Dietary Reconstruction As Seen in Coprolites

Kristin D Sobolik. Cambridge World History of Food. Editor: Kenneth F Kiple & Kriemhild Conee Ornelas. Volume 1. Cambridge, UK: Cambridge University Press, 2000.

The question of prehistoric dietary practices has become an important one. Coprolites (desiccated or mineralized feces) are a unique resource for analyzing prehistoric diet because their constituents are mainly the undigested or incompletely digested remains of food items that were actually eaten. Thus they contain direct evidence of dietary intake (Bryant 1974b, 1990; Spaulding 1974; Fry 1985; Scott 1987; Sobolik 1991a, 1994a, 1994b). In addition they can reveal important information on the health, nutrition, possible food preparation methods, and overall food economy and subsistence of a group of people (Sobolik 1991b; Reinhard and Bryant 1992).

Coprolites are mainly preserved in dry, arid environments or in the frozen arctic (Carbone and Keel 1985). Caves and enclosed areas are the best places for preserved samples and there are also samples associated with mummies. Unfortunately, conditions that help provide such samples are not observed in all archaeological sites.

Coprolite analysis is important in the determination of prehistoric diets for two significant reasons. First, the constituents of a coprolite are mainly the remains of intentionally eaten food items. This type of precise sample cannot be replicated as accurately from animal or plant debris recovered from archaeological sites. Second, coprolites tend to preserve small, fragile remains, mainly because of their compact nature, which tends to keep the constituents separated from the site matrix. These remains are typically recovered by normal coprolitic processing techniques, which involve screening with micron mesh screens rather than the larger screens used during archaeological excavations.

The limitations of coprolites are also twofold (Sobolik 1994a). First, even though the analysis of coprolites indicates the ingestion of food items, their constituents do not contain the entire diet of an individual or a population. In fact, because the different food items ingested pass through the digestive system at different rates, coprolite contents do not reflect one specific meal. The problem with coprolites is that they contain the indigestible portion of foods. The actual digested portion has been absorbed by the body. Thus, it has been estimated that meat protein may be completely absorbed during the digestion process, often leaving few traces in the coprolite (Fry 1985). However, recent protein residue analyses conducted on coprolites have indicated that some protein may survive (Newman et al. 1993).

A second limitation is that coprolites can reflect seasonal or short-term dietary intake. Individual coprolites often reflect either items a person ate earlier that day or what may have been eaten up to a month before (Williams-Dean 1978; Sobolik 1988). Thus, determining year-round dietary intake using coprolites, even with a large sample, becomes risky and inconclusive.

The History of Coprolite Research

The first observations of coprolites were those from animals of early geologic age; the Cretaceous in England (Mantell 1822;Agassiz 1833-43) and North America (Dekay 1830); the Lower Jurassic in England (Buckland 1829); and the Eocene in France (Robert 1832-33). Later works in North America include coprolites of the ground sloth (Laudermilk and Munz 1934, 1938; Martin, Sabels, and Shutler 1961; Thompson et al. 1980) and other Pleistocene animals (Davis et al. 1984).

The potential of human coprolites as dietary indicators was realized by J.W. Harshberger in 1896.The first analyses, however, were not conducted until after the beginning of the twentieth century. These initial studies were conducted by G. E. Smith and F. W. Jones (1910), who examined the dried fecal remains from Nubian mummies, and by B. H. Young (1910), L. L. Loud and M. R. Harrington (1929), and Volney H. Jones (1936), who studied materials from North American caves. Early coprolite analyses also included samples from Danger Cave (Jennings 1957), sites in Tamaulipas, Mexico (MacNeish 1958), caves in eastern Kentucky (Webb and Baby 1957), and colon contents from a mummy (Wakefield and Dellinger 1936).

The processing techniques for these early analyses consisted of either cutting open the dry coprolites and observing the large, visible contents, or grinding the samples through screens, in the process breaking much of the material. Improved techniques for analyzing coprolites were later developed by Eric O. Callen and T.W.M. Cameron (1960).

