AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 116:13–25 (2001) Craniodental Variation in Paranthropus boisei: A Developmental and Functional Perspective Bernard Wood1* and Daniel E. Lieberman2 1Department of Anthropology, George Washington University, Washington, DC 20052, and Human Origins Program, National Museum of Natural History, Smithsonian Institution, Washington, DC 20560 2Department of Anthropology, Harvard University, Peabody Museum, Cambridge Massachusetts 02138 KEY WORDS Paranthropus boisei; craniodental; taxonomy; coefficient of variation; species ABSTRACT What levels and patterns of craniodental size variables are intermediate in terms of their reliabil- variation among a fossil hypodigm are necessary to reject ity. P. boisei is found to have a low degree of variability the null hypothesis that only a single species is sampled? relative to extant primates for variables shown to be gen- We suggest how developmental and functional criteria can erally useful for testing taxonomic hypotheses. Contrary be used to predict where in the skeleton of fossil hominins to the claims of Suwa et al. ([1997] Nature 389:489–492), we should expect more, or less, within-species variation. the recently discovered material from Konso falls within We present and test three hypotheses about the factors the range of variation of the “pre-Konso” hypodigm of P. contributing to craniodental variation in extant primate boisei for available conventional metrical variables. Those taxa, and then apply these results to the interpretation of aspects of the Konso material that appear to extend the the P. boisei hypodigm. Within the comparative samples of range of the P. boisei hypodigm involve regions of the skull extant Homo, Pan, Gorilla, Pongo, and Colobus, variables predicted to be prone to high levels of within-species vari- from the cranial base, neurocranium, and face that are not ation. The approach used in this study focuses on cranio- subject to high magnitudes of strain have consistently dental data, but it is applicable to other regions of the lower levels of intraspecific variation than variables from skeleton. Am J Phys Anthropol 116:13–25, 2001. regions of the face subject to high levels of strain. Dental © 2001 Wiley-Liss, Inc. A sound, species-level, or alpha, taxonomy, based (e.g., a large brain, a retracted face, and small ca- on reliable inferences about the taxonomic signifi- nines), along with unique combinations of plesi- cance of morphological variation, is a prerequisite omorphic features, make modern human cranioden- for reconstructing human evolutionary history. Be- tal material distinctive. Thus, the pattern of cause of taphonomic factors, the majority of the intraspecific variation in the most closely related hominin fossil record is cranial, mandibular, and extant primate taxa (e.g., Wood et al., 1991) may not dental (hereafter referred to as “craniodental”). always be appropriate for investigating the taxo- Thus, hominid paleontologists must devise effective nomic significance of variation among hominin fos- methods for sorting craniodental remains into spe- sils. cies. No hominin fossils are alike, but how different This paper explores another approach for predict- do crania, jaws, and teeth have to be for researchers ing the pattern of intraspecific variation within ex- to reject the hypothesis that they are conspecific? tinct early hominin taxa. Specifically, we use devel- Conventionally, the fossil record is sorted into hy- opmental and functional information to generate pothetical species by using modern species taxa as hypotheses about the regional distribution of in- analogues. Analyses of the degree and pattern of traspecific variation. We then test those hypotheses within-species variation in modern human and by analyzing variation across a range of extant an- closely related nonhuman primate populations are thropoid primates, including those that are often typically used as the sources of empirical evidence used as comparators for fossil hominins. Are there about intraspecific craniodental variation in homi- variables that one would predict, on the basis of nins. However, the utility and testability of these developmental and functional criteria, to show rela- models is limited because all but the modern human comparative analogues lack the unique craniodental features, such as relatively small canines, reduced Grant sponsor: Henry Luce Foundation. prognathism, and centrally-located foramen mag- num, that set hominins apart from the other apes. *Correspondence to: Bernard Wood, Department of Anthropology, George Washington University, 2110 G St. NW, Washington, DC Moreover, even Homo sapiens is of limited utility as 20052. E-mail: [email protected] a comparator because generally low levels of sexual