Migration and maize in the Virú Valley: Understanding life histories through multi-tissue carbon, nitrogen, sulfur, and strontium isotope analyses

Objectives: Stable isotope analysis can provide crucial insight into the function and development of early state-level societies on the north coast of Peru. Materials and Methods: Multi-tissue (bone collagen, tooth enamel, hair, nail, skin, and tendon) stable isotope analyses (carbon, nitrogen, sulfur, and strontium) were conducted for 13 individuals from the lower Virú Valley. Results: Non-seasonal changes in a predominantly C 4 -based terrestrial diet, with minimal inputs of marine foods were identified. One individual (Burial 5), however, had a stable isotope signature unlike any previously found on the north coast of Peru, indicating both a large contribution of C 3 -terrestrial resources to their diet and an 87 Sr/ 86 Sr value suggestive of highland residence during childhood. Discussion: This research provides the first strong stable isotope evidence of a highland individual within a coastal burial in northern Peru, new insight into the ritual killing event at Huaca Santa Clara during the late middle horizon and supporting evidence of the importance of C 4 terrestrial resources to the developing Virú polity during the early intermediate period.

The Virú Valley has a long history of research focused on the development of early states (Millaire, 2010). The sites of Huaca Gallinazo and Huaca Santa Clara were important centers of the Virú polity during the EIP as the polity's capital and a regional administrative center, respectively ( Figure 1). In addition to EIP burials at both sites, a unique ritual killing event exclusively involving young humans and camelids was performed at Huaca Santa Clara during the late middle horizon (LMH) (600-1000 CE) long after the site's abandonment (Millaire, 2015).
Understanding the dietary practices and potential places of childhood residence for these individuals can provide insight into the functioning of the Virú polity as an early state society, particularly with respect to interactions between coastal and highland groups during the LMH.

| Geography and cultural history
The Andean Cordillera is a vast mountain range that produces distinct ecological zones at different altitudes that span from the western coastal deserts to the eastern Amazonian jungles (Pulgar Vidal, 1996).
The north coast of Peru falls under the rain shadow produced by the Andean Cordillera and the cold Humboldt current which results in almost no rainfall along the coast and excellent preservation of organic remains. The Humboldt current supports one of the world's most productive coastal ecosystems and marine resources were likely important for the development of the earliest societies along the coast (Moseley, 1975;Moseley, 1992). Complex cultures in this region resided in fertile river valleys fed by glacial meltwater and seasonal rainfall from the highlands. Distinct archaeological cultures developed within many of these river valleys, however, inter-valley trading networks resulted in a larger sphere of influence that produced unified north coast ideologies and worldviews (Burger, 2008;Jones, 2010).
By the end of the EIP exponential population growth ceased and the first archaic states had appeared in northern Peru (Millaire, 2010;Topic, 1982). These early states were characterized by tiered political structures, settlement hierarchies, and outposts in other valleys (Lau, 2004;Millaire, 2010;Strong, 1957;Tello, 1929Tello, , 1940Topic, 1977Topic, , 1982. It was during the EIP that the Virú Valley's population peaked, irrigation networks were expanded to support farming, and the Virú Polity established control over the valley's resources (Millaire, 2010;Millaire et al., 2016). In contrast, during the MH the old states lost their influence as new cultures, such as Wari and Tiwanaku, rose to prominence (Isbell, 2008;Shimada et al., 1991). The strong connections with the Moche state declined in the Virú Valley as the influence of the Moche state waned (Bawden, 2008;Castillo & Rengifo, 2008;Chapdelaine, 2010). The rise of prominent highland states during the LMH and their influence on the north coast has been suggested in the Virú Valley by the increased frequency of "highland" cranial modifications and ceramic styles (Dillon, 2015).

