Skip to main content

Reconstructing Climate Change and Food Production in the Ancient Peruvian Andes Using Stable Isotope Analysis


Analyses of stable oxygen and carbon isotope ratios were used to test the hypothesis that there was a severe drought and related change in human diet in ancient Peru ca. 1100 C.E. Previous studies of ice cores from the Andes and paleobotanical studies of food refuse suggest that the change in climate had a major impact on human food consumption patterns. The ratio of stable oxygen isotopes (18O/16O) and carbon isotopes (13C/12C) were obtained from the hydroxyapatite in dental enamel from human re­mains excavated from archaeology sites in the Peruvian Andes. The18O was used to reconstruct ancient precipitation levels before and after 1100 C.E. and 13C ratios were used to document the amount of C3 vs. C4 plants in hu­man diets in both time periods. Similar 18O values were found, in both time periods, suggesting no change in precipitation, but the 13C values indicate a decrease in C4 plant consumption (e.g., maize) from one era to the next. The isotopic data from this study were compared to other isotopic datasets from the Andes to examine general trends in precipitation and its relation­ship to food production and consumption.


Bone collagen and dental enamel (tooth crowns), are some of the best preserv­ers of information in organisms. This process allows archaeologists and anthro­pologists to record patterns of human behavior from long ago. This study inves­tigates the effectiveness of using stable isotope analysis, a method for testing the anatomical make-up of ancient samples, to address questions arising from the analysis of ancient artifacts and skeletons, and test the idea that climate change (i.e., a severe drought) contributed to significant changes in agricultural pro­duction and thus, food consumption and nutritional status among prehispanic peoples in the Andes.

The samples for this study come from human skeletons collected from the cen­tral highlands of modern day Peru, dating back to the Middle Horizon (MH, 600 – 1000 C.E.) and the subsequent Late Intermediate Period (LIP, 1000 – 1400 C.E.). To examine whether precipitation levels significantly changed from one era to the next (MH to the LIP), and whether climate change affected food production and consumption, the stable oxygen and carbon isotope ratios were analyzed. If there are significant changes in the oxygen isotope ratio (18O/16O) and carbon isotope ratio (13C/12C), reported as18O and 13C values, from one period to the next, then this would suggest changes in precipitation levels and changes in food production and consumption.

Previous stable isotope studies have shown that changes in stable oxygen ra­tios indicate ancient climate changes, specifically as it relates to rainfall: a less negative 18O value indicates less rainfall and a more negative value indicates more rainfall, with some exceptions that relate to distance from equator, alti­tude, and contributions from glacial melt water [1]. Because of the locations samples were excavated from high in the Peruvian mountains, most, but not all, of the problems mentioned are eliminated. The value of oxygen isotope ratios is determined based on the presence of a heavier isotope (18O), compared to the lighter (16O). When in the form rain or water vapor, 18O tends to condense and precipitate out more quickly than its counterpart, 16O. Conversely, once collected by runoff in a riverbed or stream, the lighter 16O will evaporate more readily than 18O [1]. The 18O values were retrieved from the hydroxyapatite, crystalline structure of calcium carbonate in the body, of human dental enamel. Thus, making it possible to retrieve a permanent isotope signature based on origin of drinking water of organisms. Enamel hydroxyapatite, the preserved record of water and food used in this study, is more durable than bone collagen enamel. It is also formed mainly during an organism’s juvenile period due to the anatomy of tooth development, so only water consumed during the organism’s childhood is reflected.

Changes in carbon isotope ratios reveal changes in plant foods consumed. Differences in carbon isotope values stem from the complexity of pathways adapted by plants to obtain oxygen and nutrients from the environment. Two different pathways, the Calvin cycle (C3) and the Hatch-Slack pathway (C4), lead to different 13C values once digested by animals or humans. This is due to the efficiency of carbon absorption for production of glucose; C3 plants do not utilize as much carbon per metabolic cycle as do C4 plants. Certain plants, such as Zea mays (corn), utilize the Hatch-Slack (C4) pathway, resulting in a less neg­ative value than C3 plants; this makes C3 and C4 plants easily differentiated in the consumption patterns of mammals. A more negative 13C value indicates less C4 plant consumption and a more positive13C value indicates more C4 food in the diet [2]. Thus, the plant, the animal, and the human that eats those plants and animals will exhibit similar carbon isotope ratios within dental enamel, and types of food eaten can be determined.

