JS is an employee at D.M.D. Family and Cosmetic Dentistry. This does not alter our adherence to PLOS ONE policies of sharing data and materials.
Conventional wisdom holds that a decline in oral health accompanies the transition from hunting and gathering to agriculture, given increased consumption of carbohydrates. This widely touted example of the mismatch between our biology and modern lifestyle has been intuited largely from the bioarchaeological record of the Neolithic Revolution in the New World. Recent studies of other populations have, however, challenged the universality of this assertion. Here, we present the first comprehensive study of oral health among a living population in transition from the bush to village life, the Hadza hunter-gatherers of Tanzania, to test the hypothesis that the shift from foraging to farming, or agricultural intensification, inevitably leads to increased periodontal disease, caries, and orthodontic disorders. Our results showed that women living in villages consuming a mostly agricultural diet exhibited more caries and periodontal disease than those living in the bush consuming a mostly wild-food diet. Furthermore, men living in the bush consuming mostly a wild-food diet had more than those living in the village consuming a mostly agricultural diet. These findings are explained by the high incidence of maize consumption in village settings, along with previously recognized variation in rate of caries between men and women. The unexpected discovery of high caries incidences for men in the bush is likely explained by heavy reliance on honey, and perhaps differential access to tobacco and marijuana. These data support the notions that mechanisms of cariogenesis are multifactorial and that the relationships between oral health and the shift from a predominantly wild-food diet to one dominated by cultigens are nuanced.
The transition from hunting and gathering to agriculture is routinely associated with declines in oral health, given increased consumption of carbohydrates and growth of bacterial colonies in dental plaque linked with the development of dental caries (e.g.
While diet may be the “main driver” of dental decay [
Dental diseases have been particularly important to anthropologists in studies of bioarchaeology and paleopathology, largely because teeth are resistant to degradation compared to soft tissues [
That said, the effect of agriculture on caries incidence varied across the globe [
Indeed, some hunter-gatherers had rather high caries rates related to carbohydrate-rich, and especially sugar-rich, diets. Mesolithic peoples from Sicily and Portugal, for example consumed cariogenic foods such as honey, dates, and figs [
To date, however, few studies have documented oral health among living small-scale populations [
The Hadza reside in a 4000km2 area around the shores of Lake Eyasi in Northern Tanzania, East Africa. The total population size, approximately 1000 individuals, has shown no major disruption during the past 100 years. Only around 150 individuals, however, currently practice a predominantly hunting and gathering way of life, meaning that the bulk of their diet is derived from wild plant foods and game animals [
Hadza living in the bush consume a diverse diet composed of a wide variety of plant foods, a vast array of bird species, small- to large-size game, and the larvae and honey of both stingless and stinging bees [
Only a handful of quantitative studies of diet composition are available, yet they have remained consistent over the past 20 years in the estimated contribution of particular food types to the Hadza bush diet [
Over the past two decades, the introduction of ethnotourism has greatly increased the number of Hadza who live in or near villages, where individuals make a living by taking tourists on short daily hunting and gathering treks and performing Hadza songs and dances [
Human Research Subjects approval was obtained from both the University of Nevada, Las Vegas Institutional Review Board (IRB) in the office of Research Integrity and the University of Arkansas Office of Research Compliance. Informed consent was obtained orally from all participants, as the Hadza are a non-literate population. Participant consent was recorded and witnessed affidavits were obtained. The IRB offices at both universities and the necessary Tanzanian government agencies approved the consent procedure. All data were collected with the permission of the Tanzanian Commission for Science and Technology (COSTECH) and the Tanzanian National Institute for Medical Research (NIMR).
A total of 75 adult individuals (37 women; 38 men), age 18 years or more, were included in this study. These included people residing in bush camps located on Gideru Ridge (
For each individual, name, sex, and age were recorded (only sex and age are reported in
Based on interview responses, all participants were binned into one of the following three residential groups, which we used as a proxy for diet composition: (1) bush (2) intermediate, and (3) village. The category of “bush” is a proxy for a diet low in agricultural products; the category “intermediate” is a proxy for a diet that is variable in agricultural products (with an intermediate proportion of cultigens over a long timeframe); and the category “village” is a proxy for a diet that is high in agricultural products.
