Migration into the ancient city of Teotihuacan (AD 1–550) is one of the most complex and least understood demographic developments of any prehispanic urban center (Smith Reference Smith, Carballo, Hirth and Arroyo2020). Past research demonstrates that Teotihuacan attracted migrants in its early years of urban development (Millon Reference Millon1973, Reference Millon and Sabloff1981) and that in-migrants inhabited specific neighborhoods throughout the occupation of the city (Clayton Reference Clayton2005; Gómez Chávez Reference Gómez Chávez and Ruiz2002; Manzanilla Reference Manzanilla2011; Michelet and Pereira Reference Michelet, Pereira, Olguín, Alcántara and Castillo2009; Rattray Reference Rattray, Ruiz and Pascual2004; Solís Pichardo et al. Reference Solís Pichardo, Schaaf, Treviño, Lailson, Manzanilla, Horn and Manzanilla2017; Spence et al. Reference Spence, White, Rattray, Longstaffe and Blanton2005; Taube Reference Taube and Braswell2003; White, Spence, et al. Reference White, Spence, Longstaffe and Law2004; White, Storey, et al. Reference White, Storey, Longstaffe and Spence2004). However, archaeometry data combining radiocarbon dates and isotopic measurements have never been formally tested to determine whether in-migration at Teotihuacan was highest at the city's founding or remained constant across the centuries as an integral component of sustaining the urban population. Additionally, we do not yet understand whether specific demographic groups were more inclined than others to immigrate to Teotihuacan.
This study reevaluates previous estimates of migration at the Tlajinga district of Teotihuacan (Figure 1) through a dual-isotope approach using strontium (87Sr/86Sr) and stable oxygen-phosphate (δ18Op) isotope data from human remains (n = 23). Refinements are made to the local 87Sr/86Sr ratio baseline ranges for the Teotihuacan Valley. We couple isotopic data with previously reported high-precision radiocarbon dates (Storey et al. Reference Storey, Buckley and Kennett2019:155) to observe diachronic migration in three compounds across the Tlajinga district. We also test for significant differences in isotopic signatures between residential and demographic groups to identify specific types of people who migrated to Tlajinga. For the first time at Teotihuacan, we consider how imported maize-enhancing materials used in the nixtamalization process potentially skew the 87Sr/86Sr signatures of individuals in this region.
Teotihuacan with apartment compounds Tlajinga 17, 18, and 33 shown in relation to the greater city.
Teotihuacan is located 45 km northeast of modern-day Mexico City and is spread across 25 km2 in the Teotihuacan Valley, an extension of the greater Basin of Mexico. It was the first city in the Americas to reach a population of 100,000, which it did during the Classic period (AD 200–550; Cowgill Reference Cowgill2015:143; Smith et al. Reference Smith, Chatterjee, Huster and Stewart2019). It is hypothesized that migrants were attracted to the city from across Mesoamerica, because Teotihuacan offered economic opportunity and the potential for individuals to rise in social status (Spence et al. Reference Spence, White, Rattray, Longstaffe and Blanton2005). Residents of Teotihuacan constructed apartment compounds, which were not seen before in the Americas nor again before Spanish contact. These residential structures were large, multifamily complexes, typically with shared lineal descent, cultural practices, and forms of craft specialization (Carballo Reference Carballo2011; Cowgill Reference Cowgill2015:142; Manzanilla Reference Manzanilla1996).
Teotihuacan comprised a vast network of residential districts, as is described in the Teotihuacan Mapping Project (Millon Reference Millon1973). Accordingly, research on in-migration has historically been conducted at the neighborhood level (Manzanilla Reference Manzanilla2017; Spence et al. Reference Spence, White, Rattray, Longstaffe and Blanton2005; Storey Reference Storey1992; White, Spence, et al. Reference White, Spence, Longstaffe and Law2004; White, Storey, et al. Reference White, Storey, Longstaffe and Spence2004), and the results were then consolidated to allow citywide comparisons of demographic events and mortuary practices. Here, we continue this practice and revisit the southernmost district of Tlajinga. We offer additional isotopic analyses for Compound 33:S3W1 (“Tlajinga 33”), along with new data from two recently excavated residential units: Compounds 17:S3E1 (“Tlajinga 17”) and 18:S3E1 (“Tlajinga 18”).
Located on the southern periphery of the city, the Tlajinga district was connected to the central urban hub by an extension of the Street of the Dead (Figure 1). Unlike apartment compounds near the epicenter, those in Tlajinga were spaced farther apart and characterized by a mix of adobe and stone walls, the minimal use of lime plaster, and a lack of murals. These features imply that the district was made up of lower-status residents (Carballo et al. Reference Carballo, Hirth, Sariñana, Buckley, Ramón and Kennett2019; Storey Reference Storey1992; Widmer Reference Widmer, de Tapia and Rattray1987). However, several burials from Tlajinga 33 contain elaborate mortuary offerings or are distinguished by a large tomb interment, thereby signifying the existence of compound leaders (Storey et al. Reference Storey, Buckley and Kennett2019:158; Figure 2). This compound also appeared to be a focal point of ceramic production, particularly of the San Martín Orange ware that was distributed citywide and was popular in the later phases of the city's occupation (Cowgill Reference Cowgill2015:173; Sheehy Reference Sheehy1992). Additionally, recent excavations identify Tlajinga 17 as the location of a large obsidian workshop (Hirth et al. Reference Hirth, Carballo, Mark Dennison, Imfeld and Dyrdahl2019) and the source for large outputs of lapidary crafts (Widmer Reference Widmer2019).
The Tlajinga 33 compound. Burials mentioned in the text are labeled. Adapted from Storey (Reference Storey1992:Figure 3-2). The distribution of burials at Tlajinga 17 and 18 can be viewed in Storey and colleagues (Reference Storey, Buckley and Kennett2019:Figure 1).
