A strontium isoscape for the Conchucos region of highland Peru and its application to Andean archaeology

Strontium isotope (87Sr/86Sr) analysis of human skeletal remains is an important method in archaeology to examine past human mobility and landscape use. 87Sr/86Sr signatures of a given location are largely determined by the underlying bedrock, and these geology specific isotope signatures are incorporated into skeletal tissue through food and water, often permitting the differentiation of local and non-local individuals in past human populations. This study presents the results of a systematic survey of modern flora and fauna (n = 100) from 14 locations to map the bioavailable 87Sr/86Sr signatures of the Conchucos region, an area where the extent of geologic variability was previously unknown. We illustrate the necessity to examine the variation in 87Sr/86Sr values of the different geological formations available to human land use to document the range of possible local 87Sr/86Sr values. Within the Conchucos region we found significant variation in environmental 87Sr/86Sr values (0.7078–0.7214). The resulting isoscape represents the largest regionally specific bioavailable 87Sr/86Sr map (3,840 km2) to date for the Andes, and will serve as a baseline for future archaeological studies of human mobility in this part of the Peruvian highlands.


Introduction
The study of mobility and migration are important topics in contemporary archaeology [1]. While human mobility can be studied using a variety of archaeological indicators, recent years have witnessed a marked increase in investigations that employ isotopic analyses of human remains to study ancient population movements [2][3][4][5][6][7][8][9][10][11]. Molecular methods enable researchers to focus primarily on the individual [12][13][14][15][16], and can elucidate aspects of human behavior such as mobility and landscape utilization that are otherwise difficult to observe [15]. Because strontium has the unique ability to substitute for calcium (Ca) in the hydroxyapatite of bone and tooth enamel, the use of strontium isotopes ( 87 Sr/ 86 Sr) analysis in skeletal remains can provide insight into past human and other animal movement throughout a landscape. When locally available nutrients are consumed, 87 Sr/ 86  bioavailable (i.e., only the strontium which makes its way into the food chain; see further discussion below) signature of the immediate geological location in which an individual lived. Nevertheless, the use of 87 Sr/ 86 Sr isotope analysis to identify non-local individuals and their potential place of origin, relies on an accurate characterization of local 87 Sr/ 86 Sr ranges, either through statistical spatial modeling or by testing modern/archaeological proxy materials to establish local baselines.
In the Andes, 87 Sr/ 86 Sr isotope analysis has been used to address a wide range of fundamentally important questions surrounding human life and interaction . Significant variation in the geology of the Andes makes the use of strontium isotopic analysis a useful tool in determining "local" vs "non-local" inhabitants of an archaeological site [15,45,53]. The Andes mountains are composed of many folded geological layers [54][55][56] that generally run in parallel from north to south and these geological formations can be relatively narrow and stacked close together. As a result, inhabitants of a specific archaeological site may have encountered (or frequented) multiple geological formations, thus making the identification of potential migrants more challenging. For example, the physical location of an archaeological site may not have been located in the same geological formation in which food was cultivated or herding and hunting was conducted.
In regions where the extent of geologic variation and the range of 87 Sr/ 86 Sr values are known, 87 Sr/ 86 Sr analysis can be used to track mobility and illuminate processes of interaction. In Peru, the majority of 87 Sr/ 86 Sr isotope studies have been situated along the Pacific coast, in the southern Andes and/or west of the Cordillera Blanca. The limited number of 87 Sr/ 86 Sr studies within the highland valley systems of the north central Andes have resulted in an underestimation of the geologic complexity throughout the region. Here we add to the growing body of 87 Sr/ 86 Sr studies in the Andes, and present the first regional map of the variation in bioavailable 87 Sr/ 86 Sr values of the Conchucos region of highland Ancash (Fig 1), a region with a rich archaeological history [57]. Our study also raises questions related to what may constitute isotope-based determinations of local and non-local populations in Andean archaeology.

