Early Hominids--Diversity or Distortion?

See allHide authors and affiliations

Science  28 Mar 2003:
Vol. 299, Issue 5615, pp. 1994-1997
DOI: 10.1126/science.1078294

Ernst Mayr once described hominid taxonomy as a “bewildering diversity of names” (1). George Gaylord Simpson later lamented “the chaos of anthropological nomenclature” (2). These founding fathers of modern taxonomy brought paleoanthropological and zoological systematics into conformity—but only temporarily.

Many of today's paleoanthropologists herald each new fossil as evidence of ancient hominid biodiversity, following Gould's prediction of a “bushy” hominid tree (3). Completing the circle, the resulting taxonomic inflation is widely proclaimed as evidence of “a many-branched bush of diversity” [(4), p. 33]. But whether judged from fossil evidence or zoological considerations, the metaphor of an early hominid bush seems seriously misplaced.

For example, last summer's announcement of the “Toumai” hominid cranium from Chad (5) was enthusiastically greeted as “the tip of an iceberg of taxonomic diversity during hominid evolution 5–7 million years ago” (6). The same author even predicted a Late Miocene “African ape equivalent of the Burgess Shale” (6). How could a single fossil from a previously unknown period warrant such claims?

Paleoanthropology has readily adopted “diversity” systematics over the last 10 years, partly because of new fossils. But as Wilford astutely noted (7), the embrace of ethnic diversity among contemporary academics may be creating a peculiar form of politically correct paleoanthropology. Hominid phylogenies now regularly recognize up to 20 species. New hominid fossils are routinely given new species names such as Ardipithecus ramidus, Australopithecus anamensis, Australopithecus garhi, and Homo antecessor. At the same time, long-abandoned names such as H. heidelbergensis and H. rhodesiensis have recently been resurrected. Textbook authors and publishers eagerly adopt these taxa. But does the resultant nomenclature accurately reflect early hominid species diversity?

To evaluate the biological importance of such taxonomic claims, we must consider normal variation within biological species. Humans (and presumably their ancestors and close relatives) vary considerably in their skeletal and dental anatomy. Such variation is well documented and stems from ontogenetic, sexual, geographic, and idiosyncratic (individual) sources. Populations and species lineages also change through time, introducing an additional dimension of variation.

As an example of how biological variation is currently assessed in fossil hominids, consider the African middle Pliocene (between 3.5 and 4.0 million years ago). Until recently, a substantial collection of fossils from eastern Africa dated to between 4.1 and 3.0 million years ago was thought to sample an evolving species lineage (Australopithecus anamensis to Australopithecus afarensis). The fossils showed substantial size and shape variation, but this variation readily conforms to that in modern analog species such as the African apes.

Then, in 2001, a new Kenyan cranium, KNM-WT 40000, was interpreted by its discoverers to represent a second middle Pliocene hominid lineage. Named Kenyanthropus platyops (“flat-faced man from Kenya”) by Leakey and colleagues, this fossil was heralded as evidence of an early diet-driven adaptive radiation (8).

There are two questions to be asked in considering whether the fossil constitutes evidence of early hominid species diversity. First, are the described morphological differences from the A. anamensis to A. afarensis lineage real, or are they merely artifacts of postmortem fossilization processes? Second, does the putatively new morphology lie outside the expected range of phenotypic variation of this lineage? Fortunately, the history of vertebrate paleontology provides a largely unappreciated but critically important perspective on the first question. Modern primate skeletal collections help to address the second.

In the early 1900s, thousands of fossils from the western badlands flooded America's museums. Among them were the oreodonts—medium-sized pig relatives from 33.7 to 23.8 million years ago. Paleontologists named a multitude of oreodont species and inferred an adaptive radiation from the “diversity” they created. In retrospect, “oreodont taxonomy and concepts of phylogeny became chaotic” until subsequent analysis trimmed the oreodont tree to realistic proportions [(9), p. 498].

