Hominid Mandibular Corpus Shape Variation and Its Utility For

J. Anat. (2008) 213, pp670–685 doi: 10.1111/j.1469-7580.2008.00989.x

HominidBlackwell Publishing Ltd mandibular corpus shape variation and its utility for recognizing species diversity within fossil Homo Michael R. Lague,1* Nicole J. Collard,2* Brian G. Richmond3 and Bernard A. Wood3 1Natural Sciences & Mathematics, The Richard Stockton College of New Jersey, Pomona, NJ, USA 2Sources Archaeological and Heritage Research Inc., Vancouver, British Columbia, Canada 3Department of Anthropology, The George Washington University, Washington, DC, USA

Abstract

Mandibular corpora are well represented in the hominin fossil record, yet few studies have rigorously assessed the utility of mandibular corpus morphology for species recognition, particularly with respect to the linear dimensions that are most commonly available. In this study, we explored the extent to which commonly preserved mandibular corpus morphology can be used to: (i) discriminate among extant hominid taxa and (ii) support species designations among fossil specimens assigned to the genus Homo. In the first part of the study, discriminant analysis was used to test for significant differences in mandibular corpus shape at different taxonomic levels (genus, species and subspecies) among extant hominid taxa (i.e. Homo, Pan, Gorilla, Pongo). In the second part of the study, we examined shape variation among fossil mandibles assigned to Homo (including H. habilis sensu stricto, H. rudolfensis, early African H. erectus/H. ergaster, late African H. erectus, Asian H. erectus, H. heidelbergensis, H. neanderthalensis and H. sapiens). A novel randomization procedure designed for small samples (and using group ‘distinctness values’) was used to determine whether shape variation among the fossils is consistent with conventional taxonomy (or alternatively, whether a priori taxonomic groupings are completely random with respect to mandibular morphology). The randomization of ‘distinctness values’ was also used on the extant samples to assess the ability of the test to recognize known taxa. The discriminant analysis results demonstrated that, even for a relatively modest set of traditional mandibular corpus measurements, we can detect significant differences among extant hominids at the genus and species levels, and, in some cases, also at the subspecies level. Although the randomization of ‘distinctness values’ test is more conservative than discriminant analysis (based on comparisons with extant specimens), we were able to detect at least four distinct groups among the fossil specimens (i.e. H. sapiens, H. heidelbergensis, Asian H. erectus and a combined ‘African Homo’ group consisting of H. habilis sensu stricto, H. rudolfensis, early African H. erectus/H. ergaster and late African H. erectus). These four groups appear to be distinct at a level similar to, or greater than, that of modern hominid species. In addition, the mandibular corpora of H. neanderthalensis could be distinguished from those of ‘African Homo’, although not from those of H. sapiens, H. heidelbergensis, or the Asian H. erectus group. The results suggest that the features most commonly preserved on the hominin mandibular corpus have some taxonomic utility, although they are unlikely to be useful in generating a reliable alpha taxonomy for early African members of the genus Homo. Key words discriminant analysis; distinctness values; hominin; Homo; mandible; randomization; taxonomy.

the total) comprises well-preserved mandibles or recogniz- Introduction able fragments of the mandibular corpus. As a result of The hominin fossil record contains a relatively large this biased preservation, mandibular data have played an number of mandibular specimens. For example, out of 126 important role in the definition of a number of hominin hominin cranial fossils from Koobi Fora examined by fossil species. Almost 50% of the hominin species identi- Wood (1991), the largest single subset (making up 40% of fied to date have a mandibular specimen as their holotype [e.g. Australopithecus anamensis, Leakey et al. 1995; A. Correspondence afarensis, Johanson et al. 1978; A. bahrelghazali, Brunet Michael R. Lague, Natural Sciences & Mathematics, The Richard et al. 1996; Paranthropus aethiopicus, Arambourg & Stockton College of New Jersey, Pomona, NJ 08240-0195, USA. Coppens, 1968, Chamberlain & Wood, 1985; P. crassidens E: [email protected] (now P. robustus, Broom, 1938), Broom, 1949; Homo ergaster, *These authors contributed equally to this work. Groves & Mazák, 1975; H. heidelbergensis, Schoetensack, Accepted for publication 28 August 2008 1908; H. antecessor, Bermúdez de Castro et al. 1997].

Hominid mandibular corpus variation, M. R. Lague et al. 671

Accordingly, it is important to verify the ability of mandi- sort extant hominid taxa. We cannot test the taxonomic bular morphology to identify taxonomic affiliations, parti- value of mandibular variables using the fossil record, as we cularly those aspects of morphology most frequently have no independent means of determining the taxonomy preserved on fossil specimens. of the specimens concerned. We can, however, investigate Discussion regarding the utility of hominid mandibular the taxonomic utility of mandibular corpus variables in morphology for delineating species and reconstructing their extant hominoid taxa closely related to the hominin clade relationships has continued from the early 1900s to the (e.g. Taylor & Groves, 2003; Wildman, et al. 2003). We are present. Early studies focused on modern human mandibular aware of the arguments suggesting that, with respect to morphology with a view to gaining an understanding fossil hominins, genetic propinquity is not the only criterion of inter- and intrapopulation variation, including sexual to use for selecting appropriate extant analogues (e.g. dimorphism (e.g. Harrower, 1928; Martin, 1936; Morant Aiello et al. 2000; Jolly, 2001; Plavcan, 2002). Nonetheless, et al. 1936; Cleaver, 1937; HrdliCka 1940a,b). As computa- we believe that it is unlikely that reliable taxonomic deci- tional power increased, researchers applied multivariate sions about the mandibular corpora of fossil hominin taxa statistics and other new methodologies, such as geometric can be made if mandibular corpus morphology is not morphometrics, to describe and quantify mandibular taxonomically informative in closely related extant taxa. variation in modern human populations (Humphrey et al. In the second part of this study, we employed a novel 1999; Oettlé et al. 2005; Nicholson & Havarti, 2006; Schmitt- probabilistic approach to assess within- and between- buhl et al. 2007) and fossil hominin species (Chamberlain group variation of mandibles assigned to several species of & Wood, 1985; Wood & Lieberman, 2001; Rak et al. 2002; Homo, including H. habilis sensu stricto, H. rudolfensis, Kaifu et al. 2005; Nicholson & Havarti, 2006). Other recent early African H. erectus/H. ergaster, late African H. erectus, studies have used multivariate mandibular data to evaluate Asian H. erectus, H. heidelbergensis, H. neanderthalensis a fossil specimen’s taxonomic identity, test hypotheses of and H. sapiens. More specifically, we used randomization species integrity, or reconstruct phylogenetic relationships of ‘distinctness values’ (RDV) to examine whether shape (Bromage et al. 1995; Rosas, 1995; Lam et al. 1996; Rosas & variation in mandibular morphology is consistent with Bermúdez de Castro, 1998; Stefan & Trinkaus, 1998a,b; Rosas conventional taxonomy. Following the argument of & Bermúdez de Castro, 1999; Schwartz & Tattersall, 2000; Tattersall (1986, p. 166) that ‘what is important in distin- Silverman et al. 2000, 2001; Quam et al. 2001; Rosas, 2001; guishing among species is between-species variation’, we Rak et al. 2002; Rosas & Bastir, 2004; Rightmire et al. 2006; examined the cohesion of mandibular morphology within Skinner et al. 2006). A number of studies examining great a proposed taxonomic group relative to between-group ape mandibular morphology have also been published variation. It is worth noting that our purpose was not to (Aitchison, 1963, 1965; Kinzey, 1970; Wood, 1985; Daeg- overturn conventional species designations but rather to ling & Jungers, 2000; Taylor, 2002, 2003; Taylor & Groves assess the extent to which shape variation of the mandi- 2003; Taylor, 2005, 2006a,b,c; Schmittbuhl et al. 2007). bular corpus can be used as a reliable taxonomic indicator Several of the above studies have analysed the ability of within the hominin clade. mandibular morphology to accurately predict group membership of known specimens from extant groups (e.g. Humphrey et al. 1999; Silverman et al. 2000; Taylor & Groves Materials and methods 2003; Schmittbuhl et al. 2007) but only rarely do those studies The extant hominid sample comprises 457 adult individuals (both subsequently use those validated measures to assess male and female specimens) representing four genera: Gorilla, species diversity in the fossil record. One reason for this is that Homo, Pan and Pongo (Table 1). All of the comparative specimens the measurements used on the extant specimens (e.g. are adult, based on the presence of wear facets on M3. Measure- measures of the mandibular ramus) are often not readily ments were taken on mandibles with the teeth present or with available from the fossil specimens due to poor preservation. pristine alveoli; mandibles with substantial alveolar bone resorp- In this study, we assessed the taxonomic utility of man- tion [e.g. cases where more than half of the tooth root(s) is exposed] were excluded from the study (cf. Vinter et al. 1996). dibular morphology by focusing on the traditional linear The 34 fossil hominin mandibles included in the study are all measurements that are most widely available in the conventionally included in the genus Homo (Weidenreich, 1936; hominin fossil record, namely those of the mandibular Day & Leakey, 1973; Wood, 1991; Wood & Richmond, 2000; Rosas, corpus. More specifically, we explored the extent to which 2001). Based on previous studies, we have divided the fossils into mandibular corpus morphology can be used to: (i) discrimin- nine taxonomic groups (see Table 2). We recognize that not all ate among extant hominid taxa and (ii) support species researchers will agree on the extent of diversity represented by designations (largely based on non-mandibular evidence) these specimens (e.g. Wood, 1985, Stringer, 1986; Bilsborough & Wood, 1988; Lieberman et al. 1988; Rightmire, 1990; Miller, 1991; among fossil specimens assigned to the genus Homo. Tobias, 1991; Wood, 1991; Bräuer & Mbua, 1992; Wood, 1992; In the first part of this study, we used discriminant analysis Bräuer, 1994; Wolpoff et al. 1994; Wood, 1994; Kramer et al. 1995; (DA) to test the hypothesis that size-adjusted linear Grine et al. 1996; Wolpoff, 1996; Rightmire, 1998; Wood & Collard, measurements of the mandibular corpus can be used to 1999; Miller, 2000). Nonetheless, as our goal was to test group