The demography and health of the inhabitants of Tlajinga 33 are discussed in Storey (Reference Storey1992). Residents of Tlajinga 33 fought an uphill battle with morbidity: the population suffered a steady decline due to high infant mortality rates caused mostly by prenatal malnutrition or illness (Storey Reference Storey1992:255). Skeletal observations from Compounds 17 and 18, although small in sample size (n = 9), show similar morbidity patterns (Storey et al. Reference Storey, Buckley and Kennett2019:Figure 1, 151). Therefore, we support Storey's (Reference Storey1992) hypothesis that continual migration into the Tlajinga district was necessary to maintain population stability.
In an earlier test of this hypothesis, White, Storey, and colleagues (Reference White, Storey, Longstaffe and Spence2004) measured the oxygen isotope composition of bioapatite phosphate (δ18Op) in human bone and tooth enamel for 25 individuals. Those results indicated that ~30% of sampled individuals from Tlajinga 33 had signatures outside the local δ18Op baseline range. Perhaps most surprising, this study found that higher-status individuals tended to be those who “spent time in a foreign location” (White, Storey, et al. Reference White, Storey, Longstaffe and Spence2004:189).
Until recently, δ18Op analysis was the predominant method of identifying migrant individuals in Teotihuacan (White et al. Reference White, Spence, Longstaffe, Stuart-Williams and Law2002, Reference White, Douglas Price and Longstaffe2007; White, Spence, et al. Reference White, Spence, Longstaffe and Law2004; White, Storey, et al. Reference White, Spence, Longstaffe and Law2004). A multi-isotope approach has now been shown to more reliably identify nonlocal individuals within a region (e.g., Wright Reference Wright2012). Thus, earlier studies based on a single isotopic system can benefit from the examination of additional isotopic systems at any region or site. Futhermore, the acquisition of high-precision accelerated mass spectrometry (AMS) radiocarbon (14C) dates in tandem with isotopic analysis provides key insights to migration over time, but this coupling of datasets has not been previously realized at Teotihuacan.
Benefits of a Dual-Isotope Approach (87Sr/86Sr and δ18Op)
The benefits of a dual-isotope approach stem from the fact that strontium (Sr) and oxygen (O) from the surrounding environment are absorbed into human tissues from different aspects of the diet: Sr from food (and water, to a lesser extent) and O from water ingested over the life cycle. Thus, each isotopic system provides unique provenance measurements that can be used to determine patterns of residential mobility throughout an individual's life span (see Bentley Reference Bentley2006; Pederzani and Britton Reference Pederzani and Britton2019).
In Mesoamerica, the combination of strontium and oxygen isotope analyses has already been a significant contributor to understanding paleomobility. Wright (Reference Wright2012) analyzed these isotopic systems at the ancient Maya city of Tikal to demonstrate that in-migration of people from all socioeconomic tiers contributed to the city's rapid population growth. In Teotihuacan, White and colleagues (Reference White, Douglas Price and Longstaffe2007) first used strontium and oxygen isotopes to identify many of the sacrificial individuals from the Moon Pyramid as foreign-born. More recently, Price and coauthors (Reference Price, Spence and Longstaffe2020) revisited the sacrificial individuals of the Feathered Serpent Pyramid at Teotihuacan, initially analyzed for δ18Op by White and colleagues (Reference White, Spence, Longstaffe, Stuart-Williams and Law2002), and applied new strontium and oxygen-carbonate isotope analyses. The new findings suggest that some of these sacrificed individuals were local to the central Highlands of Mexico rather than part of an ethnically diverse military, as was suggested in the intial study (White et al. Reference White, Spence, Longstaffe, Stuart-Williams and Law2002).
When using isotope analyses to detect paleomobility, researchers must consider whether individuals were eating foods and drinking water obtained from the same geological and environmental region in which they resided. Therefore, one must first understand a population's diet before successfully implementing isotopic studies to identify migration (Nado et al. Reference Nado, Zolotova and Knudson2017; Wright Reference Wright2012). At Tlajinga, past studies of stable carbon (δ13C) and nitrogen (δ15N) isotopes suggest that residents from all three compounds consumed a maize-heavy diet and did not import a significant amount of marine-derived food (Storey et al. Reference Storey, Buckley and Kennett2019; White, Storey, et al. Reference White, Spence, Longstaffe and Law2004). Additionally, Teotihuacan was surrounded by maize-producing agricultural fields (Nichols et al. Reference Nichols, Spence and Borland1991). Therefore, we are confident in assuming that Tlajinga residents ingested primarily locally produced foods.
Materials and Methods
Numerous studies have found that tooth enamel bioapatite more reliably preserves original biogenic isotopic compositions than bone (e.g., Grimes and Pellegrini Reference Grimes and Pellegrini2013). Furthermore, different teeth within the human mouth develop at various known ages during the early years of human development (Hillson Reference Hillson1996). Once tooth crown formation is completed, human enamel does not undergo further turnover. Human bone, by comparison, continuously remodels throughout the life span. Enamel's resistance to remodeling in vivo after tooth crown formation is complete provides researchers with a tissue that essentially locks in isotopic signatures acquired during childhood development (Montgomery et al. Reference Montgomery, Evans, Chenery, Pashley and Killgrove2010). Therefore, in this study, we analyzed human tooth enamel for 87Sr/86Sr and δ18Op, rather than bone.