While an exhaustive list of all camps each individual ever resided in would be impossible to confirm, a series of general questions did allow us to characterize the overall adult pattern of residence and diet. Our questions were similar to those asked by Frank Marlowe [
What is the name of the camp where you were born?
Is that camp located in the bush or near a village?
When you were a child, would you say that you lived most of your childhood in the bush, in the village, or going back and forth between the two?
Would you say that during your childhood you at mostly bush foods, village foods, or both?
Would you say that for most of your adult life you have lived in the bush, in the village, or going back and forth between the two?
Would you say that for most of your adult life you have you eaten mostly bush foods, village foods, or both?
Do you go to the village often?
Each participant was binned into their respective groups (i.e. “bush”, “intermediate”, or “village”) in the field by ANC and MP based on the responses to the interview questions. All oral health measures were scored by JS later in a blind study (see below), without knowledge of group affiliation for any participant.
Oral health data were collected from individuals participating in the study (n = 75) using the procedure outlined below. While the total sample size is 75 individuals, some individuals were missing teeth in some positions, precluding them from participating in all orthodontic measurements. When the sample size is less than 75, it is noted in the texts. The participant first cleaned teeth for approximately two minutes using a pre-pasted toothbrush, and then rinsed with water as necessary. Fissures in teeth were further cleaned with a dental sickle scaler by PSU prior to study as necessary to remove tartar to allow visualization and laser fluorescence analysis of the crown.
Initial oral health data were collected by PSU, following training in a clinical setting with licensed clinical dentist, JS at his practice in New York. Dental photographs (occlusal and buccal views of all teeth), periodontal pocket depths, and enamel mineral density values were recorded for each individual. Presence of periodontitis, dental crowding (including malocclusion and third molar occlusion), number of identifiable caries, and gross wear were all assessed from photographs following return from the field by JS.
A standard suite of oral photographs was taken with a 16 Mp Olympus digital camera modified for intraoral photography (Lester A. Dine, Inc). Cheek retractors and intra-oral mirrors were used following convention for clinical dental photography. A total of nine images were taken of each subject: anterior, right buccal and left buccal views of teeth in occlusion, and upper and lower, left and right cheek teeth and anterior teeth in occlusal view. This allowed assessment of premolar and molar occlusal surface caries.
A calibrated standard Michigan O Probe with Williams markings periodontal probe was placed in the recesses around the C1 and M1 of each individual. Clinical experience suggests that recesses around these teeth are excellent proxies for overall periodontal health. Periodontitis score was calculated as the summed depth in millimeters of the six recesses (mesiobuccal, buccal, distobuccal, mesiolingual, lingual, and distolingual) around each tooth.
Laser fluorescence of crown enamel was measured using a battery operated DIAGNOdent Classic Laser probe (KaVo). This instrument is designed to detect incipient caries. Occlusal surfaces of all cheek teeth (R+LP3-M3, R+LP3-M3, corresponding to numbers 1–5, 12–21, and 28–32 in the universal numbering system) were scanned, focusing on fissures, and points with fluorescence values exceeding 30 (the threshold value for demineralization reported by the manufacturer) were recorded and mapped on a dental chart for each individual.
The combination of dental photographs, periodontal pocket measures, and laser fluorescence allowed us to characterize and compare multiple aspects of oral health. For each participant, the following measures were scored and recorded.
Orthodontic condition was documented using several criteria. These included:
Anterior dental crowding (Crowd Measure). This was measured on dental photographs as the ratio of the summed mesiodistal lengths of the occlusal surfaces of all lower anterior teeth (R+LI1-C1, corresponding to numbers 6–11 and 22–27 in the universal numbering system) divided by the straight-line distance between the distal edges of RC1 and LC1 crown tips. When teeth were present, crowding was also scored using a 0–3 scale (Crowd Code): 0) no crowding; 1) minimal crowding; 2) moderate crowding, and 3) severe crowding.
Angle’s Classification of malocclusion. This was assessed from photographs using Mosby’s Medical Dictionary [
Third molar occlusion. This was scored from photographs as: 1) no M3s erupted; 2) one or more M3 erupted, but not in occlusion; and 3) all M3s in occlusion.