Still used today, these techniques revolutionized the science of coprolite analysis. They involved rehydrating the sample in tri-sodium phosphate, a strong detergent, in order to gently break apart the materials for ease in screening. Processing with tri-sodium phosphate also allowed for the recovery of polleniferous and parasitic materials from the samples, and increased the recovery of smaller, fragile macromaterials. Direct coprolite pollen analyses were soon followed by the first published investigation conducted by Paul S. Martin and F. W. Sharrock (1964) on material from Glen Canyon. Subsequently, there have been other innovative studies (Hill and Hevly 1968; Bryant 1974a; Bryant and Williams-Dean 1975; Hevly et al. 1979).

Coprolite Constituents

Coprolites represent such an unusual data source that their analysis is usually undertaken by specialists, generally by paleoethnobotanists (Ford 1979).A thorough coprolite analysis, however, involves the identification and interpretation of all types of botanical remains (Bryant 1974b, 1986; Fry 1985), such as fiber, seeds, and pollen, as well as nonbotanical remains, such as animal bone and hair, insects, fish and reptile scales, and parasites (to name a few of the many coprolitic constituents). Some recent studies have also identified the presence of wax and lipids through gas chromatography and mass spectrometry analyses and the analysis of phytoliths (Wales, Evans, and Leeds 1991; Danielson 1993). Clearly, then, coprolite analysis covers a myriad of sciences besides paleoethnobotany.

Yet, because coprolite analyses have tended to be conducted by paleoethnobotanists, such studies have tended to focus on the botanical remains. Recently, however, researchers have realized that the botanical portion represents a biased sample of prehistoric diet, and, consequently, studies of the nonbotanical macroremains from coprolites are becoming more prevalent.

A significant early analysis that included the identification and interpretation of a variety of coprolitic constituents was conducted on 50 coprolite samples from Lovelock Cave, Nevada (Napton 1970). As a part of this investigation, Charles L. Douglas (1969) identified eight different animal species through analysis of hair content, and Lewis K. Napton and O. A. Brunetti (1969) identified feathers from a wide variety of birds, most significantly, the mud hen.

A more recent study has focused on small animal remains recovered from coprolites excavated throughout North America (Sobolik 1993).This effort indicates that animals that have been considered noncultural or site-contaminants actually served as human food (Munson, Parmalee, and Yarnell 1971; Parmalee, Paloumpis, and Wilson 1972; Smith 1975; Cordell 1977; Lyman 1982). The large number of coprolites analyzed from North America reveals direct ingestion of small animals, suggesting that small animal remains from sites do, in fact, reflect human dietary patterns, and that reptiles, birds, bats, and a large variety of rodents were an important and prevalent component of the prehistoric diet.

A variety of microremains can be analyzed from coprolites. These constituents include spores and fungi (Reinhard et al. 1989), bacteria (Stiger 1977), viruses (Williams-Dean 1978), and, recently, phytoliths (Bryant 1969; Cummings 1989). The most frequently analyzed microremains from coprolites, however, are pollen and parasites.


Pollen is a unique resource in the analysis of coprolites because it can provide information not obtained from the macroremains. If a flower type is frequently ingested, the soft flower parts will most likely be digested. Pollen, depending on size and structure, becomes caught in the intestinal lumen, permitting it to be excreted in fecal samples for up to one month after ingestion. Therefore, the pollen content of coprolites does not reflect one meal, but can reflect numerous meals with a variety of pollen types (Williams-Dean 1978; Sobolik 1988).

Pollen in coprolites can occur through the intentional eating of flowers or seeds, through the unintentional ingestion of pollen in medicinal teas, or by the consumption of plants to which pollen adheres. Pollen, in this context, is considered “economic” because it is actually associated with food or a medicinal item. But it may also become ingested during respiration, with contaminated water supplies, and with food, especially if the food is prepared in an open area (Bryant 1974b, 1987). Such occurrences can be especially prevalent during the pollinating season of a specific plant, such as pine and oak in the spring or ragweed and juniper in the fall. Pollen, in this context, is considered “background” because it was accidentally ingested and was not associated with a particular food or medicinal item.