The early systematic exuberance of oreodont paleontologists is a textbook example of postmortem deformation driving the creation of invalid fossil taxa. Distorted oreodont crania were selected as type specimens for Platychoerus platycephalus (flat-headed flat pig) and Stenopsochoerus (narrow pig) (10). But the fossils had been flattened and narrowed by geological deformation, not natural selection. Typological systematics had combined with postmortem distortion to create a diversity that proved chimeric (9).

Early hominid crania are usually rare and isolated, in contrast to abundant oreodonts. The KNM-WT 40000 specimen is the only known cranium of Kenyanthropus. Its describers claim it to be distinguished by a supposedly diagnostic combination of craniofacial characters (8). However, these characters have virtually all been influenced to varying degrees by expanding matrix distortion (EMD) (see the first figure).


(A to C) Oreodonts from the John Day Formation, Oregon, illustrate the effects and progression of a particular form of postmortem distortion formalized here as EMD. Small bone fragments are separated by matrix-filled cracks of varying dimensions and geometry. This series typifies different stages of EMD, from slight [(A), stage 1] to severe [(C), stage 5]. Scale bar, 2 cm. (D) The cranium of Kenyanthropus platyops (EMD stage 4). The anterior facial surface contains about 1100 matrix-separated pieces of bone. A conservative estimate is that the overall specimen comprises about 4000 such individual bone particles.


The most insidious aspect of EMD is its ability to radically alter morphology in a nonlinear manner. Because matrix expansion does not enlarge all dimensions equally, it often causes highly complex distortion such as that seen in Kenyanthropus. The published photographs of KNM-WT 40000 (5) show that this fossil suffered EMD stage 4 distortion, with its original surface anatomy splintered into individual bone particles of mostly subcentimeter size (panel D, first figure).

There are about 1100 separate bone pieces in the anterior projection of the Kenyanthropus face alone, each isolated by varying thicknesses of matrix fill (11). Even the single preserved tooth crown is distorted (12). Comparable high-stage EMD is rare in other medium-sized vertebrate crania, even in collections where EMD is relatively common (13). In the hominid fossil record, only the Pleistocene Yunxian crania are so severely distorted (14). Unfortunately, it is apparently not yet possible to reliably correct for EMD of this magnitude by computerized tomography or other imaging techniques (8).

Because the only known cranium of Kenyanthropus suffers such a high level of distortion, additional fossil evidence from 3.5 million years ago in the Turkana basin is required to establish whether it represents a valid taxon or is simply an early Kenyan variant of A. afarensis. Given the extensive deformation of this specimen and the known cranial variation in early hominid species and among modern apes and humans (see the second figure), proclamations that it signals Pliocene hominid diversity seem premature at best.

Is there really a great diversity of hominid lineages waiting to be found and recognized in Africa? Was this diversity like that in extant Anopheles mosquitoes (about 500 species), Old World fruit bats (173 species), cercopithecid monkeys (94 species), or even African soft-furred rats (8 species)? Or did just a few demographically expansive and cosmopolitan hominid species expand their ranges and invade new habitats during the Pliocene (5.3 to 1.8 million years ago)?


Craniofacial variation in a modest (n = 30) sample of bonobos (an extant African ape). The two female specimens occupy opposite positions in the observed ranges of variation but are not outliers of the sample. This variation is normal in a single sex of an extant species; even more variation is present in other extant ape species.


As Mayr and Simpson appreciated, species recognition is at the core of the paleontological enterprise and is an essential component in building an accurate understanding of evolution. As the hominid fossil record expands, we should not forget their cautions about typological thinking. Confusing true biological species diversity with analytical mistakes (15, 16), preservational artifacts, diachronic evolution, or normal biological variation grossly distorts our understanding of human evolution. Past hominid diversity should be established by the canons of modern biology, not by a populist zeal for diversity.

References and Notes

View Abstract

Navigate This Article