Introduction

Both present-day African Pygmy hunter-gatherers (PHGs), characterized by a reduced adult stature (<160 cm) [1–3], and their non-Pygmy Bantu-speaking neighbors live in the tropical rainforests throughout the Congo Basin in a complex network of economic and social contacts with different subsistence strategies [2,4,5]. In contrast to PHGs, whose lifestyle and culture are forest based, non-Pygmy Bantu speakers have an agro-pastoralist, sedentary lifestyle [1,5–7]. Genetic analyses suggest that gene flow between the two populations is limited and asymmetrical due to sociocultural and demographic constraints [8–11]. Linguistic differences have also been demonstrated [2,12], and phenotypic peculiarities in PHGs are not limited to adult height [13]. Central African rainforest environments show great plant and animal biodiversity [14], including aboveground edible plants and starch-rich plant underground storage organs (USOs), as well as many accessible prey animals. Ethnographic evidence has shown that foraging activities, mainly providing wild yam tubers (Dioscorea spp.), supply the bulk of the diet among PHGs, who rarely spend time cultivating plant foods [15–19]. Agricultural resources are obtained mainly by exchange for forest products (meat and honey for iron tools and starchy foods) with Bantu-speaking farmers (BSFs) [20–24]. Reports on dental health [25] and investigations with stable isotopes [26] corroborate ethnographic data that indicate significant dietary differences in animal protein and starch-grain consumption between PHGs and agro-pastoralist Bantu speakers.

Buccal dental microwear, the pattern of microscopic use-wear on nonworking enamel surfaces of premolar and molar teeth, has been shown to reflect the physical properties of chewed foodstuffs and long-term trends in dietary preferences [27]. During food chewing, scratches of different length and orientations are formed across buccal enamel surfaces by the indentation effect of micrometer-scale (10⁻⁶ m) particles that are harder than enamel, such as plant phytoliths, grit, or quartz dust [28–31]. Thus, the type and amount of chewed abrasives are critical in the formation of tooth wear and microscopic scratch patterns [27,28,31–33]. Distinct buccal microwear patterns have been shown to distinguish nonhuman primates and fossil hominins [34,35], as well as foraging and horticultural populations with distinct dietary habits and food processing methods [36–39]. However, reliable buccal microwear patterns of modern human populations with known diets are still lacking, so the effect of preferred diets and the physical properties of ingested foodstuffs on dental microwear patterns need to be better understood [25,40]. Here we present analyses of buccal microwear patterns of both PHGs and non-Pygmy Bantu-speaking populations from Central Africa with distinct and well-characterized diets, including BSFs, inhabiting the same forest environment as PHGs, and Bantu-speaking pastoralists (BSPs) Maasai populations from savanna habitats [7,14]. We first assessed if the populations’ distinct reported diets (foraging, farmer, and pastoralist subsistence economies) are reflected in buccal microwear patterns. We then analyzed if the differential access to cultivated products among the PHGs was apparent in their buccal microwear patterns, which might provide new insights on the abrasive effect of the consumption of USOs and silica-based aboveground plant foods [34,35,41,42] for the characterization of hominin dietary preferences and adaptations.

