The robust jaws and large, thick-enameled molars of the Plio–Pleistocene hominins Australopithecus and Paranthropus have long been interpreted as adaptations for hard-object feeding. Recent studies of dental microwear indicate that only Paranthropus robustus regularly ate hard items, suggesting that the dentognathic anatomy of other australopiths reflects rare, seasonal exploitation of hard fallback foods. Here, we show that hard-object feeding cannot explain the extreme morphology of Paranthropus boisei. Rather, analysis of long-term dietary plasticity in an animal model suggests year-round reliance on tough foods requiring prolonged postcanine processing in P. boisei. Increased consumption of such items may have marked the earlier transition from Ardipithecus to Australopithecus, with routine hard-object feeding in P. robustus representing a novel behaviour.
The australopiths (genera Australopithecus and Paranthropus) represent the earliest well-documented diversification of the hominin lineage and include the ancestor of our own genus, Homo (figure 1) . Understanding the palaeobiological and phylogenetic implications of phenotypic variation in this group is therefore a critical step in the study of human origins [2–4]. Australopiths were characterized by numerous apomorphic craniodental features—including robust jaws and large postcanine teeth with thick enamel caps—that have long been interpreted as adaptations for countering powerful masticatory stresses associated with a diet of mechanically challenging foods, particularly hard objects [5–8]. However, this scenario has been challenged by recent studies of dental microwear, which have failed to detect signs of routine postcanine processing of hard items in most australopiths [9–11].
One hypothesis invoked to explain this discrepancy between anatomy and microwear is that australopith craniodental adaptations reflect hard fallback foods critical to surviving seasonal periods when easier-to-process preferred resources were scarce [9–10]. Accordingly, the dearth of fossil australopith teeth preserving evidence of hard-object consumption simply reflects the rarity of such fallback items  in the diets of these hominins. However, the role of dietary seasonality in shaping the masticatory apparatus of primates and other mammals is unclear. Indeed, there is little evidence that the highly robust jaws of the australopiths Paranthropus robustus and especially Paranthropus boisei would be required of an organism that relies only seasonally on mechanically challenging foods . Here, we report the results of a long-term diet-manipulation experiment conducted using an animal model that examines adaptive plasticity [14,15] in craniomandibular development vis-à-vis temporal variation in food mechanical properties. Our naturalistic, longitudinal data provide a novel perspective on debates over early hominin palaeoecology and have significant implications for understanding phenotypic variation in extant and fossil species that experience resource seasonality.
2. Material and methods
White rabbits (Oryctolagus cuniculus) exhibit several characteristics that make them excellent models for investigating questions regarding masticatory biomechanics in primates, including a vertically deep face; a temporomandibular joint situated high above the occlusal plane, capable of rotational and translational movements; intracortical bone remodelling; and a similar pattern of covariation among jaw–muscle activity, jaw loading and dietary properties [16–20]. Our sample contained 30 five-week-old weanling male rabbits divided equally into three dietary cohorts and raised for 48 weeks. Control subjects were fed a diet consisting solely of rabbit pellets. Annual rabbits were given hay in addition to pellets throughout the experiment, starting with two hay cubes daily for the first 18 weeks and then six hay cubes daily for the next six weeks. This 24 week schedule was then repeated. Seasonal rabbits received pellets and three hay cubes daily for the first six weeks and were then switched to an all-pellet diet for the subsequent 18 weeks, mimicking seasonal reliance on fallback foods. This schedule was repeated in the final 24 weeks.
The mechanical properties of these diets fall within the range of values for foods ingested by wild primates [21,22]. Hay and pellets result in similar levels of bone strain along the rabbit mandibular corpus . However, hay is stiffer than pellets  and therefore presents a greater masticatory challenge characterized by longer loading durations and greater cyclical loading. Compared with pellets, hay takes rabbits approximately three times as many chewing cycles and, correspondingly, three times longer to process (M Ravosa, J Scott and K McAbee, 2013, unpublished data). Dynamic alterations in dietary behaviours related to differences in the properties of experimental foods are posited to induce osteogenic responses and corresponding ontogenetic changes in the proportions of the feeding apparatus [17,22,23].
Each animal was imaged using micro-CT (Bioscan/Mediso X-CT; settings: 70 kVp, 100 μA, with 71 μm reconstructed isometric voxel size) upon arrival and every two weeks thereafter until week 24, when they became too large to image. Rabbits were imaged a final time following sacrifice at week 48. This dataset allowed us to track skull development longitudinally from weaning to mature adult stages. Using the segmenting tools available in the program PMOD, we quantified bone cross-sectional areas at three mandibular sites (symphysis, condyle and corpus) and one on the cranium (palate) involved in load resistance during chewing. To control for differences in organismal size that may confound our ability to detect a dietary signal among cohorts at a given age, shape ratios were computed by dividing the square root of each subject's cross-sectional area at a given time point by its cranial length. These ratios were logged for analysis. Statistical comparisons were performed using the bootstrap to generate confidence intervals for differences between groups . A more detailed description of the measurements and procedures is available in the electronic supplementary material.