Pollen types are divided into insect pollinated plants (zoophilous) and wind pollinated plants (anemophilous). Insect pollinated plants produce few pollen grains and are usually insect specific to ensure a high rate of pollination. Indeed, such plants generally produce fewer than 10,000 pollen grains per anther (Faegri and Iversen 1964) and are rarely observed in the pollen record.

Wind pollinated plants, by contrast, produce large amounts of pollen to ensure pollination and are frequently found in the pollen record. The enormous quantity of pollen produced by some plants was highlighted by the study of R. N. Mack and Vaughn M. Bryant (1974) in which they found over 50 percent Pinus pollen in areas where the nearest pine tree is more than 100 miles away. Knut Faegri and J. Iversen (1964) state that an average pine can produce approximately 350 million pollen grains per tree.

In coprolite analyses, this division between pollination types is essential because a high frequency of wind pollinated pollen types in a sample may indicate not diet but rather accidental ingestion from contaminated food or water supplies. A high frequency of insect pollinated pollen types, however, often indicates the intentional ingestion of food containing pollen (economic pollen), since it is unlikely that many grains of this type are accidental contaminants. Bryant (1975) has shown from field experiments that for some of the common insect pollinated types in the lower Pecos region a frequency greater than 2 percent in a coprolite suggests a strong possibility of intentional ingestion of flowers and certain seed types that still have pollen attached, and that a frequency of 10 percent should be interpreted as positive evidence of intentional ingestion.

Parasites and Nutrition

The presence of parasites observed in coprolites can help determine the amount of disease present in populations and indicate much about the subsistence and general quality of life. Examples include studies conducted by Henry J. Hall (1972) and Karl J. Reinhard (1985) in which differences were noted between the prevalence of parasitic disease in hunter-gatherers and in agriculturalists.

Agriculturalists and hunter-gatherers have very different subsistence bases and lifeways, which affect the types of diseases and parasites infecting each group. Hunter-gatherers were (and still are) mobile people who generally lived and moved in small groups and probably had limited contact with outsiders. They lived in temporary dwellings, usually moving in a seasonal pattern, and tended to enjoy a well-balanced diet (Dunn 1968). Subsistence centered on the environment and what it provided, making it the most important aspect of their existence (Nelson 1967; Hayden 1971).

Agriculturalists, by contrast, are generally sedentary and live in larger groups because of the increase in population that an agricultural subsistence base can support. Their dwellings are more permanent structures, and they have extensive contacts with groups because of a more complex society and because of extensive trading networks that link those societies.

Although population increase and sedentary agriculture seem linked, the tendency of sedentary agriculturalists to concentrate their diets largely on a single crop can adversely affect health, even though their numbers increase. Corn, for example, is known to be a poor source of iron and deficient in the essential amino acids lysine and tryptophan. Moreover, the phytic acid present in corn inhibits intestinal absorption of nutrients, all of which can lead to undernourishment and anemia (El-Najjar 1976; Walker 1985). Thus, the adoption of agriculture seems to have been accompanied by a decrease in nutritional status, although such a general proposition demands analysis in local or regional settings (Palkovich 1984; Rose et al. 1984).

Nutritional status, however, can also be affected by parasite infection, which sedentism tends to encourage (Nelson 1967). In the past, as people became more sedentary and population increased, human wastes were increasingly difficult to dispose of, poor sanitation methods increased the chances of food contamination, and water supplies were fouled (Walker 1985). Many parasites thrive in fecal material and create a breeding ground for disease. The problem is exacerbated as feces are used for fertilizer to produce larger crop yields. Irrigation is sometimes used to increase production, which promotes the proliferation of waterborne parasites and also aids the dispersal of bacteria (Cockburn 1967; Dunn 1968; Alland 1969; Fenner 1970; McNeill 1979).

In addition, as animals were domesticated they brought their own suite of parasites to the increasing pool of pathogens (Cockburn 1967;Alland 1969; Fenner 1970; McNeill 1979). And finally, the storage of grains and the disturbance of the local environment, which accompanies agricultural subsistence, can stimulate an increase in rodents and wild animals, respectively, and consequently an increase in their facultative parasites as well (Reinhard 1985).

