Chemical Approaches to Dietary Representation

Ted A Rathbun. Cambridge World History of Food. Editor: Kenneth F Kiple & Kriemhild Conee Ornelas. Volume 1. Cambridge, UK: Cambridge University Press, 2000.

Dietary reconstruction for past populations holds significant interest as it relates to biological and cultural adaptation, stability, and change. Although archaeological recovery of floral and faunal remains within a prehistoric or historical context provides some direct evidence of the presence (and sometimes quantity) of potential food resources, indirect evidence for the dietary significance of such foodstuffs frequently must be deduced from other bioarchaeological data.

The types of data with dietary significance range from recovered plant and animal remains through evidence of pathology associated with diet, growth disruption patterns, and coprolite contents. Other traditional approaches involving the people themselves—as represented by skeletal remains—include demographic (Buikstra and Mielke 1985) and metabolic (Gilbert 1985) stress patterns.

In addition to bioanthropological analyses, reconstruction of environmental factors and the availability and limits of food species and their distribution for a population with a particular size, technology, and subsistence base are typical components within an archaeological reconstruction. Although these physical aspects are significant, the distribution, or more likely the restriction, of particular foodstuffs from certain segments of the population (because of sex, age, status, food avoidance, or food taboos) may be important cultural system features. The seasonal availability of food and its procurement, preservation, and preparation may also have influenced group dietary patterns and nutritional status (Wing and Brown 1979).

Analysis of skeletal remains may also provide some direct evidence of diet. Type and adequacy of diet have long been of interest to physical anthropologists, especially osteologists and paleopathologists (Gilbert and Mielke 1985; Larsen 1987). More recently, direct chemical analysis of bones and teeth has been attempted in an effort to assess the body’s metabolism and storage of nutritive minerals and other elements. L. L. Klepinger (1984) has reviewed the potential application of this approach for nutritional assessment and summarized the early findings reported in the anthropological literature. (In addition, see Volume 14 of the Journal of Human Evolution [1985], which contains significant research surveys to that date.)

General Approaches and Assumptions

The pioneering anthropological work in bone chemical analysis and dietary reconstruction can be attributed to A. B. Brown (1973), who examined strontium concentrations in relation to meat and vegetation, and to R. I. Gilbert (1975), who explored the concentrations of five other elements in relation to prehistoric Native American samples in Illinois. The enthusiastic early promise of a relatively easy, straightforward approach to diet reconstruction from elemental concentrations in bone has more recently been tempered by recognition of the biodynamic complexity, methodological problems, and contextual changes occurring through diagenesis of buried bone. Nonetheless, a number of publications in article, dissertation, and book form have appeared in the anthropological literature from the early 1970s to the current time. Although the emphasis, samples, and time frame have varied considerably, the approaches generally share the assumptions that bone is the vital feature for mineral homeostasis and a reservoir for critical elements, and that variations in bone concentrations by individual and group reflect past intakes of dietary concentrations, which in turn reflect local environmental and cultural milieus. A. C. Aufderheide (1989) and M. K. Sandford (1992, 1993a) have provided excellent reviews of the basic premises, biogenic-diagenetic continuum, diversity of methods, and sampling and analytical protocols. Two recent, extremely important, edited volumes (Price 1989; Sandford 1993b) contain the most comprehensive bibliographies currently available and include syntheses of recent findings, remaining problems, and potential research trajectories and protocols.

Initial anthropological bone chemical research focused primarily on the inorganic mineral phase of bone—which typically makes up 75 to 80 percent of the dry weight—and was concerned with the dietary contrasts and trophic levels of early hominids, hunter-gatherers, and agriculturalists. Analysis of the stable isotopes in the organic collagen component (20 to 25 percent) is now frequently undertaken to investigate the relative importance of C3/C4 plants and reliance on maize in the Americas through a consideration of its carbon content (Bumstead 1984). Such analysis is also focused on the equally troublesome problem of the marine/terrestrial protein components of the diet by investigation of nitrogen isotopes. Other isotopes with possible relevance to aspects of dietary reconstruction include those of oxygen, sulfur, and strontium.W. F. Keegan (1989), M.A. Katzenberg (1992), and S. H. Ambrose (1993) provide excellent introductory reviews of the use of isotopes in the analysis of prehistoric diet. H. P. Schwarcz and M. J. Schoeninger (1991) provide a more esoteric review of the theory and technical details of isotopic analysis for reconstructing human nutritional ecology.

