediments (herries et al. 2020; Figure 1).the youngest units at DMQ (above −1.6 m) can therefore be dated precisely to within the reversal at ~1.95 Ma (Figure 1). the oldest excavated layers (−5.5-5.2 m below datum) have been dated to 2.04 ± 0.24 Ma based on the Us-esR dating of a bovid tooth from the base of the reversed polarity Jangi Buttress (herries et al. 2020; Figure 1).While the uncertainty of this Us-esR age means the deposits could date to as old as 2.28 Ma, or between 2.04 and 1.95 Ma, the median age is considered a good age estimate based on the fact that the higher Us-esR median ages correlate well with the age of the reversal.these oldest units have yielded the DNh 134 Homo aff.erectus cranium (herries et al. 2020) as well as some of the new fossils described here.Deposits between this reversal and the base of the Jangi Buttress (−5.5 to −1.6 m) are thus dated to between ~2.04 and 1.95 Ma (herries et al. 2020).this includes the recently discovered DNh 155 and DNh 152 P. robustus crania that were recovered from in situ breccia deposits (herries et al. 2020;Martin et al. 2021; Figure 1).the DNh 7 cranium as well as many of the hominin teeth previously described by Moggi-cecchi et al. (2010) andleece et al. (2022) were excavated between 1992 and 2008 by andre Keyser's team and come from ex situ collapsed breccia deposits within the cea (see Figure 1).there is no reason to believe that this material came from a different unit than that sampled in the currently exposed in situ western breccia deposits and thus of any age other than between ~2.04 and 1.95 Ma. excavation of the DNh 7 block indicates that it came from basal talus cone deposits and was thus likely from a period prior to 1.95 Ma (herries et al. 2020).
a number of the teeth in this current analysis derive from decalcified sediments along the eastern edge of the in situ western breccia deposits, with a cluster of specimens (DNh 106,121,129,136) emanating from where the cea abuts this breccia (Figure 1).this material sits between −3.43 m and −2.DNh 122 was recovered from a distinctive brown secondary cave fill in Warthog cave.Warthog cave is a secondary, younger cave that formed at the interface between the southern dolomite wall of the palaeocavern and the palaeocave deposits.secondary formation of caves is common at such contacts.Warthog cave formed both into the dolomite itself and the edge of the sandstone and siltstone deposits, eroding the upper portion of the Warthog cave section (herries et al. 2020).DNh 122 was obviously eroded out of the sandstone and siltstone deposits and incorporated into the Warthog cave fill, occurring within a few centimetres of the eroded surface.this is the only tooth to have been found both in such a clear secondary sedimentary context and seemingly having been sourced from sandstone and siltstone deposits rather than the central talus cone breccia.it is also the only fossil recovered from the Warthog cave sediment.in contrast, the bulk of the teeth come from in situ clearly decalcified breccia that are rich in other fossils including clusters of micromammal remains.
the hominin fossils discussed above have all been recovered from between −5.46 and −2.47 m below datum and thus sit stratigraphically beneath the Olduvai basal magnetic reversal at 1.95 Ma, which begins at −1.6 m below datum in the Walls of Jericho Pinnacle.as such, most can be confidently dated to between ~2.04 and 1.95 Ma, although those from the collapsed area of the cea are less securely associated.two other teeth, DNh 138 and DNh 149, were both recovered in 2018 from previously excavated but unsorted sediments from andre Keyser's 1997 and 2002 excavations respectively.they were not identified as hominins at the time of their excavation, and the DNh 138 tooth had no coordinate data associated with it.as such, its provenience is unknown.the DNh 149 tooth comes from the cea above the DNh 7 block, in an area that had already been excavated away in the 1990s, so its precise association is also unknown.

Swartkrans sample
the only other site within the coh that has a comparable sample size to DMQ is swartkrans, located around 6 km to the southwest of DMQ.swartkrans has yielded P. robustus fossils from four depositional units, and specimens attributed to early Homo from three. the oldest P. robustus-and Oldowan tool-bearing unit is the Member 1 lower Bank (sM1lB), which has been dated to between ~2.2 and 1.8 Ma (Pickering et al. 2011(Pickering et al. , 2019;;Gibbon et al. 2014;Kuman et al. 2021;herries 2022).the lower Bank has yielded mostly isolated Paranthropus teeth (28 specimens, not all of which are measurable; Kuman et al. 2021) with only one partial mandible attributed to Homo (Grine 1993; Kuman et al. 2021;herries 2022). in contrast, the majority of the more complete crania of P. robustus and early Homo have been recovered from sM1hR, which has not yielded any stone tools.sM1hR has yielded 229 measurable dental specimens.as such, the two units are quite distinctive in terms of their inclusions and sediments, with the hanging Remnant consisting of calcified pink breccia and lower Bank un(de)calcified brown sediments.While it has always been considered that sM1lB formed as a single lithostratigraphic unit with sM1hR, the two are not stratigraphically connected having been partly eroded before the deposition of Member 2 between the two units.the lower Bank dates to ~1.9-1.8Ma based on the cosmogenic nuclide burial dating of the older lower Bank, U-Pb ages of underlying and overlying speleothem, and Us-esR ages (herries 2022;Grine and Daegling, 1993).as such, a portion of the lower Bank hominin sample may be older than DMQ, especially if the base of DMQ is ~2.04Ma rather than 2.28 Ma. the sM1hR sample, however, is substantially younger than DMQ, perhaps by ~250,000-150,000 years.
the age of the P. robustus and early Homo-bearing Member 2 is not well resolved with esR ages of only a few hundred thousand years (curnoe et al. 2001) and a single U-Pb age on a tooth of 1.36 ± 0.29 Ma (Balter et al. 2008).stratigraphically, it formed after Member 1 as a whole.however, what has been termed as Member 2 has changed over the years.some areas once considered to be Member 2 based on the colour of the deposits are now thought to represent a younger deposit, likely explaining the very young esR ages (herries 2022).as such, there is little confidence that all the hominin teeth (24 Paranthropus teeth, not all of which are measurable; Grine 1988;Brain et al. 1988) attributed to Member 2 come from a single chronological unit.there may also be mixing between Member 1 and 2 with certain specimens, such as sKW 12, assumed to have eroded out from the older Member 1 lower Bank into a later channel fill (Brain 1981).this indicates that there is at least some mixing of hominin fossils from older into younger units.
the remaining early Pleistocene unit at swartkrans is Member 3, which contains only P. robustus specimens (12 dental specimens, not all measurable; Brain et al. 1988) among the comparatively small number of fossils preserved (Grine 1993).these deposits are important as potentially being some of the latest occurring P. robustus fossils at 1.0-0.8Ma based on U-Pb (on fossils), esR and cosmogenic nuclide dating (herries 2022).Whether these hominin fossils could have been reworked from older deposits is something that needs to be considered.