Materials and Methods

Results

Overall differences in total scratch densities and average length (in micrometers) were observed between PHGs and Bantu speakers (Figure 1 and Table 1). The PHGs showed the smallest scratch density (NT = 64.47 ± 27.31 –mean ± SD-, n = 51), followed by the BSPs (NT = 77.00 ± 9.12, n = 12), with the BSFs having the highest scratch density value (NT = 151.25 ± 30.62, n = 80). Moreover, the PHGs showed the largest scratch length (XT = 146.37 ± 28.06 μm) (Figure 2A). Within the PHG groups, the Babongo showed the highest density values (NT = 121.83 ± 31.44) and the Mbuti had the lowest XT value (97.37 ± 17.80 μm), close to that of the BSFs (87.11 ± 18.50 μm) (Figure 2B). In an analysis of scratch density and length variables by orientation (see Methods and Table 1), we found that BSFs had higher densities of shorter scratches than the PHGs. Otherwise, the BSPs presented the lowest between-group length values. Among PHGs, the Babongo had the largest scratch densities by orientation, whereas the Mbuti showed the shortest. Significant differences in overall scratch density (NT) and length (XT) were observed among the three main socioeconomic groups: PHGs, BSFs, and BSPs (F = 23.44, p < 0.001; Wilk’s λ = 0.129, partial ε² = 0.641, MANOVA; Table 2). The univariate ANOVAs showed that among-group differences were significant for all 10 microwear variables (p < 0.001). Pairwise comparisons (Tukey’s HSD post hoc test) showed (Table 3) that PHGs had significantly (p < 0.05) lower scratch densities and larger lengths than the BSFs for all orientations, while the PHGs and BSPs significantly differed for NV and all the length variables, while the BSFs and BSPs differed for all the density variables, except NV, and for XMD and XH (Table 2). The forager group (PHGs), including the Aka, Baka, Babongo, and Mbuti, was characterized by low scratch densities and high scratch lengths, a microwear pattern that was clearly opposite that of the farmer group (BSFs). Although the PHGs showed a more homogenous microwear pattern compared with the Bantu speakers, significant differences were observed among populations when the Baka, Aka, Babongo, and Mbuti were considered separately (Table S2 in File S1). All microwear variables showed significant among-group differences (p < 0.001, ANOVA). However, in a comparison of the Aka and Baka (Tukey’s HSD; p < 0.05), no significant differences in any microwear variables were found. Instead, the Baka statistically differed from the Mbuti for XT (p < 0.001) and XV (p = 0.003), and from the Babongo for NT (p < 0.001), NMD (p = 0.043), and NH (p < 0.001), whereas significant differences between the Aka and Babongo were found only for NT (p = 0.018). Finally, the Babongo differed from the Mbuti only for NT (p = 0.015). The Babongo clearly showed the most distinct microwear pattern among PHGs, resembling the BSFs with regard to high scratch densities. The other PHG groups showed similar scratch densities, but the Mbuti had shorter average scratch lengths, which resulted in a microwear pattern resembling that of the BSPs (Figure 2B)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Buccal enamel surfaces showing different microwear patterns related to abrasive properties of chewed foodstuffs.

Baka Pygmy hunter-gatherer (left) and Pahouin (#9715, Musée de l'Homme, Paris) agriculturalist from Gabon (right). Both individuals are adult females. Note highly abraded enamel surface in the Bantu-speaking farmer. Each micrograph represents a buccal enamel area of 0.56 mm² on mandibular first molars at 100×. Scale bar: 100 μm.

https://doi.org/10.1371/journal.pone.0084804.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Buccal microwear pattern variability among the analyzed populations.

(A) Box-plots showing scratch density (NT) and their average length (XT, in micrometers) among dietary groups. Boxes enclose 25%–75% percentile values, the mean and median are indicated with a circle and horizontal bar, respectively, and whiskers denote minimum–maximum values. (B) NT and XT mean values plotted within PHGs (Aka, Baka, Babongo, and Mbuti) and Bantu farmer (BSFs) and pastoralist (BSPs) populations. Error bars denote ±2 standard error (SEM). The number of individuals (n) in each group is indicated.

https://doi.org/10.1371/journal.pone.0084804.g002

The PCA of the eight independent microwear variables (excluding NT and XT, see Methods) yielded two PCs with Eigenvalues larger than 1 that explained 63.19% of the total variance (Figure 3 and Table S3 in File S1). PC1 explained 44.37% of the total variance and was positively correlated with the length variables (Pearson r ranging from 0.72 to 0.83) and negatively correlated with the density ones (r values ranging from −0.42 to −0.60). PC2 explained 18.82% of the total variance and was mainly correlated with NH (r = 0.69), NMD (r = 0.53), and NV (r = −0.52). Univariate ANOVAs for PC1 and PC2 by dietary groups (PHGs, BSFs, and BSPs) showed significant differences in both cases (p < 0.0001), which indicates that PC1 distinguished PHGs for having longer and less abundant scratches than BSFs and BSPs, and PC2 distinguished BSPs for having a lower scratch density than BSFs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Dispersion Principal Components (PC) 1 versus 2, derived from buccal microwear variables, by subsistence economy.