In this case, corn is considered a target foodstuff, and the possibility for confu­sion lies with other C4 plants locally consumed, or marine foods, which can make 13C less negative [3]. The ancient human samples in this study have been shown to have no marine foods in the diet [4], and as of yet, no other C4 plants have been identified as food sources in this area of the ancient Andes. By testing the theories concerning the drought that may have befallen the Wari empire, anthropologists gain understanding of the government structure and ways of life of an ancient society that was lost long ago through isotope analysis.


18O and 13C isotope ratios were collected from the enamel hydroxyapatite of two camelid teeth and 16 human teeth from the archaeological sites of Conchopata, Huari-Cheqo Wasi, Huari-Vegachayoq Moqo, and Nawinpukio (Figure 2). Each tooth was sampled using a Dremel© hand drill and carbide dental bures, drilling a line perpendicular to the growth axis of the tooth on the buccal or lin­gual enamel surface. All samples were first dated to their according time periods using radio carbon dating and certain inferences based on their locations and conditions, in order to establish certainty of origin. These tests were completed previous to this study.

Each powdered sample was pretreated with 30% hydrogen peroxide for 24 hours and 0.1 N acetic acid for 18 hours to remove organics and secondary carbonates, respectively [5]. All samples were subsequently analyzed (~1mg) in a Finnigan Delta Plus XP mass spectrometer at the Stable Isotope Facility at the University of Wyoming for both carbon and oxygen isotopes. All data were normalized to normal buffered saline (NBS-19) for carbon (13C) and oxygen (18O) and the standard is VPDB (Vienna Pee Dee Belemnite standard)(6). 13C (parts per mil, ‰) = ((Rsample/Rstandard)-1)*1000 and R = 13C/12C. 18O (parts per mil, ‰) = ((Rsample/Rstandard)-1)*1000, and R =18O/16O; and the standard is VPDB [6].

The data from this investigation were compared to oxygen and carbon isotope ratios from previously published studies from culturally and temporally rele­vant areas of ancient Peru. The comparative sites includes: the Tiwanaku capital and smaller Tiwanaku sites (MH and LIP), Wari site of Conchopata (MH), Inca site of Machu Picchu (LH), and sites from the Nasca drainage of coastal Peru (EIP, MH, and LIP). The data was compared to the above other site popu­lations to compare precipitation levels and diet in diverse geographic areas, and during different time periods.


Among the 19 samples submitted for stable isotope analysis in the Finnigan Delta Plus XP mass spectrometer, 15 yielded human results (Supplemental Table 1). The remaining 2 were camelid samples, and were left out of data analy­sis for simplicity.

Stable Oxygen Isotope Analysis.

The mean human 18O is -8.47 and the oxygen isotope data show a range from -3.82 to -10.01. The standard deviation of the data is 1.43 (Supplemental Table 2). Human mean and standard deviation was also calculated for each site. Th e site of Conchopata has a mean of -8.63, and a standard deviation of 0.382. Huari Cheqo Wasi has a mean of -8.81, and a standard deviation of 0.846. Vegachyoq Moqo has a mean of -9.09, and a standard deviation of 0.460. The sites of Nawinpukio and Monqachayoq both only have one data result, so no mean or standard deviation could be calculated for either.

Stable Carbon Isotope Analysis.

The mean 13C values from human samples was found to be -3.71 with a range from -12.57 to -.034. The standard deviation of these values is 3.58 (Supplementary Table 3). The human mean and standard deviations for each site was calculated. The Middle Horizon site of Conchopata has a mean of -3.93, and standard deviation of 0.262, and the Middle Horizon site of Huari-Cheqo Wasi has a mean of -1.87, and a standard deviation of 2.029. The Late Intermediate samples from Huari-Vegachayoq Moqo have a mean of -11.18, and a standard deviation of 1.966 (Supplemental Table 2). The sites of Nawinpukio (MH) and Monqachayoq (LIP) have only one data point, so no mean and stan­dard deviation could be calculated.