Gross Wear. This was scored for the LM1 and LM2 of each individual using the following coding scheme: 0) unworn; 1) facet development with slight dentin exposure; and 2) essentially flat surface with considerable enamel loss and exposed dentin. Values were summed for the two teeth. This was used in lieu of more traditional techniques [
Caries were scored using two methods: (1) visual identification in the field by PSU and ANC, and diagnostic confirmation using photographs in the laboratory by JS; and (2) using the DIAGNOdent Classic dental laser in the field by PSU with corroboration by JS in the lab using photographs.
Visual inspection entailed documenting each affected occlusal surface on the molars and premolar teeth, yielding a decayed, missing, and filled surfaces (DMFS) index [
The DIAGNOdent values recorded in the field and their locations on the dental charts were compared by JS against the digital photographs in the lab. This allowed us to check for ‘false positives’ that might have occurred due to staining, tartar, or dentin exposure. Visible caries were confirmed in all cases with high DIAGNOdent values. When both values were recorded, only one was counted. In all cases where a carie was seen in the field, it was obvious also on the photographs; there were no false positives. The aggregate total number of caries for each individual was calculated (CariesAll) by combining the DIAGNOdent data with the visual inspection data divided by the total number of cheek teeth in each mouth.
In addition to aggregate periodontal pocket depth, periodontal condition was scored according to a 0–4 for PerioPast and PerioPresent: 0) no evidence of periodontitis; 1) evidence for mild periodontitis; 2) evidence of moderate periodontitis; 3) evidence of severe periodontitis. PerioPast was visible evidence of past periodontal disease, whereas PerioPresent was visible evidence of periodontal disease at the time of measurement.
Tests for differences in proportion between groups were conducted using the Pearson’s chi squared test or Fisher’s exact, given small sample sizes. T-tests, ANOVAs and correlation coefficients (or the nonparametric equivalents) were used to identify mean differences between groups and correlations between variables. Logistic, linear and multinomial regression were used to test for main and interaction effects.
When the categorical dependent variables had greater than two categories (Gross Wear, PerioPast, and PerioPresent) a point-biserial (PB) correlation [
Responses to the interviews allowed us to bin each participant into one of the following three groups: (1) bush (2) intermediate, and (3) village. The bush group (n = 25; 12 women, 13 men) consisted of people residing in bush camps located on the Gideru Ridge, (
A total of 65 individuals had enough teeth to assess anterior dental crowding.
No significant differences were found across ages (Pearson’s PB p = 0.420), between the sexes (Fisher’s Exact p = .098), or between groups (Fisher’s Exact p = 0.767). The interaction effect for the regression model of sex x location for Crowd Measure was also not significant (Chi-square = 5.563; p = 0.936).
A total of 74 individuals were assessed by Angle’s Classification of malocclusion. Across all groups, the overwhelming majority (89%) of individuals were classified with neutrocclusion (Class 1). See
A total of 74 individuals had the presence and/or eruption of the third molar scored. Across all groups, the majority (61%) had a score of 3 (all M3s in and in occlusion), 32% had a score of 2 (one or more M3 erupted, but not in occlusion), and only 7% had a score of 1 (no M3s). See
A total of 73 individuals had enough teeth present to measure gross wear. Across all residential groups, 45% exhibited severe wear, wherein the molars measured (LM1 and LM2) had flat surfaces with exposed dentin, 48% exhibited moderate wear with some facet development, and 7% showed no evident gross wear (
A total of 23 individuals had missing or broken teeth, with the highest rates of missing/broken teeth in men and those living in the bush with a low agricultural diet. A total of 75 individuals had enough teeth present to measure rate of caries. Across all groups, 36% (24% of women and 49% of men) had at least one visible carious lesion (Caries Vis) (see
The interaction effect model for sex x location for Caries All was significant (t = 3.693, p <0.001) (β = 0.48). Males living in the bush consuming a diet with few cultigens show a significantly higher caries incidence than those living in the village consuming a high agricultural diet (t = -3.444, p = 0.001) (β = -0.26). Splitting time between the bush and the village was not a significant predictor for increased or decreased caries for men (t = -1.820, p = 0.73) (β = -0.17). Women living in the bush and consuming a low agricultural diet were at significantly lower risk of caries than those living in the village (t = -3.693, p = 0.042) (β = 0.21). Splitting time between the bush and village (intermediate category) was also not associated with increased or decreased caries prevalence for women (t = -1.705, p = 0.93) (β = -0.23). See
A total of 75 individuals had enough teeth present to assess periodontitis in the past and at the time of measurement. Across all groups, 66% of individuals showed evidence of periodontitis in the past (i.e. mild, moderate, or severe) and 80% showed some degree of periodontitis at the time of measurement. A significant correlation was found between increase in age and increase in degree of periodontitis for PerioPast (Pearson’s PB p<0.001; Kendall’s Tau p<0.001) and PerioPresent (Pearson PB p<0.002; Kendall’s Tau p < 0.001). Across the entire sample (bush, intermediate, and village camps), women showed significantly lower rates of PerioPast (Chi-square = 13.199; p = 0.001) and PerioPresent (Chi-square = 12.959; p = 0.002) when compared to men. Residential setting alone, however (bush, intermediate, village), was not a significant predictor of PerioPast (Chi-square = 7.077; p = 0.132) or PerioPresent (Fisher’s Exact p = 0.2558).