| Stable isotope analysis
Stable carbon (δ 13 C), nitrogen (δ 15 N), and sulfur (δ 34 S) isotope analyses provide information about the types of foods people choose to consume. δ 13 C values best distinguish the contributions of C 3 and C 4 plants.
In the Andes, δ 13 C values are especially useful for identifying the consumption of C 4 crops (particularly maize but also amaranth in contexts where this crop was economically important) compared to wild and cultivated C 3 plants such as legumes, fruits, peppers, and gourds (Szpak et al., 2013;Tieszen et al., 1992). δ 13 C values also tend to be higher in marine animals relative to terrestrial animals that consume C 3 plants (Chisholm et al., 1982). δ 15 N values increase with trophic level (Minagawa & Wada, 1984) and can be useful for distinguishing diets F I G U R E 1 Map of the Virú Valley, Peru, and the sites of Huaca Gallinazo and Huaca Santa Clara predominantly based on plants from diets that include meat and secondary animal products. While δ 13 C and δ 15 N values can be used in combination to estimate the contribution of marine foods (high δ 13 C and δ 15 N values) (Hansen et al., 2012;Schoeninger & DeNiro, 1984), δ 34 S values can further distinguish terrestrial and marine-based diets (Richards et al., 2003;Tieszen et al., 1992). Most marine animals have uniformly high δ 34 S values of +15-20 ‰ whereas plants and animals from terrestrial and freshwater ecosystems tend to have much lower δ 34 S values around $0 to 10 ‰ (Nehlich, 2015).
Strontium isotope ratios may also provide some indication of the contribution of marine-based diets because the 87 Sr/ 86 Sr of the ocean (0.7092) (Elderfield, 1986;Veizer, 1989) can differ greatly from the 87 Sr/ 86 Sr value of the terrestrial environment (Sealy et al., 1991). The geological history of the Andes has resulted in longitudinal bands of bedrock with distinct 87 Sr/ 86 Sr values (Knudson et al., 2014;Scaffidi & Knudson, 2020) that can distinguish non-local individuals who lived at varying distances inland.
This method is, however, less suited to distinguish north-south migrations between river valleys.

| Multi-tissue approach
The preservation of a wide range of human tissue in the arid environment of the north coast of Peru allows for the creation of a longitudinal data set that can characterize how individuals' diets changed over their lifetimes. As tissues grow and remodel they incorporate the isotopic signature of their diet and by analyzing tissues with different formation times and turnover rates, the changes in their diet or residency at different periods can be determined (Tieszen et al., 1983). Bone collagen grows and remodels quickly during childhood but then slowly continues to remodel throughout adulthood producing stable isotope values that represent long-term diet over years to decades, depending on the age of the individual (Hedges et al., 2007). Tendon and skin are also actively maintained but have a turnover rate that reflects only the last 2-3 months and 3.5-4 months respectively before death (Babraj et al., 2005;El-Harake et al., 1998;Tieszen et al., 1983). Human hair and nail are metabolically inert once formed, and their isotopic compositions reflect only the period of formation (Tieszen et al., 1983). Human hair and nail grow at approximately 1.0 cm and 2.1 mm per month (Saitoh, 1969;Yaemsiri et al., 2010) although an individual's tissue growth rate may be impacted by illness, malnutrition, and the indiscriminate sampling of hair in both active and inactive growth phases (Harkey, 1993;L. J. Williams et al., 2011). Tooth enamel is inert once formed and represents the diet and environment of early childhood only (Harris & Buck, 2002;Reid & Dean, 2006). The range of tissue types analyzed in this research is rare and has provided a wealth of information about the individuals buried in the Virú Valley.

| Materials
The carbon, nitrogen, sulfur, and strontium isotopic compositions of the tissues of 13 individuals were analyzed. Individuals from the sites of Huaca Gallinazo (n = 2) and Huaca Santa Clara (n = 11), dating to the EIP (n = 7) and LMH (n = 6) ( Table 1) were included in this analysis.
T A B L E 1 Age and sex estimation, and burial context for individuals analyzed in this research (Dillon, 2015)  Many of the EIP burials from Huaca Santa Clara are principal and retainer pairings, while all LMH burials from Huaca Santa Clara are children associated with a single ritual killing event that also included juvenile camelids (Dillon, 2015;Millaire, 2015). Additional information on the samples is available in Data S1, Table S1.