Sample Selection
Tooth enamel data for a total of 23 individuals are reported in this study (n = 39 tooth enamel samples in total): 18 individuals (n = 30 enamel samples) for 87Sr/86Sr ratios and 14 individuals (n = 22 enamel samples) for δ18Op. The δ18Op values for eight of these individuals (n = 9 enamel samples) were initially reported by White, Storey, and colleagues (Reference White, Storey, Longstaffe and Spence2004:180–181). Consistent with past δ18Op studies conducted at Teotihuacan (White, Spence, et al. Reference White, Spence, Longstaffe and Law2004; White, Storey, et al. Reference White, Storey, Longstaffe and Spence2004), the measured δ18Op values were adjusted downward by 0.35‰ for premolars, and 0.7‰ for all teeth developed before three to four years of age, because this is the accepted prehistoric weaning age in this region (Wright Reference Wright2013). The δ18O of bioapatite phosphate and structural carbonate in enamel wholly formed during breastfeeding is commonly higher by ~0.7‰ compared to similar tissues formed or remodeled after the breastfeeding period (White et al. Reference White, Spence, Longstaffe and Law2000; Wright and Schwartz Reference Wright and Schwarcz1999). This difference reflects the trophic step associated with the incorporation of maternal body water during breastfeeding.
Samples were obtained from apartment compounds within the Tlajinga district, Tlajinga 17 (n = 2), Tlajinga 18 (n = 3), and Tlajinga 33 (n = 18). Individuals analyzed by White, Storey, and colleagues (Reference White, Spence, Longstaffe and Law2004) who only had δ18Op values for bone were excluded from this study, so that we compared only oxygen isotope results reflecting early life developmental phases. When available, two teeth from each adult individual were selected for isotopic analysis based on the known ages of tooth crown formation and completion (Hillson Reference Hillson1996). We identify these phases as early childhood (EC), representing birth through age 5, and late childhood/adolescence (LCA), representing ages 12–16. Due to the lack of material for some specimens included in this study, Sr and O isotope results are not available for all individuals.
The use of isotopes in migration studies requires the availability of a baseline. For strontium isotopes, the most robust technique for estimating regional baselines is to gather a range of isotopic ratios that reflect any possible geological (bedrock), environmental (soil, water), or bioavailable (plants, fauna, human) signatures that could be considered local to the region of study (Adams et al. Reference Adams, Rainer Grün, Jian-Xin Zhao, Ai Nguyen, Michael Quaresimin, Collard and Westaway2019; Grimstead et al. Reference Grimstead, Nugent and Whipple2017). To complement the existing baseline for local 87Sr/86Sr ratios in the Teotihuacan Valley (see Price et al. Reference Price, Manzanilla and Middleton2000; Schaaf et al. Reference Schaaf, Solís, Manzanilla, Hernández, Lailson, Horn and Manzanilla2012), we analyzed additional faunal samples collected from the 2013–2014 Proyecto Arqueológico Tlajinga Teotihuacan (PATT) excavations and water samples from groundwater across the Teotihuacan Valley (n =14). We chose to use tooth enamel from rabbits and rodents, because these animals have small home ranges (see Sommerville et al. Reference Sommerville, Froehle and Schoeninger2018).
Statistical Analysis
All statistical analyses were conducted in the R statistical software package (Core Team, 2017). We used hierarchical cluster analysis to observe significantly different groups of individuals for which there are measurements of both 87Sr/86Sr and δ18Op. Here, we tested models that imposed two, three, four, and five clusters. One-way ANOVA was then performed to test for significant differences in mean isotopic compositions across each cluster. The R code for this analysis is included in Supplemental Text 1. For four burials—33-15, 33-50d, 33-56, and 33-57—EC and LCA enamel were included from the same individual to identify whether isotopic measurements were significantly different throughout childhood.
We created groups based on the following demographic characteristics: time period based on the ceramic chronology (Tlamimilolpa/Xolalpan phases), compound (Tlajinga 17 and 18/Tlajinga 33), status (higher/lower), and sex (female/male). We also tested for differences across time using 14C dates (reported in Storey et al. Reference Storey, Buckley and Kennett2019:155). Here, we divide individuals into three groups designated by the dominant economic production efforts of Tlajinga 33 at the time: lapidary (crafts; cal AD 200–350), transition (from lapidary crafts to San Martín Orange ware production; cal AD 300–450), and San Martín (Orange ware; cal AD 400–550). Shapiro-Wilk and Levene's tests were conducted to determine the normality of sample distributions and assess the equality of variances. We then performed parametric Student's t-tests or nonparametric Mann-Whitney U tests to determine whether there were significant differences between group means. One-way ANOVA or Kruskal-Wallis tests were used to detect significant differences between combined demographic data (e.g., compound and status, compound and time periods) and between the time groups using 14C dates. Supplemental Text 2 describes the methodology used to obtain the isotopic data in this study.
Results and Discussion
Isotopic Baselines in the Teotihuacan Valley
The 87Sr/86Sr ratios measured in this study for the regional baseline of the Teotihuacan Valley range from 0.70432 to 0.70499 (which we rounded to 0.7043 and 0.7050; see Supplemental Table 1). The highest measured 87Sr/86Sr ratio is from one water sample procured from a contemporary well near Teotihuacan. Without this sample, the maximum value of the range would be closer to 0.7048. Previously reported ranges of the 87Sr/86Sr baseline from faunal samples were from 0.7046 to 0.7047 (Price et al. Reference Price, Manzanilla and Middleton2000). In a recent study, Price and colleagues (Reference Price, Spence and Longstaffe2020) concluded that Teotihuacan and its nearby neighbors should have 87Sr/86Sr ratios between 0.7040 and 0.7050, supporting our new assessment of the baseline.
For Teotihuacan, the accepted δ18Op baseline range is +14 to +16‰, as first reported by Levinson and colleagues (Reference Levinson, Luz and Kolodny1987). White, Storey, and others (Reference White, Storey, Longstaffe and Spence2004) corroborated this range, noting that individuals from the Tlajinga district had a mean δ18Op of +15‰ and that the typical regional variation for locals is about 2‰.
Identifying Migrants at Teotihuacan
Table 1 reports the isotopic measurements of tooth enamel specimens and the demographic data for each individual. The strontium and oxygen isotope results indicate that 10 individuals have both or either 87Sr/86Sr ratios or δ18Op values that fall outside the local isotope baselines for the Teotihuacan Valley (also see Table 2).
Isotopic Parameters by Individual and Compound.