672 Hominid mandibular corpus variation, M. R. Lague et al.

Table 1 List and composition of extant taxa Extant taxon N Males Females

Total Gorilla: 146 79 67 Gorilla gorilla beringei 23 13 10 Gorilla gorilla gorilla 54 32 22 Gorilla gorilla graueri 69 34 35 Total Homo: 91 –– Homo sapiens (Terry Collection, NMNH) 52 26 26 Homo sapiens (Tel Aviv University) 39 – – Total Pan: 164 77 87 Pan paniscus 42 17 25 Pan troglodytes schweinfurthi 67 30 37 Pan troglodytes troglodytes 55 30 25 Total Pongo: 56 29 27 Pongo pygmaeus abelii 12 6 6 Pongo pygmaeus pygmaeus 44 23 21

All measurements made by N.J.C. NMNH = National Museum of Natural History.

Table 2 List and composition of fossil taxa

Fossil taxon Code N Measured by

Homo sapiens sapi 11 N.J.C.* Combe Capelle, Choukoutien/PA 101, Choukoutien/PA 104, Choukoutien/PA 109, Eyasi I, FishHoek I, Minatogawa I, Predmosti, Qafzeh 9, Skhul IV, Tabun II Homo neanderthalensis nean 8 N.J.C.† Amud, de la Naulette, Kebara H 2, Krapina E, Krapina H, Shanidar I, Spy I, Tabun I Homo heidelbergensis heid 2 N.J.C.* Arago II, Mauer Asian Homo erectus (Zhoukoudian) zhou 2 N.J.C.* Ckn. G1.6, Ckn. H1.12 Asian Homo erectus (Sangiran) sang 2 B.A.W.‡ Sangiran 1, Sangiran 9 Late African Homo erectus lahe 2 B.A.W.‡ KNM-BK 8518, OH 22 Early African Homo erectus/H. ergaster erga 2 B.A.W.‡ KNM-ER 730, KNM-ER 992 Homo habilis habi 2 B.A.W.‡ OH 13, OH 37 Homo rudolfensis rudo 3 B.A.W.‡ KNM-ER 1482, KNM-ER 1483, KNM-ER 1802

*Measured from casts (Smithsonian Institution, National Museum of Natural History). †Amud and Kebara measured from original fossils (Tel Aviv University), remainder measured from casts (Smithsonian Institution, National Museum of Natural History). ‡Data taken from Wood (1991).

integrity (based on mandibular corpus morphology), we began consisting of two specimens whose taxonomic allocation is con- our analyses by ‘splitting’ our fossil groups and subsequently sidered unresolved by some authors (e.g. Rosas, 2001). Finally, ‘lumping’ those for which no mandibular corpus morphological we began by considering the Asian H. erectus specimens as two justification for separation could be found (i.e. one cannot assess separate groups (Sangiran and Zhoukoudian), as these samples whether groups are distinct if they are not initially considered as differ both geographically and temporally. separate groups). For example, although specimens from both Linear dimensions (Table 3) were chosen to capture morpholog- Africa and Asia have been assigned to H. erectus, we thought it ical information from that part of the mandible (i.e. the corpus) prudent to maintain the geographic integrity of the samples and that is most often preserved in the early hominin fossil record. initially consider them as separate groups. We also began with a Measurements were taken with digital calipers using the measure- separate Middle Pleistocene ‘late African H. erectus’ group ment definitions given in Wood (1991). In order to maximize

Table 3 Measurements of the mandibular corpus

Measurement Definition*

Symphyseal depth 142: Maximum depth, at right angles to symphyseal height (Wood, 1991, pp. 295)

Corpus height at P4 147: Minimum distance between the most inferior point on the base and

the lingual alveolar margin at the midpoint of P4 (Wood, 1991, pp. 295)

Corpus width at P4 148: Maximum width at right angles to 147, taken at the midpoint of P4 (Wood, 1991, pp. 296)

Corpus height at M1 150: Same as 147 (Wood, 1991)

Corpus width at M1 151: Same as 148 (Wood, 1991) Canine socket (labiolingual length) 164: Maximum internal breadth of the canine alveolus in the labiolingual axis (Wood, 1991, pp. 296) Canine socket (mesiodistal length) 165: Maximum internal breadth of the canine alveolus in the mesiodistal axis (Wood, 1991, pp. 296)

P3–P4 alveolar length 167: Minimum chord distance between the midpoints of the interalveolar

septa between C/P3 and P4/M1 (Wood, 1991, pp. 97)

*The number at the beginning of the definitions refers to the measurement number in Wood (1991).

the fossil hominin sample size, only eight variables were selected the most meaningful for group separation. We also used posterior from a larger data set of 19 corpus variables. Interobserver probabilities to assess whether individuals could be correctly (B.A.W. and N.J.C.) and intraobserver (N.J.C.) measurement errors assigned to a given taxon. were < 3% in all cases. Measurements were taken on the original For purposes of significance testing, additional tests were run fossils where possible (see Table 2). The values of Weidenreich in which data were ranked following the rank transformation (1936) for the originals of Ckn. G1.6 and Ckn. H1.12 were not used approach of Conover & Iman (1981). This non-parametric approach because Weidenreich employed different landmarks than those relaxes the assumption of normality without significant loss of employed in this study. power. As the results based on ranked vs unranked data do not differ, we present only the latter results here.

Assessment of extant taxa: DA Assessment of fossil taxa: randomization analysis Discriminant function analysis (Klecka, 1980; Rencher, 1995) was (RDV test) applied to the extant specimens to assess whether metrical data from the mandibular corpus can be used to discriminate among The small sizes of the fossil hominin samples preclude the use of extant hominid genera, species, and subspecies. Size-adjusted, or DA for exploring group differences, as significance testing requires ‘shape’, values were generated by dividing each variable (of a more cases within each group than the total number of variables. given specimen) by that specimen’s geometric mean (cf. Darroch As an alternative, we employed an RDV test adapted from Sokal & Mosimann, 1985; Jungers et al. 1995). Taxa included in the & Rohlf (1995). This is a non-parametric probabilistic approach genus level analysis were Gorilla, Homo, Pan, and Pongo. The that assesses whether the a priori fossil groups are random with species level analysis was confined to Pan, and involved only respect to mandibular morphology. The RDV test assesses the P. paniscus and P. troglodytes. The subspecies level analyses exam- cohesiveness of a group of individuals by calculating a ‘distinctness ined three sets of subspecies, including three subspecies of Gorilla value’ (DV) defined by Sokal & Rohlf (1995, p. 806) as ‘... a measure (G. gorilla beringei, G. gorilla graueri and G. gorilla gorilla), two of homogeneity or cohesion of the members of a group relative subspecies of Pan (P. troglodytes schweinfurthi and P. troglodytes to their similarity with other groups’. The DV for a given group is troglodytes) and two subspecies of Pongo (P. pygmaeus pygmaeus calculated as the average correlation within the group (i.e. aver- and P. pygmaeus abelii). We have opted for a conventional taxonomy age of all pairwise within-group correlation coefficients) minus in the absence of a firm consensus about an alternative (for Gorilla the average correlation between groups (i.e. average of all possible see Groves 1967, 1970, 1989, Albrecht et al. 2003; Thalmann et al. correlation coefficients between members and non-members). 2007; for Pongo see Groves, 1971; Courtenay et al. 1988; Muir Hence, high positive DVs indicate that the chosen specimens form et al. 1998, 2000; Zhang et al. 2001). The DAs were performed a distinct group (relative to the other specimens) in which members using Statistica (Statsoft, Inc.,Tulsa, OK, USA) as well as an algorithm are more similar in shape to each other than they are to non-members. written by M.R.L. for MATLAB software (R2006a, version 7.2.0.232; (Note that the use of correlation coefficients implicitly adjusts for The Mathworks, Inc., Natick, MA, USA). scale, although it does not control for size-related shape variation.) Discriminant analysis is computationally equivalent to MANOVA Negative values indicate that members of the chosen group are and, for each analysis, we tested the null hypothesis that there is generally more similar in shape to outside members than they are no difference among groups. We also used ‘structure coefficients’ to one another. (i.e. the product-moment correlation between a given variable In the procedure described by Sokal & Rohlf (1995), randomiza- and a given canonical function) to identify those variables that are tion is used to determine whether the DV for a given group can most closely associated with group discrimination along a given be considered significantly high. We required something differ- canonical function. Variables with larger structure coefficients ent, which was to establish whether there is any morphological (i.e. ≥ | 0.40 | by convention; also see Schneider, 2006) are considered justification for a particular group configuration (e.g. 34 fossil