Drimolen hominin sample
this study describes 19 new hominin dental specimens.in addition to the 62 specimens published by Moggi-cecchi et al. (2010) and the 24 specimens published by leece et al. (2022), this brings the total for the DMQ dental assemblage to 105 specimens.the majority of the specimens presented here were excavated between 2007 and 2018, with the material presented in Moggi-cecchi et al. (2010Moggi-cecchi et al. ( ) having been discovered between 1992Moggi-cecchi et al. ( and 1999Moggi-cecchi et al. ( , and material presented in leece et al. (2022) ) having been discovered between 1999 and 2008.the exception is DNh 138, which was found in an unlabelled bag of previously excavated but unsorted sediment from excavations in 1997.the majority of the collection can be provenanced using either square and height data, or full 3D co-ordinates (see Figure 1 and supplementary material, table s1).Of the 19 specimens presented here, many contain multiple teeth.this includes two specimens (DNh 107 and DNh 108) with nearly complete dental arcades. in total, these 19 specimens add 47 new individual teeth to the DMQ assemblage.Of these, 16 have been given taxonomic attributions in this paper, substantially expanding the sample aiding further inquiries into variation within P. robustus.a list of DMQ specimens presented in this study is provided in table 1.

Comparative sample
the previous analyses of Moggi-cecchi et al. (2010) and leece et al. ( 2022) compared the DMQ P. robustus material to a combined sample from KBM4-7 (i.e.< 1.95 Ma; herries 2022) and all swartkrans members.Owing to the small sample size of Members 2-3 from swartkrans (and its large temporal range) and KBM4-7, as in Martin et al. (2021), this analysis only compares the DMQ P. robustus material to material from sM1hR, which consists of 229 specimens; the comparative sample is outlined in tables s2 and s3 (supplementary material).

Morphological descriptions
all specimens were examined and described following leece et al. (2022).Distinct morphological features reported by leece et al. (2022) and Moggi-cecchi et al. (2010), as well as those from earlier studies of south african P. robustus and early Homo (Robinson 1956;tobias 1965, 1967, 1991;clarke 1977;howell 1978;Grine 1984Grine , 1989Grine , 2005;;Grine and strait 1994), were utilised for the purposes of assessing occlusal crown anatomy.Basic metric data (mesiodistal and buccolingual dimensions) were taken following the protocols of Wood (1991).specimens were examined using both a hand lens with 10× and 20× magnification and a low-powered binocular microscope.While leece et al. ( 2022) outlined key morphological features of the teeth assessed, not all dental elements (i.e.tooth positions) were presented in leece et al. ( 2022); as such, a description of these additional diagnostic morphological traits are presented below: Maxillary permanent teeth Premolars: Paranthropus: Mesiobuccal projection in P 3 ; simple occlusal groove pattern; robust and complex root system (often three or even four radicular canals); distinct mesiodistal expansion in P 4 (molarisation); broad and shallow mesiobuccal groove; central fovea is narrow but deep.
Homo: lack of molarisation -overall reduction in size; distinct ridging creating almost continuous mesial marginal ridge (MMR) and distal marginal ridge (DMR) across apex of buccal cusp; reduction in degree of root development; deep and narrow mesiobuccal groove.

Statistical analyses
Descriptive statistics are provided following the methodology of Moggi-cecchi et al. (2010).For all elements where n ≤ 5 but >1, an adjusted cV* was calculated following sokal and Braumann (1980).
Given that sample sizes for each tooth class rarely exceed 10, non-parametric Mann-Whitney U-tests were performed with alpha set at 0.05.these analyses were conducted using sPss version 28.0.1.0.Given the small number of incisors in these two fossil assemblages, comparisons were limited to the permanent canine and postcanine teeth.the goal of this analysis was to compare mesiodistal and buccolingual dimensions across P. robustus assemblages to evaluate the hypotheses that (1) metric differences in the dentitions between DMQ and sM1hR are the result of sexual dimorphism in a species differently represented between sites due to taphonomic biases (Moggi-cecchi et al. 2010;Pickering et al. 2016), or (2) metric differences in the dentitions between DMQ and sM1hR are the result of micro-evolutionary differences between sites, as suggested by analyses of craniofacial morphology (Martin et al. 2021).the statistical null hypothesis is that the DMQ and sM1hR assemblages are not significantly different.analyses yielding p-values below 0.05 would provide support for refuting the null hypothesis and thus suggest a low likelihood that the DMQ assemblage means could be chosen at random from the sM1hR assemblage.

Results
Dental metrics for each newly described tooth are presented in table 1, while summary statistics are presented in tables 2 and 3.Not all specimens have been measured due to fragmentation.

LP 3 (Figure 2a-c)
this element preserves only the crown, which has broken off the still-brecciated maxilla.the distobuccal enamel face is missing.While both cusps are of similar size, the protocone is taller than the paracone.the anterior fovea is a circular pit located buccally helping to delineate a distinct mesiostyle.a wide but shallow groove on the mesial portion of the buccal face also delineates the mesiostyle from the paracone.a small but distinct epiconule is present, located lingual to the anterior fovea and centrally on the tooth.the central fovea is narrow but deep, supporting an attribution to P. robustus.the posterior fovea is split into buccal and lingual sections by distally curving enamel extensions from both the protocone and paracone.Despite being fragmentary, an incipient paramolar tubercle is also present distal to the paracone.slight distal expansion is present in the form of the enamel extensions described above as well as a thick DMR consistent with an attribution to P. robustus.

LP 4 (Figure 2d)
this specimen is still articulated to the maxilla and as such is greatly obscured by both preserved maxillary bone and adhering breccia matrix, limiting a complete morphological evaluation.as with the lP 3 , while both cusps are a similar size, the protocone is much taller than the paracone.the central fovea is large, joined with both the anterior fovea and the posterior fovea by a marked longitudinal groove.the DMR is thick, rising lingually to a paramolar tubercle contributing to a clear distal expansion of the tooth consistent with P. robustus material.

LM 1 (Figure 2e-g)
this specimen is highly fragmented.the crown has separated from the roots and preserved maxilla at the cervical margin, and the paracone and a portion of the metacone are missing.the preserved protocone is moderately worn, having lost nearly all crown height, and the preserved hypocone shows heavy cusp polishing.the central fovea has been reduced to a faint and shallow basin by the advanced stage of wear. the posterior fovea appears deep and long, supporting an attribution to P. robustus, but is obscured by adhering breccia.the thickness of the DMR cannot be assessed due to adhering breccia, but a small postentoconule can be observed.a distolingual groove originating from the posterior fovea delineates the hypocone and gradually terminates halfway down the lingual face of the crown.a short but deep mesiolingual groove is present and terminates near the occlusal surface in a small pit.Only a small portion of the mesial icF is preserved.

LM 2 (Figure 2h-j)
this is the best-preserved tooth of this specimen, although the crown has detached from the roots and preserved maxilla.all four main cusps are well developed, and multiple accessory cusps are present.the central fovea is deep but narrow, clearly delineating the principal cusps.the anterior fovea is small but deep and positioned buccally and does not extend centrally past the paracone.two small but distinct mesiostyles are present on the MMR along with a third incipient mesiostyle.a large epiconule similar in size to the main cusps is present mesiobuccal to the protocone, delineated by a deep mesiolingual groove terminating in a deep pit. a large metaconule, also similar in size to the main cusps, is present centrally, extending buccally from the protocone.the distolingual groove presents as a broad V-shaped fissure occupying a large portion of the side of the crown from the apex of the protocone to the apex of the hypocone and terminates at the cervical margin consistent with P. robustus material.the buccal groove is wide and terminates just above the cervical margin.the posterior fovea is deep and large, bounded by a poorly developed and low DMR. a small but distinct distostyle is present on the distal side of the metacone, a large postentoconule is present buccal to the posterior fovea, and a second incipient postentoconule is present lingual to the posterior fovea.the slight molarisation of the two premolars and the deep occlusal foveae and curved buccal and lingual faces present on both molars support an attribution to P. robustus.