Mean shape ratios for the three dietary cohorts were statistically indistinguishable at the onset of the experiment (week 0). By week 6, seasonal and annual rabbits had significantly greater shape ratios at all sites—indicating relatively larger cross-sectional areas—than control rabbits, while being similar to each other (table 1 and figure 2). Annual and seasonal rabbits diverged rapidly following the latter's shift to a less-challenging all-pellet diet after week 6. By week 12, means for the annual cohort were significantly larger than those for the seasonal and control cohorts at three of the four sites. Notably, seasonal rabbits differed from control animals only in relative palatal cross-sectional area. This general pattern persisted, with slight differences, through the first half of the experiment (week 24) and characterized the final set of comparisons: at week 48, annual rabbits had significantly larger symphyseal, palatal and corporal relative cross-sectional areas than seasonal and control groups; the latter two cohorts differed only at the symphysis (seasonal > control). Ratios for the annual group at this stage were 6–17% and 4–11% larger than the ratios for the control and seasonal groups, respectively. Such differences fall within the range of variation observed in closely related primate species that differ in diet (see electronic supplementary material, table S4), indicating that phenotypic plasticity is probably an important source of interspecific adaptive variation.
Seasonal hay consumption resulted in adult phenotypes clearly distinct from those of animals that ate hay throughout the experiment, but only minimally differentiated from those associated with the less-challenging all-pellet diet. This finding indicates that the relationship between dietary properties and craniomandibular morphology is highly dependent on loading history, specifically the extent to which a structure is exposed to a behavioural stimulus during development. With respect to linking australopith jaw robusticity to seasonal consumption of hard objects, this observation implies that the more extreme australopiths would have relied on such foods for a greater portion of the year than the more generalized species. It is notable, therefore, that P. boisei, the apex of the australopith trend toward increased jaw robusticity [1,6], presents molar microwear suggesting that it processed hard foods less frequently than the closely related but less-specialized P. robustus [10,25], the only australopith with microwear consistent with at least seasonal hard-object feeding . Our results therefore suggest that the apomorphic masticatory apparatus of P. boisei cannot be explained by a scenario in which this species fed mainly on relatively easy-to-process foods throughout the year while relying on hard objects during fallback episodes that were shorter in duration than was the case for P. robustus. Instead, the remarkable jaws of P. boisei probably reflect regular consumption of items that required intensive postcanine processing, resulting in masticatory stresses that exceeded those experienced by P. robustus.
Considered within the broader context of australopith variation, the link between morphology and markedly seasonal hard-object feeding in P. boisei appears even more tenuous. In jaw robusticity, species of Australopithecus fall between Paranthropus and the geologically older and more plesiomorphic Ardipithecus ramidus [1,3,4]. If the morphological differences between Ardipithecus and basal australopiths signal an adaptive shift to seasonal exploitation of hard objects in the latter, then our experimental evidence suggests that exaggeration of australopith craniodental features in Paranthropus implies increased reliance on such items. Microwear results for P. robustus fit this scenario, whereas those for P. boisei contradict it [10,25]. Indeed, the microwear signatures of P. boisei, Australopithecus anamensis, and Australopithecus afarensis are striking in their uniform lack of evidence for consumption of very hard or very tough items [10,11].
The functional significance of this microwear signature remains enigmatic, but it may result from prolonged milling and grinding, which would have been necessary when australopiths consumed tough foods because their low-cusped molars are not as well suited for shearing such items as are the high-cusped molars of extant folivorous primates . Although speculative, this interpretation is supported by our results: because the microwear data reject frequent hard-object feeding in P. boisei, this species must have masticated considerably tougher foods on a regular basis. The properties of foods eaten by species of Australopithecus are more difficult to infer, but differences between these hominins and Ar. ramidus in jaw robusticity, megadontia, and microwear [3,4], combined with the absence of a hard-object microwear signature , suggests tough-object feeding , but not to the degree inferred for P. boisei. We posit, therefore, that increases in jaw robusticity from Ardipithecus to Australopithecus to P. boisei reflect progressively greater reliance on tough, probably 13C-enriched  foods and concomitantly elevated masticatory stresses owing to higher repetitive loading and longer load durations resulting from extended bouts of milling and grinding . Under this scenario, the hard-object feeding evident in the microwear of P. robustus represents a novel feeding strategy, perhaps indicating adoption of a broader niche facilitated by a masticatory apparatus initially shaped by a diet of tough foods but nevertheless capable of processing objects with a wide range of mechanical properties.
All subjects were housed at the University of Notre Dame's animal care facility, Freimann Life Science Center, which is USDA-licensed, AAALAC-accredited, and subject to periodic inspections. Day-to-day care of the animals, including monitoring of their health, was handled by trained veterinary staff. All procedures used in this study were approved by the University of Notre Dame's Institutional Animal Care and Use Committee.
This work was financially supported by NSF grant BCS-1029149/1214767 to M.J.R.
We thank E. Franks, A. Remer, S. Stack, D. Daegling, the Notre Dame Integrated Imaging Facility, and the staff of Freimann Life Science Center.
- Received September 9, 2013.
- Accepted December 3, 2013.
- © 2014 The Author(s) Published by the Royal Society. All rights reserved.