Molar microwear textures and the diets of Australopithecus anamensis and Australopithecus afarensis

Peter S. Ungar, Robert S. Scott, Frederick E. Grine, Mark F. Teaford


Many researchers have suggested that Australopithecus anamensis and Australopithecus afarensis were among the earliest hominins to have diets that included hard, brittle items. Here we examine dental microwear textures of these hominins for evidence of this. The molars of three Au. anamensis and 19 Au. afarensis specimens examined preserve unobscured antemortem microwear. Microwear textures of these individuals closely resemble those of Paranthropus boisei, having lower complexity values than Australopithecus africanus and especially Paranthropus robustus. The microwear texture complexity values for Au. anamensis and Au. afarensis are similar to those of the grass-eating Theropithecus gelada and folivorous Alouatta palliata and Trachypithecus cristatus. This implies that these Au. anamensis and Au. afarensis individuals did not have diets dominated by hard, brittle foods shortly before their deaths. On the other hand, microwear texture anisotropy values for these taxa are lower on average than those of Theropithecus, Alouatta or Trachypithecus. This suggests that the fossil taxa did not have diets dominated by tough foods either, or if they did that directions of tooth–tooth movement were less constrained than in higher cusped and sharper crested extant primate grass eaters and folivores.

1. Introduction

Researchers have recognized for more than three decades that patterns of microscopic use wear on teeth hold the potential to provide information about the diets of early hominins (e.g. Grine 1977, 1981, 1986; Puech 1979; Ryan 1980; Walker 1981; Ungar et al. 2006). Many studies of hominin dental microwear have been published over the past thirty years, including feature-based quantitative analyses of molar occlusal surface microwear in Australopithecus afarensis1 and Australopithecus anamensis (Grine et al. 2006a,b). Data presented here extend this work by comparing the microwear textures of these species with those of other early hominins and recent primate taxa with known diets. Results indicate that sampled Au. anamensis and Au. afarensis individuals tend to have relatively simple microwear surface textures varying in degree of anisotropy. This pattern is comparable to that previously reported for Paranthropus boisei (Ungar et al. 2008), but differs from that of Au. africanus, and especially of Paranthropus robustus (Scott et al. 2005; see also Grine 1986).

(a) Background

Australopithecus anamensis and Au. afarensis have traditionally been argued to be the earliest hominins to show an adaptive shift from diets dominated by soft, sugary forest fruits to hard and brittle or abrasive foods (Ward et al. 1999; Teaford & Ungar 2000; White et al. 2000; Wood & Richmond 2000; Walker 2002; Macho et al. 2005). Australopithecus afarensis, for example, has long been noted to have thickly enamelled, large and flat crowned cheek teeth and robustly constructed crania and mandibles, at least when compared with our nearest living relatives, the chimpanzees (e.g. McHenry 1984; Hylander 1988; White et al. 2000). This enhanced craniodental toolkit has led workers to suggest that ‘nuts, seeds, and hard fruit may have been an important component to the diet of this species’ (Wood & Richmond 2000).

These hominins have also been thought to have had dietary adaptations intermediate between those of frugivorous forest apes and later hominins. White et al. (2000), for example, considered them to have taken the ‘initial functional steps that would eventually culminate in the far more derived, specialized masticatory apparatus of later hominid species’ such as Au. africanus and especially Par. boisei and Par. robustus. As White et al. (1981) noted, while Au. africanus showed several craniodental features foreshadowing the functional specializations seen in Paranthropus, Au. afarensis retained at least some more primitive traits, such as relatively larger front teeth and less swollen and inflated cusps on their postcanines. Picq (1990) further suggested that the height and thickness of the Au. afarensis mandibular corpus, and robusticity of the symphysis, were intermediate between those of chimpanzees on the one hand and those of Au. africanus and especially Paranthropus species on the other. He argued the same for the height of the ramus and size of the mandibular condyle.

Picq (1990) used these lines of evidence to build a scenario in which Au. afarensis was still dependent on the fruits of the forest, but seasonally sought sustenance in more open settings with ‘tougher foods containing abrasive structures’ and ‘nuts protected by skins or a hard shell’. Ungar & Teaford (2001) called this a ‘mixed forest–savanna resource adaptation’. Ungar (2004) further noted that the degree of differences in occlusal slope and relief between Au. afarensis and chimpanzees are as expected for differences in ‘fallback’ foods, suggesting that the early hominins may have preferred soft, sugar-rich fruits, but had the ability to make more effective use of hard, brittle resources as seasonal availabilities required.