Despite differing diets, Hadza foragers living in the bush and the village both show evidence of caries, periodontal disease, and gross wear. Not surprisingly, a significant increase in age mapped onto all of these orthodontic measures, which supports the findings of previous studies on oral health among small-scale societies [
We did not find any notable differences in gross wear patterns between the sexes, though different teeth might have yielded a different pattern [
Strong sex differences in caries prevalence and periodontitis emerged, which mapped onto residence in different ways. When compared to men, Hadza women across all residential settings exhibited the lowest rates of periodontal disease, both in the past and at the time of measurement. Women residing in the bush eating a diet of mostly wild foods exhibited the best oral health overall, exhibiting the lowest rate of caries. Women living in the village consuming a maize dominated diet, however, showed a higher caries prevalence than their female counterparts in the bush or their male counterparts in the village. Hadza men residing in the bush exhibited the worst oral health overall; they exhibited significantly higher rates of broken/missing teeth, higher prevalence of caries, and were significantly more likely to have periodontal disease at the time of study and evidence of it in the past. Interestingly, an intermediate diet composed of both wild foods and domesticated products did not significantly affect caries prevalence for either men or women. This suggests that a diet composed of
Numerous studies demonstrate that women have significantly higher prevalence rates of caries compared to men across subsistence regimes, both in recent populations [
The most surprising and initially unexpected finding presented here is that Hadza men residing in the bush exhibit the worst oral health. Unfortunately, there are few studies in which to compare our results. The findings that Hadza men in the bush have greater rates of caries support conclusions drawn from earlier work among indigenous Guaraní Indians of Brazil [
Hadza men and women residing in the bush exhibit strong sex differences, not only in foraging behavior [
It is possible that smoking may also contribute to the high caries rate. Estimates suggest that men residing both in the bush and the village have been smoking marijuana for at least two centuries [
The low rates of periodontal disease and caries in women residing in the bush may also be associated with dietary differences. Hadza women consume significantly more tubers than their male counterparts [
Another possible contributing factor to the low incidence of caries among women consuming a low agricultural diet may be sex differences in oral microbiome, which have been linked to diet. It has been proposed that Hadza women house bacterial species in their gut that allow them to more efficiently digest tubers [
It should be noted that some Hadza, residing in both the bush and the village, anecdotally report using small branches from the
Limitations of this study include small sample size and the inability to quantitatively document diet composition throughout the life course. As periodontitis and caries prevalence are associated with lifetime diet [
In conclusion, this study, the first to report oral health and prevalence of caries and periodontal disease among the Hadza, supports findings in the burgeoning literature that suggest that foragers do not always have superior oral health when compared to agriculturalists. These data reinforce the notion that the mechanisms of cariogenesis are dependent on a suite of factors that may include diet, behavior, and the oral microbiome. Future work among the Hadza will aim to tease apart such connections.
(DOC)
(DOC)
(CSV)
We would like to thank all of our Hadza participants and the Tanzanian Commission for Science and Technology (COSTECH) and the Tanzanian National Institute for Medical Research (NIMR) for research approval. We would also like to acknowledge the Center for Academic Research and Training in Anthropogeny (CARTA) at the University of California, San Diego for hosting the symposium on the “Evolution of Human Nutrition” that was the inspiration for the current project. Finally, we would like to thank the two anonymous reviewers and the editor for helpful comments on earlier versions of this paper.