The proportion of the variance explained by the PCs is indicated in parentheses. The ellipses include 95% confidence regions of the dietary groups considered. Pygmy hunter-gatherers (PHGs) have low density of longer scratches, whereas variation in Bantu-speaking farmers (BSFs) and pastoralists (BSPs) are reflected in high (BSFs) and low (BSPs) density pattern of shorter scratches. All individuals analyzed are plotted. See Materials and Methods sections for the details of methodology and sample compositions.

https://doi.org/10.1371/journal.pone.0084804.g003

Discussion

Buccal microwear patterns reflect abrasive properties of ingested foodstuffs [36–38], even in present-day living people [33,39]. Meat is not hard enough to damage enamel surfaces during chewing [59], whereas both plant phytoliths and gritty-quartz particles have been shown to cause enamel etching [28,29,31]. However, the effect of specific abrasive agents on enamel surfaces is difficult to characterize because differences in the load-bearing capacity of teeth might be a function of morphology (tooth size and geometry or enamel thickness) rather than underlying mechanical properties [30,32,59,60]. Nonetheless, dental microwear formation seems to be strongly influenced by the amount of abrasive particles ingested with food and their intrinsic hardness compared to enamel [29,31]. Plant silica phytoliths and grit particles have been shown to be involved in the formation and remodeling of microwear patterns [27,29], with phytoliths being responsible for long-term processes, mainly forming small microwear features, and exogenous quartz particles (~2.5-fold harder than enamel) causing significant biting stresses, even with modest loadings, resulting in microwear features larger in size [27,31,32].

The variability of the buccal microwear pattern observed between the BSF and PHG groups was clearly linked to the inclusion of agricultural products in their diets, which likely had a significantly higher content of abrasive particles than foraged fruits and USOs. These microwear patterns are representative of two distinct dietary habits, one mainly vegetarian and including highly abrasive particles and the other having a higher content of hunted meat and foraged plant foods. Otherwise, the two Bantu-speaking groups (BSFs and BSPs) resembled each other in having short scratches; however, the pastoralist Maasai showed a distinct pattern, with both a low density of scratches and short average lengths, likely due to their soft and less abrasive diet [7]. Both PHG and Maasai populations have limited access to cultivated goods (a significant source of phytoliths and dust grit) compared with BSFs [2,4,7,19–21]. Moreover, the PHGs overall showed lower scratch densities and longer average lengths than the BSFs. However, among PHGs the microwear pattern of the Babongo, the most sedentary of the forager populations analyzed, overlapping that of the BSFs, whose samples showed the highest scratch densities and shorter striation lengths. The buccal microwear variability of the Babongo (though represented by a rather small sample) was much smaller than that of the BSFs, showing similarities with the PHG microwear pattern that might suggest they did not fully abandon a forager economy [20].

As a whole, the feeding strategies of the PHGs clearly translate into a distinct low-density pattern of buccal microwear that depends on the amount of etching particles incorporated in foodstuffs. Our microwear findings suggest that PHGs mainly consume foraged plant foods with reduced abrasiveness compared to harvested foods. Yam (including over 600 varieties, 95% of which are endemic to Africa), a monocot plant that differs from sweet potatoes (a dicot called “yams” in some parts of the world), is the most consumed starch-rich tuber, either gathered or cultivated, by African populations [41,42], including PHGs [18,19,24,40]. Yam tubers lack phytoliths, useful for taxonomic classification [30], but show the highest levels of fracture toughness (J m⁻²) among 33 analyzed root tubers [42]. If cooked or roasted, their mechanical resistance decreases, while the energetic gain increases, allowing for less forceful mastication and easier digestion [42,61]. If abrasive particles are not incorporated during processing prior to ingestion, yam consumption in rainforest environments does not cause enamel indentation [32,59]. In contrast, USOs from savannah or desert habitats, including a wider range of edible tubers, can be consumed raw [41] and frequently incorporate grit-soil particles that cause higher tooth-wear rates in savannah foragers, such as the Hadzabe in northern Tanzania, than those seen in farmer populations [62]. In PHG culture, animal resources are frequently smoked [4,49,54,63] and tubers, either wild or cultivated, are harvested with a digging stick [18,50] and soaked in water for a few days before being roasted or boiled for consumption [15,19,25,49]. Thereby, both meat and tubers may incorporate abrasive particles from ashes or dust that, despite usually being removed before ingestion, may contribute to the formation of the buccal microwear patterns observed in the PHGs. Otherwise, the highly abraded microwear patterns shown by the Bantu farmers, who consume gritty particles from stoneground foods [5,6], resemble those observed in horticulturalist populations [38,39] and differ from the groups that consume considerable quantities of meat [37]. The consistency of the association seen between food abrasiveness and buccal microwear pattern supports the hypothesis that PHG and BSP populations share an overall soft diet, with significant consumption of meat and other animal-derived products. A diet including some exogenous, coarse, and brittle particles, likely larger in diameter than those in Maasai pastoralist diets, would explain the distinct microwear pattern of the foragers compared with the pastoralist group.