L and R dm 1 (Figure 3a-f)
the right element is missing the protoconid and a portion of the mesial root, while the left element is missing the hypoconid, the hypoconulid, and most of the distal root plate.as such, the following description is a composite from both antimeres.the protoconid and metaconid are aligned transversely.the metaconid is the largest cusp, the hypoconid is larger than the protoconid, and the hypoconulid is poorly developed.as such, the trigonid is mesiodistally dominant over the talonid.a weakly expressed tuberculum molare is present.the specimens are heavily worn with a large dentine exposure on the hypoconid and hypoconulid, small dentine exposures on the protoconid and metaconid, and a tiny dentine exposure on the entoconid.a well-developed mesioconulid also shows a very small dentine exposure.the central fossa is not pronounced, although this is likely due to the advanced stage of wear. the buccal groove is shallow and faint but visible.the lingual groove is more distinct and runs to the cervical margin.the anterior fovea is short but deep, presenting as a single buccolingually oriented fissure, the position of which is lingually skewed.the MMR is cut by a well-developed mesioconulid.a clear distal trigonid ridge bounds the anterior fovea on its distal side.the MMR is well developed and completely enclosed the anterior fovea consistent with P. robustus material.the posterior fovea is obscured by the advanced stage of wear. the metaconid and entoconid are clearly delineated by a deep fovea.Both the mesial and distal icFs are large and reach the occlusal surface.Both the mesial and the distal roots are fused but mesiodistal compression makes it clear that two radicular canals are present in both.Both roots show some resorption.

R and Ldm 2 (Figure 3g-l)
Both elements are well preserved, although the right element is missing portions of the entoconid and hypoconulid.the following description is drawn from the left element.the metaconid is the largest cusp, while the other four main cusps are of a similar size.the protoconid and metaconid are aligned transversely.this specimen is moderately worn with small dentine exposures visible on the protoconid, hypoconid, hypoconulid, and entoconid.the metaconid shows a fair degree of cusp polishing.a moderately well-developed metastylid (c7) is visible on the distal side of the metaconid delineated by a faint groove.a well-developed c6 is also present.the central fovea is deep and narrow, supporting an attribution to P. robustus.the primary occlusal fissure shape is y-shaped.the longitudinal groove is indistinguishable from the central fovea and is only stopped from joining with the posterior fovea by enamel extensions from both the entoconid and hypoconulid, nearly forming a postentocristid.the posterior fovea is a deep pit bounded by a thick DMR also supporting an attribution to P. robustus.the anterior fovea is narrow but deeply incised and bounded by a thick MMR and a distal trigonid crest of equal sizes.the mesiobuccal groove is short and terminates abruptly in a deep pit. the distobuccal groove and the lingual groove are also short but do not terminate in a pit. the mesial icF is large and reaches the occlusal surface.the distobuccal root is the only unbroken root on either antimere.this root shows some degree of resorption.

RI 1 (Figure 3m-n)
this tooth is well preserved and unerupted.the labial face is tall and straight comparing well to KB 5223/5383.the mesial and distal walls are straight and run parallel to the incisal margin with no divergence.Perikymata are clearly visible down the labial face.the lingual face shows only a weakly developed cervical eminence that is skewed slightly distally supporting an attribution to P. robustus.Both the MMR and the DMR are faint leaving a relatively straight and even lingual face.three clear mamelons are visible along the incisal edge.the root is still incomplete and terminates 2.9 mm from the cervical margin.

R and LI 2 (Figure 3o-r)
Both antimeres are fractured mesiodistally with only the labial face preserved.these teeth are also unerupted, although root development cannot be assessed due to fragmentation. the distal wall diverges slightly while the mesial wall is relatively straight.Perikymata are clearly visible down the preserved labial face.three faint mamelons are visible along the incisal edge.

R C (Figure 3s-t)
this specimen is fractured labiolingually preserving only the mesial half of the crown.this specimen is unworn and root development is likely incomplete although fragmentation limits this assessment.there is no distinct MMR, and the preserved portion of the lingual cervical eminence suggests a distal skew consistent with P. robustus material.the fragmentary nature of this specimen prevents further morphological description.

R and LP 3 (Figure 3u-z)
Both antimeres are well preserved although broken at the cervical margin preserving only the unworn crown.the crown is complete though root development cannot be assessed.the following description is drawn from the left element.the protoconid is significantly larger than the metaconid.the anterior fovea is a deep pit bounded by a low but distinct MMR and a thin but high distal talonid ridge.the central fovea is joined with the posterior fovea by a wide and deep longitudinal groove forming a large, deep basin.the DMR is thick but low. a distinct buccal ridge that continues to the central fovea delineates the protoconid from a moderately developed talonid consistent with P. robustus material.a lingual extension of the central fovea delineates the metaconid from a well-developed postmetaconulid.

RP 4 (Figure 3aa-cc)
this is a poorly preserved and fragmentary crown.the mesial portion of the crown is missing.crown formation is not yet complete, and the specimen still exhibits some pavement cracking.the central fovea is deep and wide, occupying most of the preserved occlusal surface.the posterior fovea is deep and long bounded by a thick but low DMR and a thin, tall, and incised postentocristid.Both distal cusps are well developed and clearly delineated from the preserved portions of the protoconid and metaconid.a shallow but wide groove runs down the buccal face terminating just short of the cervical margin.

RM 1 (Figure 3dd-ff)
this is a well-preserved tooth with a fragment of enamel missing from the metaconid near the cervical margin and showing cusp polishing on all five major cusps.all five major cusps are well developed and similar in size.the central fossa is deep and narrow, supporting an attribution to P. robustus, clearly delineating all cusps.a small but well-delineated c7 is present.the anterior fovea is deep but small, bounded by a thick but low and incised MMR and a weak incised distal talonid ridge.the posterior fovea is deep and of moderate size, bounded by a low DMR that is perforated by a groove that continues down the distal face.a distinct postentocristid bounds the posterior fovea mesially.Both buccal grooves terminate halfway down the buccal face in pronounced pits.the lingual groove also terminates high but without a pit. a large enamel extension protrudes lingually from the hypoconid and occupies the central portion of the occlusal surface.Preserved mandibular bone obscures assessment of the lingual roots, but the buccal roots are visible.Root development is incomplete.

RM 2 (Figure 3gg-ii)
this element is poorly preserved with the entoconid and a portion of the hypoconulid missing.crown development is incomplete with the specimen still exhibiting distinct pavement cracking.the central fovea is wide and indistinguishable from the longitudinal groove, clearly delineating the five main cusps.the anterior fovea is deep, bounded by a thick but low MMR and an incised distal talonid ridge.the mesiobuccal ridge is deep and well pronounced, while the distobuccal ridge is faint.an enamel extension originating from the hypoconid occupies the centre of the central fovea.