Homo specimens divided into nine taxa), i.e. based on the mor- complete samples and (ii) another set of tests using small samples phometric information that we have captured, do the species (three to four specimens for each group). The latter set of tests designations represent the ‘best’ way to sort n specimens into N was designed to assess the efficacy of the RDV test on sample sizes groups or are these a priori groups random with respect to the similar to those of the fossil taxa. A small number of individuals captured morphology? Hence, rather than test the ‘distinctness’ was randomly chosen from each extant group. Only these ran- of one group at a time, the algorithm was modified to consider domly chosen individuals were used to calculate the average DV the average distinctness of multiple groups. The average DV is and the randomized distribution. This procedure was repeated simply the average of the N DVs of the N groups (e.g. average of 1000 times (using different randomly chosen individuals each nine distinctness values of nine groups of fossil Homo). The time), resulting in a total of 1000 RDV tests for a given group more ‘distinct’ each particular group is, the greater the value of configuration. We then computed the percentage of significant the average DV. Therefore, a high positive DV indicates that the results (P ≤ 0.05) out of these 1000 RDV tests to assess the probability particular group configuration under consideration is largely of rejecting the null hypothesis when using small samples. To con- supported by the mandibular corpus morphology captured by our serve computing time, we used a maximum of 1000 (rather than eight linear measurements. In contrast, group configurations with 10 000) iterations per RDV test. negative values indicate that, on average, between-group correlations are higher than within-group correlations and therefore the particular group configuration is not supported by mandibular Sexual dimorphism corpus morphology. Previous studies have demonstrated that sexual dimorphism is an The null hypothesis of the RDV test is that the average DV (of important component of intraspecific mandibular corpus size the given group configuration) is not significantly higher than and shape variation among hominoids (Wood, 1976; Smith, 1983; expected by chance alone. The observed average DV is compared Chamberlain & Wood, 1985; Kimbel & White, 1988; Wood et al. 1991; with a distribution of average DVs obtained via a randomization Humphrey et al. 1999; Plavcan, 2002; Taylor, 2006c). Nonetheless, procedure. For each iteration, the sample is split into a random we did not conduct separate analyses of male and female extant number of groups (N ≥ 2), whereby each group has at least two specimens, particularly given that the goal of this study was to members (the minimum necessary to calculate a within-group examine the extent to which mandibular corpus shape is distinct correlation coefficient). The average DV is calculated from this among taxa, including those with substantial sexual dimorphism. random group configuration and the procedure is repeated for up As our focus was on mandibular shape, it is of interest to note those to 10 000 iterations. In those cases where the number of possible groups known to be characterized by significant mandibular shape novel combinations was less than 10 000 (e.g. there are only 2079 dimorphism (especially as related to the mandibular corpus), as ways to combine eight specimens into multiple groups of two or such taxa may have an impact on our results. In a recent study of more), exact randomization was used. hominoid mandibles by Taylor (2006c), Pongo and Gorilla were We began by assessing a group configuration in which 34 fossil found to exhibit significant shape dimorphism, although only one hominin specimens were divided into the nine groups indicated in of the significant shape dimensions (i.e. corpus width relative to Table 2, i.e. we tested whether the average DV based on this corpus depth at M1) used by Taylor is considered in the present a priori taxonomic grouping is significantly higher than one would study. As our ability to recognize taxa (species and subspecies) is expect for a random taxonomic allocation in which the 34 speci- partly predicated on the extent of shape dimorphism present in any mens are divided into a random number of groups (N ≥ 2) of random given taxon, it is conceivable that substantial shape dimorphism sample size (n ≥ 2). We subsequently used the RDV procedure on could compromise our ability to define taxonomic boundaries. smaller subsets of the fossil data (e.g. all pairwise comparisons Nevertheless, even given significant mandibular corpus shape as well as tests suggested either by previous taxonomic hypo- dimorphism, our analyses will recognize distinct taxa as long as theses or by the RDV results themselves). We used principal any sex-related shape variation does not exceed taxonomic varia- components analysis (PCA) of group means to visually depict the tion (such that males and females of a given taxon are more similar morphological affinities among the different fossil taxa being tested. in shape to one another than to members of other taxa). The RDV tests and PCA were performed using algorithms written by M.R.L. for MATLAB software (R2006a, version 7.2.0.232; The Mathworks, Inc.). Results It is known that correlation coefficients (r) are distributed in an asymmetrical fashion. To assess the potential effect of such asym- Extant hominids: DAs metry on our results, we also ran all of the fossil RDV tests described below using Fisher’s z-transformation of r (see Sokal Genus level & Rohlf, 1995). Although the use of z instead of r changed the The DA based on the four-genus configuration (Homo, resulting P-values somewhat, without exception, all of the results Pan, Gorilla and Pongo) produced a significant result were the same in terms of whether or not statistical significance 2 (P < 0.05) was observed. Only the results based on untransformed (Λ = 0.04, χ = 1497, P < 0.001) and sorted the extant genera correlation coefficients are presented below. with 80% success. On the corresponding plot (Fig. 1a), Homo is distinguished from Pan, Pongo and Gorilla on the first canonical axis, which accounts for the large majority Validation of the RDV test (93%) of the variation among groups. There is considerable overlap along the second axis (only 6% of the In order to validate the method, and compare the results with those obtained using more traditional DA, we applied the RDV variation), although overall generalized distances between test to a number of extant group configurations. Two sets of RDV group centroids are all significantly large (P < 0.001). Based tests were run using the same extant data: (i) one set of tests using on the structure coefficients of the first discriminant axis

Fig. 1 Canonical variates plots based on size-adjusted data for (a) four extant hominid genera and (b) the same taxa with the exclusion of Homo. Genera are indicated as follows: H, Homo; P, Pan; G, Gorilla; O, Pongo.

Table 4 Structure coefficients for genus-level analysis*

All four genera Homo excluded

Variables Axis 1 Axis 2 Axis 1 Axis 2

Symphyseal depth –0.519 0.395 0.362 0.498

Height at P4 0.556 0.121 0.170 –0.477

Width at P4 0.186 –0.047 –0.060 0.578

Height at M1 0.534 0.183 0.238 –0.299

Width at M1 0.346 0.078 0.087 0.316 Canine socket (labiolingual) –0.100 –0.198 –0.187 –0.354 Canine socket (mesiodistal) –0.268 –0.748 –0.729 –0.235

P3–P4 alveolar length –0.105 0.499 0.505 0.076 Percentage of explained variation 93.1 6.3 85.4 14.6

*Values > | 0.400 | are shown in bold to highlight the most important discriminating variables.

(Table 4), mandibular corpus height is relatively larger in As most of the shape variation among extant genera is extant modern humans (than in non-human hominids) accounted for by differences between modern humans at both P4 and M1, whereas the symphyseal depth of the and non-human hominids, we ran an additional analysis in modern human mandibles is relatively smaller. which Homo was excluded. Even without the influence of

Table 5 Structure coefficients for species-level Species Subspecies and two subspecies-level analyses*

Variables Pan Pan Pongo

Symphyseal depth 0.408 0.163 0.135

Height at P4 0.066 –0.112 0.586

Width at P4 –0.184 0.083 0.058

Height at M1 –0.158 –0.194 0.250

Width at M1 0.053 –0.045 –0.380 Canine socket (labiolingual) 0.149 0.425 –0.296 Canine socket (mesiodistal) 0.034 –0.335 –0.171

P3–P4 alveolar length –0.567 –0.047 0.056

Group centroids† Group 1 0.69 –0.952 0.295 Group 2 –2.00 0.782 –1.081

*Values > | 0.400 | are shown in bold to highlight the most important discriminating variables. †‘Group 1’ designated as follows: P. troglodytes for Pan (species), P. t. troglodytes for Pan (subspecies), and P. p. pygmaeus for Pongo.