Strontium geochemistry
The trace element strontium (Sr) is found in extremely low concentrations in bedrock, groundwater, soil, plants, and animals. Sr is composed of different percentages of the following four stable isotopes: 84 Sr (~0.56%), 86 Sr (~9.87%), 87 Sr (~7.04%) and 88 Sr (~82.53%) [74,75]. Of these four isotopes, 87 Sr is radiogenic and formed over time by the radioactive decay of rubidium ( 87 Rb) in the bedrock, which has a half-life of~4.88 x 10 10 years. As a result, specific concentrations of 87 Sr in the environment are a result of a bedrock's age and Rb content [58][59][60]. Sr enters the biosphere through uptake from the substrate by plants and cycles through food webs, into for example, the tissues of both animals and humans.
However, not all Sr in bedrock is uniformly weathered into the soil and water [15,61,62]. Various minerals found within a single bedrock unit can have considerable variability in their 87 Sr/ 86 Sr values. For example, granite can have two feldspars with radically differing 87 Sr/ 86 Sr values (plagioclase and potassium feldspars) depending on which section is measured [15]. As such, biologically available 87 Sr/ 86 Sr, which is soluble and is taken up by biotic agents, can substantially differ in its values between the lithosphere and the biosphere [12,15,63,64]. As a result, direct bedrock 87 Sr/ 86 Sr measurements typically conducted for geological dating studies [65][66][67][68][69][70] are not necessarily accurate for applications in archaeological science. Besides Sr deriving from the weathering of local bedrock, atmospheric and surface sources, such as rainfall, rivers, sea-spray, and wind-dust, also contribute to the bioavailable Sr in the food chain [15,59,[71][72][73][74]. Modern anthropogenic Sr contaminations can be introduced through industrial fertilizers and even via dust from large scale construction sites [15,59,71,75].
As organisms consume locally available food and water, these sources of Sr are mixed and incorporated into the organism's tissue [12,42,63]. In contrast to many commonly utilized light isotope systems, the isotopic composition of Sr does not change or fractionate during biological processes [63]. This is because the mass differences between the four Sr isotopes are relatively small [13,62,63,76]. As a result, the 87 Sr/ 86 Sr values measured in flora and fauna vary mainly based on the age of the bedrock on which they sourced their nutrients. Very old bedrock with high Rb/Sr ratios will have the highest 87 Sr/ 86 Sr values today [58,77]. Examples of geological deposits that have relatively high Rb/Sr ratios include clay-rich rocks such as shale, or igneous rocks that have high silica content, such as granite, with 87  PLOS ONE [13]. In contrast, geologically young rocks and sediments will have low Rb/Sr ratios and typically have 87 Sr/ 86 Sr less than 0.706 [e.g., 78]. To infer the biologically available 87 Sr/ 86 Sr values in an area, recent studies commonly use samples of uncontaminated environmental sources of local origin such as plants and small animals [59,75,79,80], water [72,81,82] and soil [59] samples.
As 87 Sr/ 86 Sr isotope analysis has been applied to address a wide range of archaeologically significant questions, methods for determining local 87 Sr/ 86 Sr ranges in the environment also continue to improve. Originally, researchers determined the local range of 87 Sr/ 86 Sr values as a two-standard deviation (±2σ) range around the average 87 Sr/ 86 Sr value measured in all archaeological samples from a site, characterizing outliers as non-local individuals [15,16]. This tends to produce a conservative estimate of non-locals in a population and may inadvertently underestimate the number of non-locals in a sample [15,42,98].
Given the potential challenges with defining local ranges based on mean calculations of the 87 Sr/ 86 Sr values of an ancient (and potentially highly mobile) human population, researchers now commonly sample local (both archaeological and modern) fauna and flora as proxies for locally bioavailable 87 Sr/ 86 Sr [5, 15-16, 20, 22, 32, 44, 98-101]. There are however, several considerations that should be made in sample selection [16]. In archaeological fauna, it is often unclear if animals were kept locally, remotely, or if they were subject to exchange. Depending on the source, modern domestic fauna may not reflect local 87 Sr/ 86 Sr values if they were fed imported, non-local foods, and/or if fodder was exposed to exogenic Sr though industrial fertilizers [15,16,59,62]. Animals, or animal products, purchased from local markets where their geographic origin and/or the origin of their fodder may be unclear, can make associating the obtained 87 Sr/ 86 Sr data to a specific geological formation with the necessary certainty difficult [15,31,32].
In recent years there has also been an increased effort to create large-scale isoscapes, a spatially explicit prediction of isotopic variation across landscapes [42, [102][103][104][105]]. An isoscape considers all published 87 Sr/ 86 Sr data for a given region and uses this dataset to extrapolate the extent of possible 87 Sr/ 86 Sr values across large geographic areas [71,75,[106][107][108][109][110][111]. While these studies provide invaluable insight into the nature of past mobility on a population-wide panregional scale, they are dependent on the amount and quality of data used to generate the isoscape.
There are many approaches to conducting 87 Sr/ 86 Sr isotope research and each of these methods have advantages and limitations depending on the research questions and resolution of the data. In this study we present a detailed regional mapping project that emphasizes the collection of environmental samples of biologically available 87 Sr/ 86 Sr, both within archaeological sites, as well as from the surrounding geological formations. This regional isoscape can then be applied to the study of human and animal mobility within the region.