| Sample preparation
Bone collagen was extracted as described previously (Szpak et al., 2014). The skin and tendon samples (n = 9) were prepared for stable carbon and nitrogen isotope analysis following Finucane (2007).
Approximately 100 mg of skin or tendon were extracted from the available remains using a scalpel cleaned in acetone. External debris was removed from the samples by sonication in 10 mL of Type I water Hair was pulled in line with the hair roots using acetone-cleaned tweezers and placed in a clean surface of aluminum foil. The straightened hair was aligned, wrapped in aluminum foil, and 1 cm increments were marked and cut along the foil wrapping. Adhering dirt was removed by sonication in 10 mL of Type I water for 30 min. Lipids were removed with two rinses in 8 mL of 2:1 chloroform: methanol (v/v) solution sonicated for 30 min. Two rinses of 10 mL Type I water with ultrasonication for 30 min removed any remaining 2:1 chloroform: methanol (v/v) solution. The final Type I water rinse was decanted, and samples were dried overnight at 50 C. The hair samples were minced using an acetone-cleaned razor blade and weighed into tin capsules for stable carbon and nitrogen isotope analysis. A subset of hair samples was also analyzed for δ 34 S and for these, vanadium pentoxide was added to $1 mg of hair in tin capsules.
Tooth enamel was prepared based on the techniques of S. Ambrose et al. (2018). External dirt was removed from the tooth by sonication in 10 mL of 0.1 M acetic acid for 2 × 20 min. Teeth were then cleaned of visible calculus/plaque using a dental drill. The tooth crown was separated from the root and the dentine within the crown was removed from large enamel pieces. The enamel pieces were powdered using an industrial mortar and pestle and 150-200 mg of enamel powder was weighed into perfluoroalkoxy alkane (PFA) vials.
Additional technical information concerning the operation and conditions of the analytical equipment is available in Data S1.

| Data treatment
To make direct intra-individual comparisons between different tissue types all sample δ 13 C and δ 15 N values were adjusted to be directly comparable to bone collagen using the proposed inter-tissue isotopic  (Józsa et al., 1984;Kucharz, 1992) and are compositionally similar to that of bone collagen, so no inter-tissue isotopic adjustments were made (White & Schwarcz, 1994). A Kruskal-Wallis test was performed to compare the mean isotopic ratios among individuals and groups. To examine trends in changing stable isotope values over time non-parametric Spearman's ρ was performed.

| Sample preservation
The bone collagen isotopic compositions included in this analysis passed all quality control criteria (wt% collagen >1%, 3.13 < C: N atomic < 3.38, wt% C > 13.0%, wt% N > 4.5%) (  Ambrose, 1990;DeNiro, 1985;van Klinken, 1999). All hair segments had C:N atomic ratios between 3.36 and 3.95 (Table S2) which fit well with the range for modern human hair (2.9-3.8) determined by O'Connell and Hedges (1999). The range of C:N atomic for the nail samples was between 3.78 and 4.01 (Table S3)  Changes in C:N atomic , wt% C, and wt% N for soft tissue samples (skin and tendon) with each successive pre-treatment phase suggested that this procedure successfully removed some carbon-rich contaminants. For samples with initially high C:N atomic ratios (>5.00) after only the Type I water rinse, the gelatinization step further reduced the C:N atomic ratios by an average of −1.68 ± 1.51 (n = 4) (- Table S3). Four soft tissue samples had C:N atomic > 3.6 even after the complete preparation procedure, and these values were excluded from further analysis (Table S3). The skin and tendon samples with C: N atomic < 3.6 had the highest δ 15 N values of all tissues sampled (skin: +11.68 ± 0.88 ‰, n = 2; tendon: +11.74 ± 0.71 ‰, n = 4; nail: +9.99 ± 0.92 ‰, n = 4; bone collagen: +9.8 ± 0.9 ‰, n = 13; hair: +9.09 ± 1.35 ‰, n = 114). When tissues from Burial 7 were adjusted for inter-isotopic offsets, tendon, and skin δ 15 N values were higher than the nail and incremental hair δ 15 N values that corresponded with the 2-3 months (tendon) and 3.5-4 months (skin) before death