Homo (and despite considerable overlap), the overall result is significant (Λ = 0.43, χ2 = 305, P < 0.001) and distances between group centroids are found to be significantly large (P < 0.001). Specimens were allocated to the correct genus with 74% success. Most of the variation between groups (ca 85%) is accounted for by the first axis, which is similar to the second axis of the previous analysis (which included Homo) in that Pan is somewhat separated from the other two genera (Fig. 1b); as expected, the most influential variables associated with these axes are the same (see Table 4).

Species level The two species of Pan were found to differ significantly (T2 = 1.389, F = 26.9, P < 0.001) and 91% of the individuals were allocated correctly. Based on structure coefficients

(Table 5), P. troglodytes has a relatively smaller P3–P4 alveolar length and a relatively taller symphysis than P. paniscus.

Subspecies level Fig. 2 Canonical variates plot based on size-adjusted data for the three Tests of the three Gorilla subspecies produced a significant gorilla subspecies. Group centroids are indicated with black dots. Subspecies result (Λ = 0.48, χ2 = 102, P < 0.001), with all pairwise inter- are indicated as follows: g, G. g. gorilla; b, G. g. beringei; i, G. g. graueri. centroid distances being significantly large (P < 0.001). Specimens were allocated to the correct subspecies with 76% allocated with a success rate of 80%. Structure coefficients success. The corresponding canonical variates plot is presented are listed in Table 5. in Fig. 2; structure coefficients are provided in Table 6. The two subspecies of P. troglodytes were also found to RDV differ significantly (T2 = 0.75, F = 10.68, P < 0.001). The extent of successful allocation to the correct taxon was Validation 80%. Based on structure coefficients (Table 5), discrimina- The DAs indicate that significant differences in mandibu- tion between Pan subspecies appears to be based largely lar corpus morphology exist among extant taxa even at on the shape of the canine socket, which has a relatively the subspecific level (with the exception of Pongo). Hence, larger labiolingual dimension in P. t. troglodytes. it is reasonable to consider whether mandibular corpus In contrast to the above analyses, no significant differ- morphology is of value for distinguishing fossil hominin ence was found between the two subspecies of Pongo taxa. As noted above, before proceeding with RDV tests (T2 = 0.33, F = 0.95, P > 0.05), although specimens were on the fossil specimens, we applied the RDV test to a

Table 6 Structure coefficients for subspecies-level analysis of Gorilla* above (21%). Nonetheless, it is clear that the chances of rejecting a ‘false’ (based on full data) null hypothesis is not Variables Axis 1 Axis 2 particularly large when utilizing small samples. Increasing group sample sizes to four individuals each increases the Symphyseal depth –0.034 –0.456 percentage of significant results (to 29% in both cases), Height at P 0.409 –0.064 4 albeit not substantially. Width at P4 –0.353 0.204 The results of the above tests based on extant data Height at M1 0.352 0.331

Width at M1 –0.408 –0.023 demonstrate that the RDV test is more conservative than Canine socket (labiolingual) –0.245 –0.056 DA, even with large samples. At small sample sizes (such as Canine socket (mesiodistal) 0.177 0.092 those characterizing the fossils), there is a good chance P3–P4 alveolar length 0.323 0.043 that the null hypothesis will be falsely accepted unless Percentage of explained variation 66.6 33.4 the groups under consideration are particularly ‘distinct’. Hence, with respect to the small fossil samples examined *Values > | 0.400 | are shown in bold to highlight the most below, rejection of the null hypothesis for a given RDV test important discriminating variables. can be accepted as strong evidence that the groups involved are morphologically distinct at a level that is number of extant group configurations based on com- equal to or greater than that of modern hominid species. plete data sets, as well as on small random samples. The RDV results based on complete samples indicate Fossil Homo that, in comparison to DA, the RDV test is more conservative The average DV for the complete nine-group configura- (Table 7). None of the three RDV tests based on subspecies tion of fossils (DV = 0.0182) is found to be significantly indicate a significant difference among groups. In contrast, high based on sampled randomization; only 11 of the none of the 1000 randomized DVs for the two species of 10 000 random group configurations produced an average Pan is higher than the observed value, indicating that DV higher than that observed for the nine a priori groups these two species are significantly distinct with respect (i.e. P = 0.0011). Hence, there is some morphological to mandibular corpus morphology. The same result is support from the mandibular corpus for the way in which obtained for a four-group configuration of extant non- specimens have been assigned to fossil Homo taxa. human hominids in which all of the subspecies are The above result does not necessarily indicate that all combined into their respective species (i.e. Gorilla, P. nine groups are significantly distinct from one another but troglodytes, P. paniscus and Pongo). it does suggest that further analysis is worthwhile. We pro- For the RDV tests using small random samples (n = 3), ceeded by considering three logical pairwise comparisons one is extremely unlikely to reject the null hypothesis for involving groups that have been (or are) considered to be any of the hominid subspecies; based on 1000 tests for conspecific: (i) the two sets of H. erectus specimens from each set of subspecies, the probability of doing so equals Asia (Sangiran vs Zhoukoudian), (ii) early African H. erec- < 0.001 (0.1%) in all three cases. The small sample test for tus/ergaster vs late African H. erectus and (iii) H. habilis vs the two species of Pan yields a higher percentage of H. rudolfensis. The sample sizes of all six of the aforemen- rejected null hypotheses (21%), as does the test for the tioned groups are very small (n = 2 or 3); hence, these com- four-group configuration of non-human hominids noted parisons do not lend themselves to the randomization test

Table 7 Results of randomization of ‘distinctness values’ (RDV) tests using extant taxa

Complete sample sizes Small samples

Groups tested N of groups Total n Average DV‡ P value* n per group Total n Iterations per RDV test % sig. result

Pongo subspecies 2 56 –0.0005 N/A† 3 6 70 0.022 Pan subspecies 2 122 0.0017 0.236 3 6 70 0.062 Gorilla subspecies 3 146 0.0010 0.259 3 9 1000 0.070 Pan species 2 164 0.0104 0 3 6 70 20.5 – – – – 4 8 1000 28.9 Gorilla, P. pan., 4 366 0.0064 0 3 12 1000 21.4 P. trog., Pongo – – – – 4 16 1000 28.8

*Based on 1000 iterations. †Not applicable. No RDV test necessary for negative average distinctness value. ‡DV = distinctness values.

Table 8 Correlation matrix for select fossil hominin groups

Sang. Sang. Ckn. Ckn. ER ER ER OH OH ER ER BK OH 1 9 G1.6 H1.12 1482 1483 1802 37 13 730 992 8518 22

H. erectus (Asian) Sangiran 1 1 Sangiran 9 0.965 1 Ckn.G1.6 0.981 0.977 1 Ckn.H1.12 0.978 0.980 0.998 1 H. rudolfensis ER 1482 0.951 0.982 0.971 0.983 1 ER 1483 0.958 0.989 0.978 0.981 0.984 1 ER 1802 0.978 0.978 0.987 0.993 0.986 0.980 1 H. habilis OH 37 0.977 0.989 0.982 0.984 0.980 0.994 0.979 1 OH 13 0.945 0.972 0.956 0.969 0.983 0.984 0.978 0.977 1 H. ergaster ER 730 0.978 0.982 0.966 0.974 0.982 0.975 0.988 0.980 0.981 1 ER 992 0.962 0.976 0.965 0.977 0.989 0.985 0.986 0.984 0.996 0.990 1 Late African BK 8518 0.935 0.932 0.930 0.943 0.955 0.959 0.967 0.952 0.983 0.967 0.984 1 H. erectus OH 22 0.935 0.975 0.950 0.963 0.984 0.990 0.973 0.979 0.997 0.977 0.993 0.978 1

Fig. 3 Principal components analysis (PCA) plot of group means for select fossil hominin groups based on the variance-covariance matrix of size- adjusted data.

