The demise of “Nutcracker Man”

More than 50 y since the first announcement of Zinjanthropus boisei at Olduvai Gorge (1, 2), the diet of Paranthropus boisei, as it is now known, has continued to be the focus of study and debate. In part, this is because it has such distinctive features, shared in some degree with the South African australopith, Paranthropus robustus: a dished, buttressed face, strongly developed muscle attachments, huge molars and premolars and diminutive anterior teeth, and thick enamel caps, all pointing to a diet requiring heavy, forceful oral processing (from whence the term “Nutcracker Man”). Additionally, P. boisei shared the East African landscape with the earliest members of the genus Homo for more than a half million years, which strongly suggests that they occupied distinct dietary niches. What was P. boisei eating that required such heavy-duty “equipment,” and how did its diet differ from that of early Homo? Until recently, we thought we knew the answers: according to conventional interpretations of the craniodental morphology, they were specialists in small, hard fruits and nuts (3). However, recent research, culminating in the stable isotope study by Cerling et al. in PNAS (4), shows this interpretation to be seriously off the mark. Their study demonstrates firmly and unequivocally that the bulk of P. boisei's diet was from C₄ plant sources. These data, combined with the results of microwear texture studies showing no evidence for hard object feeding (5, 6), place a rather different slant on the diet of this hominin. Equally important perhaps, the results suggest that we need to exercise more caution in interpreting the morphological data in terms of paleodietary ecology.

The results are not entirely new, or even completely unexpected, as a preceding study by van der Merwe et al. (7) gave essentially the same result, but it was difficult to know how much reliance to place on data from just two P. boisei specimens. Both studies now show that P. boisei tooth enamel δ¹³C values average −1.3 ± 0.9‰. A simple calculation yields a C₄ contribution to the diet of approximately 75% to 80%. An important contribution of the study of Cerling et al. (4) is that it allows a good deal of confidence to be placed in the results, based as it is on a decent sample size (N = 24 from 22 individuals) drawn from several sites and periods. Given the sample size and the tight clustering of data toward the positive, C₄ end of the spectrum, there can be no equivocation about the quality of the data and what the results mean.

What Do These Data Tell Us About the Dietary Ecology of P. boisei?

The carbon isotope composition of tooth enamel reflects all sources of carbon in the diet, with the primary difference coming from isotopically distinct C₃ or C₄ plants at the base of the foodweb. Since in most tropical African environments, trees, shrubs, rootstocks, and herbs follow the C₃ photosynthetic pathway, whereas almost

The consistently positive tooth enamel δ¹³C values for P. boisei indicate a direct reliance on C₄ plants.

all grasses and many sedges are C₄, the δ¹³C dietary distinction is really about whether the carbon source is from tropical grasses or sedges versus the edible parts of all other plants. Cerling et al. (4) make a compelling argument that the consistently positive tooth enamel δ¹³C values for P. boisei indicate a direct reliance on C₄ plants—grasses or sedges—because indirect consumption via C₄-consuming animals would require impossible amounts of animal food. That does not rule out C₄-consuming animals in the diet, but plants must form the bulk of the C₄ carbon contribution. Leaving aside the source of carbon for the moment, it is worth emphasizing that such a strong reliance on C₄ resources in a higher primate is well-nigh unique. It is rivaled only by the isotopic composition of the extinct Gelada baboon, Theropithecus oswaldi (figure 1 in ref. 4). That P. boisei was even more heavily reliant on C₄ resources than this grass-eating baboon is remarkable.

We can also infer with some confidence that these C₄ foods were reliable throughout the year because, although no intratooth measurements were made, δ¹³C values near zero allow little room for much seasonal variation. This observation stands in contrast with the high variability seen in the diets of the South African australopith, P. robustus (7). Such a strong emphasis also precludes notions that the C₄ resources formed “fall-back” foods—they must be regarded as staples, and moreover they seem to have formed a remarkably consistent part of the diet for at least a half million years (4).

Grasses or Sedges?