Molar Fragment (Figure 3jj-kk)
a molar fragment is also associated with this specimen.it likely represents the upper M1 preserving what is potentially the mesial face.the fragmentary nature of this specimen precludes morphological description.
the molarisation of the third premolars, the deep and narrow central fovea of the permanent molars, and the distal and mesial ridging of the deciduous molars is supportive of an attribution to P. robustus.
Ldm 1 (Figure 4a-c) this specimen is well preserved aside from a chip of enamel missing from the lingual side of the distal face and heavy wear leaving only the hypocone with notable morphology.the paracone, metacone and protocone are all obscured by one large dentine exposure, which has worn these three cusps down to the cervical margin.a small dentine exposure is also visible on the remaining hypocone.this advanced degree of wear greatly hinders morphological description.Despite this, a rhomboidal crown outline is consistent with P. robustus material.the distal icF is large, occupying most of the distal face and contacting the occlusal surface.a faint and shallow ridge is present on the buccal face.the lingual root is greatly diverged from the buccal roots.all three roots show a degree of resorption.

R and Ldm 2 (Figure 4d-f and y)
the ldm 2 is fragmentary with the distal half of the crown missing, while the Rdm 2 is well preserved.On both specimens, the protocone is obscured by a large dentine exposure and a small dentine exposure is present on the paracone.the following crown description will be based in the right antimere.On the right element, a small dentine exposure is also visible on the hypocone.the cusps are all of a similar size, though the crown outline is asymmetrical due to a mesiobuccal projection of the paracone.the posterior fovea is deep and short, bounded by a thick DMR supporting an attribution to P. robustus.the central fovea is deep and extends buccally delineating the metacone from the paracone terminating halfway down the buccal face without a pit. the anterior fovea is faint and bounded by a thick MMR. the distal limb of the trigon basin is better developed than the mesial fissure.at the occlusobuccal margin of the trigon basin, the buccal limb cuts a deep, V-shaped notch.a fracture runs through the lingual ridge obscuring any detailed morphological description.Both the mesial and distal icFs are large with the former showing a distinct concavity.the lingual groove is narrow and terminates gradually halfway down the crown.the mesiobuccal groove is weakly expressed.as the roots of the right antimere are obscured by breccia, detail here is drawn from the left antimere.the lingual root is greatly diverged from the buccal roots.Both lingual roots show compression internally.the mesiobuccal root is broken 5.6 mm from the cervical margin.

RI 1 (Figure 4g-h)
this element is well preserved with only the tip of the root missing.heavy wear has left a large, rectangular dentine exposure occupying the majority of the incisal edge and greatly reducing crown height. in labial view, the incisal edge is dipped, creating a concave outline.Multiple hypoplastic lines are visible on the labial face.Only the lowest portion of the mesial icF is preserved.On the lingual face, the faint cervical eminence is slightly mesial skewed.the presence of a cervical eminence supports an attribution to P. robustus.Neither the MMR nor the DMR is distinct leaving the lingual face with a slight concavity rather than with distinct bounding ridges.the root is broken 14.1 mm from the cervical margin.

R and L C (Figure 4i-l)
Both antimeres are well preserved and unworn.the left element has a single crack running centrally through the crown.as the right element is obscured by adhering breccia and remaining maxillary bone, the following description is based on the left element.the labial crown outline is asymmetrical with the distal edge sloping more steeply than the mesial.a small but distinct ridge runs down the labial face just distal of midline.a clear hypoplastic line is visible on the lower 1/3 of the labial face.Both the DMR and the MMR are weakly developed.On the lingual face, the cervical eminence is small and located centrally.the presence of a cervical eminence supports an attribution to P. robustus.

R and LP 3 (Figure 4k-o)
Both antimeres are preserved, but as the right element has a crack running mesiodistally through the paracone and is obscured by adhering breccia matrix and remaining maxillary bone, the following description is drawn from the left element.Both main cusps are well developed.the central fossa is deeply incised, supporting an attribution to P. robustus, and joins with the anterior fovea.the posterior fovea, although deep and large, is only separated from the central fossa by enamel extensions coming from both the protocone and paracone.the MMR is extremely faint, while the DMR is low but well developed.Multiple incipient cuspules are present on the DMR. the talonid is poorly developed, as is the trigonid, though the latter is more distinct than the former.Both buccal grooves are faint to the point of absence.the lingual root is broken at the cervical margin.the buccal root is incomplete.the buccal root exhibits marked buccolingual compression with a clear groove running down its length.

R and LP 4 (Figure 4p-u)
Both antimeres are unworn and have broken at the cervical margin and do not preserve any roots.the left antimere is also missing the buccal half of the paracone and the following description is based on the right antimere.the talonid is moderately developed, consistent with P. robustus material, with a well-developed incipient cuspule on the distal side of the paracone delineated by a faint groove on the buccal face.the central fovea is deeply incised, supporting an attribution to P. robustus, and is joined with both the anterior and posterior fovea by a deep and narrow longitudinal groove.Both the MMR and DMR are large, although the latter is higher than the former.an enamel extension protrudes distobuccally from the protocone towards the posterior fovea.No grooves are present on the buccal face, though a faint distolingual groove is present.the entirety of the occlusal central to the two cusps is heavily crenulated.

R and LM 1 (Figure 4v-y)
Both antimeres are well preserved, but the right is obscured by breccia, and the following description is drawn from the left.this specimen shows moderate wear with notable loss of cusp height in the protocone, paracone, and hypocone as well as some degree of cusp polishing visible on the metacone.the protocone is the largest cusp followed by the metacone and paracone which are of equivalent size.the hypocone is the smallest cusp though still well developed.the MMR is very thick though height cannot be assessed due to wear. the anterior fovea is small and deep, partly enclosed by an incised epicrista.the central fossa is deep and large, supporting an attribution to P. robustus, and joins with both the anterior and posterior fovea.Deep and narrow foveae delineate the paracone from the metacone and the protocone from the hypocone.these foveae continue to grooves on the buccal and lingual faces, both terminating halfway down the crown without pits.short buccally oriented extensions are present from on the protocone and the hypocone terminating at the central fovea.a faint groove is present on the mesial side of the protocone delineating a moderately expressed epiconule.the DMR is thick but relatively low and contains an incipient distostyle and an incipient postentoconule.Faint hypoplastic pits are visible on the buccal face.the mesiobuccal root is missing, and the lingual root is broken 9.3 mm from the cervical margin.

RM 2 (Figure 4y)
this tooth is not yet fully erupted and is obscured by both adhering breccia matrix and the remaining maxillary bone.the protocone is visible as is a weakly expressed hypocone.the distal portion of the metacone is also visible.the central fovea is shallow and wide, joining with the posterior fovea.the DMR is cut centrally by an accessory cuspule.the small hypocone is delineated by a groove running from the central fovea for the lingual face.although obscured, root development appears incomplete.
Despite little to no molarisation of the third premolars, molarisation of the fourth premolars, deep and narrow central fovea of all premolars and permanent molars, clear cervical eminences on incisors and the canine, and marked development of the DMRs of the deciduous molars support attribution to P. robustus.

