Technical Comments

Comment on “A Complete Skull from Dmanisi, Georgia, and the Evolutionary Biology of Early Homo

+ See all authors and affiliations

Science  25 Apr 2014:
Vol. 344, Issue 6182, pp. 360
DOI: 10.1126/science.1250056


Lordkipanidze et al. (Research Article, 18 October 2013, p. 326) conclude, from gross morphological comparisons and geometric-morphometric analysis of general shape, that the five hominid crania from Dmanisi in Georgia represent a single regional variant of Homo erectus. However, dental, mandibular, and cranial morphologies all suggest taxic diversity and, in particular, validate the previously named H. georgicus.

Lordkipanidze et al. (1) confirm the importance of the early Pleistocene hominin site of Dmanisi, with the addition of a fifth cranium (D4500) to the sample (24). The authors conclude that the large morphological differences among the five crania and four mandibles represent not taxic diversity but variation within a paleodeme of a single evolving Homo erectus lineage. They suggest classifying the East African specimens often identified as H. ergaster as the subspecies H. erectus ergaster and their proposed Dmanisi scion as the sub-subspecies H. erectus ergaster georgicus. They also hint that the long-accepted species H. habilis and H. rudolfensis might be folded into this lineage. Spoor (5) responded that taxonomic quadrinomials are invalid, and neither the three-dimensional geometric-morphometric analysis (GMA) nor the morphological comparisons captured detail relevant to phylogenetic analysis. Here, we review mandibular, dental, and cranial morphologies not considered by Lordkipanidze et al. These indicate that the Dmanisi sample presents evidence of taxic diversity rather than within-species variation, with at least one species being the previously named H. georgicus.

The Dmanisi fossils are assumed to sample a single population, primarily because they come from the same site and a relatively short time period. By a priori defining “difference” as intraspecies “variation,” this permits focusing uniquely on general shape and gross morphology. Thus, when differences between mandibles D211 and D2600 exceeded those in humans and chimpanzees, variability in highly sexually dimorphic Gorilla was taken to affirm the single-deme assumption (4). Now, despite D4500 “stand[ing] apart from…other…fossil Homo…[in] combin[ing] a small braincase with a large prognathic face” (p. 326), the enlarged cranial sample is said to mirror variation in humans and chimpanzees (1). However, if D211 is associated with skull 2 (D2282), and D2600 with D4500, as Lordkipanidze et al. (1) argue, their claims of Gorilla-like mandibular but human/Pan-like cranial variability (notwithstanding that the Pan sample includes two species) cannot both be correct. Given this contradiction, morphological detail might prove enlightening in deciphering the systematic identities of the Dmanisi hominins.

Differences among mandibles D2735, D211, and D2600 are attributed to age, tooth wear, and pathology (6). However, while there is postcanine abscessing and resorption in D2600, its disease-free symphyseal alveolar bone is unaltered (7) and, as seen in Neandertals and modern humans (8), taxon-specific adult mandibular morphologies are present in juveniles. Thus, systematically informative differences are found in symphyseal morphology (tall and broadly arced in D2600 but short and tightly curved in D2735 and D211) and in postincisal plane configuration (tall, steep, and slightly bowed in D2600 but shorter and more sloping in D2735 and D211) (Fig. 1). Further, since dental attrition does not change or totally obliterate crown morphology (D2600); reconfigure lower first, second, and third molar size gradients (M1 > M2 > M3 in D211 versus M1 ≤ M2 < M3 in D2600); convert long, narrow, rectangular molars (D2735) into roundedly squat teeth (D211) and then back again (D2600); or add a metaconid (D2735) to a metaconid-free lower first premolar (P1) (D211), these differences are also systematically relevant (Fig. 1).

Fig. 1 Comparison of Dmanisi mandibles D211, D2735, and D4500.

Note the obvious morphological differences in bone and tooth morphology (not to scale).

Regarding other taxically distinctive dental features, D2600’s obliquely oriented lower canines (C1s) are ovoid in cross section (versus buccolingually compressed in D2735); its P1s bear entoconids (absent in D211 and D2735); and its M3 differs from D211’s in size and shape and in having a large, more mesially extensive entoconid and a large, more buccally expanded hypoconid. Because root number has taxonomic valence, it is noteworthy that D2600’s P1 bears three robust, long, separate roots, whereas D211 and D2735’s P1 is single-rooted, with grooves above a bifid tip. D2735’s M3 roots are expectedly less splayed than M1’s (7), whereas D2600’s M3 roots are unusually well separated and D211’s M2-3 roots are single (6). Because root divergence appears if and when interradicular tongues grow into the cervical loop (7), these differences, which exceed variation in H. sapiens (7), are important.

Potentially species-distinguishing features are also obvious in the suggested cranial counterparts of these mandibles in vault size and lateral and posterior outline and supraorbital detail [e.g., double-arced/nonprotruding (D2735), versus thin/straight across/anteriorly protruding (D2282), versus tall/barlike/superiorly expanded, with posttoral sulcus (D4500) (1, 9, 10)]. These differences are further highlighted by comparison not only with the type specimen but also with the array of supposed H. erectus specimens, which clearly comprise a visually and morphologically incoherent assemblage (11) (Fig. 2).

Fig. 2 Array of specimens commonly deemed H. erectus.

The type specimen is Trinil 2 (Java), and the only specimens similar to it in derived features are from Sangiran (Java). Otherwise, note the obvious morphological differences, not only between the Dmanisi specimens but also in the assemblage as a whole. ZKD, Zhoukoudian; ER, East Turkana; WT, West Turkana; r, reversed (not to scale).

Why did the GMA of 78 landmarks (1) not capture the visually obvious differences between the Dmanisi crania and specimens commonly subsumed in H. erectus? Difficulties might include information loss due to (i) representing landmarks by their coordinates and using too few and unconnected single points (absence of semi-landmarks) to represent averages of shape; (ii) emphasis on the greatest variation by using principal components analysis of general shape [facial profile (first component), brain size (second component)]; (iii) meshing of models and nonisotropic measurement-induced variance caused by poorly defined landmarks (“Pinocchio effect”); and (iv) lack of a relative warp analysis of large-scaled deformation (especially of the first component) (12). Notwithstanding that only ~45% of the 20 “erectus” specimens compared by GMA preserve structure below the supraorbital region, and only ~40% of these specimens preserve the lower face, one wonders how phylogenetically reliable a method can be that does not reflect even easily visible gross morphological differences.

Lordkipanidze et al. (1) acknowledge that if the Dmanisi crania came from five different sites, they would be allocated to as many taxa. Yet, unlike morphology, neither time nor place is necessarily related to systematic identity (13). Because the Dmanisi fossils were possibly deposited over as many as several hundred years, there was ample time for faunal migration and/or replacement. The D2600 mandible is both highly idiosyncratic and the holotype of the morphologically very distinctive species H. georgicus (14), to which its equally unusual cranial counterpart D4500 can also be referred. To deny this hominin a distinct identity is effectively to deny the utility of morphology in systematics, a radical proposition to which few would subscribe. We predict that continued study of the extraordinary Dmanisi hominins will provide further evidence of their taxic diversity. As Goldschmidt (15) remarked, biology should be studied with, but not as, mathematics.

References and Notes

  1. Acknowledgments: J.H.S. and I.T. thank D. Lordkipanidze for access to Dmanisi specimens and willingness to discuss alternatives.
View Abstract

Related Content


Navigate This Article