Samples, Instrumentation, and Variables

It is probably unrealistic to expect a finely detailed reconstruction of past diets from skeletal chemical data because of the nature of our evidence. A number of factors influence the survival and recovery of human remains. These include mortuary method, climatic conditions, soil chemistry, decomposition rates, and archaeological methods and goals. Although a single individual, may reflect important aspects of the biocultural past, including diet, we should remember that each individual, and the circumstances and context of the recovery of that individual, is unique.

Moreover, because the bone is analyzed not in life, but after death, bone turnover rates must be viewed in relation to age and health status at the time of that death. For aggregate data, especially for statistical comparisons, the representativeness, comparable size, and composition of the samples are also of concern. Chemical analysis should be done only after thorough and professional osteological analysis is conducted. Useful standardized guides in such analysis are the Paleopathology Association’s Skeletal Database Committee Recommendations (1991) and the collective recommendations for standard skeletal data collection (Buikstra and Ubelaker 1994).

Accurate demographic profiles are especially important for later considerations of age, sex, health/disease, and perhaps status categories within the population sample. Individual bone samples may be taken for later analysis, especially now that many invaluable skeletal collections are being reburied. Because as little as 1 to 2 grams of bone may be used for chemical analysis, depending upon the instrumentation and method, the removal and laboratory destruction of this amount of bone represents a loss of even less than that of a typical tooth.

Chemical concentrations within bone vary from one bone to another and even in different portions of an individual bone; understandably, earlier comparative studies were difficult before this fact was recognized. Current recommendations are to use cortical bone, preferably from femur midshafts. The particular technique chosen for quantitative elemental analysis will depend upon the number of elements to be analyzed, the cost, and the degree of precision required. Aufderheide (1989) and Sandford (1992) review the theoretical foundations and relative advantage of a number of options. Current laboratory analytical techniques include electroanalysis, light spectrometry, scanning electron microscopy, neutron activation analysis, mass spectrometry, and the widely used inductively coupled plasma (ICP) optical spectrometry for multiple element analysis. Results are typically reported in parts per million of bone or bone ash, the latter preferred (Price et al. 1989).

Although elemental analysis may be conducted directly on bone or in solution, isotopic analysis first requires decalcification, extraction of collagen from bone—approximately 5 milligrams (mg) of collagen is needed—and then analysis through mass spectrometry. Katzenberg (1992) provides a capsular review of the process and cites B. S. Chisholm (1989) and Schoeninger and colleagues (1989) as current basic references. Stringent laboratory conditions and lengthy preparation techniques, frequently with elevated costs, are necessary for isotopic analysis (Ambrose 1990).

Diagenesis appears to be less of a problem with isotopic analysis; however, it must still be considered. Much recent research in elemental analysis has attempted to document and cope with the problems of chemical and physical changes that may occur in buried bone through leaching, contextual contamination, and chemical reactions of bone outside the living body. In addition to the bone, samples of the matrix of soil must be collected and analyzed so that potential contamination may be identified.

Of course, postmortem influences must be detected, but physiological processes may also influence the incorporation and utilization of elements present in a particular diet. Absorption of particular elements ingested may be enhanced or reduced through chemical processes or physiological regulation by competing substances in the diet. Phytates found in some cereals, for example, may bind zinc and iron and reduce the absorption of these elements.

In addition, not all elements are distributed equally through body tissues, and some, such as lead or strontium, may be deposited differentially into bone. Metabolism differences for particular elements may also compound the interpretation. Retention of ingested elements is variable as well in certain tissues as is the rate of bone turnover at different ages and under variable health conditions.

Finally, excretion rates for particular elements depend upon physiological processes and the nature of the element itself. Although there are a great number of variables in the incorporation, retention, and analysis of any particular chemical element in bone, a cautious application of trace element studies with appropriate samples, methods, and situations or research goals continues to hold promise for some aspects of dietary reconstruction. Indeed, despite severe critical evaluation of earlier trace element studies (Radosevich 1993), improved and refined multidisciplinary research and laboratory protocols should prevent the necessity of throwing the baby out with the bathwater.

Anthropological Dietary Chemical Reconstructions

The following sampling of past contributions within anthropological bone chemical research reflects three major thrusts related to diet: trophic level, temporal change in subsistence, and distinctive chemical elements. The general trophic level of past human diets was first investigated by strontium and strontium/calcium ratios. The basic premise was that the relative reliance on meat derived from animals higher on the food chain would be reflected by a lower concentration of strontium in the human bone because of its differential presence and absorption (relative to calcium) along the food chain. For paleoanthropologists concerned with the physical and cultural development of humans and the hunting complex (especially Australopithecus, Homo habilis, and Homo erectus), it first appeared that the answers could be derived in a relatively straightforward manner (Sillen and Kavanagh 1982). However, the fossilization process itself appears to alter the initial concentration, and other diagenic processes may confound the interpretation (Sillen, Sealy, and van der Merwe 1989).