Is there any evidence to favor either grasses or sedges? On present evidence, there is little to choose between them. Cerling et al. weigh both options but are unable to draw a conclusion. The outcome carries important implications for the ecology, biogeography, and social structure of P. boisei, so it is worth pursuing further. Both grasses and sedges are low-quality foods, that is, they are low in protein and high in cellulose (9). Grasses are widely and fairly homogeneously distributed in African savannas, and they can support a relatively high biomass, as Cerling et al. point out (4). However, the more nutritious period of new growth and seed production is seasonally restricted to the wet season; in the dry season, grass blades quickly become desiccated, leaving the rhizomes as the remaining grassy resource. Modern grass-eating suids (i.e., warthogs) use the full suite of grass resources, as do modern Theropithecus gelada, which utilize grass rhizomes and herbs when their favored grass blades are unavailable (10). P. boisei could possibly have followed a similar strategy, as Jolly originally suggested (10). However, the idea of seed- and rhizome-gathering as a viable strategy has been regarded with some skepticism, possibly because it requires large energy expenditure for relatively poor energy returns. C₄ sedges, on the contrary, are perennials available throughout the year, but they are patchily distributed because most sedges prefer permanently damp ground or wetland (12). van der Merwe et al. favored exploitation of C₄ sedges as a plausible explanation for the high δ¹³C values observed in P. boisei, pointing out that the raw culms and large, fleshy rhizomes of the common, highly productive C₄ sedge, Cyperus papyrus, are good sources of carbohydrates still used by people in the Okavango Delta (7).

Can other, independent evidence be brought to bear on this question? Analyses of preferred habitat associations produce highly variable results (13, 14) and are thus not very helpful to address the question. Microwear texture analysis on P. boisei specimens are reported to show similar complexity profiles—and dissimilar directionality profiles—as T. oswaldi (6), suggesting that the mechanical properties were similar despite differing directionality imposed by tooth morphology. However, similar complexity profiles, and indeed steep macrowear profiles, in P. boisei could just as easily reflect repetitive forceful chewing of fibrous sedge culms and rhizomes. One of the problems faced by methods that investigate wear fabrics is the requirement for appropriate comparative models among primates. Yet, in the case of P. boisei, there are none, bar perhaps Theropithecus.

One other clue may reside in the enamel oxygen isotope (δ¹⁸O) composition. The δ¹⁸O values reported for P. boisei are low (−1.3 ± 1.3‰) compared with all other fauna except the hippopotamus (4). The authors suggest that this observation indicates high water dependency for P. boisei, which is almost certainly the case. However, these values are lower than those of the other obligate-drinking, grass-eating animals, such as the suids and equids (figure 2 in ref. 4). The low δ¹⁸O values universally observed in hippopotamus are linked to a combination of several factors: in addition to their requirements for copious amounts of drinking water, their semiaquatic lifestyle regulates body temperature without loss of water vapor, and nocturnal feeding means that internal water in their plant foods has relatively low δ¹⁸O values (15). It could be speculated that the low P. boisei δ¹⁸O values indicate more than just a requirement for drinking water; might their preferred plant foods have low values too? If that is the case, sedges would best fit the profile, as they are not subject to the same pressures of evapotranspiration as purely terrestrial plants. However, at present, we lack supporting evidence, as there are no published δ¹⁸O values for sedges, or sedge-consumers, and little information on the precise distribution of C₄ sedges in East African wetlands. We also lack δ¹⁸O and δ¹³C data for East African T. oswaldi, to provide a helpful terrestrial C₄ grass forager primate comparison for P. boisei. These gaps require further research.

What Are the Implications for P. boisei and P. robustus Monophyly?

Both stable carbon isotope and microwear data have shown that the diets of P. boisei in East Africa and P. robustus in South Africa are fundamentally different despite the apparent similarities in their craniofacial morphology. This result alone suggests that we should be more cautious in transferring observations between the two forms and look more carefully at the differences. There are no obvious external explanations for the stark difference in diet, and indeed δ¹³C values for early Homo in Swartkrans (16) and Olduvai (7) remain broadly similar. Cerling et al. (4) suggest an adaptive divergence in which P. boisei focuses on C₄ grasses and/or sedges whereas P. robustus continues to follow a more traditional hominoid mixed diet expanded to include some C₄ resources. This explanation is less than satisfactory given the timing. P. boisei sensu stricto possesses more derived morphological features than P. robustus, and yet appears earlier in East Africa (14). That would imply that the form with the more derived set of dentognathic features evolved before the less derived form, which makes little evolutionary sense. The important question, now given added impetus by the demonstration of the distinct dietary ecologies, is whether the two forms evolved from a common ancestor or independently in the two regions. Certainly there are suitable candidates for precursors in both regions—Paranthropus aethiopicus in East Africa and Australopithecus africanus in South Africa. Independent origins would constitute, as noted by Wood and Constantino (ref. 14, p. 118), “a striking case of homoplasy with both lineages exhibiting a morphological trend towards masticatory hypertrophy and extreme postcanine megadontia,” and further, all in the absence of close similarity in dietary ecology.

Certainly the new isotope data for P. boisei has unleashed many intriguing questions and raised new avenues for future research. That is the way it should be!