| Major components of diet
The δ 13 C values for both long-term (bone collagen) and short term (hair keratin, nail keratin, tendon, and skin) tissues were relatively high and support a diet predominantly based on C 4 plants (likely maize) for the burials at both Huaca Gallinazo (bone collagen: −11.74 ± 0.46 ‰, n = 2) and Huaca Santa Clara (bone collagen: −12.30 ± 2.06 ‰, n = 11) ( Table 2  values from incremental hair samples ranged from +1.7 to +6.9 ‰ (Table 3, Figures 2 and 4), further supporting the notion that marine resources were not a quantitatively important part of their diet. If marine food sources were a significant component of the diet of these individuals, much higher δ 15 N (c. +15 to +25 ‰) and δ 34 S (c. +20 ‰) values would be expected (e.g., King et al., 2018;Santana-Sagredo et al., 2015;Tieszen et al., 1992;Tomczak, 2003) (Table 3 and Figure 4). The strontium isotope results also support a diet with minimal input of marine foods during childhood. Marine diets alter tooth enamel 87 Sr/ 86 Sr signatures to approximate the oceanic 87 Sr/ 86 Sr value of 0.70987 (Elderfield, 1986;Veizer, 1989), however the 87 Sr/ 86 Sr ratios from Huaca Gallinazo and Huaca Santa Clara were all below the oceanic 87 Sr/ 86 Sr value and ranged from 0.70569 to 0.70868 (Table 4).
Human bone collagen, tendon, skin, nail and hair δ 13 C and δ 15 N values were adjusted for trophic enrichment and compared to the Virú Valley archaeological camelids (Szpak et al., 2014)

| Isotopic changes in the months before death
The isotopic compositions of the hair, nail, skin, and tendon suggest that the individuals analyzed in this study consumed a predominantly C 4 terrestrial plant and animal-based diet, consistent with the bone collagen data. After hair, nail, skin, and tendon isotopic compositions were normalized relative to bone collagen (adjusted for inter-tissue isotopic offsets) there were only slight differences between long-term and short-term diet (Figures 2, 5, and 6). To some extent, this variation may also be driven by the uncertainty in the inter-tissue offsets that were used in this study.
The isotopic compositions of the soft tissues were quite variable (δ 13 C max range: 6.74 ‰ for Burial 7 hair samples; δ 15 N max range: 3.85 ‰ for Burial 8 hair samples) and showed that the diet of these individuals changed within the months before their death (Figures 2   and 7 and Table S3). The change in incremental hair δ 13 C, δ 15 N, and δ 34 S values did not repeat over a yearly cycle (assuming $1 cm of scalp hair growth per month) nor was it patterned in any way to suggest that the processes causing these changes were seasonal  F I G U R E 4 Incremental hair δ 34 S, δ 13 C (a) and δ 15 N (b) values from Huaca Gallinazo and Huaca Santa Clara compared to the expected range for a marine-based diet: $+20 ‰ for δ 34 S, $−15 to −8 ‰ for δ 13 C (a), and $+15 to +25 ‰ for δ 15 N (b)   (Figure 9). Neither was there a consistent change in the short-term diet, represented by hair samples, relative to the long-term diets, represented by bone collagen (Figure 6).
There is, therefore, no isotopic evidence for a change in diet in the months before the ritual killing event. There may, however, have been some significant dietary changes that occurred for these individuals before death that were simply not detectable using these isotopic methods. Stable carbon and nitrogen isotope analysis can only distinguish changes in diet that relate to isotopically distinct food sources.
For example, if individuals began consuming maize beer (chica) instead of maize or higher quality cuts of camelid meat rather than low-quality cuts, this would not be visible isotopically.

