Australia: The Land Where Time Began

A biography of the Australian continent 

Ancient Teeth, Phenetic Affinities, African Hominins, Where Homo naledi Fits in

In 2015 a new species of Homo, Homo Naledi, was described, that was based on the remains of a hominin Skelton uncovered from the Dinaledi Chamber of the Rising Star cave system, South Africa. In this paper a new quantitative analysis is presented in which 78 nonmetric crown and root traits of the permanent dentition were compared among samples of H naledi, which included remains from the Lesedi chamber, which were discovered recently, as well as 8 other species from Africa: Australopithecus afarensis, Australopithecus Africanis, Paranthropus boisei, Paranthropus robustus, Homo habilis, Homo erectus, Homo sp. from the Middle Pleistocene, and  Homo sapiens from the Pleistocene and Holocene. Phenetic affinities were calculated among samples to evaluate interspecific relatedness, by using the mean measure of divergence distance statistic. Comparison of the results with those that had been obtained previously, to further assess the taxonomic validity of the Rising Star hominin species, was the aim of this study.  The former species is characterised, however, by its retention and full expression of features that related to the main cusps, as well as the number of roots, with almost no accessory traits, which included many that are plesiomorphic in both extinct and extant African hominins, based on various cladistics studies. As such, additional support is provided by the findings presented here for the taxonomic validity of H. naledi as a distinct Homo species.

In the taxonomic description of skeletal and dental remains (Berger et al., 2015) found what they considered to be enough similarities Homo habilis, Homo rudolfensis, Homo erectus and  Homo heidelbergensis to assign a new fossil to a the genus Homo, though with enough distinctions to merit a separate species. Academic discussion concerning the validity of H. naledi as a new and distinct species followed (Randolph-Quinney, 2015a: 2; Von Mirbach, 2015; Schwartz, 2015; among other issues, e.g., Val, 2016), and a researcher even suggested that it be assigned to the genus Australopithecus as being more appropriate (see discussion in Randolph-Quinney, 2015a:2), (at that time 1) a phylogenetic analysis had at that time not been conducted (see Randolph-Quinney, 2015b:2) the remains that had been recovered from Lesedi Chamber of the Rising Star cave system, whose morphology supports the original taxonomic description (Hawkes et al., 3017), and 3) date the H. naledi recovered from the Dinaledi chamber (Berger et al., 2015), about 335-236, was not known (Dirks et al., 2017).

In a quantitative study that was based on cranial data that had been published, it was agreed (Thackeray, 2015) that it was justified to assign a distinct species status within the genus Homo. Of Thackeray’s 11 African samples, H. naledi appears to be most similar to H. habilis and H. rudolfensis and H. erectus, though to a lesser degree. These findings were based on a phenetic method (Thackeray, 2005) that yielded standard errors of m-coefficients from pairwise comparisons of specimens in regression equations. Criticism was not lacking of this methodological approach, as the results are said to be dependent on variable number, as well as other issues (Gordon & Wood, 2013). Bayesian phylogenetic analyses was used more recently on a supermatrix of craniodental data in H. naledi and 19 hominin species from Africa, Europe and Asia (Dembo et al., 2016). They concurred that H. naledi is a separate species, in that it forms a clade with members of the genus Homo, though its position is ambiguous relative to species within that clade (Dembo et al., 2016). Irish et al. say this was an ambitious first step in attempting a cladistic analysis of the species that was newly described using many comparative hominin samples and characters. As was acknowledged by the authors, however, there are several methodological concerns that involve the treatment of various characters and character states.

In this paper Irish et al. present the results of a phenetic analysis that was based on 78 nonmetric dental traits, in order to explore further how H. naledi may relate to other hominin species. These traits have been observed in a range of fossil hominins, driven to a large extent by geographic provenance, the focus of the study was H. naledi and other African species, like (Thackeray, 2015) and, for the most part (Berger et al., 2015) and (Hawks et al., 2017). The extent of the skeletal assemblage from the Dinaledi Chamber (≥15) individuals; Berger et al., 2015) and Lesedi Camber (≥3 individuals; Hawks et al., 2017) in the Rising Star cave system also influences the approach. There were 10 samples that comprised 8 different species of Australopithecus, Paranthropus and Homo that were compared with H. naledi, though only 8 of these contained enough specimens to record the traits multiple times, at least to the extent possible, were analysed quantitatively here; the aim is to consider at least some of the variation that occurs within groups) or species in this case) relative to that between groups (e.g. Relethford, 1994, 2001; Templeton, 1999; Leigh et al., 2003).