Overall, the average 18O value for the second half of the Middle Horizon era (800 – 1000 C.E.) was -8.85667 (s.d. = 0.782924), and it slightly declined to -8.89000 (s.d. = 0.468722) in the first half of the subsequent Late Intermediate Period (1000 – 1200 C.E.) (Supplementary Table 4).

The average 13C value for the second half of the Middle Horizon era was -2.40667 (s.d. = 1.985407), and it decreased to -8.93333 (s.d. = 4.132146) in the first half of the subsequent Late Intermediate Period (Supplemental Table 4). Oxygen and carbon isotope ratios in this study were plotted against each other to explore possible relationships between the two variables (Figure 1). However, no relationship was found.


Figure 1. Relationship between oxygen and carbon isotope ratios from hydroxya­pitite dental enamel.



Although sample sizes are small, oxygen isotope ratios show little to no change between sites (Supplementary Table 2). This suggests sites analyzed within the same area generally received the same precipitation, and had low variability. On the other hand, carbon isotope ratios varied greatly between sites. The site of Huari-Cheqo Wasi—an elite mortuary area—has carbon isotope values indica­tive of large amounts of C4 plant (maize) consumption, while the site of Huari-Vegachayoq Moqo—a funerary area with injured and dismembered bodies [7]— exhibits a more C3 plant based diet (Supplemental Table 3).

The changes in oxygen isotope values over time have very little variation, sug­gesting that climate change was minimal in the central, highland Andes be­tween the MH and LIP; this contradicts previous theories (Supplementary Table 4). Because of this discrepancy, the authors suggest that the limitations of oxygen isotope analysis are related to the numerous variables that can affect the outcomes. The authors suggest that further research is necessary to understand the complex ways that variables such as rainfall, glacial water melt, altitude, and temperature can affect oxygen isotope values.

In contrast, the carbon isotope ratios in this study are highly variable, suggest­ing that human diets were quite distinct between sites of the same time period, and that they changed dramatically from one era to the next (Supplementary Table 4). Because of these significant differences and agreement with previ­ous studies, the data suggests that carbon isotope analysis is a reliable scientific technique to document ancient dietary practices.

Overall the two variables together, oxygen and carbon isotope ratios, show no correlation (Figure 1). Considering the hypothesis, the data provided by this study suggests that climate change was not the cause of transformations in diet in the central, highland Andes. Based on these findings, further investigation must be done to understand not only the reasons behind such dramatic diet changes.

Comparisons to Other Andean Sites.

Values collected in this study were analyzed and compared with previously pub­lished data from the Andes to strengthen the analyses (Supplementary Table 5). Data was compared to human samples from Conchopata [4, 8], Machu Picchu, [9, 10], Nasca sites [1], from the ancient capital city of Tiwanaku [11], and from Chanka sites in the nearby site of Andahuaylas [12].


Figure 2. Box plots grouped by site and time; greater 13C values mean more C4 plant consumption.


Figure 3. δ18O box plots grouped by site and time. CHNK=Chanka, Con=Conchopata, HCW=Huari Cheqo Wasi, HMQ=Huari Moqachayoq, HVM=Huari Vegachayoq Moqo, MP=Machu Piccu, Nas=Nasca, TK=Tiwanaku.


In the study area examined here, the 13C values change significantly from the MH to the LIP (Figure 2, Supplemental Table 4), indicating a major change in diet. Compared to the MH and LIP samples from Tiwanaku in the southern Andes (Bolivian highlands), there is a similar change from one time period to the next. The Tiwanaku samples also exhibit less negative 13C values in the MH, meaning more maize in the diet, and exhibit more negative 13C values in the LIP (less maize) (Figure 2). In that region, the change in diet may be correlated with a decline in precipitation, as evidenced by the decline in the oxygen isotope ra­tio. Together, those two datasets suggest a correlation between rainfall and the productivity of certain crops. This does not appear to be the case in the central, highland Andes, where this study was carried out.