Following its initial description, Au. anamensis quickly took on the role of ‘intermediate form’ in terms of dietary adaptation between the earlier Ardipithecus and Au. afarensis. Like Au. afarensis, Au. anamensis had thicker post-canine tooth enamel and larger average cheek teeth than Ardipithecus, again suggesting a dietary shift towards harder foods or more abrasive ones (Ward et al. 1999; Teaford & Ungar 2000; Wood & Richmond 2000; Walker 2002; Suwa et al. 2009). Still, in comparison with Au. afarensis, Au. anamensis molars were ‘not very high-crowned’ (Walker 2002), and the larger values for enamel thickness in Au. anamensis when compared with Ar. ramidus may reflect, at least in part, increased tooth size in the former (Suwa et al. 2009, supporting online material). Further, Au. anamensis lacked changes in the geometry of the mandible and maxilla seen in Au. afarensis, Au. africanus and especially Paranthropus (Ward et al. 1999), such as greater mandibular corpus robusticity, which might buttress the jaw against higher peak force magnitudes or repetitive loading in mastication (Teaford & Ungar 2000). These differences led Teaford & Ungar (2000) to speculate that hard and perhaps abrasive foods may have become even more important components of the diet of Au. afarensis. Walker (2002) suggested that changes from Au. anamensis to Au. afarensis were carried to extremes in Au. africanus and especially Paranthropus.

The microwear evidence paints a somewhat more complex picture. One might predict heavy pitting associated with hard-object consumption in Au. anamensis and Au. afarensis and an increase in pitting from the earlier to the later species. Contrary to expectation though, both show fairly fine features and microwear surfaces dominated by striations rather than pits (Grine et al. 2006a,b; Suwa et al. 2009, supporting online material). This pattern is more similar to that seen in the tough food folivore Gorilla gorilla beringei than in hard-object feeders, such as the grey-cheeked mangabey Lophocebus albigena and the brown capuchin, Cebus apella. Further, the microwear pattern for these hominins is remarkably homogeneous between specimens across both time and paleoecological context. Grine et al. (2006a,b) suggest that these results highlight the difference between ‘faculty’ and ‘biological role’ (Bock & von Wahlert 1965) or dietary potential and what an animal eats on a day-to-day basis. As a result, Grine et al. (2006a,b) suggested that the shift in diet-related adaptive morphology in Au. anamensis and Au. afarensis may relate more to occasional but critical fallback food exploitation than to preferred resources.

We might predict, if Au. africanus and especially Par. boisei and Par. robustus show further craniodental specializations for the consumption of hard, brittle foods, that samples of these species should have more individuals showing high levels of microwear pitting or surface complexity with fewer fine, parallel striations and with texture anisotropy, compared with Au. afarensis and especially Au. anamensis.

Here we present the first microwear texture analysis of Au. afarensis and Au. anamensis for comparison with Au. africanus, Par. boisei and Par. robustus. Microwear texture analysis has proved to provide a three-dimensional characterization of microwear surfaces free from the need to identify and measure individual features (e.g. Ungar et al. 2003, 2007a,b, 2008; Scott et al. 2005, 2006, 2009a; El-Zaatari 2008; Krueger et al. 2008; Ungar & Scott 2009; Krueger & Ungar in press). While microwear texture analysis and conventional feature-based analyses to date have yielded similar results for other hominins (compare Grine (1986) with Scott et al. (2005) and Ungar et al. (2006) with Ungar & Scott (2009)), texture analyses are particularly well suited for between-studies comparisons, because data collected are free from observer measurement error. Further, microwear texture results for Au. africanus, Par. robustus and Par. boisei are available in the literature for comparison (Scott et al. 2005; Ungar et al. 2008). Indeed, only texture data are available for Par. boisei.