Geology of the Peruvian Andes
The Central Andes are divided into the Cordillera Occidental to the west and the Cordillera Oriental to the east. The Cordillera Occidental is largely composed of late Cenozoic volcanic rocks such as andesites and Mesozoic formations. Age of the Cenozoic volcanic rock increases from the northern Andes to the southern Andes, and as a result the 87 Sr/ 86 Sr values are generally higher in the southern part of the Andes [78,112]. 87 Sr/ 86 Sr values reported from late Cenozoic volcanic rocks in Ecuador exhibit 87 Sr/ 86 Sr values of 0.70431±0.00016 (1σ, n = 23) [112], while exposed bedrock samples from similar geologic formations in northern Chile exhibit mean 87 Sr/ 86 Sr values of 0.70646 ±0.00020 (1σ, n = 8) [78]. The Cordillera Oriental in the east is mainly comprised of Paleozoic geology. These formations generally have higher 87 Sr/ 86 Sr values than the western Cordillera; however, their 87 Sr/ 86 Sr values have not yet been measured in bedrock [61,113]. In addition, on a broad pan-regional scale, 87

Material and study region
The study area consists of a broad swath of the eastern highlands of north-central Peru known as the Conchucos region. Conchucos is an intermontane valley system situated on the southeastern side of the Cordillera Blanca and is characterized by several rivers that drain into the Marañón River, one of the major tributaries of the Amazon. Our study focuses on sample collection over an area of 2,640 km 2 that includes the Huaritambo, Mosna/Puccha, and Marañón rivers. This region is archaeologically rich [e.g., 70], with archaeological sites dating from ca. 1100 BCE until the 16 th century [e.g., [114][115][116][117][118][119][120][121][122][123][124][125][126].
As illustrated in Fig 2, the Conchucos region is geologically diverse. The predominant geology comprises folded Mesozoic sedimentary rock formations, including sandstones, dark shales, and carbonates (limestone, marls, and dolomites), as well as metamorphic rocks like quartzite and slate [127][128][129]. The entire region is shaped by these folded and uplifted layers of bedrock that causes the repetition of specific geologic units over a broad area. This is important to consider when defining the categories of local and non-local populations in the archaeological record based on 87 Sr/ 86 Sr values, as similar geological units can be found throughout the landscape. Towards the north and east the study area is bordered by the geological Marañón Group. Dating to the Proterozoic, Marañón Group rocks are much older than the other formations and consist of meta-sedimentary schists, gneiss, and red sandstone [130].
To assess bioavailable 87 Sr/ 86 Sr values, we collected empty shells of modern terrestrial snails (Bulimulidae), as well as wild perennial grasses abundant in the Peruvian highlands (i.e. Stipa ichu) ( Table 1; Fig 2). Snail shells are plentiful on the landscape and make it unnecessary to obtain live animals. Snails are additionally limited in the extent of their movement throughout their lifetime and can therefore be considered representative of local variability in bioavailable 87 Sr/ 86 Sr [83,131,132]. Sr is deposited in the snail shell, where it substitutes for its main component Ca [133]. Plant 87 Sr/ 86 Sr values reflect the 87 Sr/ 86 Sr values in the immediate local soil, as well as 87 Sr/ 86 Sr admixture introduced by rainwater and atmospheric dust [131].
During field sampling, major geological formations in the region were identified using a geological map [129]. We obtained 100 modern environmental reference samples from 14 sampling sites in six geological units covering a 3,840 km 2 area of the Conchucos region. In each geological unit, we selected sampling locations where anthropogenic contamination through fertilizers or other pollutants were unlikely, as there were no signs of use through agriculture and there was considerable distance to roads and/or towns. At each location we collected snail shells from the surface alongside several samples of Stipa ichu (3-10 plants/unit).