Notes: The oxygen isotope composition was adjusted downward by 0.35% for premolars and 0.7% for tooth enamel developed before the age of 3–4 (incisors, canines, and molars). Values shown in bold font are outside of the local baseline range. EC = Early Childhood; LCA = Late Childhood/Adolescence developmental phases.
a Originally reported in White et al. Reference White, Storey, Longstaffe and Spence2004:Table 4
b These values represent the average of a duplicate or triplicate analysis.
Summary of Demographic Totals and Percentages.
a Percentage = [# of migrants]/[total sample]. Total sample = 23.
Strontium Isotopes. 87Sr/86Sr ratios for tooth enamel samples range from 0.70418 to 0.70570, with a mean of 0.70499 ± 0.00030. Figure 3 illustrates the chronological order of the 87Sr/86Sr ratios based on the ceramic chronology of mortuary offerings and previously reported AMS 14C dates (Storey et al. Reference Storey, Buckley and Kennett2019:155). Here we see that ~40% (n = 7) of individuals have EC enamel 87Sr/86Sr ratios indicative of migrants from outside the Teotihuacan Valley (Burials 33-56, 33-14, 33-15, 33-43, 33-25, 33-50d, and 33-50b; Table 2). Burials 33-56 and 33-15 have similar LCA enamel 87Sr/86Sr ratios, suggesting residential mobility by late childhood (0.70500 and 0.70498, respectively). Both of these individuals likely originated near the Basin of Mexico (see Pacheco-Forés et al. Reference Pacheco-Forés, Gordon and Knudson2020). Burials 33-25 and 33-50b have a second EC enamel 87Sr/86Sr ratio that indicates movement into the region during early childhood years. Various forms of evidence suggest that both individuals came from Michoacán in west Mexico. Burial 33-25 has low 87Sr/86Sr ratios consistent with this region, ranging from 0.7040 to 0.7048 (Juarez Reference Juarez2008; White et al. Reference White, Douglas Price and Longstaffe2007:163). Burial 33-50b has a much higher EC 87Sr/86Sr ratio (0.70512) but was interred in a shaft tomb unlike any other at Tlajinga but that was common to west Mexico (Widmer Reference Widmer, de Tapia and Rattray1987).
Chronological representations of individual 87Sr/86Sr ratios for all samples in this study. Vertical, dotted lines join multiple enamel samples for the same individual. (*) indicates higher-status individuals. The AMS 14C dates were taken from Storey and colleagues (Reference Storey, Buckley and Kennett2019:155) and are reported in calibrated years AD.
Burials 33-43 and 33-50d each have EC and LCA 87Sr/86Sr ratios that are nearly identical, suggesting that these individuals did not move into the Teotihuacan region before adolescence (Table 1). The 87Sr/86Sr ratios for Burial 33-43 indicate an origin outside the central Highlands (White et al. Reference White, Douglas Price and Longstaffe2007:163). Burial 33-50d, interred in the shaft tomb with 33-50b, does not have 87Sr/86Sr ratios that support in-migration from west Mexico. Burial 33-14 has an LCA 87Sr/86Sr ratio (0.70513) lower than the EC ratio (0.70530), which suggests an origin from within the central Highlands. This Sr-isotope composition may represent a mixing of foods consumed from both the place of origin and the Teotihuacan Valley, thus suggesting relocation of this individual during tooth enamel crown formation of the third molar.
All seven outlier individuals except for Burial 33-43 have been classified as higher status based on their burial contexts and mortuary offerings (Storey Reference Storey1992). This result supports White, Storey, and colleagues' (Reference White, Storey, Longstaffe and Spence2004) hypothesis that higher-status individuals were more likely to have immigrated from outside of Teotihuacan than those of lower status (see also Table 2). Furthermore, Burial 33-57, interpreted as a master artisan, is the only individual defined as a possible compound leader, with local 87Sr/86Sr ratios for both childhood phases (others include 33-56, 33-25, 33-50d, and 33-50b; Widmer and Storey Reference Widmer, Storey, Charlotte Arnauld, Manzanilla and Smith2012). This finding has two implications. First, it suggests that master artisans were essential members of Tlajinga society and that status was obtained through skilled craft specialization in this district. Second, it appears that compound leaders and higher-status members of Tlajinga were of both native and migrant origin, thus suggesting that all members of society could achieve a higher ranking within the social order. Outsiders, or those born outside of Teotihuacan, were not prevented from becoming more affluent members of society.
Chronologically, the highest number of individuals with nonlocal 87Sr/86Sr ratios (n = 5) occurs during the Tlamimilolpa phase (Figure 3). However, this result may be explained by discrepancies in sample size (Tlamimilolpa = 13; Xolalpan = 3, respectively). Of these individuals, Burial 33-25 dates to cal AD 315–395 (2σ; Storey et al. Reference Storey, Buckley and Kennett2019), representing the transition period between the Tlamimilolpa and Xolalpan phases, as well as the transition between lapidary and San Martín production. When observing these individuals by 14C phases, we also note that the earliest 14C time range, the lapidary phase, has the highest number of migrants (n = 4).
Oxygen Isotopes. Oxygen isotope results range from +11.3 to +18.1‰, with a mean of +15.4 ± 1.4‰. Figure 4 displays the chronological order of δ18Op values based on previously reported ceramic mortuary offerings and AMS 14C dates (Storey et al. Reference Storey, Buckley and Kennett2019:155). Here, 50% (n = 7) have enamel compositions that are outside the local δ18Op baseline for Teotihuacan (Burials 33-6, 33-14, 33-43, 33-44, 33-45a, 33-50d, and 33-50b; see also Table 2).
Chronological representations of individual δ18Op values for all samples in this study. Vertical, dotted lines join multiple enamel samples for the same individual. (*) indicates higher-status individuals. The AMS 14C dates were taken from Storey and colleagues (Reference Storey, Buckley and Kennett2019:155) and are reported in calibrated years AD.