It seems clear that the quality of the diet declined and parasite load increased as the transition was made to sedentary agriculture. This is not to say that hunter-gatherers were parasite-free. Rather, there was a parallel evolution of some parasites along with humankind’s evolution from ancestral nonhuman primates to Homo sapiens (Kliks 1983).Then, as human mobility increased, some parasites were lost because of their specific habitat range and because of changes in environment and temperature. But as these latter changes occurred, new parasites were picked up. Probably such changes took place as humans migrated across the cold arctic environment of the Bering Strait into North America. Some parasites would have made the journey with their human hosts, whereas others were left behind (Cockburn 1967; McNeill 1979).

New Approaches to Dietary Reconstruction

Regional Syntheses

As more researchers are analyzing coprolites, regional syntheses of diet are becoming possible. Paul E. Minnis (1989), for example, has condensed a large coprolite data set from Anasazi populations in the Four Corners Region of the southwestern United States, covering a time period from Basketmaker III to Pueblo III (A.D. 500-1300). He observed that for the sample area, local resource structure seemed to be more important for determining diet than chronological differences. For example, domesticated plants, particularly corn, were a consistent dietary item from Basketmaker III to Pueblo III; and there was a “generally stable dietary regime” during the time periods studied, although small-scale changes were also noted (Minnis 1989: 559).

In another example from the Lower Pecos Region of southwestern Texas and northern New Mexico, a total of 359 coprolite samples have been studied (Sobolik 1991a, 1994b). Analysis indicates that the prehistoric populations of the region relied on a wide variety of dietary items for their subsistence. A substantial amount of fiber was provided by the diet, particularly that derived from prickly pear, onion, and the desert succulents (agave, yucca, sotol). A large number of seed and nut types were also ingested, although prickly pear and grass seeds were the most frequent. Animal remains, especially bone and fur, were also observed with a high frequency in the coprolites, indicating that these prehistoric people were eating a variety of animals (e.g., rodents, fish, reptiles, birds, and rabbits).The ingestion of an extremely wide variety of flowers and inflorescences is also indicated by the coprolite pollen data.

Significant differences were also observed in the dietary components of the coprolites. These differences might be attributable to changes in dietary practice, particularly to an increase in the variety of the prehistoric diet. A more plausible explanation, however, is that such differences are a result of the different locations of the archaeological sites from which the coprolite samples were excavated.

These sites are located on a south-north and a west-east gradient. Sites located in the southwestern portion of the region are in a dryer, more desert environment (the Chihuahuan Desert) with little access to water, whereas the sites located in the northeastern portion of the region are closer to the more mesic Edwards Plateau, which contains a diversity of plants and trees and is close to a continuous water supply. Thus, dietary change reflected in the coprolites most likely represents spatial differences rather than temporal fluctuations (Sobolik 1991a).

Nutritional Analyses

Coprolites are extremely useful in providing dietary and nutritional data, although information from botanical, faunal, and human skeletal remains is also needed in any attempt to characterize the nutrition of a prehistoric population (Sobolik 1990, 1994a). A recent study involving the nutritional analysis of 49 coprolites from Nubia was conducted by Linda S. Cummings (1989). This analysis was unique in that the coprolites were taken from skeletal remains buried in cemeteries representing two distinct time periods, including the early Christian period (A.D. 550-750) and the late Christian period (up to A.D. 1450). It is rare that prehistoric coprolites can actually be attributed to specific people, and the health of individuals can be assessed through both the coprolite remains and the human skeletal material, with one method illuminating the other.

In this case, the human skeletal material suggested that cribra orbitalia indicating anemia was the major indicator of nutritional stress. Coprolite analysis by Cummings (1989) revealed that there was probably a synergistic relationship in the diet of the population between iron-deficiency anemia and deficiencies of other nutrients, mainly folacin, vitamin C, vitamin B6, and vitamin B12. Cummings also noted differences in the diet and health of the two populations, including differences between males and females and older and younger members.