Standard coding criteria does not address adequately the variability of character (see concerns of Asfaw et al., 1999) and (Curnoe & Thorne, 2003), also below – given the major increase in trait number, as well as larger sample sizes relative to these prior studies, though a cladistic analysis of nonmetric dental data is possible (e.g. Bailey, 2002a) and (Irish et al., 2013, 2104a). Alternate strategies for the coding of character states are available and have been investigated (e.g., Bailey, 2002a). All have weaknesses (see below), however, to match their strengths (Thiele, 1993; Stringer et al., 1997; Wiens, 1995, 2001; Ried & Sidwell, 2002; Schols et al., 2004). Therefore a non-phylogenetic method, the mean measure of divergence distance statistic (MMD), was applied here to trait frequencies from the 9 largest samples in order to calculate phenetic frequencies among H. naledi as well as these other African hominins (Sjøvold, 1977; Irish, 2010).

In previous studies the MMD has been used to compare dental data in species dated to the Pliocene-Pleistocene, as was allowed by sample size (Irish, 1998; Bailey, 2000, 2002a; Martinôn-Torres et al., 2012). MMD is preferred over other distance measures, including the pseudo-Mahalanobis D2, by many researchers who are studying recent humans. The most critical among several advantages is that MMD uses summary data instead of individual trait scores to address missing data that characterise fragmentary specimens in archaeological and, as above, palaeontological, contexts (see Materials and methods in original paper). In order to visualise patterning in intersample distance a supplementary technique, such as multidimensional scaling in this study. It is almost certain that these phenotypic patterns are a reflection of underlying genetic variation (Rightmire & Deacon, 1999: 2); it is therefore assumed that a reasonable approximation of genetic relatedness is provided by phenetic similarity, at least based on findings of recent human studies (e.g. Scott et al., 1983; Larsen, 1997; Scott & Turner, 1997; Martinôn-Torres et al., 2007, 2012; Hughes & Townsend, 2013a, b; Rathmann et al., 2017).

In summary, the intent of this study was 2-fold:

1)    To use the dental nonmetric data to assess interspecific relationships based on overall similarity among the 11 African samples, and

2)    Compare these results with those from Thackeray (2015) and (Dembo et al., 2016), based on different data and methods.

The overarching objective, then is to continue to work towards an improved understanding of where and how H. naledi fits in relative to other hominin species.

Discussion and conclusions

Beginning with the quantitative comparisons between H. naledi and the 8 other largest samples, the effectiveness of the dental traits and MMD (in spite of many small sample sizes of n < 10) is evidenced by the similarity of groupings to those for the same species in earlier studies that used cladistic analyses and alternate data (Strait et al., 1997; Strait & Grine, 2004; Smith & Grine, 2004; and Smith & Grine, 2008; Berger et al., 2010). The consistency in patterning from MMD vs PCA using different numbers of traits further suggests that the results are ‘real’ and not a byproduct of method or illustrative device. It appears H. naledi is most like others in the genus Homo of all the comparative samples, though it is distinct from them in, most notably, its extreme position on the Dimension 2 and Factor Score 2 axes. It is found by an inspection of individual MMD distances, focused on the 45 trait matrix that not only are all pairwise differences significant, but that H. naledi is the most divergent of all samples that are non-H. Sapiens. The average pairwise distance between H. naledi and the 6 other samples from the Pliocene-Pleistocene is 0.445; the same value among the latter 6 is 0.111. In the taxonomic description of H. naledi from the Dinaledi Chamber, a summary has been provided (Berger et al., 2015:20, 24) of features that are likely to have been responsible for this divergence, i.e., the teeth, as well as being small, that have a crown morphology that is remarkably simple, lacking many derived features that are shared by Homo and H. sapiens from the Middle and Late Pleistocene (below). Crowns of specimens from the Lesedi Chamber display the same patterning (Hawks et al., 2017).