Microwear texture analysis studies indicate that hard-object feeding extant primates (and other mammals) tend to have higher average levels of texture complexity and lower levels of anisotropy on their molar occlusal surfaces than seen in tough food eaters (see Ungar et al. 2007b for a review). Here we test the hypothesis that interpretations of craniodental functional morphology described above are reflected in microwear texture patterns. If so, Au. afarensis and especially Au. anamensis specimens should show less surface complexity and more anisotropy than those of Au. africanus, and especially Par. boisei and Par. robustus.

2. Material and methods

Dental microwear texture data are presented here for the molar teeth of Au. anamensis from Kanapoi and Allia Bay in Kenya and Au. afarensis from the Laetolil Beds in Tanzania and the Hadar Formation in Ethiopia. Specimens included in this study are the same as those employed in feature-based SEM microwear analyses reported by Grine et al. (2006a,b).

All permanent molars available to us were first examined to determine the suitability for microwear analysis. Those that preserved wear facets were cleaned with cotton swabs soaked in alcohol or acetone to remove adherent dirt or preservatives. Moulds of occlusal surfaces were then made using President Jet Regular Body vinyl dental impression material (Coltène-Whaledent Corp.), and casts were produced from these molds using Epotek 301 epoxy and hardener (Epoxy Technologies Inc.).

Replicas were examined by light microscopy and SEM as necessary to determine the suitability for microwear analysis following the criteria described by Teaford (1988). Most specimens had occlusal surfaces obscured by taphonomic damage and so had to be excluded from this study. In the end, the molars of only three of the Au. anamensis specimens and 19 Au. afarensis individuals available to us were found to preserve unobscured antemortem occlusal surface microwear. A list of all specimens considered can be found in Grine et al. (2006a,b) and those included in this study are presented in table 1.

View this table:
Table 1.

Microwear texture analysis data for Au. anamensis and Au. afarensis.

All specimens included in this study were analysed using a Sensofar Plµ white-light confocal profiler (Solarius, Inc.) with an integrated vertical scanning interferometer. Three-dimensional point clouds were collected for ‘phase II’ facets (the buccal occlusal surfaces of lower molars and the lingual occlusal surface of uppers) using a 100× long working distance objective. The point clouds sampled elevations at intervals of 0.18 µm along the x- and y-axes, with a vertical resolution of 0.005 µm. Data were obtained for four adjacent fields on facets 9 or 10n, for a combined work envelope of 276 µm × 204 µm. The feature-based study of these specimens reported in Grine et al. (2006a,b) used a pixel resolution of 0.25 µm and combined sample area of 0.04 mm2.

Resulting point clouds were analysed using ToothFrax and SFrax scale-sensitive fractal analysis (SSFA) software (Surfract Corp.). SSFA as applied to microwear research is described in detail elsewhere (e.g. Scott et al. 2006). The basic premise is that surface texture varies with scale of observation, and that this variation can be used to characterize functionally relevant aspects of microwear. SSFA texture variables included in this study are area-scale fractal complexity (Asfc), anisotropy (epLsar), scale of maximum complexity (Smc) and heterogeneity of complexity (HAsfc). Values for individual specimens are reported as medians of the four fields sampled for each specimen.

Area-scale fractal complexity is a measure of change in roughness with scale. The faster a measured surface area increases with resolution, the more complex the surface. Anisotropy is characterized as variation in lengths of transect lines measured at a given scale (we use 1.8 µm) with orientations sampled at 5° intervals across a surface. An anisotropic surface will have shorter transects in the direction of the surface pattern than perpendicular to it (e.g. a transect that cross-cuts parallel scratches must trace the peaks and valleys of each individual feature). Thus, a heavily pitted surface typically has high Asfc and low epLsar values, whereas one dominated by homogeneous, parallel striations has low Asfc and high epLsar values. Other variables used to characterize microwear surface texture include Smc, the scale range over which Asfc is calculated, and HAsfc, variation of Asfc across a surface (in this case, each field of view was divided into a 3 × 3 grid = HAsfc9, and a 9 × 9 grid = HAsfc81). High Smc values should correspond to more complex coarse features. High HAsfc values are observed for surfaces that vary in complexity across a facet.