| Isotopic evidence of non-locals
The strontium isotope data provide evidence for the inclusion of nonlocal individuals in the lower Virú Valley burials. Within this research the term "highland" refers to extremely high elevation zones above the suni/jalca (3500-4000 masl) while "coastal" is used in reference to the chala zone (0 to $500 masl) (Pulgar Vidal, 1996). Additionally, ethnicity is assumed to be innate based on the area of childhood resi- had "coastal" diets ( Figure 3 and Table 2). Burial 8 was identified as a potential young adult male warrior dating to the EIP and may have been born in the highlands based on his high 87 Sr/ 86 Sr ratio (Table 4 and Figure 10). He appears to have spent most of his adult life living on the coast or moving between different coastal areas, based on his isotopically coastal diet (Table 2 and Figures 3 and 8). Burial 11, however, was a younger individual (12 years ±21 months) and either similarly spent a short amount of their childhood in the highlands before permanently moving to the coast, or consumed a coastal diet throughout their life while living in the highlands. The rapid rate of collagen turnover during growth experienced by adolescents (Hedges et al., 2007) precludes the identification of one of these scenarios as more or less likely on the basis of the isotopic data.
Burials 4 and 6 had cranial modifications suggestive of highland origins ( Figure 11), but enamel 87 Sr/ 86 Sr values consistent with a childhood residence in the lower Valley and δ 13 C and δ 15 N values similar to a "coastal" diet (Table 2 and Figures 3 and 10). Burials 4 and in the LMH burials supports that coastal and highland groups were interacting during this period (L. Masur, 2012;Millaire, 2015). A child with highland cranial modifications (Burial 6) was selected as the principal burial (Dillon, 2015) and the retainer burials represented children from potentially intermediate, highland, and coastal populations. Juvenile llamas that were interpreted as having been raised locally on the coast accompanied these burials (Millaire, 2015;Szpak et al., 2014) and may have been selected to further establish the importance of coastal-highland cooperation in this ritual. Previous interpretations of this ritual killing event suggested that it may have been enacted at Huaca Santa Clara after the site's abandonment to establish and legitimize the power of the ruling elites over the population (Dillon, 2015;Millaire, 2015). 3.6 | Why fish when you could farm?
Contrary to expectations, all the individuals analyzed from the lower Virú Valley relied heavily, perhaps nearly exclusively, on terrestrial resources.
Archaeological sites along the coast cannot be assumed to have relied on marine resources (Carmichael et al., 2014;Falabella & Sanhueza, 2019), however the Humboldt current supports one of the world's most productive marine ecosystems (Chavez et al., 2008). So why were the people of the Virú Valley, who lived so close to this rich source of marine food, relying so heavily on agro-pastoral resources during the EIP?
The individuals analyzed in this study could represent a subset of the population that relied on terrestrial and agricultural resources to a greater extent than the population writ large. Agricultural products and camelid meat could have been high-status foods that were principally available at the larger centers of Huaca Gallinazo and Huaca Santa Clara, while marine resources were more significant to smaller commoners' settlements (Reitz, 1979). Fishing may also have been practiced as a distinct economic specialization which limited its availability to predominantly agricultural communities (e.g., Tomczak, 2003). Different skillsets, knowledge, and technologies are required for agriculture and fishing, which might have prevented certain individuals or groups from practicing both F I G U R E 8 Hair increment δ 13 C and δ 15 N over time for Burial 7 and 8. Time is months before death based on 1.0 cm per month hair growth F I G U R E 9 δ 13 C and δ 15 N over time for Burials 3, 4, and 10 from the ritual killing event at Huaca Santa Clara (LMH). Time is months before death based on 1.0 cm per month hair growth F I G U R E 1 0 Huaca Gallinazo and Huaca Santa Clara strontium values from tooth enamel compared to the strontium baseline model for the Andes produced by Scaffidi and Knudson (2020) successfully at larger scales (Moseley, 1992, p. 22). Ethnographic accounts from the north coast of Peru provide evidence that in the Late Horizon and post-contact periods, fishing and farming communities lived separately. Fisherfolk had a unique language distinct from the dialect of agriculturalists (Rabinowitz, 1983), members of the fisherfolk communities did not intermarry with agriculturalists, and fishing communities were governed by their own hereditary lords (Moseley, 1992:22;Netherly, 1977 (Millaire, 2010;Willey, 1953). The large investments required to maintain and expand irrigation networks may quickly have led to the dependency on agricultural food for large portions of the valley, especially once the population outgrew the carrying capacity of local marine resources.