Methods
Sample preparation was conducted in the Primate Ecology and Molecular Anthropology laboratory (PEMA) at the University of California at Santa Cruz (UCSC). Snail shells were repeatedly rinsed with ddH 2 O in an ultrasonic bath to remove any attached sediment. Snail shells were then broken into smaller fragments, placed in individual beakers with ultrapure acetone, rinsed in an ultrasonic bath for another 15 minutes to remove any potential contaminants on the shell surface and were then set to dry. Plant samples (2g of well-dried plant material) and snail shells (~300mg) were then ashed at 800˚C for 12 hours in a muffle furnace.

PLOS ONE
remaining sample, again re-dissolved in 5% HNO 3 was dip checked on the Thermo Finnigan Neptune™ MC-ICP-MS instrument to check the concentration of Sr in each sample. Any sample that had a v 88 SR value above 40ppm was diluted down to~40ppm (v 88 SR). Samples were then measured parallel to the SRM 987 standard, procedural blanks (one/every batch of 9 samples), as well as one clean acid blank after every 5 samples, in a Thermo Finnigan Neptune™ MC-ICP-MS.

Results
We measured 87 Fig 2). Mean 87 Sr/ 86 Sr values for each sampling location as well as more detailed information on each geological unit are presented in Table 1.

Discussion
The use of 87 Sr/ 86 Sr isoscapes: What does it mean to be a "local"?
Within relatively short distances between sampling locations, we documented considerable differences in mean 87 Sr/ 86 Sr values per geological unit that range from as low as 0.7078 to 0.7212 within 10km distance (Fig 2). This suggests that in these geological settings, ancient farming, animal husbandry and hunting would likely result in the utilization of several larger geological units with distinct geological ages and 87 Sr/ 86 Sr values. We can extend this statement to other locations within the Conchucos region. This finding has important implications for archaeological research interested in understanding past human mobility not only in this specific region, but throughout the Andes. The analysis of 87 Sr/ 86 Sr values in human skeletal remains is a powerful tool to reconstruct past human mobility. However, the interpretation of 87 Sr/ 86 Sr data are not always straightforward. Individuals with 87 Sr/ 86 Sr values outside the estimated local 87 Sr/ 86 Sr range of a given site are commonly described as having consumed non-local sources of Sr, either by they themselves being non-locals or by consuming non-local foods (i.e. through trade or consuming foods farmed in a geologically distinct region) [1, 7, 8, 30-36, 43, 44]. Whereas those with 87 Sr/ 86 Sr values matching those of the immediate vicinity of the site are considered to have consumed local sources of Sr and were therefore potential residents of that site [1,7,8,[30][31][32][33][34][35][36]59]. To address questions surrounding residential mobility requires not only 87 Sr/ 86 Sr data but also nuanced interpretations of archaeological context and potentially the use of light isotopes (i.e. Carbon and Nitrogen) to estimate diet. Our data illustrates that even locally residing individuals can potentially have a range of sources of 87 Sr/ 86 Sr values within a discrete area, depending on where they farmed their plants and produced their animal food. If the 87 Sr/ 86 Sr values measured in enamel are represented in the region surrounding an archaeological site, they may be considered potentially local.
Because the 87 Sr/ 86 Sr value of a given tissue (i.e., bone or tooth enamel) is an average of all the bioavailable 87 Sr/ 86 Sr ingested over the duration of that tissues' formation [59,109,151], the extent of landscape-use related mobility should be considered, especially within regions that are as ecologically and geologically complex as the Andes. If enamel of late forming teeth is used and individuals are frequently consuming dietary items of different geological origin, their 87 Sr/ 86 Sr values will be a mix of the 87 Sr/ 86 Sr values of these consumed food sources. For example, if an adult individual's 87 Sr/ 86 Sr value does not fit within the bioavailable 87 Sr/ 86 Sr range of a given site that does not necessarily mean this person should be considered nonlocal. Rather, this may indicate a higher degree of local mobility within the framework of farming and hunting. On the other hand, even though a region is geologically diverse, if individuals were not utilizing the entire landscape human 87 Sr/ 86 Sr values may not be variable. It is for this reason that establishing baseline environmental 87 Sr/ 86 Sr isotope data from within archaeological sites as well as from the surrounding landscape is crucial to a more thorough examination of past human mobility.
For example, in a recent study Slovak and colleagues [44] report the first 87 Sr/ 86 Sr signatures from five human Mariash-Recuay individuals (ca. AD 1-700) buried at the Peruvian highland ceremonial center of Chavín de Huántar (3,180 masl) located in our study region (Fig 2). To establish a local bioavailable 87 Sr/ 86 Sr range, several soil, animal and plant samples collected from within and around the ceremonial center (~2ha) were analyzed [44]. Based on these reference samples, three Chavín human individuals were classified to be of local origin (CdH_38, 39, 40 87 Sr/ 86 Sr = 0.7111-0.7113), whereas two others with 87 Sr/ 86 Sr values outside the estimated local range (CdH_36 87 Sr/ 86 Sr = 0.708; CdH_37 87 Sr/ 86 Sr = 0.706) were considered to be of non-local origin. Slovak and colleagues [44] report potential regions of origin that range from the central coast to the Atacama Desert.
Based on our data, we propose that while it is possible that individual CdH_36 ( 87 Sr/ 86 Sr value of 0.708) may have migrated to Chavín de Huántar from much further distances, this individual may have had a life history background in the Conchucos region and moved to Chavín de Huántar after early childhood (as premolars and second molars were used in this study). 87 Sr/ 86 Sr values similar to this individual can be found in the vicinity of Chavín de Huántar such as within the Jumasha and Cajamarca formations, only 10 km away, where we report 87 Sr/ 86 Sr values of 0.708 ±0.0017 (Fig 2).