Statistical analyses focused on comparisons of microwear textures of Au. afarensis with those of Au. africanus and Par. robustus from South Africa as reported in Scott et al. (2005) and with Par. boisei from eastern Africa as reported by Ungar et al. (2008). Australopithecus anamensis data could not be compared statistically with those of the other hominins given an available sample of only three specimens, although newly recovered specimens (e.g. Haile-Selassie, this volume) hold the potential for larger microwear datasets in the future. First, differences in central tendencies between taxa were assessed using a MANOVA performed on ranked data (Conover & Iman 1981) for all variables (Asfc, epLsar, Smc, HAsfc9 and HAsfc81). Individual ANOVAs and multiple comparisons tests were used to determine the sources of significant variation. Both Tukey's honestly significant difference (HSD) and Fisher's least significant difference (LSD) tests were used to balance risks of type I and type II errors (Cook & Farewell 1996). Values of p < 0.05 for Tukey's HSD tests may be assigned significance with some confidence, whereas values of p < 0.05 on Fisher's LSD but not Tukey's LSD tests are considered suggestive but of marginal significance.

Degree of variance in microwear textures within taxa may be as important for distinguishing species as differences in central tendencies, especially given differences in foraging and feeding strategies between primates. With this in mind, raw data for each variable were transformed for Levene's test following the procedure described by Plavcan & Cope (2001) to compare distribution variances between taxa. A MANOVA was used to assess the variation in variance between taxa, and as with the comparisons of central tendencies, ANOVAs and multiple comparisons tests were used to determine the sources of significant variation as needed.

Complexity and anisotropy results for Au. anamensis and Au. afarensis were also compared with those for a series of extant primates with known differences in feeding behaviors to put the microwear texture analysis results for these hominins in the context of modern primate diets. The baseline series includes: (i) the mantled howler, Alouatta palliata and the silvered leaf monkey, Trachypithecus cristatus, C. apella and L. albigena from the dataset reported by Scott et al. (2005); (ii) G. gorilla beringei, G. g. gorilla, Pan troglodytes and the orangutan Pongo pygmaeus from Ungar et al. (2007b); and (iii) the yellow baboon Papio cynocephalus and the gelada baboon Theropithecus gelada from Scott et al. (2009b). The mantled howler and silvered leaf monkey are typically characterized as folivores, whereas the brown capuchin and grey-cheeked mangabey are considered to be hard-object fallback feeders. Among the great apes, chimpanzees are the most frugivorous, and gorillas, especially the G. g. beringei sample considered here (the Fossey collection at the US National Museum of Natural History), are the most folivorous (see references in Ungar et al. 2007b). Orangutans are intermediate in their diets. Finally, geladas are specialized grass eaters, whereas yellow baboons have a more catholic diet including fruits, leaves and animal prey (see Post 1982; Norton et al. 1987; Dunbar 1988; Altmann 1998; Pochron 2000; Bentley-Condit 2009).

3. Results

Results for Au. anamensis and Au. afarensis are presented in tables 14 and are illustrated in figures 1 and 2.

View this table:
Table 2.

Summary statistics for early hominins.

View this table:
Table 3.

Analyses of hominin microwear texture data (central tendencies). All data rank transformed to mitigate violation of assumptions inherent to parametric statistics (Conover & Iman 1981). *p < 0.05 for Fisher's LSD test, **p < 0.05 for both Tukey's HSD and Fisher's LSD tests (shown in italic).

View this table:
Table 4.

Analyses of hominin microwear texture data (variance). Microwear data transformed for Levene's test (X ′ = |X—mean (X)|) following Plavcan & Cope (2001). *p < 0.05 for Fisher's LSD test, **p < 0.05 for both Tukey's HSD and Fisher's LSD tests (shown in italic).

Figure 1.

Photosimulations of molar microwear surfaces of (a) Au. anamensis and (b) Au. afarensis generated from point cloud data collected using the white-light confocal profiler. Each image represents a surface 208 µm × 280 µm.

Figure 2.

Plots of (a) anisotropy versus complexity and (b) heterogeneity of complexity versus scale of maximum complexity (below) for early hominin individuals considered by taxon. Open triangle, Au. afarensis; filled triangle, Au. africanus; open circle, Par. boisei; filled circle, Par. robustus; diamond, Au. anamensis.