Pollen Concentration Studies

Although the determination of pollen concentration values has not been attempted in many coprolite studies, such values are important in determining which pollen types were most likely ingested. Studies show that after ingestion, pollen can be excreted for many days as the grains become caught in the intestinal folds. Experiments have also demonstrated that the concentration of intentionally ingested pollen can vary considerably in sequentially produced fecal samples (Kelso 1976;Williams-Dean 1978).

Glenna Williams-Dean (1978) conducted a modern fecal study that analyzed Brassicaceae and Prosopis pollen as a small component of pollen ingestion. It was revealed that Brassicaceae pollen was retained in the digestive system for much longer periods of time (up to one month after ingestion) than Prosopis pollen. Brassicaceae pollen is an extremely small grain with an average size of 12 micrometers (•m) and has a finely defined outer-wall sculpturing pattern. Both traits would most likely increase the retention of this pollen type in the folds of the intestine, allowing for it to be observed in many fecal samples. Prosopis pollen is a spherical, medium-sized grain (average size 30•m), with a smooth exine sculpturing pattern. The larger size of this grain and the decreased resistance provided by the smooth exine would permit this pollen type to pass more quickly through the intestinal folds without retention.

In light of this study, it can be predicted that larger pollen grains, such as corn (Zea) and cactus (Cactaceae), and pollen with little exine sculpturing, such as juniper (Juniperus), will move quickly through the human digestive system. Thus, these pollen types would be observed in fewer sequential fecal samples than those of other types. By contrast, smaller pollen grains with significant exine sculpturing, such as sunflower pollen (high-spine Asteraceae), can be predicted to move more slowly through the digestive system, become frequently caught in the intestinal lumen, and thus observed in fecal samples many days after initial ingestion.

Such predictions were subsequently applied in an examination of prehistoric coprolites (Sobolik 1988). This investigation revealed that a high pollen concentration value in coprolite samples should indicate that the economic pollen types observed in the samples were ingested recently. Concentration values of over 100,000 pollen grains/gram of material usually contain recently ingested pollen. But samples that contain less than 100,000 pollen grains/gram of material may contain economic pollen types that were intentionally ingested many days before the sample was deposited (Sobolik 1988). Such samples will also contain a wide variety of unintentionally ingested, background pollen types. Therefore, it is more difficult to recognize intentionally ingested pollen types from samples that contain less than 100,000 pollen grains/gram of material.

Modern fecal studies are, thus, invaluable as guides in the interpretation of prehistoric coprolite pollen content and in indicating the limitations of the data. Many more such investigations are needed to determine both pollen percentage and concentration. The diet of the participants will have to be stringently regulated in order to minimize the influence of outside pollen contaminants, particularly those in bread and canned foods (Williams-Dean 1978). Ideally, such studies will include many people and, thus, as diverse a population of digestive systems as possible over a long period of time.An important addition to such a study would be the observance of the effect of a high-fiber and a high-meat diet on pollen output and fecal output in general.

Medicinal Plant Usage

Documenting prehistoric medicinal plant usage is problematic because it is difficult to distinguish between plants that were consumed for dietary purposes and those consumed for medicinal purposes. Indeed, in many instances plants were probably used both dietarily and medicinally.

Nonetheless, the analysis of plant remains from archaeological sites is often employed to suggest dietary and medicinal intake. Such remains can be deposited through a number of channels, most significantly by contamination from outside sources (i.e., water, wind, matrix shifts, and animals). Plants also were used prehistorically as clothing, shelter, baskets, and twining, and these, when deposited into archaeological contexts, can become mistaken for food or medicinal items. Here, then, is a reason why coprolites, which are a direct indication of diet, can provide new insights into prehistoric medicinal usage (Reinhard, Hamilton, and Hevly 1991).

In an analysis of the pollen content of 32 coprolites recovered from Caldwell Cave, Culberson County, Texas, Kristin D. Sobolik and Deborah J. Gerick (1992) revealed a direct correlation between the presence of plants useful for alleviating diarrhea and coprolites that were severely diarrhetic. This correlation suggests that the prehistoric population of Caldwell Cave was ingesting medicinal plants to help alleviate chronic diarrhea. These plants, identified through analysis of the pollen content of the coprolites, included Ephedra (Mormon tea) and Prosopis (mesquite). Interestingly, this investigation confirmed the study conducted by Richard G. Holloway (1983), which indicated that the Caldwell Cave occupants were possibly using Ephedra and Larrea (creosote bush) in medicinal teas to help cure chronic diarrhea (Holloway 1983).