Burials 33-6, 33-14, and 33-43 all have LCA enamel δ18Op values that fall outside the local baseline (Table 1). It is not clear from the δ18Op values alone whether these individuals were born outside the region, given the lack of an EC enamel sample. For instance, 33-50b, interred in the shaft tomb, has an EC δ18Op value within the local range of variation for Teotihuacan (+14.7‰), although it should be noted that these values also coincide with local δ18Op ranges at Michoacán (White et al. Reference White, Douglas Price and Longstaffe2007:163). However, the LCA δ18Op value from the same individual indicates possible movement outside this region (+17.2‰). Burial 33-50b may have been born in Teotihuacan, moved during childhood, and then returned in his adult years. Or, he may have been born in west Mexico, moved outside of the central Highlands, and then arrived at Teotihuacan as an adult. Burial 50d, also interred in the shaft tomb, seems to have had a different life trajectory. The δ18Op signatures from EC (+16.3‰) and LCA (+17.3‰) suggest that 33-50d was born and lived outside the city for the majority of his childhood (White et al. Reference White, Douglas Price and Longstaffe2007:163). By early adolescence, his third molar had acquired a lower δ18Op value (+14.3‰), indicating his eventual move into Teotihuacan or, plausibly, Michoacán. If the latter is true, he may have arrived in Tlajinga as an adult, similar to 33-50b with whom he was buried.
Chronological observations of the δ18Op data show that the highest number of nonlocal individuals occur in the Tlamimilolpa sample (n = 5; Figure 4). As with the 87Sr/86Sr data, this finding may not be representative of both time periods, given that the Tlamimilolpa sample is much larger than the Xolalpan (n = 11 and 3, respectively). Here, the lapidary group is also considerably larger than the other two 14C phase groups (n = 10, 2, and 2, respectively). Nevertheless, it is noteworthy that many migrants fall along the transitional points between the Tlamimilolpa and Xolalpan ceramic phases and across the transitions between the three 14C phase groups (Figure 4).
Overall, we conclude that, of the 23 individuals in this study, ~45% (n = 10) were migrants from outside Teotihuacan (Burials 33-6, 33-14, 33-15, 33-25, 33-43, 33-44, 33-45, 33-50b, 33-50d, and 33-56; Table 2). This proportion is higher than reported by White, Storey, and colleagues (Reference White, Storey, Longstaffe and Spence2004), who estimated nonlocals as ~30% of the Tlajinga 33 population using a similar sample size. All migrants identified thus far come from Tlajinga 33, but the sample from Tlajinga 17 and 18 is too small to determine whether only local individuals resided there.
Dual-Isotope Approach
Figure 5 illustrates the covariation of tooth enamel Sr- and O-isotope compositions for the nine individuals for whom both isotopic systems were analyzed. We used hierarchical cluster analysis to determine significant groupings among individuals and to identify potential in-migration patterns from similar points of origin. This analysis showed that the optimal number of clusters in this sample set was only two (Figure 6) and that the two-cluster model produced the most significant results (Supplemental Table 2a).
Biplot of 87Sr/86Sr and δ18Op results. The sample ID is provided next to data points, and lines connect early and late childhood tooth enamel samples from the same individual.
Hierarchical cluster analysis results. The dendrogram displays the average linkage between groups based on Sr and O isotope parameters for all individuals undergoing both analytical procedures (n = 9).
The first cluster comprises the majority of individuals analyzed for both Sr and O isotopic compositions (n = 6), with an 87Sr/86Sr range of 0.70487 to 0.70523 (mean = 0.70503 ± 0.00010, 1σ) and a δ18Op range of +14.3 to +15.9‰ (mean = +15.2 ± 0.6‰, 1σ). This cluster contains those individuals with isotopic signatures significantly similar to those in the Teotihuacan Valley. However, this result does not indicate that all individuals in the first cluster are those native to Teotihuacan, but rather that they are from isotopically similar regions. For example, individuals from the greater Basin of Mexico but outside the Teotihuacan Valley would be included in this grouping (see Pacheco-Forés et al. Reference Pacheco-Forés, Gordon and Knudson2020).
The second cluster includes samples from three individuals with an 87Sr/86Sr range of 0.70513 to 0.70570 (mean = 0.70544 ± 0.00029, 1σ) and a δ18Op range of +15.0 to +18.1‰ (mean = +16.5 ± 1.6‰, 1σ). Rather than a true “cluster” of isotopically similar individuals, this second group consists of distinct entities, all of which are the farthest-outlying isotopic signatures in the sample set compared to the local Teotihuacan baselines. For example, Burial 33-14 has the highest δ18Op value (+18.1‰) but only slightly higher-than-local 87Sr/86Sr ratios (0.70530, 0.70513). As discussed earlier, the 87Sr/86Sr ratios for 33-14 suggest origins from within the central Highlands. However, the δ18Op value is too high for this region, and the combined isotopic compositions best match a possible origin in the southern Highlands near Kaminaljuyu (White et al. Reference White, Douglas Price and Longstaffe2007:163). In contrast, Burial 33-43 has the highest 87Sr/86Sr ratios (0.70565, 0.70570) but is only slightly enriched in 18O relative to the local baseline (δ18Op = +16.4‰), making it unclear where he was born. For Burial 33-15, we see a significant difference between early and late childhood isotopic compositions. Although the EC 87Sr/86Sr ratio lies outside the local baseline (0.70550), the δ18Op value is within the local range of variation (+15.0‰). The LCA 87Sr/86Sr ratio then decreases to a local value (0.70498), thus indicating a significant change in the geological provenance of the food resources available to her; this ratio falls within Cluster 1. Likely, this individual was born within the central Highlands and moved into Teotihuacan by adolescence.