(a) Comparisons with other fossil hominins

The early hominins are well-separated from one another by microwear texture complexity. First, the specimens from South Africa have higher Asfc values on average than those from eastern Africa, regardless of the species considered. Within the eastern African sample, Au. anamensis may have slightly higher complexity on average than Au. afarensis or Par. boisei, but larger samples of Au. anamensis are really needed to evaluate this (only one Au. anamensis specimen is outside the range of Au. afarensis or Par. boisei). No significant variation in complexity is noted between Au. afarensis and Par. boisei. Within the South African sample, Par. robustus has more complex microwear surfaces on average than Au. africanus. The Par. robustus sample also shows significantly greater variation in its complexity values than do any of the other taxa. All other taxa have similar levels of within-species variation in complexity.

The species do not differ as much in anisotropy as they do in complexity, though Par. robustus has both a lower average epLsar and less variability in this variable than either Au. africanus or Au. afarensis. The Par. robustus sample also has a lower average scale of maximum complexity than Au. afarensis, Au. africanus or Par. boisei, though no other significant differences in Smc central tendencies are noted. In addition, Au. africanus has significantly greater variation in Smc values than Au. afarensis, Par. robustus and (marginally) both Au. anamensis and Par. boisei. It should be noted though that the high Au. africanus variance is driven by a single outlier with an extremely high Smc value. The Par. robustus sample also has higher average HAsfc values than Au. anamensis, Au. afarensis, Par. boisei or (at least marginally) Au. africanus. In addition, Au. africanus has higher HAsfc values than Au. afarensis. Further, Par. boisei has marginally lower HAsfc values than Au. africanus, but marginally higher heterogeneity (at least for HAsfc81) than Au. afarensis. Finally, Par. robustus has more variation in its HAsfc values than Au. anamensis, Au. afarensis, Au. africanus and Par. boisei.

In summary, Au. afarensis microwear textures are most similar to those of Par. boisei, with the only difference being a marginally higher average value for one of the heterogeneity measures. Australopithecus afarensis also has lower complexity on average and less heterogeneity than Au. africanus, as well as less variation in scale of maximum complexity (though the latter result is driven largely by a single outlier). Australopithecus afarensis differs most markedly from Par. robustus. Australopithecus afarensis has lower complexity and heterogeneity of complexity, higher scale of maximum complexity, and marginally higher anisotropy. The degree of variation in anisotropy values is also greater for Au. afarensis, though the variation in complexity and heterogeneity are greater in Par. robustus. In fact, Par. robustus is very much the ‘outlier’ compared with the other taxa in most cases. This hominin tends towards a greater spread in and larger average values for complexity, scale of maximum complexity and heterogeneity of complexity, though less spread in, and lower values for anisotropy.

Results for Au. anamensis are more difficult to have confidence in, given its sample size, though these do separate clearly from Par. robustus. There are some additional ‘hints’ suggested by the data if the patterns hold with larger samples. Australopithecus anamensis has marginally higher average complexity and lower variation in anisotropy than either Au. afarensis or Par. boisei, as well as marginally higher heterogeneity than Au. afarensis. These results will be of limited interpretability until larger samples are available for study.

(b) Comparisons with extant primates

Microwear complexity and anisotropy results for Au. anamensis and Au. afarensis are illustrated alongside data for living primates in figure 3. These hominins are on the low end for both Asfc and epLsar, both in their means and in their variations, when considered in light of the extant primate baseline. The distributions and central tendencies of complexity values for these hominins are most comparable to those reported for A. palliata, T. gelada and T. cristatus. They lack the degrees of dispersion in Asfc seen in the other primates, especially C. apella, P. cynocephalus and L. albigena. In contrast, the anisotropy values for the hominins are most different from those for A. palliata, T. gelada and T. cristatus and are more similar to those of the other primates, especially P. troglodytes and P. cynocephalus.

Figure 3.

Plots of anisotropy versus complexity for various extant primates, compared with data for Au. anamensis and Au. afarensis.