(as there are only three possible ways to combine four H. erectus’ group. With respect to the two African H. erec- specimens). Nonetheless, an assessment of the ‘distinctness’ tus groups (DV = 0.0042), only one of the four specimens of each group can be made by simply considering the (KNM-ER 730) is most highly correlated with the specimen average DV (i.e. is it positive or negative?) and the pattern belonging to the same group (KNM-ER 992). As such, the of correlations. With respect to the latter, we can justify two African H. erectus groups were considered as a single maintaining a distinction between two groups if both (or group (‘African H. erectus’) in subsequent analyses. all three) members of each group are more highly corre- To visually depict the morphological affinities among lated with another group member than they are to mem- the resulting six groups of fossil hominins, we used a PCA bers of the alternative group. (Fig. 3) based on the group means for size-adjusted data. The average DV for the comparison between H. habilis and Along the first axis (91% of the variation), the African H. rudolfensis is negative (DV = –0.0028). Hence, no justifica- H. erectus and ‘early Homo’ groups fall to the right side of tion can be made for maintaining a distinction between these the graph. Three taxa collectively referred to as ‘later Homo’ two groups based on our data; in the remainder of this work, (H. sapiens, H. heidelbergensis and H. neanderthalensis) these two taxa are referred to collectively as ‘early Homo’. fall to the opposite side of the graph, whereas the Asian Although the average DVs for the other two compari- H. erectus group has an intermediate position. Interestingly, sons are positive, their respective correlation patterns are the African H. erectus group does not show particular affin- not supportive of a morphological distinction between the ities with the Asian H. erectus group; the former is on the two groups in question (Table 8). In the case of the two extreme right side of the graph, separated from the Asian Asian H. erectus groups (DV = 0.0024), both Sangiran speci- H. erectus group by the two taxa of the ‘early Homo’ group. mens are more highly correlated with one of the Zhou- We performed RDV tests for all possible pairwise com- koudian specimens than they are to one another. Hence, in parisons of the six fossil groups noted above (see Table 9). the analyses below, they are considered as a single ‘Asian Of particular interest are those pairs of taxa on opposite

Table 9 Randomization of ‘distinctness values’ (RDV) results for pairwise comparisons*

N Number of iterations Average DV‡ P value nean sapi 19 10 000 0.0011 0.3329 ns heid 10 10 000 0.0025 0.2854 ns [zhou, sang] 12 N/A† –0.0011 N/A† ns [rudo, habi] 13 10 000 0.0174 0.0054 ** [erga, lahe] 12 10 000 0.0392 0.0005 *** sapi heid 13 10 000 0.0047 0.0334 * [zhou, sang] 15 10 000 0.0053 0.0449 * [rudo, habi] 16 10 000 0.0310 0.0001 *** [erga, lahe] 15 10 000 0.0600 0.0003 *** heid [zhou, sang] 6 70 0.0045 0.0857 ns [rudo, habi] 7 266 0.0189 0 *** [erga, lahe] 6 70 0.0461 0 *** [zhou, sang][rudo, habi] 9 10 000 0.0061 0.0750 ns [erga, lahe] 8 2 079 0.0216 0.0029 ** [rudo, habi][erga, lahe] 9 10 000 0.00283 0.2019 ns

*Taxa included within brackets were analysed as a single group: [zhou, sang] = ‘Asian H. erectus’, [rudo, habi] = ‘early Homo’, [erga, lahe] = ‘African H. erectus’. †Not applicable. No test necessary for negative average distinctness value. ‡DV = distinctness value. sides of the PCA plot (i.e. those that should be most the influence of H. neanderthalensis, which, unlike the ‘distinct’ from one another and are most likely to yield other two species of ‘later Homo’, has a negative DV. significant RDV results). Based on pairwise testing, all Indeed, pairwise comparisons (Table 9) indicate that three of the ‘later Homo’ taxa (nean, sapi, heid) are although H. neanderthalensis is not significantly distinct significantly distinct from both of the groups on the from either H. sapiens or H. heidelbergensis, the latter opposite side of the plot (i.e. ‘early Homo’ and African two species are significantly different from one another H. erectus). The intermediate position of Asian H. erectus (P = 0.033). Based on our eight dimensions, shape varia- (as suggested by the PCA) is also reflected in the randomization of the H. neanderthalensis mandibular corpus appears tion results; with respect to pairwise comparisons, Asian to overlap that of H. sapiens, H. heidelbergensis and Asian H. erectus is not distinct from two of the three ‘later Homo’ H. erectus. groups (i.e. H. neanderthalensis and H. heidelbergensis) As noted above, the PCA (and pairwise RDV tests) sug- or from ‘early Homo’. If we impose a more inclusive two- gests that Asian H. erectus is morphologically intermediate group configuration of ‘later Homo’ (hsap, nean, heid) between the ‘later Homo’and ‘African Homo’ groups. Com- vs ‘African Homo’ (rudo, habi, erga, lahe) while excluding parison of the ‘African Homo’ and Asian H. erectus groups the intermediate ‘Asian H. erectus’ individuals, the (in a two-group configuration) yields a significantly high resulting average DV (0.035) is significantly high (P = 0). average DV (P = 0.005; Table 10). In contrast, when Asian Hence, the RDV results complement the visual PCA results H. erectus is compared with the combined ‘later Homo’ and suggest that the fossil specimens can be split into at group (in a different two-group configuration), the least two distinct morphological groups (i.e. ‘later Homo’ average DV is non-significant (P = 0.184). However, when and ‘African Homo’), within which further testing can the Neanderthals are removed from the ‘later Homo’ be done. group (for reasons described above), the average DV is Among the ‘African Homo’ group (right side of the PCA higher and significant (P = 0.049) in a similar two-group plot), the ‘early Homo’ group (habi and rudo) is not found test. In addition, the average DV of a three-group confi- to be distinct (DV = 0.0028, P = 0.202) from the African guration consisting of H. sapiens, H. heidelbergensis and H. erectus group (erga and lahe). Hence, on the basis of Asian H. erectus (but not H. neanderthalensis) is also sig- traditional linear measurements of the mandibular corpus, nificantly high (P = 0.027). The pairwise RDV test between the two groups should be treated as a single group (i.e. Asian H. erectus and H. sapiens indicates that these two ‘African Homo’). groups are also significantly distinct. Although the same Similarly, the three ‘later Homo’ taxa (left side of the cannot be said for the pairwise test between Asian H. PCA plot) also do not produce a significantly high average erectus and H. heidelbergensis, the associated average DV (DV = 0.0031, P = 0.117) when compared in a three- DV does have a very low probability of being randomly group configuration (Table 10). The low average DV for sampled (P = 0.086; Table 9). Hence, among the ‘later this configuration, however, appears to be due mainly to Homo’ species, although H. neanderthalensis is not distinct

Table 10 Results of additional randomization of ‘distinctness values’ (RDV) tests*

Groups included† N (groups) n (specimens) Average DV‡ P value

All nine groups 9 34 0.0182 0.0011 ** ‘Later Homo’ group sapi, heid, nean 3 21 0.0031 0.1173 ns Tests with ‘Asian H. erectus’ [zhou, sang] [zhou, sang], [habi, rudo, erga, lahe] 2 13 0.0122 0.0049 ** [zhou, sang], [sapi, heid, nean] 2 25 0.0022 0.1836 ns [zhou, sang], [sapi, heid] 2 17 0.0045 0.0489 * [zhou, sang], sapi, heid 3 17 0.0053 0.0269 * Tests with ‘African Homo’ [habi, rudo, erga, lahe], sapi 2 20 0.0431 0 *** [habi, rudo, erga, lahe], heid 2 11 0.0303 0 *** Final four ‘distinct’ groups sapi, heid, [zhou, sang], [habi, rudo, erga, lahe] 4 26 0.0215 0.0012 **

*10 000 iterations used in all cases. †Taxa included within brackets were analysed as a single group: [sapi, heid, nean] = ‘later Homo’; [zhou, sang] = ‘Asian H. erectus’, [habi, rudo, erga, lahe] = ‘African Homo’. ‡DV = distinctness value.

from either H. heidelbergensis or H. sapiens, the latter Table 11 Structure coefficients for canonical variates analysis (CVA) of two species are distinct from one another, as well as from the four ‘distinct’ fossil groups depicted in Fig. 4a (based on ‘shape’ data)* Asian H. erectus. In summary, evidence from the randomization analyses Variables Axis 1 Axis 2 suggests that shape variation of the mandibular corpus, Symphyseal depth 0.177 0.523 based on traditional linear measurements, is not random Height at P4 –0.501 –0.210 among fossil Homo taxa. Among the 34 fossil mandibles Width at P4 0.351 0.338

examined here, there is sufficient evidence to recognize Height at M1 –0.487 –0.112

at least four distinct groups [i.e. modern H. sapiens, H. hei- Width at M1 0.291 –0.086 delbergensis, Asian H. erectus and a group consisting of Canine socket (labiolingual) –0.053 0.004 exclusively African species (H. habilis, H. rudolfensis, early Canine socket (mesiodistal) –0.205 –0.117 P –P alveolar length 0.442 –0.292 African H. erectus/H. ergaster and late African H. erectus)]. 3 4 With the exception of the comparison between H. heidel- Percentage of explained variation 88.5 8.8 bergensis and Asian H. erectus (P = 0.086), these four groups are all significantly distinct from one another *Values > | 0.400 | are shown in bold to highlight the most important discriminating variables. (Tables 9 and 10). A four-group RDV test based on the above four groups also yields a significantly high average