When attempting to connect migrants to their region of origin in Mesoamerica, a complicating factor is that multiple regions can present similar isotopic compositions. This is particularly true for Teotihuacan and Michoacán, which have overlapping 87Sr/86Sr and δ18Op baselines. These similar isotopic signatures complicate the interpretation of isotopic data given that material evidence of migration from west Mexico is prevalent across all Tlajinga compounds. Along with the shaft tomb at Tlajinga 33, clay figurines with stylistic attributes from west Mexico were found at Tlajinga 17 and 18 (Carballo et al. Reference Carballo, Hirth, Sariñana, Buckley, Ramón and Kennett2019:Figure 14j). Furthermore, there is a need to expand the mapping of isotopic compositions across all of Mesoamerica; the baselines currently available are few and far between (see Hodell et al. Reference Hodell, Quinn, Brenner and Kamenov2004; Pacheco-Forés et al. Reference Pacheco-Forés, Gordon and Knudson2020). Therefore, we exercise caution when attempting to pinpoint origins for outlier individuals based solely on isotopic data and without multiple lines of evidence.
Patterns of Migration among Demographic Groups
This study tested for significant differences between residential and demographic groups to identify specific groups of people who migrated to Tlajinga. As stated earlier, these groups are divided according to time phases (ceramic and 14C dating), compound, status, and sex. Obtaining meaningful results is challenging in a study with such a small sample size. Therefore, we assess these results with caution, and in this section, we discuss only those tests with significant differences between group means (p < 0.05). Only the tests involving 87Sr/86Sr group means are significant. Complete results for all statistical tests are provided in Supplemental Tables 2b–e.
For all tooth enamel samples, there is no difference in the mean 87Sr/86Sr ratio of 23 EC and 7 LCA enamel specimens (Mann-Whitney U; W = 48, p = 0.12), nor is there a difference in the mean δ18Op signatures of 12 EC and 9 LCA samples (Student t-test; t = 0.014, df = 19, p = 0.99). Therefore, data from both developmental phases can be considered together for statistical purposes. However, we provide statistical analysis of demographic groups and tooth development phases in Supplemental Table 2b.
We tested for differences in 87Sr/86Sr group means between Tlajinga 33 (n = 24) and Tlajinga 17 and 18 (combined due to small sample size, n = 6). The δ18Op sample size is too small for statistical analysis. Tlajinga 17 and 18 have a narrow range of 87Sr/86Sr ratios compared to the broader range for Tlajinga 33, although this is likely due to the small sample size (Figure 7a). A Mann-Whitney U test indicates that 87Sr/86Sr group means are significantly different (W = 118, p = 0.02; see Supplemental Table 2c): Tlajinga 33 has a higher 87Sr/86Sr mean than Tlajinga 17 and 18 by 0.00032 on average. These data suggest that residents of Tlajinga 33 were eating foods with slightly higher 87Sr/86Sr ratios.
(a) The distribution of 87Sr/86Sr ratios by compound (left) and by sex and compound (right). “Tla” = Tlajinga. (b) The distribution of 87Sr/86Sr ratios by 14C date phases for all samples (left) and by 14C date phase and status (right).
A previous examination of diet using stable carbon isotopes found that residents of Tlajinga 33 were slightly more reliant on a maize-based diet than those at Compounds 17 and 18 (Storey et al. Reference Storey, Buckley and Kennett2019:153−154). Mano and metate fragments were abundant at Tlajinga 33 (Widmer Reference Widmer, de Tapia and Rattray1987) but less so at Tlajinga 17 and 18. Tlajinga 33 residents may have processed their maize for consumption only within this compound, but perhaps the other two residential groups obtained this staple crop from a different source, explaining the discrepancy in both strontium and carbon isotope compositions.
Across compounds, there are also significant differences between the 87Sr/86Sr means of Tlajinga 33 males (0.70516), Tlajinga 17–18 males (0.70488), and Tlajinga 17–18 females (0.70487; Supplemental Tables 2d-e). However, Tlajinga 33 females (n = 7), many of whom have nonlocal 87Sr/86Sr ratios, have a broader range of signatures than those measured for males (n = 11; Figure 7a). For example, 57% of women in this study were migrants, based on their 87Sr/86Sr and δ18Op signatures (Table 2). Male migrants, by comparison, constitute 46% of the total male population.
Of the females with nonlocal isotopic compositions, all are also of higher status (n = 4; Burials 33-6, 33-14, 33-15, and 33-25). These data may indicate that women of prestige migrated into Tlajinga after tooth development was complete, perhaps for marriage or childbearing. Returning to the original hypothesis that poor health conditions would have required a constant influx of migrants, it is plausible that women were recruited from outside the city for reproductive purposes and to maintain the population over time. Given that this would have been an immense responsibility, these women may have been revered after death, as demonstrated by the more elaborate burial offerings present in the archaeological record. More pragmatically, migrant women probably had larger families than local women because of the malnutrition common among native residents. The offspring of nonlocal women represented a larger portion of the population in subsequent generations and, thereby, a larger number of individuals contributing to the compound economy. Perhaps, the sheer number of offspring memorializing the graves of nonlocal women is the reason why these mortuary remains appear to be of higher status than those of local women.
A similar pattern of female paleomobility has been noted in the Oaxaca Barrio, a Teotihuacan enclave of Zapotec immigrants. Here, the δ18Op of bone and tooth enamel suggests that women were sojourning from various diasporic nodes while breastfeeding (White, Spence, et al. Reference White, Spence, Longstaffe and Law2004); it is hypothesized that this movement was essential to the sociopolitical construction and maintenance of Zapotec identity among new generations while living outside the Oaxaca homeland.