In the case of more recent human groups, however, an analysis of strontium bone content has proven more fruitful. The anthropological significance of the strontium content of human bone was initially investigated by Brown (1973), and subsequent studies of strontium content suggest that dietary differences may reflect social stratification. Schoeninger (1979), for example, determined that at a prehistoric Mexican site, higher-ranking individuals—as indicated by interment with more grave goods—had lower levels of strontium and, hence, presumably a greater access to animal protein. Other studies, such as that by A.A. Geidel (1982), appear to confirm the value of strontium analysis in this respect, even though diagenetic change must be evaluated.

Temporal dietary changes and the relative amounts of meat and plants in the diet (perhaps related to population size as well as technological complexity) have been documented in a number of regions. Gilbert (1975, 1977), for Late Woodland Mississippian groups in the Midwest, and T. D. Price and M. Kavanagh (1982), for the same area, document an increasing strontium concentration among groups as they experienced an increasing reliance on cereals and a concomitant decrease in meat availability. Katzenberg (1984) determined similar temporal changes among Canadian groups, as did Schoeninger (1981) for areas of the Middle East.

It should be noted, however, that bone strontium concentrations are strongly influenced by ingestion of marine foods—such as shellfish—and some nuts. J. H. Burton and Price (1990) suggest that low barium/ strontium ratios distinguish consumption of marine resources. It should also be noted that soil and water concentrations of strontium and, hence, plant absorption of it, also vary geographically. A final caveat is the documentation of the influences of physiological processes such as weaning (Sillen and Smith 1984) and pregnancy and lactation (Blakely 1989), which elevate bone strontium and depress maternal bone calcium concentrations.

A number of other elements found in food and water (Ca, Na, Sr, Cu, Fe, Mn, Mg, Zn, Al, Fe, Ba) have the potential for assisting in dietary reconstruction. These elements have been analyzed in skeletal samples with varying degrees of success in delineating food categories, temporal changes, and subsample variations related to age, gender, or class. Like strontium, these elements are subject to many of the same modifications and processes from ingestion to deposition into bone, and frequently to the same diagenic processes after death, so the same caveats apply to their analysis and interpretation.

In addition, when these various elements are incorporated together in diets (and later deposited in bone), they may be antagonistic to each other, or enhanced when ingested together or as part of the same diet. In anthropological analysis, although the major elements such as calcium or phosphorous may be significant, the majority of research has been concerned with trace elements in either their total concentration for dietary categories, or as deficiencies related to particular diseases, or at toxic levels, such as lead poisoning.

Although there is a relatively abundant literature on individual trace elements and their role in human metabolism and nutrition in the medical and nutrition literature (Underwood 1977; Prasad 1978; Rennert and Chan 1984), these studies tend to focus on Western diets and modern food standards and samples. The major emphasis within anthropological elemental studies has been with meat and vegetable dietary questions and temporal change, especially in the prehistoric American Southeast and Middle West. Research in other world areas has included Europe (Grupe and Herrmann 1988), Southwest Asia (Sillen and Smith 1984), Sicily (Klepinger, Kuhn, and Williams 1986), Tunisia (Sandford, Repke, and Earle 1988), Australia (Kyle 1986), and Peru (Edward and Benfer 1993).

The theoretical premise behind such investigations is based on the different concentration levels of particular elements in dietary resources that then should vary in the human skeletal concentrations. Meat, for example, is typically associated with increased concentrations of iron, zinc, copper, molybdenum, and selenium. Plants, however, generally have greater amounts of strontium, magnesium, manganese, cobalt, and nickel. Unfortunately, a single plant or animal species rarely possesses a unique chemical signature. Besides the problem of mixed dietary resources, many of the prevailing trace elements overlap (Gilbert 1985), and nuts present special problems (Buikstra et al. 1989). Synthetic critical reviews of relevant literature have been provided by Price (1989), Sandford (1992, 1993a, 1993b), Aufderheide (1989), and J. E. Buikstra and colleagues (1989).

The emerging consensus is that elemental and isotopic studies may indeed be significant in circumstantial dietary reconstructions of past populations. But additional research is necessary to cope with the numerous problems and issues connected with such studies.Among these are diagenesis, laboratory analysis and sample preparation, expansion to skeletal samples of more recent origin, wider geographical representations and inclusions, feeding experiments, and more sophisticated statistical and interpretative techniques.