4. Discussion

(a) Microwear and the diets of Au. anamensis and Au. afarensis

Dental microwear texture analysis results suggest that neither the Au. anamensis nor the Au. afarensis individuals included in this study had diets dominated by hard, brittle foods in the days, weeks or perhaps even months prior to their deaths. While these species have been suggested to show an adaptive shift from diets dominated by soft forest fruits to hard, brittle foods (e.g. Ward et al. 1999; Teaford & Ungar 2000; White et al. 2000; Wood & Richmond 2000; Walker 2002), none of the specimens examined exhibit the high microwear surface texture complexity expected of a hard-object feeder. The distribution of Asfc values more closely resembles those of the grass-eating T. gelada and the folivores A. palliata and T. cristatus than hard-object feeding L. albigena or C. apella.

The anisotropy results for Au. anamensis and Au. afarensis are, on the other hand, most different from those of A. palliata, T. cristatus and T. gelada among the baseline series. These three extant primates have higher average anisotropy values than the early hominins. High anisotropy is often taken as a proxy for tough food consumption and repetitive chew cycles with opposing teeth moving past one another along constrained paths. At first glance, this might suggest that Au. anamensis and Au. afarensis specimens sampled also avoided tough foods in the period prior to death, although the combination of low anisotropy and low complexity averages in Au. anamensis and Au. afarensis is unusual for primates. Most extant primate samples published to date have either high anisotropy averages combined with low complexity values, associated with the consumption of tough foods, or low anisotropy combined with high complexity averages consistent with a hard-brittle item diet.

We propose that Au. anamensis and Au. afarensis may have indeed consumed tough foods, but that their anisotropy values are low because their dentognathic morphology did not limit occlusal movements to the degree presumed for primates with steeper occlusal surfaces, such as A. palliata, T. cristatus and T. gelada. The flatter teeth of early hominins offer fewer constraints to masticatory movements during food fracture and may therefore allow ‘grinding’ action (sensu Simpson 1933; Kay & Hiiemae 1974), wherein food items are milled between opposing surfaces. The combination of flat teeth and tough foods would be expected to result in a combination of low complexity and low anisotropy. Still, the relationship between microwear feature anisotropy and occlusal topographic relief remains to be investigated fully.

(b) Comparisons with other early hominins

Studies of craniodental functional morphology have suggested to several researchers a tendency towards increasing specialization for hard objects from Au. anamensis to Au. afarensis, Au. africanus and finally Par. robustus and Par. boisei (see White et al. 1981, 2000; Ward et al. 1999; Teaford & Ungar 2000; Wood & Richmond 2000; Walker 2002). Australopithecus africanus evinces a higher average Asfc than Au. afarensis, and Par. robustus has an even higher average Asfc value. The same is true for heterogeneity and variation in the scale of maximum complexity (though the latter result appears to be driven by a single outlier). These findings are all consistent with an increasing component of hard, brittle items in the diet of Au. africanus compared with Au. afarensis and Par. robustus compared with Au. africanus.

On the other hand, the distribution of Asfc values for Par. boisei is very similar to that of Au. afarensis. Further, while the sample for Au. anamensis is too small for a reasonable comparison with other taxa, its mean Asfc value and heterogeneity are, if anything, slightly higher than that of Au. afarensis. Thus, if high texture complexity is considered to be a proxy for the consumption of hard, brittle items, there is no evidence for an increase in the role of such foods in the diet from Au. anamensis to Au. afarensis to Par. boisei.

Ungar et al. (2008) remarked on the apparent discordance between microwear in Par. boisei and biomechanical models for this species based on craniodental functional morphology. They proposed that Par. boisei may represent a hominin example of Liem's Paradox, wherein craniodental specializations developed as an adaptation to processing less preferred, mechanically challenging foods, even though the microwear suggests that these hominins rarely consumed such foods. This does not, however, explain either the microwear differences between Par. boisei and Par. robustus (given similarities in their gnathodental adaptations) or the similarities between Par. boisei, Au. anamensis and Au. afarensis (given differences in their gnathodental adaptations).

The story becomes even more complicated when we consider these taxa in their presumed phylogenetic contexts, especially the purported anagenetic lineage leading from Au. anamensis to Au. afarensis, Par. aethopicus and finally Par. boisei (Kimbel et al. 2006; Rak et al. 2007). Many workers have suggested an ancestor–descendant relationship between Au. anamensis and Au. afarensis (Ward et al. 1999, 2001; White et al. 2000, 2006; Walker 2002). Such relationships are consistent with numerical cladistic studies (Strait et al. 1997; Strait & Grine 2004) and with changes in dentognathic characters among temporally successive samples from Kanapoi, Allia Bay, Laetoli and Hadar (Kimbel et al. 2006).