dimorphism, and the presence of unique features Received 31 August 2000; accepted 31 May 2001. © 2001 WILEY-LISS, INC. 14 B. WOOD AND D.E. LIEBERMAN tively little within-species variation? If so, substan- than in the extant species, then the null hypothesis tial levels of variation in these variables in a fossil that the specimens sample one species is rejected. A hypodigm would strongly suggest taxonomic heter- fundamental problem with this type of comparative ogeneity. Conversely, are there classes of variables analysis is the choice of analogue species, which is that, for developmental or functional reasons, one usually determined on the basis of observations of, would expect to show higher levels of intraspecific or inferences about, genetic propinquity. Thus mod- variation? If so, high levels of variation in these ern humans, and the other African great apes, are variables would not be reliable indicators of taxo- traditionally the preferred analogues for investiga- nomic heterogeneity. Newly discovered fossil evi- tions of intraspecific variation within fossil homi- dence of existing species can then be used to “test” nins. However, the specialized nature of the homi- these hypotheses. Variability within a distinctive nin cranium and dentition mean that nonhuman fossil species should be concentrated on morphology extant higher primate analogues are not always an that is prone to higher than average levels of in- appropriate model. Most early hominin species prob- traspecific variation in extant comparators. For rea- ably have unique patterns and components of their sons set out above and below, this study applies variation that result in both type I and type II sta- these principles to the craniodental remains attrib- tistical errors. For example, P. boisei was probably uted to Paranthropus boisei. considerably sexually dimorphic in body mass (in- For a given feature, the total sum of variation cluding skull size), but not for canine size (Wood, ␴ within a living species, total, can be broken down 1991; Wood et al., 1991; Plavcan and van Schaik, into several contributory factors (Vaisnys et al., 1992). Such a pattern of sexual dimorphism is not 1984): present in any extant higher primate species. VARIATION IN PARANTHROPUS BOISEI ␴total ϭ ␴ontogenic ϩ ␴geographic ϩ ␴temporal The epistemological problem of which cranioden- ϩ ␴intersexual ϩ ␴intrasexual. tal characters are more, or less, reliable as indica- For a fossil species, taphonomy also contributes to tors of species diversity is a general one for all fossil the observed variation (Wood, 1991; Silverman et hominin taxa. However, this study focuses on al., 2001). The extent to which these factors contrib- Paranthropus boisei sensu stricto (hereafter referred ute to intraspecific variation is expected to differ for to as P. boisei) for several, related reasons. Most each fossil species. In the case of P. boisei, variation importantly, P. boisei is a distinctive, relatively well- due to sampling individuals from different ontoge- sampled, early hominin taxon, whose hypodigm in- netic stages can be controlled for by including only cludes several substantially complete fossil crania. mature adult specimens in an analysis. Variation Although the type specimen of P. boisei is the well- due to sampling individuals from geographically dis- preserved OH5 cranium from Olduvai Gorge, discov- persed populations is probably not a significant eries made subsequently at Koobi Fora and West problem for P. boisei because the hypodigm is re- Turkana comprise the majority of the craniodental stricted to a reasonably small geographic area, East hypodigm of P. boisei (Wood, 1991). The first good Africa, with only one specimen coming from a site evidence of a mandible from the same taxon as OH5 elsewhere, in Malawi. Variation due to sampling came with the recovery of the well-preserved Peninj individuals from different time periods is a potential 1 mandible from Lake Natron in 1964 (Leakey and problem for all fossil taxa. There is reason, however, Leakey, 1964). Since then, an impressive mandibu- to suspect that time is apparently not a major source lar sample has been accumulated from collections of variation in P. boisei. Wood et al. (1994) demon- made at the Shungura Formation, in the Omo re- strated that, with the exception of the P4 crown, gion, and from Koobi Fora (Wood, 1991). The denti- there is little change in size of most craniodental tions of OH5 and the Peninj mandible, together with variables among P. boisei fossils dated to between evidence from associated and isolated teeth from the 2.0 and 1.4 Myr, which is most of the hypodigm. The Shungura Formation and Koobi Fora, have provided ␴ final components of total, i.e., the variation between a substantial amount of information