It is difficult to make comparisons between children in these two neighborhoods, because the two young individuals from the present Tlajinga sample, 33-21a and 33-59a, are not connected to any of the female burials. However, the 87Sr/86Sr ratios of the Tlajinga children do not suggest movement in early life but instead support local birth. These observations represent a significant social distinction between Tlajinga and the Oaxaca Barrio. Although it is clear that the people of the Oaxaca Barrio were motivated to retain their ethnic identity while living in Teotihuacan, the migrants of Tlajinga appear to have assimilated almost entirely to the local social and political norms. This conformation to Teotihuacan identity was perhaps driven by the large-scale participation of Tlajinga residents in crafting and manufacturing local goods for distribution across the city. Storey (Reference Storey1992) also suggests that singular individuals or small families migrated into Tlajinga, whereas the Oaxaca Barrio likely experienced migration en masse. Therefore, it would have been much easier for the people of Tlajinga to assimilate to local Teotihuacan customs rather than hold onto practices from their homelands.
Time phases by ceramic chronological groups (Tlamimilolpa and Xolalpan) show no significance between either 87Sr/86Sr or δ18Op group means. Nevertheless, time phases classified by 14C date ranges do have significant differences (Supplemental Tables 2c and 2e): the lapidary phase mean 87Sr/86Sr ratio (0.70521; n = 11) is significantly different from the transition phase mean (0.70483; n = 14). However, there is no significant difference between the San Martín phase mean (0.70503) and the lapidary or transition means. These results likely stem from the fact that the San Martín phase consists of only three samples. Results confirm earlier observations that individuals with 87Sr/86Sr ratios significantly higher than the local Teotihuacan baseline are most common at Tlajinga during the first phase of compound occupation (Figure 7b).
For mean 87Sr/86Sr, the lapidary and transition phases are also significantly different between status groups and sex and within the Tlajinga 33 compound (Supplemental Tables 2d–e). The sample size for Tlajinga 17–18 was too small for comparison (n = 4). The mean 87Sr/86Sr ratio (0.70510, n = 9) for all higher-status interments in the lapidary phase is significantly higher than the transition-phase mean (0.70472, n = 5; Figure 7b), although this is mostly a reflection of Burial 33-25's low ratios. Additionally, females (0.70515, n = 5) and males (0.70526, n = 6) in the lapidary phase also have higher mean 87Sr/86Sr ratios than transition-phase females (0.70455, n = 4), also due to Burial 33-25.
Combing the 87Sr/86Sr and δ18Op data, we see that during the Tlamimilolpa phase, 39% of the sample population were migrants compared to 9% during the Xolalpan (Table 2). The same is true for the 14C date phases. During the lapidary phase, 35% of the sample population have nonlocal isotopic signatures, but only 4% are nonlocal during the transition phase and 9% during the San Martín phase. Therefore, we conclude that migration was highest during the first centuries of compound occupation. However, migration did not end after this point but continued at a lower rate. Additionally, ~65% of higher-status individuals were migrants compared to only 25% of lower-status people, although mean 87Sr/86Sr ratios and δ18Op values were not significantly different between the groups.
Nixtamalization and Strontium Isotopes
Although our confidence in the isotopic results presented here is high, it is important to note an anomaly in the Tlajinga Sr-isotope data. The range of 87Sr/86Sr ratios for the Teotihuacan Valley lies between 0.7043 and 0.7050, yet most of the samples used to define this baseline are between 0.7043 and 0.7048, with only one sample having a higher value (see Supplemental Table 1). Therefore, the Sr-isotope signatures of tooth enamel from local individuals, which vary between 0.7048 and 0.7050, plot above the majority of the samples defining the baseline (Figure 8). Instead, we would expect to see the 87Sr/86Sr ratios distributed about the entire local baseline from 0.7043 to 0.7050. These data, however, suggest that some strontium in the diet comes from an additional source with different 87Sr/86Sr ratios. In Figure 8, we also see that local individuals have higher Sr concentrations (ppm) than migrants. Depending on the identification of a clear mixing line (Montgomery et al. Reference Montgomery, Evans, Chenery, Pashley and Killgrove2010), higher Sr ppm will potentially support an additional source of strontium in the diet. A possible explanation for this discrepancy could be the incorporation of strontium into the local diet via nixtamalization.
Measurements of 87Sr/86Sr ratios and Sr concentrations (1000/Sr) of human enamel from Tlajinga differentiated by early childhood measurements for local (asterisks) and migrant (squares) individuals as determined by this study. The horizontal lines represent the range of variation representative for the 87Sr/86Sr ratios in the local environmental. The dashed line represents the upper value for the local range when the anomalous water sample measured for this study is removed. Inset: Boxplot of all samples in this study used to develop the 87Sr/86Sr baseline range for the Teotihuacan Valley.
Nixtamalization—the process of adding an alkali solution of lime or wood ash to maize to enhance the nutritional value of this staple dish—was practiced across Mesoamerica (Katz et al. Reference Katz, Hediger and Valleroy1974). Wright (Reference Wright2012:339) suggested that alkaline processing using lime be considered the primary source of strontium in human remains from the Maya region, because at least 50% of the diet was maize. Additionally, the impact of the alkali solution used for nixtamalization on the 87Sr/86Sr signatures of maize consumers in the Classic period Soconusco region has also been suggested (Renson et al. Reference Renson, Navarro-Castillo, Cucina, Culleton, Kennett and Neff2019). As discussed earlier, the stable carbon isotope ratios obtained for Tlajinga individuals show that the diet here was maize-dominant (Storey et al. Reference Storey, Buckley and Kennett2019; White, Storey, et al. Reference White, Storey, Longstaffe and Spence2004). There is also good evidence from Teotihuacan that lime was primarily used for maize preparation (Biskowski Reference Biskowski2017).
Lime can be derived from limestone or shells. Although no limestone was locally accessible in the Teotihuacan Valley, 12 million square meters of architectural surfaces across the city were covered with mortar and plaster made from a lime solution (Barba and Frunz Reference Barba and Frunz1999). Biskowski (Reference Biskowski2017:32) states that the amount of lime required for nixtamalization would have been negligible compared to that needed for construction. Therefore, the limestone source used at Teotihuacan for nixtamalization could derive from the same limestone source used for construction materials.