Studies of cranial morphology have suggested that the pattern of intracranial venous sinus drainage is shared between Au. afarensis and Paranthropus species for the exclusion of Au. africanus and Homo (Falk & Conroy 1983; Kimbel 1984; Falk 1988), and additional fossils have reinforced this similarity (Kimbel et al. 2004; de Ruiter et al. 2006). Kimbel et al. (2004) have, moreover, identified an additional six cranial characters that Au. afarensis shares with one or more species of Paranthropus, although they regarded them, like the pattern of intracranial venous drainage, as being homoplastic. Most recently, similarities in ramal morphology between Au. afarensis and Par. robustus mandibles have been observed by Rak et al. (2007), who interpreted them as synapomorphies, suggesting that Au. afarensis and Par. robustus are united in a single clade and that this possibly includes Par. aethiopicus and Par. boisei, although no fossils attributable to the latter two species preserve the relevant anatomy. The evidence suggesting an ancestor–descendant relationship between Au. afarensis and Paranthropus, and particularly that for an Au. afarensisPar. aethiopicusPar. boisei lineage in eastern Africa, is not wholly inconsistent with their hypothesized cladistic relationships (Strait et al. 1997; Kimbel et al. 2004; Strait & Grine 2004).

Similarities in microwear texture results for Au. anamensis, Au. afarensis and Par. boisei may make sense if these taxa comprise an anagenetic lineage and all shared food-type preferences. One possible scenario might be increasing craniodental specializations through the lineage for repetitive loading given consumption of tough foods in the face of morphological constraints imposed by relatively flat cheek teeth. This would be consistent with a C4 isotope signature in Par. boisei (van der Merwe et al. 2008) and the consumption of tough grasses or sedges, assuming that the specimens examined thus far are representative. Carbon stable isotope studies of Au. anamensis, Au. afarensis and additional Par. boisei specimens would provide a valuable test of this hypothesis.

On the other hand, if Par. boisei and Par. robustus form a clade excluding Au. africanus, the two early hominins from South Africa probably independently increased their consumption of hard, brittle foods as evidenced by increased pit percentages in both hominins. In any case, the microwear texture patterns of the eastern African early hominins are more similar to one another than to those of the South African early hominins, independent of whether one is considering Australopithecus or Paranthropus.

So in the end, what can be said of the microwear of Au. anamensis and Au. afarensis? We may reasonably infer that specimens examined for this study did not have a diet dominated by hard and brittle foods, at least shortly before death. Picq (1990) proposed that Au. afarensis often consumed soft foods that were not fracture resistant, but had craniomandibular adaptations for seasonal consumption of hard, brittle foods. Grine et al. (2006a,b) further suggested that traditional microwear results for both Au. anamensis and Au. afarensis are best explained by food preferences for less mechanically challenging foods, though as Ungar (2004) noted, their occlusal morphology would have allowed the consumption of hard, brittle items in times of dietary stress when favored foods were unavailable. The microwear texture analysis data presented here cannot be used to falsify the notion of rare hard-object feeding, but it also provides no evidence for it. Whether or not the craniodental specializations seen in Au. anamensis and Au. afarensis are adaptations for the occasional consumption of hard, brittle foods, however, their microwear texture patterns are consistent with the regular consumption of softer and or tougher items.


We are grateful to curators at the National Museums of Ethiopia, Kenya and Tanzania for permission to examine and make moulds of specimens in their care and thank Alejandro Peréz-Peréz for his assistance with preparing these replicas. Kristin Krueger helped with the preparation of figure 1. This study was funded by the US National Science Foundation.


  • One contribution of 14 to a Discussion Meeting Issue ‘The first four million years of human evolution’.

  • 1 One of us (F.E.G.) prefers the nomen Praeanthropus afarensis, as use of the generic name Australopithecus for Au. anamensis and Au. afarensis probably violates the criterion of monophyly (Grine & Strait 2000). Because of the lack of consensus among the authors on the taxonomy of these hominins, we employ here the more commonly used nomenclature.


    View Abstract