According to Irish et al., sample size must be considered, though this description is quantified in Table 1, where 36 of 78 traits are not present in Homo naledi, most of which are present in other fossils and/or recent samples – at high frequencies in many cases. Most of these can be considered to be additive or mass-additive traits (above). There are additional traits present, though at much lower rates than in other samples, notably in older specimens. This is not to say there are no ‘accessory’ traits that occur frequently in H. naledi, e.g., double shovelling, C1 distal accessory ridge, and possibly upper premolar buccal accessory ridges. That being said, the species exhibits the highest or one of the highest percentage of traits that relate to retention and expression of principal (main) cusps, as well as roots. Though not a cladistic study, it is known that such traits are plesiomorphic (see full reference list in Irish, 1997; Irish & Guatelli-Steinberg, 2003) that is based on the presence in various fossil hominins and in extinct and extant non-human primates (as are many of those listed above in the paper). Included among them are: little or no root number reduction in P3-M3, P3, and M1 to M3, along with a lack of reduction of the M1-3 metacones and hypercones (grades 4-5) and presence of the 5th principal cusp, plus Y-pattern on M1-M­3.

The above findings are supported by qualitative comparisons between H. naledi and the 2 pencontemporaneous samples, Middle Pleistocene Homo and Middle/Late Pleistocene H. sapiens, at least based on those teeth with multiple observations. H. naledi shares a number of similar percentages with 1 or both samples. On the other hand, H. naledi displays some obvious distinctions from these Homo samples. These 2 comparative samples, as mentioned previously, are limited by the overall lack of such African specimens, in particular, Homo from the Middle Pleistocene. There are other data; e.g., (Martinôn-Torres et al., 2012) used ASUDAS to record traits in the Sima de los Huesos (SH) hominins from the Middle Pleistocene, a very small sample of other European H. heidelbergensis (HEI) and European Neanderthals (NEA), among others. As a result of, e.g., the “highly derived and Neanderthal character of the Sima de los Huesos dentitions” (Martinôn-Torres et al., 2012: 55), it was assumed that these samples are too disparate to support the data from Africa; this was found to be a valid concern based on marked percentage differences for many of the 50 traits in common between studies. Also, large, significantly different distances between SH, HEI and NEA and the 5 largest Homo samples from the present study including H. naledi, was yielded by an MMD comparison. Another MMD comparison that included Homo from the Middle Pleistocene, and H. sapiens from the Middle/Late Pleistocene, that was based on 41 traits in common, provides additional support; it is suggested that there is some similarity to the HEI sample, but sample size issues are likely contributory.

The present phenetic findings, overall, parallel those of other authors (Thackeray, 2015) that was based on cranial data, (Dembo et al., 2016), that was based on the phylogenetic analyses of craniodental characters. H. naledi groups most closely with other members of the genus Homo, when dental samples are compared simultaneously. The species, nevertheless, has combinations and expressions of traits that serve to distinguish it from the latter, as is distinguish it from the latter, as is indicated by the distances between samples. As has been described (Berger et al., 2015; Hawks et al, 2017) (also see Schroeder et al., 2017), beyond dental traits the H. naledi crania and postcrania present a mixture of shared and unique features relative to other Homo species, that includes: supraorbital torus that is well-developed and arched, that is separated from a vault by a continuous supratoral sulcus as in H. habilis and H. erectus, marked angular and occipital tori as in H. erectus, and some facial similarities to H. rudolfensis. That said, there are not many characteristics of the cranium that are of more recent Homo that include, notably, large cranial capacity (Garvin et al., 2017). Homo-like features in the in the postcranial skeleton include lower limbs that are relatively long, muscle attachments that are indicative of a striding gait, and modern aspects of the ankles, feet and hands; other traits are reminiscent of earlier species, that include phalanges that are curved, a lower thorax that is wide, and upper limbs that are ape-like (Berger et al., 2015; Harcourt-Smith et al., 2015; Kivell et al., 2015; Feuerriegel et al., 2017; Gavin et al., 2017; Williams et al., 2017; Hawks et al., 2017). As stated, interspecific patterning is consistent with that in earlier studies. The MMD distances, however, beyond indicating the individuality of H. naledi, are telling. The 45-trait matrix is again emphasised and, certainly for the specific portion of the discussion, the make-up of the pooled H. erectus sample should be considered (above). There is no significant difference between the samples of H. sapiens of EA and SA in the Holocene (MMD = 0.011, p ≥ 0.025), as may be expected with contemporary members of the same species in elative proximate geographic regions. They differ significantly from all samples from the Pliocene-Pleistocene, and are most similar to others within the genus Homo in which the MMD distances are:

·        H. erectus (with SA H. sapiens = 0.346; EA H. sapiens = 0.428),

·        H. habilis (SA H. sapiens        = 0.371; EA H. sapiens = 0.438), and finally

·        H. naledi (SA H. sapiens          = 0.586; EA H. sapiens = 0.591).

