Peter Ungar (2004) investigated the dietary adaptations of A. afarensis and early Homo by looking at the three-dimensional topography of their teeth. the shapes of the teeth are expected to reflect diet because the teeth themselves are adaptations for processing food. Among mammals there are some regular relationships between the morphology of an organism's teeth and the type of food it prefers. Looking specifically at primates, high-crowned teeth that interlock between the upper and lower jaws are adapted to shearing foods, in a manner analogous to the pinking shears of a seamstress. Shearing is necessary for cutting and puncturing through food items, which are the predominant actions necessary to process leaves and insects, respectively. Primates that specialize on other foods, notably fruits or harder objects like seeds, tends to have flatter teeth without high crests. These teeth useful for crushing or grinding foods. So and examination of the anatomy of the teeth is expected to give insights about the diets of ancient primates.
But direct examination of fossil tooth anatomy is complicated by the fact that the teeth wear during the course of an animal's life. The effective chewing characteristics of a tooth therefore change over time, as a result of progressive attrition on its occlusal surface. This means that it's difficult to measure the anatomy of teeth, as in the lengths of different crests or the area of shearing surfaces, because these characteristics are altered in individuals of different ages. This is a special problem among early hominids, whose teeth are generally heavily worn even among relatively young individuals. It also means that the chewing characteristics of younger individuals and older individuals may actually have been different, possibly reflecting ontogenetic differences in diet. The details of the anatomy of unworn teeth may have been largely irrelevant to the pattern of mastication throughout most of the individual's life.
Ungar (2004) applied three-dimensional modeling to assess the shape characteristics of teeth at different stages of wear. In this way he was able to quantify the surface characteristics of the teeth--that is, how flat or jagged they were, how high the cusps were, and any angulation or deviation from a horizontal occlusal surface. He applied the technique to a series of second mandibular molars (M two) of Australopithecus afarensis and early Homo (including specimens attributed to H. erectus, H. rudolfensis, and H. habilis). As a comparison, Unger also examined a series of gorillas and chimpanzees, not because their diets were probably similar, but because their molar cusp patterns are similar, yielding comparable topographic results. The topographic values derived from each of the specimens included the "average slope," which was a measure of the average vertical displacement between adjacent points on the tooth, and the "occlusal relief," which was the measurement of the three dimensional surface area scaled to the two dimensional occlusal surface area of the tooth. In principle, more jagged teeth with higher cusps will have a higher average slope and a higher occlusal relief.
The results indicated that the early Homo specimens had relatively high average slope, except in the most worn molars. These specimens had average slope near that of gorillas, which were the highest among the samples measured. The occlusal relief of early Homo molars was not as great as that of guerrillas, but was substantially larger than chimpanzees, except again among the most worn specimens.
In contrast, A. afarensis had by far the lowest average slope and occlusal relief among any of the samples. Chimpanzees were lower than early Homo, but the A. afarensis sample was still significantly flatter in its dental morphology. These results held true across a range of wear categories, degrading slightly among the most worn sample of teeth in all four species. But as teeth wore down in all four species, they tended to become flatter. His results perhaps isn't surprising but does inform about the chewing characteristics of the teeth of older individuals.
Ungar (2004) includes a discussion that addresses why the differences in surface characteristics may have come to characterize these samples. He begins with a discussion of the differences between chimpanzees and gorillas in their diets:
The differences in occlusal morphology between chimpanzees and gorillas evidently relate to differences in the material properties of the foods they eat, particularly their fallback foods. Central African common chimpanzees are primarily frugivorous, with soft fruits reported to constitute 70-80% of their diets (Kuroda, 1992; Tutin et al., 1997). Fruit is also commonly consumed by western lowland gorillas, making up about half of the food species found in their fecal remains (Williamson et al., 1990; Nishihara, 1992; Remis, 1997; Doran et al., 2002). Differences and similarities in food preferences are most obvious were these taxa are sympatric and have access to the same resources. At Lope, Gabon, for example, the dietary overlap is substantial, with gorillas reported to consume 73% of the food species eaten by chimpanzees (Tutin and Fernandez, 1985). Differences between the two taxa are notable at times of fruit scarcity though, when gorillas fall back on tougher, more fibrous foods (such as leaves and stems) than those eaten by chimpanzees (Tutin et al., 1991; Remis, 1997) (615).
It should be reiterated that differences in occlusal morphology between P. t. troglodytes and G. g. gorilla evidently reflect differences in fallback resources rather than preferred foods. While both tax evidently prefer soft fruits when available, differences in occlusal morphology apparently allow the gorillas to take advantage of fallback foods that are less accessible to the chimpanzees (615)
For the case of A. afarensis, Ungar concludes that the differences in occlusal morphology from chimpanzees are not as great as the differences between chimpanzees and gorillas, although in the opposite direction. In his view this implies that A. afarensis included more brittle, less deformable foods, possibly as a fallback strategy in contrast to chimpanzees.
For the case of early Homo, the conclusion is that the molars were intermediate between chimpanzees and gorillas in their occlusal characteristics. This means that, compared to chimpanzees and compared to A. afarensis, early Homo was better adapted to chewing "tough, pliant foods" (616). This has important implications for dietary reconstruction in early humans:
[T]ubers, especially larger ones (Baritelle and Hyde, 1999) are often fairly brittle, whereas mammalian soft tissues tend to be tough and elastic (Lucas and Peters, 2000). Thus, meat seems more likely to have been a key tough-food resource for early Homo then would have USOs. It has also been noted that USOs those are of limited nutritional value (Schoeninger et al., 2001), and so would not have made very good keystone resources (616-617)
Ungar does not mention a number of other arguments that might also favor this interpretation. The contrast between Homo and A. afarensis is in the same direction as the contrast in occlusal morphology between primarily meat-eating carnivores like felids and canids as opposed to more omnivorous carnivores like bears. Another observation is that meat is a major food resource of chimpanzees, although this is hardly a fallback resource. Indeed, if meat eating was indeed an important component of the behavioral repertoire of early Homo, it probably is not fair to assert that the difference in diet between Homo and Australopithecus was primarily a difference in fallback resources. It may be true that australopithecines and early Homo overlapped in their food resources, particularly in plant species consumed. But considering the likely effectiveness of early humans as predators, I think it likely that the fallback foods of early humans--when hunting was ineffective--may well have been the preferred foods of australopithecines. And when australopithecines were forced to abandon their preferred foods by early humans, they were forced to fall back upon resources that either were common or were difficult for early Homo to exploit. The disappearance of early small-bodied Homo by around 1.6 million years ago, and the ultimate extinction of the robust australopithecines after a progressive increase in their molar sizes (Wood and Lieberman 2001) indicate that this fallback strategy could not be maintained in the face of increased hunting effectiveness by large-bodied Homo.
Ungar P. 2004. Dental topography and diets of Australopithecus afarensis and early Homo. J Hum Evol 46:605-622.
Other references therein.