Mormon tea pollen, leaves, and stems are widely used as a diarrhetic and have been one of the most prevalent medicinal remedies for diarrhea both prehistorically and historically (Burlage 1968; Niethammer 1974; Moore 1979; Moerman 1986). Mesquite leaves are also useful as a medicinal tea for stomach ailments and to cleanse the digestive system (Niethammer 1974). As part of the process of preparing mesquite leaves for a medicinal tea, pollen could also become incorporated into the sample, either intentionally or unintentionally.

In another study, Reinhard and colleagues (1991) also determined medicinal plant usage through analysis of the pollen content in prehistoric coprolites. They found that willow (Salix), Mormon tea (Ephedra), and creosote (Larrea) were probably used for medicinal purposes prehistorically and that a large variety of other plants may have been used as well.

Protein Residues

It was previously mentioned that analysis of protein residues in coprolites is a relatively new advance in coprolite studies (Newman et al. 1993). This method attempts to link protein residues with the type of plant or animal that was consumed and involves the immunological analysis of tiny amounts of protein through crossover electrophoresis. The unknown protein residue from the coprolites is placed in agarose gel with known antiserum from different plants and animals. The agarose gel is then placed in an electrophoresis tank with a barbital buffer at pH 8.6, and the electrophoretic action causes the protein antigens to move toward the antibody that is not affected by the electrical action. The solution, which contains the unknown protein residue, and the matching plant or animal species antiserum form a precipitate that is easily identifiable when stained with Coomassie Blue R250 solution (Kooyman, Newman, and Ceri 1992). The samples that form a precipitate indicate that the matching plant or animal was eaten by the person who deposited the coprolite sample.

Two sample sets were selected for an analysis of the protein residues found in coprolites – seven from Lovelock Cave, Nevada, and five from an open site in the Coachella Valley of southern California (Newman et al. 1993). Protein analysis of samples from the open site was not successful. But the samples from Lovelock Cave indicated that human protein residues were found in six of the samples and protein residues from pronghorns were found in four samples. Such an initial study suggests that protein residue analysis can be a successful and important component in the determination of prehistoric diet.

Gender Specificity

A new technique that will distinguish the gender of the coprolite depositor is presently being tested on coprolite samples from Mammoth and Salts Caves, Kentucky, by Patricia Whitten of Emory University. In this technique, which has been successful with primate studies, the gonadal (sex) steroids are removed from each coprolite sample and analyzed according to the content of testosterone, the male hormone, and estradiol, the female hormone. Both steroids can be found in each sample, but their frequencies vary depending upon gender. Modern human samples will first be analyzed to determine the frequencies of each gonadal steroid expected in males and females.

DNA analysis is also being attempted from coprolite samples to determine gender (Mark Q. Sutton 1994, personal communication).This type of research should allow archaeologists to determine dietary differences between males and females in a population and should also help in reconstructing the patterns of differential access to resources.


Reconstructing prehistoric human diets is a vast process requiring a variety of assemblages and disciplines to obtain a complete picture. Coprolite analysis provides a diverse and significant insight into the prehistoric diet. When analyzing the entire diet of a population, researchers must take into consideration information gleaned from other archaeological materials. Past coprolite research has focused on developing new and innovative techniques so that recovery of the diverse data inherent in such samples can be achieved. Such development has allowed researchers not only to observe the macrobotanical and macrofaunal remains from coprolites but also to analyze their pollen, parasite, and phytolith content.

Recent advances have seen the discipline progress toward determining medicinal plant ingestion so as to permit inter- and intraregional dietary comparisons; to determine the protein content of the samples; to analyze the nutritional content of the dietary items; and to determine the gender of the depositor of the sample. Coprolite analysis has definitely advanced out of its infancy, and its contributions to the determination of prehistoric diet, health, and nutrition in the future should prove to be significant indeed.