DNH 121: Ldm 2 (Figure 5a-c)
this is a fragmentary and heavily worn tooth preserving a small segment of the distobuccal root and the crown, excluding the mesial-most portion of the crown enamel.a large lingual dentine lake occupies both the protocone and hypocone.this exposure flares buccally at its mesial extension and nearly joins a moderate dentine exposure on the paracone.small dentine exposures are also present on the metacone and along the distal marginal ridge.the remaining occlusal surface is worn nearly flat.Occlusal outline is square with a slight distal tapering as is commonly observed in P. robustus.although enamel cracking obscures the majority of crown morphology, a shallow posterior fovea is visible.the buccal groove is very faint although a clear fovea delineates the paracone and metacone.the distal interproximal contact facet is large and reaches the occlusal surface.the roots have been almost completely resorbed to the point of tooth shedding.it is unlikely this tooth would have remained in the mouth in this stage.
DNH 122: LM 2 (Figure 5d-f ) this tooth preserves the crown, excluding a distal-lingual portion of the enamel between the entoconid and the hypoconid.the roots are also preserved excluding the extreme tips.Occlusal outline is rectangular with a slight mesiolingual extension and distinct distal tapering.Occlusal surface is worn mostly flat with a small dentine exposure on the protoconid.a small anterior fovea is visible distal to a thick MMR. the distal trigonid crest is broken by a small pit.Both the central fovea and the posterior fovea are pronounced despite advanced wear. the posterior fovea is intruded upon by enamel extensions originating from the hypoconid, entoconid, hypoconulid, and DMR. the mesial interproximal contact facet is large, covering the majority of the mesial face and reaching the occlusal plane.Only a small portion of the distal interproximal contact facet is preserved, but it appears large and would likely have covered the majority of the distal face.a small enamel pit is visible on the buccal face between the hypoconid and the hypoconulid, which may be a remnant of a high distobuccal groove.another small pit is visible on the occlusal surface between the protoconid and the hypoconid, and the buccal face shows multiple hypoplastic pits.the distal roots are fused completely to their point of breakage, although mesiodistal compression suggests the presence of two distinct radicular canals.the mesial roots are fused for approximately three-quarters of their length.the absolute size of this specimen as well as the relatively thick enamel seen along fractured surfaces supports an attribution to P. robustus, despite advanced wear obscuring occlusal morphology.
DNH 125: Ldm 1 (Figure 5g-i ) this is an isolated and well-preserved deciduous tooth.Developmentally, it is at crown completion and unerupted.the protoconid, metaconid, entoconid, and hypoconid are well developed, while the hypoconulid is small yet distinct.the protoconid and metaconid are of a similar size and are larger than all other cusps.the protoconid and metaconid are slightly higher than the hypoconid and entoconid.the protoconid is positioned mesially to the metaconid.the trigonid is mesiodistally dominant over the talonid.the occlusal outline is rectangular aside from a large, protuberant, mesioconulid.indeed, the crown appears buccolingually compressed, as is often observed within south african Homo specimens.the anterior fovea presents as a broad triangular basin, is shifted slightly buccally, and is partially bordered by a thin MMR. the lingual end of the MMR is low.the MMR is separated from the metaconid by a distinct fissure and rises to a large mesioconulid.this accessory cusp is delineated by faint grooves down both the buccal and mesial faces.a weakly expressed tuberculum molare is present, as is a very small but clearly delineated c6. the posterior fovea and the deep central fovea are joined by a deep longitudinal groove.the mesiobuccal groove is deep and terminates halfway down the crown.the distobuccal groove is faint.a lingual groove is also evident down the lingual face.the mesial and distal roots are fused with marked mesiodistal compression.Both roots are in early stages of development and do not extend far from the cervical margin.

Maxillary molar fragment (Figure 5j-l)
this fragment only preserves part of the crown, although it is possible this specimen represents an upper first molar.the metacone is almost completely preserved, while only a small portion of the paracone and hypocone remain.Despite this specimen's fragmentary nature, a deep central fovea and a wide buccal groove are suggested.Further, thick enamel seen along the fracture plane suggests that this specimen should be attributed to P. robustus.this attribution is supported by the bulbous nature of the preserved cusps and the curved buccal face.the buccal and distal portions of the yet incomplete roots also preserve.

DNH 128: L c (Figure 5m-n)
this is an isolated element with enamel missing from the mesial face and a fragment of the root missing on the distal side of the tip.heavy apical wear has greatly reduced crown height and left a large dentine exposure covering the majority of the occlusal surface.the worn surface slopes linguo-distally.the mesial interproximal contact facet is not preserved, while the distal interproximal contact facet is large and lingually skewed, covering the entire distal face and reaching the occlusal surface.Despite wear, an accessory ridge is evident mesial to the midline.the labial face preserves a well-developed DMR consistent with P. robustus material.the weak cervical eminence is skewed mesially.On the lingual face, the cervical eminence is pronounced and markedly skewed mesially with a pronounced ridge on the distal edge.the root is long and tilted mesially with a groove along the distal face suggesting the possibility of a double radicular canal.the crown convexity below mid-crown is preserved and supports an attribution to P. robustus.