Both SA and EA H. sapiens are highly distinct from the earlier and/or dead-end species A. afarensis, A. Africanis, P. boisei, and robustus (range 0.689-0.967, mean MMD = .0867). Also suggestive of a temporal component, H. erectus and H. habilis are closer to the latter 4, older samples (MMD = 0.000-0.230), though there is 1 exception, H. erectus, at least, differs significantly from them. 

Beyond comparative purposes with H. naledi, it would clearly be important to obtain larger, African and other samples from the Middle Pleistocene that are more representative (though again see SOM S2-S3) to discern more fully diachronic relatedness. Also a factor in the samples that are better represented are samples of small size, most conspicuously H. habilis and its insignificant distance with P. boisei, P. robustus, and possibly Australopithecus (e.g., MMD = 0.000-0.108, p ≥ 0.025); It has been reported in previous studies (Irish, 1993,2005, 2006, 2010; Irish et al., 2014b) that analogous size-influenced affinities between samples that were otherwise documented to be divergent. I.e., individual MMD distsances can be reduced by small sample sizes, though they most often increase the standard deviation, given the significance formula (above), can yield a type II error; therefore, the patterning of MDS and PCA among samples figures to be a better, overall indicator of interspecific variation. The A. afarensis sample, as compiled in this study, nevertheless, appears to be more similar to H. habilis, as well as H. erectus than was revealed by the cladistic study that was mentioned above (Irish et al., 2013); fewer (n = 18) traits in the latter study and standard coding, which does not address intraspecific variability to the level of the present analyses, is likely to have played a role. Based on the proximity of the Australopithecus and the Paranthropus samples in all figures, the remaining affinities are not unexpected.

In conclusion, the original qualitative taxonomic description (Berger et al., 2015), later quantitative analyses (Thackeray, 2015; Dembo et al., 2016), description of additional specimens (Hawks et al., 2017), and current dental study of African samples that were nonmetric (including SOM S1-S2) all provide findings that are generally concordant to support the inclusion of H. naledi in the genus Homo; or, put another way, there is not much that suggests it does not belong in the genus. It is suggested that the dates that are relatively recent (Dirks et al., 2017) of the Dinaledi remains may also be supportive indirectly of its taxonomic classification, unless a population of relict Australopithecus that survived into the Middle Pleistocene or it is found by future phylogenetic research that the hominins are unique enough to warrant naming a separate genus. The combination of cranial, dental, and postcranial characteristics (above), relative to other member of the genus, supports the designation of H. naledi as a taxonomically valid member of Homo. According to Irish et al. it must be considered likely there is population homogeneity, but, on its own, the highly likely pattern of crown and root morphology in all specimens of H. naledi is indicative of the distinctiveness of the species. In this study (also SOM S2) none of the other samples exhibit an equivalent combination of small teeth with retention and full expression of the principal cusps on all molars, yet accessory and other traits that are known to be plesiomorphic in both African fossil and recent hominins are absent; indeed, the samples of Homo sapiens from the Holocene in the present study (Irish et al., 2014b), in which the teeth are smaller than in H. naledi (compare the diameters of the crowns in Irish et al., 2016 with those of Berger et al., 2015: Table 2), evidence such accessory, plesiomorphic features, such as labial curvature IL1, cusp 7 LM, etc. (Irish et al., 1993,1997, 1998, 2013). Again, therefore, H. naledi appears distinct, dentally, relative to these other species. Lastly, interspecific groupings of these are comparable to those in studies that were published previously. Irish et al. suggest comparative analyses in the future will benefit from the collection of additional data that is not currently presented in this paper, particularly contemporaries of H. naledi in the Middle Pleistocene. As was stated in the introduction, different quantitative and illustrative methods can then be used with these data to further discern relatedness; the goal is simply to understand where and how H. naledi fits in.

Sources & Further reading

  1. Irish, J. D., et al. (2018). "Ancient teeth, phenetic affinities, and African hominins: Another look at where Homo naledi fits in." Journal of Human Evolution 122: 108-123.


Author: M. H. Monroe
Last Updated 20/09/2018
Journey Back Through Time
Experience Australia
Aboriginal Australia
National Parks
Photo Galleries
Site Map
                                                                                           Author: M.H.Monroe  Email:     Sources & Further reading