DV (P = 0.001; Table 10). higher at both P4 and M1, and the P3–P4 alveolar length is In contrast, we cannot reject the null hypothesis that the relatively smaller. As expected, when Neanderthals are four exclusively African taxa are random with respect to added as a sixth group by projecting them onto the same mandibular corpus morphology. In addition, although the canonical vectors (Fig. 4b), they show substantial overlap mandibular corpora of H. neanderthalensis are found to with H. sapiens, H. heidelbergensis and Asian H. erectus. be distinct from those of the ‘African Homo’ group, they cannot be distinguished from those of H. sapiens, H. heidelbergensis or ‘Asian H. erectus’. Discussion To provide a visual summary of the RDV results, we per- The results of our DAs indicate that the hominid mandi- formed a canonical variates analysis (CVA) based on shape bular corpus is taxonomically informative, even when its data for the four ‘distinct’ groups noted above (Fig. 4a), each morphology is captured by eight ‘low-tech’ traditional of which was entered into the analysis as an a priori group. linear measurements commonly preserved in the hominin The majority of variation among groups is accounted for fossil record. To some extent, we can successfully discrimi- by the first axis (88.5%), with extensive overlap along the nate extant hominids at the genus, species and subspecies second axis (with the exception of H. heidelbergensis). levels (with the exception of the Pongo subspecies). Structure coefficients (Table 11) indicate that H. sapiens These results are consistent with those of previous studies and H. heidelbergensis differ from the other three fossil that used a broader set of mandibular variables (includ- hominin groups in that the mandibular corpus is relatively ing measurements of the ramus and different corpus

Fig. 4 Canonical variates plot based on size-adjusted mandibular data for four ‘distinct’ groups of fossil hominins, presented alone (a) and with Neanderthals projected onto the same canonical vectors (b). The fossil specimens were configured into four groups based on the results of randomization of ‘distinctness values’ (RDV) tests, which also indicate that Neanderthal mandibular morphology is not significantly distinct from that of H. sapiens, H. erectus or H. heidelbergensis. The ‘African Homo’ group consists of specimens assigned to H. habilis, H. rudolfensis and early African H. erectus/H. ergaster. measurements; Humphrey et al. 1999; Taylor & Groves, 2003). especially at the subspecific level. Nonetheless, despite As noted in the Introduction, it is theoretically possible that reports of significant mandibular shape dimorphism in mandibular corpus shape dimorphism may be substantial Gorilla (Taylor, 2006c), we are able to distinguish among enough to obfuscate the detection of distinct taxa, three gorilla subspecies with moderate success (using DA

but not RDV). Our inability to differentiate between the geographic distribution. Specimens range in time from as orangutan subspecies may be partly related to the rela- early as 1.7 mya (KNM-ER 730; Feibel et al. 1989) to as late tively small sample sizes for P. p. abelli or perhaps to as ~200 kya (Zhoukoudian specimens G1.6 and H1.12; significant mandibular shape dimorphism (Taylor, 2006c), Antón, 2003). Not only does this fossil group span a time although an analysis of sexual dimorphism in Pongo is depth of over 1.5 million years, it also has an extensive beyond the scope of this study. geographic distribution ranging from Africa to China and Analyses based on the extant data demonstrate that our Indonesia. Various researchers have examined the integ- RDV test is more conservative than DA, even at similar rity of H. erectus sensu lato with the understanding that sample sizes. For example, although the complete-sample the extreme time depth and geographic distribution RDV results mirror the DA results in recognizing morpho- sampled by the taxon may introduce variation that is too great logical differences among hominids at the generic and for a single hominin species (Groves & Mazák, 1975; Tyler, species levels, RDV does not return statistically significant 1991; Bräuer & Mbua, 1992; Bräuer, 1994; Rightmire, 1998; results for analyses at the subspecific level. In addition, at Antón, 2002; Kidder & Durband 2004). Our own results small sample sizes (three to four individuals per group), it suggest that mandibular corpus morphology does not is likely that the null hypothesis of RDV will be falsely allow us to discriminate between Chinese and Indonesian accepted unless the groups under consideration are parti- H. erectus specimens, nor are we able to recognize any cularly ‘distinct’ (e.g. only 29% of 1000 RDV tests indicated diagnostic differences between early and late African a significant distinction among the four hominin genera H. erectus. In contrast, the division between African when sample size was set to four). As such, with respect to and Asian H. erectus specimens is supported by our our fossil samples, we believe that any significant results analyses. obtained from the RDV analyses provide strong evidence Another issue in fossil Homo taxonomy that has long for morphological distinction among groups, probably at been of interest concerns the scope of the hypodigm of a level similar to (or greater than) that of extant conge- H. habilis. Arguments have been made both for (e.g. neric hominid species (e.g. P. troglodytes and P. paniscus). Alexeev, 1986; Wood, 1991) and against (e.g. Howell, Evidence from RDV of hominin fossil samples suggests 1978; Tobias, 1985) the division of the H. habilis sensu lato that the mandibular corpus does have limited taxonomic hypodigm. Based on our limited eight-variable data set, the utility. Among the 34 fossil Homo mandibles examined morphology of the mandibular corpus does not provide here (potentially divisible into nine groups), we can recog- any additional evidence for splitting the H. habilis hypo- nize at least four unambiguously distinct groups: (i) modern digm. However, some of the mandibular morphology that H. sapiens; (ii) H. heidelbergensis; (iii) Asian H. erectus is said to differ between H. habilis and H. rudolfensis and (iv) ‘African Homo’. The latter combined group (iv) (e.g. the size of the alveolar planum; Lieberman et al. 1996) consists of four predominately African species (H. habilis, is not captured by our analysis and much of the argument H. rudolfensis, early African H. erectus/H. ergaster and late for splitting the H. habilis hypodigm is based on dental and African H. erectus) and we cannot reject the null hypo- other non-mandibular cranial data. thesis that these ‘African Homo’ taxa are random with We deliberately limited the scope of this study to simple respect to mandibular corpus morphology (hence, their traditional linear measurements that can be taken by any collection into a single group). In addition, although the researcher on the part of the mandible (i.e. the corpus) mandibular corpora of H. neanderthalensis are found to that is best represented in the hominin fossil record. We be distinct from those of the ‘African Homo’ group, they also deliberately focused on fossil taxa within a single genus cannot be distinguished from those of H. sapiens, H. hei- and did not compare the mandibles of more distantly delbergensis or ‘Asian H. erectus’. related taxa such as H. ergaster and P. boisei (which differ We note that our results are contingent upon the small markedly in size and shape), nor did we investigate number of corpus linear dimensions that we chose to measure; mandibular corpus variation more widely within and among alternative variables may have produced significant results fossil hominoid taxa. Despite constraining the study in in those cases where the null hypothesis was accepted. For these ways, it is clear that these commonly available data example, it is well established that modern humans differ do carry a taxonomic signal and we anticipate that such from other hominins in the presence of a mental eminence a signal will have at least some taxonomic utility within and, although we did not measure this feature, it should and among fossil hominoid taxa. Although linear measure- serve to distinguish H. sapiens from the other specimens. ments of the mandibular corpus do not distinguish In addition, although it has been argued that H. neander- among all fossil Homo taxa (particularly not among early thalensis can be defined by a suite of mandibular characters African specimens), they do distinguish among several of (see Rosas, 2001), our limited set of variables did not include them. These results imply that, to a limited extent, it is most of the characters commonly cited as distinctive. possible to test hypotheses concerning hominin alpha Of all of the fossil species in our sample, H. erectus sensu taxonomy using the mandibular corpus, particularly when lato samples the longest time period and has the widest new discoveries increase the fossil sample sizes.