DNH 129: RP 3 Fragment (Figure 5o-q)
the tooth is poorly preserved with only half the crown and a portion of the lingual root remaining.Wear is moderate with small dentine exposure on the metaconid.Mesial and distal interproximal contact facets both occupy their remaining faces and reach the occlusal surface.the lingual root is broken 4.8 mm from the cervical margin.the mesiodistal metrics of this specimen and the relatively thick enamel visible on the broken surface support an attribution to P. robustus.
DNH 132: Rdm 2 (Figure 6a-c) this tooth is heavily fragmented, missing the entire distal face and most of the central crown, as well as large portions of all three roots and enamel along the entire mesial face and half of the buccal face.this tooth is heavily worn with dentine exposures on the paracone, metacone and hypocone.the protocone and hypocone are delineated by a deeply incised V-shaped groove that continues down the superior portion of the lingual face consistent with Homo material.Owing to the fragmentary nature of this tooth, much of the occlusal morphology has been lost.Despite the relatively large size of this specimen, the enamel is comparatively thin, and the crown walls are quite straight, supporting an attribution to Homo. the remaining roots are markedly splayed and display some signs of resorption.
DNH 133: RP 4 (Figure 6d-f) this well-preserved tooth is at crown completion with minor pavement cracking and no root formation.the central fovea is deep with both the protocone and paracone large and pronounced.Both the MMR and the DMR are distinct though not notably thick.a posterior fovea is clearly visible separate from the central fovea.an incipient cuspule is visible just mesial to the protocone.a distolingual groove is suggested delineating this cuspule from the protocone and continuing down the lingual face but is obscured by adhering breccia matrix.Multiple incipient cuspules are visible on the MMR consistent with P. robustus material.While talonid expansion is not extreme in this specimen, molarisation is evident. in overall crown size, this tooth sits beyond the range previously seen at Drimolen and instead falls within the range of swartkrans P. robustus.
DNH 136: Rdm 1 (Figure 6g-i ) this isolated element is moderately well preserved with small cracks running through the protocone, paracone, and metacone.Only the distobuccal root is fully preserved and is not yet at completion.Wear is minimal, and only slight cusp polishing is evident.Neither the mesial nor the distal interproximal contact facet is visible.the protocone is the largest cusp followed by the hypocone and then by the equivalently sized paracone and metacone.the occlusal outline is rhomboidal with a marked mesiobuccal extension.Further, the lingual side of the protocone shows moderate bevelling and some inflation.While this is not an extreme example, this morphology is consistent with P. robustus material.a tuberculum molare is present but development is weak, as expected within P. robustus.the MMR is thick and contains a strongly developed parastyle delineated by a short but pronounced mesiobuccal ridge.an incipient epiconule is also present.the anterior fovea presents as a single linear fissure, fully enclosed by a low epicrista.the central fovea is deep, and the buccal and lingual cusps are well delineated from one another.a Figure 6.DnH 132,133,136,138,140,143,146,147,148,and  distobuccal groove is also evident though faint and shallow.the lingual groove is deep and narrow, terminating more than halfway down the crown.No carabelli trait is evident.the posterior fovea is represented by a long fissure and is bounded by a moderately thick and high DMR consistent with P. robustus material.Orientation of the distal trigon crest changes from transverse to oblique at midline.an incipient postentoconule is present centrally on the DMR.Perikymata are clearly visible along the buccal face.the overall size of the crown and bulbous nature of the cusps as well as the deep and narrow nature of the cuspal furrows support an attribution to P. robustus.
DNH 138: L C (Figure 6j-k ) this specimen is well preserved with a small fragment of enamel missing at the occlusal edge of the labial face.Wear is heavy with a large amount of crown height loss and a large dentine exposure occupying the majority of the occlusal surface.Wear is angled lingually.Both the mesial and the distal interproximal contact facets are large with the superior portions absent due to wear. the root is long and straight exhibiting marked mesiodistal compression.longitudinal grooves are present on both the mesial and distal faces.some root resorption is evident.a marked hypoplastic line is visible on the labial surface just superior to a large cervical eminence.though much of the morphological detail has been lost to wear, this specimen compares well metrically to DNh 7, supporting an attribution to P. robustus.

DNH 140: Molar fragment (Figure 6l-n)
this molar fragment is heavily worn with dentine exposures suggested along the broken edges of the occlusal surface.the crown is broken at the cervical margin and preserves only a portion of one cusp.No morphological details can be described.thick enamel suggests an attribution to P. robustus though the lack of preserved morphological detail does not support a clear attribution.
DNH 143: Enamel fragments (Figure 6o-q ) this specimen consists of five small enamel fragments.No morphological detail can be described.this specimen has been attributed to 'cf.hominin?' based on the thickness of the enamel flakes and its possible spatial association with DNh 122.
DNH 146: Rdm 2 (Figure 6r-t ) this specimen is well preserved and moderately worn.small dentine exposures are visible on the paracone, metacone, and hypocone, two small dentine exposures on the protocone, and a large dentine exposure is visible in place of the MMR and anterior fovea.the central fovea is deep and narrow, extending lingually to delineate the protocone and hypocone and terminating at the cervical margin.the buccal groove manifests as an equally deep and narrow groove, and delineates the paracone and metacone terminating halfway down the buccal crown in a small pit. the posterior fovea is deep, bounded by a thick and tall DMR and a crista obliqua.
these occlusal features are similar to occlusal morphology observed in P. robustus specimens from both DMQ and swartkrans.this as well as the large overall size of the specimen supports an attribution to this taxon.the lingual root is greatly diverged from the buccal roots.the lingual root is broken 7.5 mm from the cervical margin.Both buccal roots show resorption.
DNH 147: RM 3 (Figure 6u-w ) this specimen is poorly preserved with only one root and the buccal face remaining.a portion of the mesial icF is visible, suggesting the tooth had reached functional occlusion.the preserved portion of the occlusal surface shows no morphology, as the entire surface has been obscured by an extremely large pool of exposed dentine.thick enamel preserved along the circumference of the crown supports an attribution to P. robustus, though the lack of morphological detail hinders a confident attribution.the preserved roots are fused, but mesiodistal compression suggests the presence of two radicular canals.
DNH 148: RM 2 (Figure 6x-z ) this specimen is well preserved and at crown completion with little to no root development.the occlusal outline is slightly rhomboid with minor mesiobuccal extension.the protocone and paracone are the largest cusps and roughly equivalent in size followed by the equivalently sized metacone and hypocone.the central fovea is deep and relatively narrow, joined with both the anterior fovea and posterior fovea by a deep longitudinal groove.Because of this, the anterior fovea is indistinguishable.the MMR is thick.a carabelli's trait is evident mesiolingually from the protocone, delineated by a deep groove down the lingual portion of the mesial face ending in a distinct pit.an epiconule is also visible on the MMR, located buccally to the carabelli's trait.a large enamel extension protrudes distobuccally from the protocone and is located centrally.a small pit is located near the apex of both the paracone and the metacone acting to partially separate the central portion of the cusps.Despite joining with the central fovea, the posterior fovea is still distinct as a deep, longitudinal pit. the DMR is low but thick.a small but distinct postentoconule is visible on the DMR. the buccal groove is deep and narrow, consistent with P. robustus material, terminating high on the buccal face.the lingual groove is also deep and narrow, terminating halfway down the lingual face.the bulbous nature of the cusps and the curved buccal and lingual faces are consistent with P. robustus material.

DNH 149: LI 1 (Figure 6aa-bb)
this specimen is well preserved with a crack running mesiodistally through the crown.heavy wear has left a large rectangular dentine exposure occupying the incisal edge and has greatly reduced crown height.the wear plane is sloped lingually and distally.Perikymata are visible on the labial face.On the lingual face, the moderately developed cervical eminence is located centrally.Both the MMR and DMR are visible, though the former is more developed.the root shows mesiolingual hooking at the tip.No mesiodistal compression is evident.Owing to mesiodistal reduction and labiolingual expansion, the apical outline of the crown is nearly square consistent with P. robustus material.

Taxonomic attributions
Based on the preserved crown and root anatomy, taxonomic designations yielded three categories: P. robustus, south african Homo, and indeterminate (cf.hominin).specimens in the third category consist of either fragmentary material or specimens without distinct diagnostic morphology.a number of morphological features were examined for the purpose of species attribution.along with the features described above, and those described in leece et al. ( 2022), we evaluated overall occlusal outline (i.e.square, rectangular, rhomboidal), angulation of enamel at the cervical margin (i.e.'bulbous' versus 'straight' lateral faces), presence or absence of accessory cusps, angulation of cusps in relation to one another, relative size and depth of fovea and other pitting, enamel thickness (when visible), and the overall buccolingual and mesiodistal dimensions.taxonomic attributions are presented in table 4.

Statistical analyses
significant results for all metric comparisons are highlighted in bold and italics, and are shown in table 5.