The ability of local elites to control and monopolize agricultural resources may further explain why terrestrial resources were so important during the height of the Virú polity in the EIP. Land in the arid coastal valleys of the Andes requires irrigation and possibly fertilization to ensure its productivity (Billman, 2002;Masuda, 1985;Moseley, 1972) and the development and maintenance of irrigation systems in coastal valleys would have allowed subsets of the population to control this increased productivity (Moseley, 1972). For example, farmers of the southern Peruvian highlands went to great lengths to access offshore guano deposits to be used as fertilizer during early Colonial times, because without this important source of nutrients, their crops would not have produced adequate yields (Julien, 1985). In the Virú Valley, crop surpluses could be easily stored allowing those who controlled the irrigation systems to accumulate wealth and increase their power and influence (Lambert et al., 2012). Large storage rooms at the site of Huaca Santa Clara (Millaire, 2010) (Szpak & Chiou, 2019). The low δ 15 N values therefore suggest a greater emphasis on irrigation rather than manuring to increase agricultural productivity (Austin & Vitousek, 1998), but additional isotopic data from both humans and botanical remains must be collected to test this hypothesis.
The consistently high contribution of C 4 resources among individuals from both Huaca Gallinazo and Huaca Santa Clara lends further support to the characterization of the Virú polity as a true state-level society, as this represents an important system of redistribution (Millaire, 2010;Millaire et al., 2016). Local elites in the Moche Valley may have been similarly monopolizing agropastoral resources to establish their power and control during both the EIP and Middle Horizon as the early Moche state also intensified agricultural production during these periods (Billman, 2002;Lambert et al., 2012). This research supports the notion that the Virú polity represents one of the oldest functioning states along the north coast of Peru and highlights the importance of agricultural resources, especially maize, to the early state-level societies along the north coast of Peru (Millaire, 2010).
F I G U R E 1 1 Crania of Burials 4 and 6 with cranial modifications suggestive of highland origins for these individuals (Dillon, 2015) 4 | CONCLUSION Stable isotope analysis provided insight into the lives of 13 individuals as well as the functions of the societies in which they lived. The analysis of soft tissues and incremental hair provided a longitudinal dataset that generated insight beyond the scope of bone collagen as a marker of long-term average diet to reveal subtle changes in their diet through time. These individuals consumed a predominantly terrestrial diet based on C 4 plants and camelid meat despite their proximity to the productive coast. Yet they also experienced large changes in their lives that were observable isotopically, including shifts in diet, changes in health, and both short and long-term migrations. These results demonstrate how dynamic individuals' lives and mobility patterns were during the EIP. A combination of distinct farming and fishing communities, a developed dependency on agricultural resources, possibly due to population pressure, and the monopolization of crops by elites help explain why terrestrial resources were the staple foods of the early Virú Polity, which continued during the LMH. The strong highland influence in the LMH ritual killing event at Huaca Santa Clara highlights the need for further research into the interactions between coastal and highland populations and the role that those living in intermediate zones played as potential mediators of these interactions.
Continued use of stable isotope analysis within the Virú Valley and along the north coast of Peru will provide enhanced insight into this important region of the Andes.

ACKNOWLEDGMENTS
The authors would like to extend their gratitude to the following people for their contribution to this research project: Esuardo La Torre Calvera, Jeisen Navarro Vega, Amedeo Sghinolfi and the staff of the

CONFLICT OF INTEREST
The author declares that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supplementary material of this article.