The closest limestone source was in the Zumpago region in the Basin of Mexico, approximately 40 km northwest of Teotihuacan. Nevertheless, no evidence thus far suggests Teotihuacan was importing from this location (Murakami et al. Reference Murakami, Hodgins and Simon2013). Barba and colleagues (Reference Barba, Blancas, Manzanilla, Ortiz, Barca, Crisci, Miriello and Pecci2009) analyzed lime plaster from Teopancazco, a Teotihuacan apartment compound, and found that it came from the Tula region, 60 km northwest of Teotihuacan in the state of Hidalgo. Transportation from Chingú to Teotihuacan would take one day on foot (Carballo Reference Carballo, Hirth and Pillsbury2013:Figures 5.8 and 5.9), a reasonable distance for a population with no wheel technology or beasts of burden. Additional research by Murakami and colleagues (Reference Murakami, Hodgins and Simon2013) suggests that in the later years of Teotihuacan occupation the Tepeaca region in the state of Puebla, nearly 150 km to the east of Teotihuacan, was another source of limestone for the city.
Limestone from both of these sources was formed during the Cretaceous and Jurassic periods and have expected Sr-isotope ratios between ~ 0.7068 and ~ 0.7078, derived from the 87Sr/86Sr evolution curve for seawater through geological time (Richter et al. Reference Richter, Rowley and Paolo1992). This range is much higher than our estimated 87Sr/86Sr ratios for the Teotihuacan Valley (0.7043–0.7050), which is surrounded by Cenozoic volcanic rocks. Therefore, we hypothesize that the addition of an alkali solution made from any of these sources of limestone to the Teotihuacano diet could have caused an upward shift in the measured Sr-ratios.
Further research and comparisons to other populations within and outside of Teotihuacan need to be conducted to confirm this hypothesis. Given that nixtamalization was a common practice across Mesoamerica, it may also be a factor in the measured 87Sr/86Sr of migrants. Unless migrant individuals were ingesting lime from geological sources local to their diets, their 87Sr/86Sr signatures are also likely skewed.
Additional discussion of nixtamalization and of its possible effect on the interpretation of Sr-isotope ratios is beyond the scope of this article. That said, further investigation of this hypothesis at Teotihuacan and other sites in Mesoamerica is an important avenue for future Sr–isotope studies.
Conclusions
Applying a dual-isotope approach (87Sr/86Sr and δ18Op) to human tooth enamel from three apartment compounds in the Tlajinga district, we find that migrants comprised ~45% of the population. Our results bolster the original study of White, Storey, and colleagues (Reference White, Storey, Longstaffe and Spence2004) that used δ18Op values only. We also refine the range of local 87Sr/86Sr ratios (0.7043–0.7050) for the Teotihuacan Valley by analyzing local faunal and water samples. The 87Sr/86Sr and δ18Op data, along with mortuary and ceramic evidence, suggest that in-migration from Michoacán in west Mexico likely occurred at Tlajinga. However, these isotopic compositions are challenging to discern from Teotihuacan, because both isotopic systems overlap considerably between regions.
Previously reported radiocarbon dates and mortuary data reveal that in-migration was highest during the first centuries of district occupation, although the sample size for later occupation at Tlajinga is small. However, in-migration occurred throughout the history of Tlajinga and was a significant demographic process throughout neighborhood occupation.
Our results support an earlier hypothesis that higher-status individuals were inclined to have nonlocal isotope signatures; more than twice as many have nonlocal isotopic ratios than those of lower status. Additionally, the range of isotopic compositions of Tlajinga 33 females is considerably broader than in males, indicating possible recruitment of women to this compound for marriage and childbearing.
Mean 87Sr/86Sr ratios are significantly higher in Tlajinga 33 than in Tlajinga 17 and 18. We hypothesize that the importation of lime from outside Teotihuacan for nixtamalization of maize may have skewed the local human 87Sr/86Sr signatures. A higher proportion of maize consumption in the diet among Tlajinga 33 residents—consistent with previous research—could have resulted in greater exposure to imported lime, thus causing the higher 87Sr/86Sr ratios found among local individuals. Further research exploring this hypothesis is needed across Teotihuacan neighborhoods and throughout Mesoamerica.
Acknowledgments
Material in this article was based on work supported by the U.S. National Science Foundation (NSF) Dissertation Improvement Grant (BCS-1927690) (GMB, KGH), the Proyecto Arqueológico Tlajinga Teotihuacan (PATT) funded by the NSF (BCS-13212447) (DMC, KGH), and a Junior Fellowship at Dumbarton Oaks Research Library and Collection (GMB). The oxygen isotope analyses were supported in part by the Canada Research Chairs Program (FJL). Permission for destructive analysis of human remains was provided by the Instituto Nacional de Antropología e Historia in Mexico City. The Archaeometry Laboratory at University of Missouri Research Reactor is supported by the NSF (grant #BCS-1912776), and a previous NSF grant (#BCS-1621158) provided Gina Buckley a Graduate Internship (Virginie Renson). We are grateful for the invaluable comments from all reviewers that improved this article. Additional thanks go to Andrés Mejía Ramón for collecting water samples, Maria Codlin for fauna identification, and Grace Yau (LSIS) for assistance with the phosphate oxygen isotope analyses. This is Western's Laboratory for Stable Isotope Science Contribution #380.
Data Availability Statement
All data presented in this article can be obtained from ScholarSphere (Penn State), https://doi.org/10.26207/483d-ve56.
Supplemental Materials
For supplemental material accompanying this article, visit https://doi.org/10.1017/laq.2021.20.
Supplemental Text 1. R Markdown Tlajinga Hierarchical Cluster Code.
Supplemental Text 2. Strontium and Oxygen Isotope Methodology.
Supplemental Table 1. Teotihuacan Valley 87Sr/86Sr Ratio Baseline Measurements.
Supplemental Tables 2a–e. Statistical Results of Hierarchal Cluster Analysis and All Significance Tests.
Open access