Anterior dentition
No incisors are present within the KBM4-7 assemblage, thereby limiting our analysis to the DMQ and sM1hR assemblages.Further interpretation of these data is limited by the small sample and the frequency with which incisors from these assemblages show a high degree of incisal wear.
No significant differences between samples are present in either dimension for canines from either dental arcade.interestingly, canines are the only elements where the mean size is larger (albeit non-significantly) within the DMQ assemblage in at least one dimension (see scatterplots, Figure 7).this is obviously not compatible with a scenario in which DMQ P. robustus represent small females within a highly dimorphic species, as suggested by Moggi-cecchi et al. (2010).conversely, it is more compatible with the micro-evolution theory presented by Martin et al. (2021) where the DMQ assemblage represents an earlier morph of P. robustus and so may not exhibit all traditional traits such as reduced canines.

Premolars
Overall, P3s from Drimolen are slightly larger than those from swartkrans, with the differences being greatest for maxillary P 3 s (table 5 and Figure 8).Mann-Whitney U-tests suggest significant differences in the buccolingual dimensions (table 5).indeed, in the P 3 , significant differences between the DMQ and sM1hR assemblages are present in both mesiodistal and buccolingual dimensions (table 5). the bivariate plots show DMQ P 3 distributions overlap slightly with only the smallest sM1hR specimens (Figure 8).
in the bivariate plot, the P 4 s of Drimolen overlap almost entirely with the sM1hR assemblage, despite showing a high degree of separation between the sample means, with the sM1hR sample showing the larger mean (Figure 9).No significant difference is reported in the P 4 s and the bivariate plots show a high degree of overlap with similar means between the samples (Figure 9; table 5).Overall, the DMQ P 3 s are among the largest specimens, a pattern that again runs contrary to the proposed dimorphic pattern (where the DMQ individuals would represent small females while the swartkrans material represents males).Premolar metric data also fail to support expectations of a micro-evolution hypothesis (where the DMQ individuals would represent a more basal member of P. robustus and should therefore possess smaller premolars) (table 5; Figure 9).

Molars
Both upper and lower M1s and M2s show a high degree of overlap in both dimensions in the bivariate plots (Figures 10 and 11).Despite this overlap, the M 1 s and M 2 s both show statistically significant differences between the two assemblages in breadth while the M 2 s show significant differences only in length, with the DMQ molars being smaller in both cases (table 5). the M 1 comparison approaches, but does not quite reach, significance (p = 0.051).Overlap in the M 3 is complete to the point that both the largest individual and smallest individual are represented by M 3 s from the DMQ P. robustus assemblage (Figure 12; table 5).conversely, overlap within the M 3 s is minimal, and significant differences were found in both the buccolingual and mesiodistal dimensions, with the DMQ M 3 s being significantly smaller (Figure 12; table 5).

Deciduous dentition
the current sample size of P. robustus deciduous dentition prohibits statistical analysis.Nevertheless, comparison between the DMQ, sM1hR, and (where possible) KBM4-7 samples provide some insight.as with the permanent dentition, the d c , dm 2 , di 1 , and dm 1 are smaller in the DMQ assemblage than in the sM1hR assemblage.the difference between   the DMQ and sM1hR dm 2 and dm 1 is negligible to absent.the most marked difference can be seen between the DMQ and sM1hR dm 1 s.contrary to the pattern seen in the permanent dentition, where the most consistent differences are seen in the buccolingual dimension, it is the mesiodistal dimension that shows more marked differences in the deciduous teeth.comparison to P. robustus deciduous dentition from KBM4-7 is more varied.the DMQ material is either no different in size (dm 1 ), smaller (d c ), or slightly larger (dm 2 ). it must be noted, of course, that the sample sizes employed in these comparisons are small, ranging between n = 5 and n = 1, and averaging n = 2.9.