The increasing reduction of natural areas and scarcity of big game have greatly altered Pygmy economy [5,12,20,21,48]. Manioc cultivation by Baka pygmies was first described at the beginning of the 20th century [64], but it is not yet clear if dietary preferences have changed between ancient and present-day pygmies [2,4,19]. No differences in buccal microwear patterns were observed between the Aka (collected at the beginning of the 20th century) and present-day Baka (collected from living individuals), probably due to similarities in their dietary regimens despite the temporal divergence [2], and few differences were observed between the Baka and the Mbuti (a 1950s collection). The buccal microwear pattern of the Mbuti, and especially that of the Babongo, slightly resembles that of the Bantu farmers, since they engage in food trade contacts [8,9,23,54], whereas the Aka and Baka have less access than the Babongo and Mbuti to cultivated products [2,24,54]. When dietary preferences shift to include greater amounts of abrasives, such as in agricultural goods, buccal enamel microwear patterns have been shown to significantly reflect the change [27,39]. The lack of information about the studied ancient individuals prior to their deaths does not permit seasonal buccal microwear inferences. Nonetheless, a relative importance of wild yam gathering and hunting activities (up to 90%) among the Aka and Baka has been reported in both dry and rainy seasons when no trade for Bantu resources occurs [2,16,19] and supports the homogenous microwear patterns found between these populations. By contrast, we have insufficient data to extend our conclusions to any other PHGs analyzed due to small sample sizes and the scarcity of similar seasonal observational surveys [15,53,54]. Nonetheless, in vivo experiments with humans have indicated that buccal-microwear might vary in populations with specific diets and ecological conditions but at the same time, long-term microwear patterns probably remain stable independent of dietary habits [27,33].

Based on overall observation, our results suggest that dietary habits of Pygmy populations have not significantly changed during the last century, as some ethnographic studies have already indicated [4,19,21]. The economic associations between Bantu speakers and Pygmy seem to be limited to specific time periods and vary according to the group, as shown between the Mbuti and Bantu-speaking groups in the Ituri forest [4,12,50,65], and do not significantly affect microwear patterns of most of the forager groups analyzed from Central Africa. Similar information from western Africa is lacking [5,50,51,53].

The implications of the buccal microwear variability of the Pygmy and Bantu populations for the characterization of hominin diets should not be dismissed. Meat and USOs may have played a significant role in dietary adaptations of early hominins [41,42] and morphological adaptations of dental traits in the earliest members of the genus Homo (2–1.5 Ma) have been related to the consumption of mechanically challenging, tough high-energy foodstuffs, likely associated with brain growth and cultural changes in stone-tool technology [66,67]. However, the USO-eating hypothesis [41,42] on tooth use and microwear has not yet been tested in modern foragers [41: 494]. Our buccal microwear analysis of Pygmy populations indicates that a diet mainly based on fire-processed meat and USOs, similar to that hypothesized for early Homo [41,67], should result in a low-density, not highly variable microwear pattern. However, occlusal microwear texture complexity analyses have shown highly variable microwear patterns for this taxon, interpreted as dietary habits characterized by the consumption of non-fracture-resistant foods and low biomechanical demands on chewing [66], consistent with cranio-dental evolutionary changes in early Homo [67]. Buccal microwear patterns of early Homo specimens, as well as of other modern forager populations, contribute to this microwear interpretation and to the characterization of the significance of staple foods in ancient human diets.

Supporting Information

Acknowledgments

We are grateful to all the Baka participants. We are also indebted to A. Froment and P. Mennecier (Musée de L'Homme, Paris), A. Sanchez-Mazas and G. Pérreard (University of Geneva), R. Orban and P. Semal (Institut Royal des Sciences Naturelles de Belgique, Bruxelles), and Ian Tattersall (American Museum of Natural History, New York) who provided access to collections under their care. Thanks to Jordi Galbany for helpful comments on earlier drafts of the manuscript and to Frank L'Engle Williams and the reviewers for valuable and detailed comments, which contributed to the improvement of this paper.