Brunet M, Beauvilain A, Coppens Y, Heintz E, Moutaye AHE, Acknowledgements Pilbeam D (1996) Australopithecus bahrelghazali, une nouvelle We would like to thank Linda Gordon, David Hunt and Rick Potts espèce d’Hominidé ancien de la région de Koro Toro (Tchad). (National Museum of Natural History, Smithsonian Institution) for CR Acad Sci IIa 322, 907–913. providing access to fossil casts and primate skeletal remains. We Chamberlain AT, Wood BA (1985) A reappraisal of variation in are grateful to Bruce Latimer and Lymen Jellema (Cleveland hominid mandibular corpus dimensions. Am J Phys Anthropol Museum of Natural History) and Wim Van Neer (Royal Museum 66, 399–405. of Central Africa, Tervuren, Belgium) for access to the primate Cleaver FH (1937) A contribution to the biometric study of the collections in their care. Thanks to Yoel Rak for allowing access to human mandible. Biometrika 29, 80–112. the fossil specimens and archaeological human skeletal remains Conover W, Iman R (1981) Rank transformations as a bridge stored at Tel Aviv University, Israel, and for his comments on a previous between parametric and nonparametric statistics. Am Stat 35, version of this manuscript. Thanks also to Mark Collard, Robbin 124–133. Chatan, Susan Antón, Dan Lieberman, and several anonymous Courtenay J, Groves C, Andrews P (1988) Inter- or intra-island reviewers for their comments. This research was funded in part by variation? An assessment of the differences between Bornean the Louis B. Cotlow Research Fund, The Henry Luce Foundation, and Sumatran Orang-utans. In Orang-utan Biology (ed. Schwartz the University of Illinois at Urbana-Champaign and Lee Silverman. JH), pp. 19–29. New York: Oxford University Press. Daegling DJ, Jungers WL (2000) Elliptical Fourier analysis of symphyseal shape in great ape mandibles. J Hum Evol 39, 107–122. References Darroch JN, Mosimann JE (1985) Canonical and principal components of shape. Biometrika 72, 241–252. Aiello LC, Collard M, Thackeray JF, Wood BA (2000) Assessing exact Day MH, Leakey REF (1973) New evidence for the genus Homo randomization-based methods for determining the taxonomic from East Rudolf, Kenya. Am J Phys Anthropol 39, 341–354. significance of variability in the human fossil record. S Afr J Sci Feibel CS, Brown FH, McDougall I (1989) Stratigraphic context of 96, 179–183. fossil hominids from the Omo Group Deposits, northern Turkana Aitchison J (1963) Comparison of chimpanzee skulls, jaws and Basin, Kenya and Ethiopia. Am J Phys Anthropol 78, 595–622. teeth with those of modern hominids. Aust Dent J 8, 477– Grine FE, Jungers WL, Schultz J (1996) Phenetic affinities among 482. early Homo crania from East and South Africa. J Hum Evol 30, Aitchison J (1965) Contrasts in the mandibles and mandibular 189–225. teeth of the chimpanzee, orang-utan and gorilla. Dent Mag Oral Groves CP (1967) Ecology and taxonomy of the Gorilla. Nature Top 81, 105–108. 213, 890–893. Albrecht GH, Gelvin BR, Miller JMA (2003) The hierarchy of Groves CP (1970) Population systematics of the Gorilla. J Zool 161, intraspecific craniometric variation in gorillas: a population- 287–300. thinking approach with implications for fossil species recogni- Groves CP (1971) Pongo pygmaeus. Mamm Species 4, 1–6. tion studies. In Gorilla Biology: A Multidisciplinary Perspective Groves CP (1989) A Theory of Human and Primate Evolution. (eds Taylor AB, Goldsmith ML), pp. 62–103. Cambridge, UK: Oxford: Clarendon Press. Cambridge University Press. Groves CP, Mazák V (1975) An approach to the taxonomy of the Alexeev VP (1986) The Origin of the Human Race. Moscow: Hominidae: gracile Villafranchian hominids of Africa. Casopis Progress Publishers. Mineral Geol 20, 225–247. Antón S (2003) Natural history of Homo erectus. Yrbk Phys Harrower G (1928) A biometric study of one hundred and ten Anthropol 46, 126–170. Asiatic mandibles. Biometrika 20B, 279–293. Antón SC (2002) Evolutionary significance of cranial variation in Howell FC (1978) Hominidae. In Evolution of African Mammals Asian Homo erectus. Am J Phys Anthropol 118, 301–323. (eds Maglio VJ, Cooke HBS), pp. 154–248. Cambridge, MA: Harvard Arambourg C, Coppens Y (1968) Découverte d’un australo- University Press. pithécien nouveau dans les gisements de l’Omo (Éthiopie). S Afr Hrdli!ka A (1940a) Lower jaw. Am J Phys Anthropol 27, 281–308. J Sci 64, 58–59. Hrdli!ka A (1940b) Lower jaw. Further studies. Am J Phys Anthropol Bermúdez de Castro JM, Arsuaga JL, Carbonell E, Rosas A, 27, 383–467. Martínez I, Mosquera M (1997) A hominid from the lower Pleis- Humphrey LT, Dean MC, Stringer CB (1999) Morphological varia- tocene of Atapuerca, Spain: possible ancestor to Neandertals tion in great ape and modern human mandibles. J Anat 195, and modern humans. Science 276, 1392–1395. 491–513. Bilsborough A, Wood BA (1988) Cranial morphometry of early Johanson DC, White TD, Coppens Y (1978) A new species of the hominids: facial region. Am J Phys Anthropol 76, 61–86. genus Australopithecus (Primates: Hominidae) from the Pliocene Bräuer G (1994) How different are Asian and African Homo erec- of eastern Africa. Kirtlandia 28, 1–14. tus? Cour Forsch-inst Senckenberg 171, 301–318. Jolly CJ (2001) A proper study for Mankind: analogies from the Bräuer G, Mbua E (1992) Homo erectus features used in cladistics Papionin monkeys and their implications for human evolution. and their variability in Asian and African hominids. J Hum Evol Yrbk Phys Anthropol 44, 177–204. 22, 79–108. Jungers W, Falsetti A, Wall C (1995) Shape, relative size, and size- Bromage TG, Schrenk F, Zonneveld FW (1995) Paleoanthropology adjustments in morphometrics. Yrbk Phys Anthropol 38, 137–161. of the Malawi Rift: an early hominid mandible from the Chiwondo Kaifu Y, Baba H, Aziz F, Indriati E, Schrenk F, Jacob T (2005) Beds, northern Malawi. J Hum Evol 28, 71–108. Taxonomic affinities and evolutionary history of the Early Broom R (1938) The Pleistocene anthropoid apes of South Africa. Pleistocene hominids of Java: dentognathic evidence. Am J Phys Nature 142, 377–379. Anthropol 128, 709–726. Broom R (1949) Another new type of fossil ape-man (Paranthropus Kidder JH, Durband AC (2004) Evaluation of the metric diversity crassidens). Nature 163, 57. within Homo erectus. J Hum Evol 46, 299–315.