A number of studies have attempted to deal with the question of diagenesis and the need for adjustments before statistical analysis (Lambert, Szpunar, and Buikstra 1979; Katzenberg 1984; Price 1989; Edward and Benfer 1993; Radosevich 1993). Multiple bone analyses, comparisons with nonhuman animals (herbivores, carnivores, and mixed feeders), more multielement surveys, and careful laboratory evaluation are recommended.

Expansion of multielement or single element studies into more recent historical periods should have the advantage of combining available historic information concerning diet and food habits with the chemical analysis of skeletal samples for a more comprehensive understanding. For example, Aufderheide and colleagues (1981, 1985, 1988) have delineated socioeconomic differences, occupational categories, and probably food storage patterns from the analysis of skeletal lead in the United States colonial period. H. A. Waldron (1981, 1983) and T. Waldron (1982, 1987) have addressed similar problems in the United Kingdom. In like fashion, J. S. Handler,Aufderheide, and R. S. Corruccini (1986) combined nineteenth-century descriptions of “dry bellyache” among Barbados slaves with an analysis of slave remains to demonstrate that “dry bellyache” was actually lead poisoning, the result of contaminated rum from stills with lead fittings. T. A. Rathbun and J. D. Scurry (1991) found regional variation in lead burdens in skeletal samples of whites and blacks from the seventeenth- and eighteenth-century eastern United States. Such variations seem to reflect differences in socioeconomic class, food preparation, and drinking patterns. Whites, who made far greater use of drinking and eating utensils, carried considerably higher lead burdens than blacks, with those of the Middle Atlantic states having slightly higher levels than those of other Southeast samples. Among blacks, females had the highest levels, indicating that they also doubtless had greater access to the whites’ lead-contaminated food and drink.

Utilizing techniques of chemical analysis, W. D. Wood, K. R. Burns, and S. R. Lee (1985) and Rathbun (1987) were able to document regional and perhaps cultural differences among rural blacks, plantation slaves, and white elites in the nineteenth-century southeastern United States. Among the findings were that whites apparently had more access to meat than did either enslaved or freed African-Americans.

Similarly, Rathbun (1987) and T. A. J. Crist (1991) found dietary variation by gender, age, and perhaps stress level among a nineteenth-century South Carolina plantation slave sample. Males seem to have relied more heavily on meats, grains, and nuts than females, whose diets consisted more of leafy and leguminous vegetables. The remains of older adults reflected diets of grains, vegetables, and seafood, whereas those of younger adults revealed the consumption of more meats and, perhaps, nuts. Analysis of historical documents concerning food allocations on the plantation suggests that much of this differential was because slaves supplemented plantation rations with food items they collected and cooked themselves. Clearly, in many instances, a combining of historical, anthropological, and chemical information has the potential for providing a richer determination of past dietary contents and the consequences of various dietary regimens.


In addition to the confounding problems of preservation, diagenesis, data collection, and analysis, if elemental and isotopic chemical analysis of skeletal material is to fulfill its potential in dietary reconstruction, insightful and appropriate avenues of interpretation are necessary. Although descriptive statistics of aggregate data drawn from a sample are useful heuristic devices, selection of appropriate analytical techniques appear to be linked to the nature of the concentration distributions.

Parametric and nonparametric—as well as univariate and multivariate—statistics have been applied to bone chemical quantitative data. The multiple problems and considerations involved have recently been discussed by Buikstra and colleagues (1989), who ultimately recommend principal component analysis. Even though mathematical rigor remains extremely important, insightful interpretations of relationships and findings still seem to require evaluation within a biocultural context.

Schoeninger (1989), for example, attempted to match food component elements and isotopes as well as skeletal analysis for prehistoric Pecos Pueblo and historic Dutch whalers to propose reasonable diets for them. Klepinger (1992) also commented on the importance of reevaluating model hypotheses that are frequently invoked in the light of new data and developing technologies.

Chemical approaches to dietary representation, especially of past groups, can be fascinating, frustrating, and fulfilling. But it seems unlikely that we will soon develop a comprehensive picture of past diets through chemical analysis alone. The complexity of the geochemical, biochemical, biological, physiological, cultural, and social systems involved require collaborative research and multidisciplinary sharing of results. Although each discipline and researcher may contribute various pieces of the puzzle, a clearer image can emerge only through integrative interpretations. The goal appears well worth the effort!