The 18O values from this study suggest that rainfall in the central, highland Andes (our study area) did not dramatically decline in the second era under study. In contrast, other geographical zones in southern Peru experienced more rainfall during the MH than during the LIP (Figure 3). Notably, populations that ex­perienced a drought also saw a shift in plant production and consumption, as revealed by carbon isotope ratios (Figure 2). This suggests that a drought in the LIP adversely affected some population’s ability to produce the much desired plant food, maize. In the study presented here, the oxygen isotope values show no change, yet there is a dramatic decline in the consumption of maize.


By evaluating the oxygen isotope ratios from human dental enamel as a proxy for rainfall, some questions about paleoclimatic change can be addressed. This geographic area in particular was thought to have been disturbed by a severe drought, which then altered plant production and consumption patterns, perhaps negatively affecting the nutritional status of populations living there. While the data presented here do show a change in plant food consumption by ancient humans in the central, highland Andes, there is no evidence to sug­gest that rainfall levels also significantly changed. This suggests that other non-climatic events contributed to the shift in food production and consumption at the beginning of the Late Intermediate Period. Most notably, this was the era when the powerful Wari Empire collapsed, so perhaps the absence of state gov­ernance contributed to the decline in agricultural production, particularly that of maize, a socially valued food in the ancient (and modern) Andes, leading to a sudden shift from a C4 diet to a simpler C3 diet.

Stable isotope data from other Andean areas, especially the Tiwanaku zone, show that there was both a decline in rainfall and maize consumption [11]. By coupling the oxygen and carbon isotope data, researchers can examine both the estimated precipitation levels and the amount of specific food sources. When both data sets are combined, scholars can infer when climate change may have occurred, how severely, and whether those environmental changes affected food production capacity.

In the future, more samples will be isotopically analyzed to confirm the findings of this study. This will facilitate a better reconstruction of the paleoclimate in central Peru. This will add more data points, and add more stability in general to the project as a whole. This study might want to also take a closer look at the differences in oxygen isotope values in order to more clearly define whether our results were due to precipitation, altitude, temperature, etc., and allow us to more accurately reconstruct the paleoclimate in central Peru using stable iso­tope analysis.


Dr. Tiffiny Tung for her vital insights to the study, the use of her lab, and her funding, Dr. Larisa Desantis for her aid in sampling, prepping, and shipping, Danielle Kurin for guiding the research in the field in Andahuaylas Peru, and her data contributions, Dr. Jonathan Creamer for his insight and feedback on the research, and the Vanderbilt School for Science and Math for providing the opportunity for this research. The project described was supported by Award Number R25RR024261 from the National Center For Research Resources. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center For Research Resources or the National Institutes of Health.


  1. K. J. Knudson. Int. Jour. Osteoarch.19,171 (2009).
  2. M. J. Schoeninger. Cam. Univ. Press. 38 (1989).
  3. U. Gowik, P. Westhoff . Plant Phy. 155, 56 (2011).
  4. B. Finucane, et al., Jou. Arch. Sci.33, 1 (2006).
  5. P. L. Koch, et al. Jou. Arch. Sci.24, 417 (1997).
  6. T. B. Coplen. Pure. App. Chem.66,273 (1994).
  7. T. A. Tung. Nawpa Pacha. 29, 101-118 (2008).
  8. T. A. Tung, E. Chambers. Report. Nat. Inst. Culture, Peru.(2011).
  9. B. L. Turner. et. al. Jou. Arch. Sci.36, 317-332 (2009).
  10. B. L. Turner, J. D. Kingston, G. J. Armelagos. Revista de Anth. Chilena.42, 515­534 (2010).
  11. C. A. Berryman. Ph. D. Dissertation. Dept. of Anth. Vanderbilt University. (2010).
  12. E. Lefaro, D. Kurin. In prep. (2011).



Posted by on Thursday, August 27, 2015 in May 2012.

Tags: , , , , ,