Kimbel W, White T (1988) Variation, sexual dimorphism and the Rightmire GP, Lordkipanidze D, Vekua A (2006) Anatomical descrip- taxonomy of Australopithecus. In Evolutionary History of the tions, comparative studies and evolutionary significance of the ‘Robust’ Australopithecines (ed. Grine F), pp. 175–192. New hominin skulls from Dmanisi, Republic of Georgia. J Hum Evol York: Aldine de Gruyter. 50, 115–141. Kinzey WG (1970) Basic rectangle of the mandible. Nature 228, Rosas A (1995) Seventeen new mandibular specimens from the 289–290. Atapuerca/Ibeas Middle Pleistocene Hominids Sample (1985– Klecka WR (1980) Discriminant Analysis. Quantitative Applications 1992). J Hum Evol 28, 533–559. in the Social Sciences Series, No. 19. Thousand Oaks, CA: Sage Rosas A (2001) Occurrence of Neanderthal features in mandibles Publications. from the Atapuerca-SH site. Am J Phys Anthropol 114, 74–91. Kramer A, Donnelly SM, Kidder JH, Ousley SD, Olah SM (1995) Rosas A, Bastir M (2004) Geometric morphometric analysis of allo- Craniometric variation in large-bodied hominids: testing the single- metric variation in the mandibular morphology of the hominids species hypothesis for Homo habilis. J Hum Evol 29, 443–462. of Atapuerca, Sima de los Huesos Site. Anat Rec A 278A, 551– Lam YM, Pearson OM, Smith CM (1996) Chin morphology and 560. sexual dimorphism in the fossil hominid mandible sample from Rosas A, Bermúdez de Castro JM (1998) On the taxonomic affini- Klasies River Mouth. Am J Phys Anthropol 100, 545–557. ties of the Dmanisi mandible (Georgia). Am J Phys Anthropol Leakey MG, Feibel CS, McDougall I, Walker AC (1995) New 107, 145–162. four-million-year-old hominid species from Kanapoi and Allia Rosas A, Bermúdez de Castro JM (1999) The ATD6–5 mandibular Bay, Kenya. Nature 376, 565–571. specimen from Gran Dolina (Atapuerca, Spain). Morphological Lieberman DE, Pilbeam DR, Wood BA (1988) A probabilistic study and phylogenetic implications. J Hum Evol 37, 567–590. approach to the problem of sexual dimorphism in Homo habilis: Schmittbuhl M, Rieger J, Le Minor J-M, Schaaf A, Guy F (2007) a comparison of KNM-ER 1470 and KNM-ER 1813. J Hum Evol 17, Variations of the mandibular shape in extant hominoids: generic, 503–511. specific, and subspecific quantification using elliptical Fourier Lieberman DE, Wood BA, Pilbeam DR (1996) Homoplasy and early analysis in lateral view. Am J Phys Anthropol 132, 119–131. Homo: an analysis of the evolutionary relationships of H. habilis Schneider M (2006) Examination of Structure Coefficient Interpre- sensu stricto and H. rudolfensis. J Hum Evol 30, 97–120. tation in Descriptive Discriminant Analysis: the Three-Group Case. Martin ES (1936) A study of an Egyptian series of mandibles with Abstract from the 2006 Joint Statistical Meetings (American special reference to mathematical methods of sexing. Biometrika Statistical Association), p. 406. 28, 149–178. Schoetensack O (1908) Der Unterkierfer des Homo heidelber- Miller JMA (1991) Does brain size variability provide evidence of gensis aus den Sanden von Mauer bei Heidelberg. Leipzig: W. multiple species in Homo habilis? Am J Phys Anthropol 84, 385–398. Engelmann. Miller JMA (2000) Craniofacial variation in Homo habilis: an Schwartz JH, Tattersall I (2000) The human chin revisited: what is analysis of the evidence for multiple species. Am J Phys Anthropol it and who has it? J Hum Evol 38, 367–409. 112, 103–128. Silverman N, Richmond B, Wood B (2000) Does the hominid man- Morant GM, Collett M, Adyanthaya NK (1936) A biometric study dibular corpus have any taxonomic utility? Am J Phys Anthropol of the human mandible. Biometrika 28, 84–122. 30, 281. Muir CC, Galdikas BMF, Beckenbach AT (1998) Is there sufficient Silverman N, Richmond B, Wood B (2001) The taxonomic integrity evidence to elevate the orangutan of Borneo and Sumatra to of Paranthropus boisei sensu stricto. Am J Phys Anthropol 115, separate species. J Mol Evol 46, 378–381. 167–178. Muir CC, Galdikas BMF, Beckenbach AT (2000) mtDNA sequence Skinner MM, Gordon AD, Collard NJ (2006) Mandibular size and diversity of orangutans from the islands of Borneo and Sumatra. shape variation in the hominins at Dmanisi, Republic of Georgia. J Mol Evol 51, 471–480. J Hum Evol 51, 36–49. Nicholson E, Havarti K (2006) Quantitative analysis of human man- Smith RJ (1983) The mandibular corpus of female primates: taxo- dibular shape using three-dimensional geometric morphometrics. nomic, dietary, and allometric correlates of interspecific varia- Am J Phys Anthropol 131, 368–383. tions in size and shape. Am J Phys Anthropol 61, 315–330. Oettlé AC, Pretorius E, Steyn M (2005) Geometric morphometric Sokal RR, Rohlf FJ (1995) Biometry, 3rd edn. New York: W.H. analysis of mandibular ramus flexure. Am J Phys Anthropol 128, Freeman Co. 623–629. Stefan VH, Trinkaus E (1998a) La Quina 9 and Neandertal mandi- Plavcan JM (2002) Taxonomic variation in the patterns of cranio- bular variability. Bull Mém Soc Anthropol Paris 10, 293–324. facial dimorphism in primates. J Hum Evol 42, 579–608. Stefan VH, Trinkaus E (1998b) Discrete trait and dental morpho- Quam RM, Arsuaga J-L, Bermúdez de Castro J-M (2001) Human metric affinities of the Tabun 2 mandible. J Hum Evol 34, 443– remains from Valdegoba Cave (Huérmeces, Burgos, Spain). J Hum 468. Evol 41, 385–435. Stringer CB (1986) The credibility of Homo habilis. In Major Topics Rak Y, Ginzburg A, Geffen E (2002) Does Homo neanderthalensis in Primate and Human Evolution (eds Wood BA, Martin LB, play a role in modern human ancestry? The mandibular evidence. Andrews P), pp. 266–294. New York: Alan R. Liss. Am J Phys Anthropol 119, 199–204. Tattersall I (1986) Species recognition in human paleontology. Rencher AC (1995) Methods of Multivariate Analysis. New York: J Hum Evol 15, 165–175. John Wiley & Sons. Taylor AB (2002) Masticatory form and function in the African apes. Rightmire GP (1990) The Evolution of Homo erectus: Comparative Am J Phys Anthropol 117, 133–156. Anatomical Studies of an Extinct Human Species. Cambridge, Taylor AB (2003) Ontogeny and function of the masticatory U.K.: Cambridge University Press. complex in Gorilla: Functional, evolutionary, and taxonomic Rightmire GP (1998) Evidence from facial morphology for similarity implications. In Gorilla Biology: A Multidisciplinary Perspective of Asian and African representatives of Homo erectus. Am J Phys (eds Taylor AB, Goldsmith ML), pp. 132–189. Cambridge, U.K.: Anthropol 106, 61–85. Cambridge University Press.

Taylor AB (2005) A comparative analysis of temporomandibular Wildman D, Uddin M, Liu G, Grossman LI, Goodman M (2003) joint morphology in the African apes. J Hum Evol 48, 555–574. Implications of natural selection in shaping 99.4% nonsynonymous Taylor AB (2006a) Diet and mandibular morphology in the African DNA identity between humans and chimpanzees: enlarging apes. Int J Primatol 27, 181–201. genus Homo. Proc Natl Acad Sci 100, 7181–7188. Taylor AB (2006b) Feeding behavior, diet, and the functional Wolpoff MH (1996) Human Evolution. New York: McGraw-Hill. consequences of jaw form in orangutans, with implications for Wolpoff MH, Thorne A, Jelinek J, Zhang Y (1994) The case for the evolution of Pongo. J Hum Evol 50, 377–393. sinking Homo erectus: 100 years of Pithecanthropus is enough! Taylor AB (2006c) Size and shape dimorphism in great ape mandibles Cour Forsch-inst Senckenberg 171, 341–361. and implications for fossil species recognition. Am J Phys Anthropol Wood BA (1976) The nature and basis of sexual dimorphism in the 129, 82–98. primate skeleton. J Zool Lond 180, 15–34. Taylor AB, Groves CP (2003) Patterns of mandibular variation in Wood BA (1985) Early Homo in Kenya, and its systematic relation- Pan and Gorilla and implications for African ape taxonomy. ships. In Ancestors: The Hard Evidence (ed. Delson E), pp. 206– J Hum Evol 44, 529–561. 214. New York: Alan R. Liss. Thalmann O, Fischer A, Lankester F, Paabo S, Vigilant L (2007) The Wood BA (1991) Koobi Fora Research Project, Vol. 4: Hominid complex evolutionary history of gorillas: insights from genomic Cranial Remains. Oxford: Clarendon Press. data. Mol Biol Evol 24, 146–158. Wood BA (1992) Early hominid species and speciation. J Hum Evol Tobias PV (1985) Single characters and the total morphological 22, 351–365. pattern redefined: the sorting effected by a selection of Wood BA (1994) Taxonomy and evolutionary relationships of morphological features of the early hominids. In Ancestors: Homo erectus. Cour Forsch-inst Senckenberg 171, 159–165. The Hard Evidence (ed. Delson E), pp. 94–101. New York: Wood BA, Collard M (1999) The human genus. Science 284, 65–71. Alan R. Liss. Wood BA, Lieberman DE (2001) Craniodental variation in Paran- Tobias PV (1991) Olduvai Gorge. Vol. 4: The Skulls, Endocasts and thropus boisei: a developmental and functional perspective. Teeth of Homo habilis. Cambridge, U.K.: Cambridge University Am J Phys Anthropol 116, 13–25. Press. Wood BA, Richmond BG (2000) Human evolution: taxonomy and Tyler DE (1991) A taxonomy of Javan hominid mandibles. Hum paleobiology. J Anat 197, 19–60. Evol 6(56), 401–420. Wood BA, Li Y, Willoughby C (1991) Intraspecific variation and Vinter I, Krmpotic-Nemanic J, Ivankovic D, Jal6ovec D (1996) The sexual dimorphism in cranial and dental variables among higher influence of the dentition on the shape of the mandible. Coll primates, and their bearing on the hominid fossil record. J Anat Antropol Zagreb 20(2), 555–560. 174, 185–205. Weidenreich F (1936) The mandibles of Sinanthropus pekinensis: a Zhang Y, Ryder OA, Zhang Y (2001) Genetic divergence of orangutan comparative study. Palaeont Sin Ser D 7, 1–162. subspecies (Pongo pygmaeus). J Mol Evol 52, 516–526.