Comparing a pan-Andean isoscape to our regionally specific 87 Sr/ 86 Sr study
In a recent publication, Scaffidi and Knudson [42] present a pan-Andean isoscape that combines all published 87 Sr/ 86 Sr data prior to 2019 from Peru, and applies geostatistical modeling to generate a predictive model for 87 Sr/ 86 Sr values found within the Andes. As discussed by Scaffidi and Knudson [42], this extensive dataset has the potential to be particularly valuable in regions where baseline environmental sampling is logistically or contextually problematic and/ or those regions of Peru lacking 87 Sr/ 86 Sr reference data. Until recently, the majority of 87 Sr/ 86 Sr studies within Peru have taken place either along the Pacific coast or in the southern Andes particularly along the western slopes, with very few studies in the eastern highlands (Fig  2; [e.g., 17-52]). In this isoscape, regions with little to no 87 Sr/ 86 Sr reference data are presented as geologically and isotopically uniform. This affects the projection of 87 Sr/ 86 Sr values for the Conchucos region, for which we present highly variable environmental data.
Scaffidi and Knudson [42] show a general pattern of a west to east gradient of lesser to greater radiogenic values, with lower 87 Sr/ 86 Sr values along the coast (i.e., 0.7038-0.70550 coastal) and generally higher values moving towards the east (0.7177-0.7239). While on a macro-scale and over large distances this distinction is observed, our study suggests that there is also considerable geological variation in the Conchucos region, a small area totaling only 0.4% of Peru. Within the Conchucos region, we document more extensive isotopic variation than initially estimated, including relatively low and relatively radiogenic 87 Sr/ 86 Sr values (0.7078-0.7212).
The geology of the Andes is comprised of closely stacked geological formations that run in parallel from north to south. Our study demonstrates that because each of these geological formations is of distinct geologic age, there are differing 87 Sr/ 86 Sr values represented within close proximity. In contrast to the pan-Andean isoscape created by Scaffidi and Knudson [42], within our localized study region there does not appear to be a west-east trend in 87 Sr/ 86 Sr values. In this region the lowest 87 Sr/ 86 Sr value (0.7078) is found in Jumasha and Cajamarca formations that run in between geological formations with higher 87 Sr/ 86 Sr values (0.7133-0.7178 to the west; 0.7196-0.7208 to the east).
The entire range of documented 87 Sr/ 86 Sr values in all archaeological Andean samples measured to date is 0.7038-0.7234 [42], which is just as broad as reported globally [59]. Interestingly, within the Conchucos isoscape we report a similar range of environmental 87 Sr/ 86 Sr values (0.7078-0.7212). Based on our data, we can predict that this environmental degree of 87 Sr/ 86 Sr variation will be present throughout the Andes. The results of our regional isoscape have the potential to fine tune the resolution of this pan-Andean isoscape.

Conclusion
This study contributes to the achievements of previous 87 Sr/ 86 Sr isotope studies within Peru by providing a novel and detailed 87 Sr/ 86 Sr isoscape for the previously understudied Conchucos region. We also address the challenge with the application of 87 Sr/ 86 Sr data in making determinations about past human mobility. Our data illustrates the need to consider a larger scope of possibilities to explain why an individual may have an 87