Discussion
ever since the first fossils were discovered at Drimolen, researchers have disagreed regarding how they fit within the overall sample of south african P. robustus.For instance, Moggi-cecchi et al. ( 2010) and lockwood et al. ( 2007) both suggested that the size differences between DMQ and swartkrans P. robustus are best explained by sexual dimorphism, with the DMQ sample being dominated by females, and the swartkrans sample dominated by males.they further argued that this partitioning could be due to taphonomic bias, with swartkrans characterised as a carnivore accumulation of mostly males, and Drimolen a sleeping site of mostly females and juveniles.in a recent analysis of new material from DMQ, Martin et al. (2021) documented the presence of a comparatively small male specimen (DNh 155) within the DMQ assemblage, which shared a unique suite of primitive and derived character traits with the female DNh 7 rather than other presumed male specimens from swartkrans.this pattern of variation between Drimolen and swartkrans  weakens support for the hypothesis that differences between the swartkrans and Drimolen P. robustus assemblages are reflective of high degrees of sexual dimorphism.lockwood et al. ( 2007) compared patterns and degrees of sexual dimorphism in P. robustus and Gorilla gorilla.studies that focus on differential site accumulation models that characterise Drimolen as a sleep site, such as Riga et al. ( 2019), hypothesise behavioural similarities between P. robustus and baboons.For this reason, dental dimorphism in these extant species should be assessed as compared with results found within this study.in large-bodied sexually dimorphic primates such as gorillas or baboons, sexual dimorphism drives the buccolingual and mesiodistal metrical differences of the maxillary and mandibular canines, third premolars, second, and third molars to the exclusion of other tooth positions (lauer 1975;Plavcan 2001).Differences demonstrated within this study are primarily in the maxillary dentition rather than the mandibular dentition and in the buccolingual dimension rather than overall size differences.specifically, in regard to tooth size, Moggi-cecchi et al. (2010) found seven tooth positions (out of 16) to be statistically significantly smaller at DMQ than at swartkrans.this is in broad agreement with our study, wherein we find eight (out of 16) tooth types to be statistically significantly smaller.Of the differences found herein, a clear pattern emerges whereby significant differences in size (in at least one dimension) are present for all postcanine maxillary teeth, whereas fewer differences exist in the mandibular arcade, where differences in size were only found to occur in the P 3 (Bl only), M 1 (MD only), and M 2 (MD only).Moreover, the most consistent pattern emerging from the dental size comparisons is that the maxillary postcanine teeth are buccolingually narrower in the DMQ sample than in the sM1hR sample.
this is not a pattern that can obviously be explained by sexual dimorphism in so far as it does not obviously reflect any known pattern in extant primate species.as dental size and shape are intrinsically linked with dietary adaptions, as are many of the morphological features identified by Martin et al. (2021), we suggest that this pattern is better explained by broader micro-evolutionary changes in the masticatory apparatus, probably as a response to changing adaptions around dietary behaviours.sponheimer et al. ( 2006) presented evidence that south african hominins moved from a more restricted, invariable diet to a more variable, opportunistic, seasonal diet. it was hypothesised that the highly derived masticatory apparatus seen in Paranthropus reflects a broader rather than narrower diet as compared with A. africanus (sponheimer et al. 2006).if true, it is possible the DMQ P. robustus specimens are at an earlier point in this adaptive change where they have not yet accrued the full suite of primitive and derived traits seen in later populations, such as those from swartkrans and Kromdraai.however, it must be noted that the sponheimer et al. ( 2006) study sampled only P. robustus specimens from swartkrans M1 and did not include any DMQ specimens.
the cV between the DMQ and swartkrans assemblages has been used to suggest that the former, with a higher cV (6.9), represents a less biased sample than the swartkrans assemblage, with a lower cV (6.1), which has been interpreted as male-dominated (Moggi-cecchi et al. 2010).however, intraspecific variation differs greatly across primate species -for example, 21 contemporary species of Cercopithecus varied less than four subspecies of Papio hamadryas (Plavcan and cope 2001).additionally, Plavcan and cope (2001) warn against the comparison of temporally disparate palaeo-populations and that testing a fossil assemblage against standard cV values of a living sample is not an analogically strong piece of evidence.the presence of temporal variation will inflate the expected cV of any fossil taxon since, to have enough fossil material to undertake this analysis, fossils from different temporal horizons are usually analysed (simpson 1961;cope 1993;Plavcan 1993;Plavcan and cope 2001).another reason for the difference in cV noted by Moggi-cecchi et al. (2010) could be that they used P. robustus material from four temporally distinct deposits at swartkrans, covering a period between 2.2 and perhaps as young as 1.0-0.8Ma. in contrast, the DMQ assemblage represents a temporally constrained palaeo population deposited within ~90 ka (herries et al. 2020).
With the addition of newly discovered teeth from both swartkrans and Drimolen, the whole assemblage from swartkrans now has a cV of 6.3 (leece et al. 2022). in this study, Drimolen has a cV of 7.1 (table 2).this suggests that the cV values are highly susceptible to change as a function of small samples sizes because the addition of a single new specimen has a material impact on the overall sample. in the case of this study, adding new specimens has resulted in the data not supporting Moggi-cecchi et al. 's (2010) original interpretation.When the swartkrans sample was limited to sM1hR in this study, the average cV was found to hold constant at 6.3, contra to Plavcan and cope's (2001) argument that limiting the temporal variation of the assemblage will lead to a drop in cV. that said, the sample is heavily weighted towards sM1hR specimens potentially skewing this comparison.it is possible that a male bias within the sM1hR assemblage accounts for the lower cV as compared to DMQ, though it must be noted that a difference of 0.8 between sM1hR and DMQ is not substantial or statistically meaningful. in fact, the values found for DMQ, swartkrans (all members), and sM1hR, all fall within the expected range for a single species with low degrees of sexual dimorphism (Plavcan and cope 2001).
the Kromdraai Member 4-7 (KBM4-7) P. robustus dental sample is prohibitively small for statistical analyses such as those presented here.While certain specimens, such as tM1517, can be confidently attributed to Members 4-7 and are thought to be younger than 1.95 Ma based on palaeomagnetism, much of the Kromdraai material cannot be definitively contextualised or dated, as they were excavated from decalcified deposits (Braga et al. 2016;herries 2022).however, bivariate plots of dental size data from KBM4-7, DMQ, and sM1hR do show that the KBM4-7 P. robustus specimens are consistently found at the point of overlap between the DMQ and sM1hR samples .the only exception to this is the maxillary P 3 , where the singular KBM4-7 specimen sits well above the range seen from DMQ (Figure 8).interpretation is limited by the small KBM4-7 sample size, though this could signify that these specimens (tM 1517, tM 1600, and tM 1601) sit at an intermediate point between the DMQ and sM1hR samples within the P. robustus lineage.
a study of semi-circular canals of P. robustus from DMQ, sM1hR, and Kromdraai Unit P (age unknown, but older than Members 4-7; herries 2022) as compared with A. africanus conducted by Braga et al. (2022) also noted variation between P. robustus samples.indeed, Braga et al. (2022) found similarities between the DMQ and Kromdraai Unit P samples, to the exclusion of sM1hR, that aligned these samples more closely with A. africanus.the apparent plesiomorphic traits identified in this study were interpreted as evidence for the DMQ and Kromdraai Unit P samples being closer to the ancestral australopithecine morph than the sM1hR sample (Braga et al. 2022).the age of Unit P at Kromdraai is not yet known, but it is suggested to be older than Member 4-7 (herries 2022).as such, the pattern of difference between DMQ, Kromdraai Unit P, and sM1hR could be due to temporal variation.Dental samples are not yet open access for analysis to compare against DMQ.
although one might not reject the sexual dimorphism hypothesis based solely on the pattern found here, in conjunction with the cranial evidence (Martin et al. 2021;Braga et al. 2022) and the chronological difference between the DMQ and sM1hR samples (herries et al. 2020), it would appear that the micro-evolution hypothesis (Martin et al. 2021) is more parsimonious than lockwood et al. 's (2007) and Moggi-cecchi et al. 's (2010) nested suite of hypotheses positing complex taphonomic processes acting on a highly sexually dimorphic population.although we do not doubt that P. robustus was sexually dimorphic (e.g.compare DNh 7 and DNh 155; Martin et al. 2021), the differences between the DMQ and sM1hR samples are not easily characterised as the differences expected between males and females in a highly sexually dimorphic population.We instead suggest (after Martin et al. 2021) that the dental size differences between the DMQ and sM1hR P. robustus samples represent a micro-evolutionary change between time-successive populations.
47 below datum and is at a similar height to the DNh 152 cranium (−3.23 to 2.99 m) from the italian Job Pinnacle.DNh 128 was recovered from lightly decalcified breccia at the base of the italian Job Pinnacle (−3.05 m below datum) close to P. robustus cranium DNh 152.two teeth (DNh 133, 146) were recovered from the base of the Jangi Buttress during the excavations that recovered the DNh 134 H. aff.erectus cranium and range between −5.46 m and −4.57below datum.the Jangi Buttress represents lightly calcified in situ breccia.DNh 132 and 148 come from significantly different depths (−3.21 and −4.79 m, respectively) in an in situ buttress of breccia (Bryn Buttress) on the NW corner of the cea.DNh 107, 108 and 125 come from decalcified deposits within the cea surrounding the DNh 7 collapsed block.

Figure 1 .
Figure 1.(A) Photogrammetry 3D model of DmQ facing nW showing the location of the hominin fossils described in this paper (orange circles) and their relationship to published crania (DnH 7, DnH 134, DnH 152, DnH 155; blue circles).major section locations as described in Herries et al. (2020) are also shown (WC = Warthog Cave, JB = Jangi Buttress, mP = marcel Pinnacle, iJ = italian Job Pinnacle, WoJ = Walls of Jericho Pinnacle).The locations of us-EsR dates are also shown (yellow circles) as well as the location of the base of the olduvai subchron at 1.95 ma.note that DnH 7's location is now in mid-air above the DnH 7 block, as this block has been reduced since it was discovered.see location in plan B. (B) Photogrammetry plan of the cave showing the location of the hominin fossil samples within DmQ.

Table 1 .
Dental metrics (in millimetres) for new DmQ specimens.

Table 2 .
summary statistics of DmQ P. robustus permanent dentition.

Table 4 .
DmQ specimens presented in this study and associated taxonomic attributions.

Table 5 .
p-values of mann-Whitney U-tests comparing DmQ and sm1HR P. robustus teeth.
significant values (p < 0.05) are shown in bold and italics.