Keywords:
geolocation, isotope mapping, geostatistics, ecology, geochemistry, bioavailable 87Sr/86Sr
Strontium isotopes (87Sr/86Sr) are increasingly used as a provenance tool in multiple disciplines. Application to biological materials requires knowledge of the variation in bioavailable 87Sr/86Sr across the landscape, potentially in the form of an isoscape (a quantitative model of spatial isotopic variability). This paper summarizes and provides advice on our current understanding of the main concerns in creating and interpreting isoscapes of bioavailable 87Sr/86Sr. Isoscape creation approaches include domain mapping, geostatistical contour mapping and machine learning, the last becoming more readily achievable with the availability of software packages. It is critically important to develop isoscapes at a resolution appropriate for addressing the research questions. Choice of sample materials depends on the research questions and availability: plants or fauna with small ranges are favoured, with some analytes (snails, soil leachates) posing challenges. Interpreting 87Sr/ 86Sr in biological tissues requires considering Sr metabolism and the timing of tissue formation, thus far underappreciated. The numerous sources of error involved in developing and applying isoscapes must be recognized to avoid over-interpreting data and spurious provenance precision. We hope this paper will help researchers investigating provenance, mobility, landscape use and migration to develop the most appropriate isoscapes for their purposes, and possible future use by others.
1. Introduction
Strontium isotopes (87Sr/86Sr) are increasingly widely used as a provenance tool across many fields, including archaeology, ecology, biology, forensics and the food industry (e.g. [1-16]). In recent years, there has been substantial development in the field, especially improvements in the creation of quantitative spatial models of isotopic variation, or 'isoscapes', and the use of probabilistic models for geographic allocation. Yet, some important challenges remain, many of which were identified in reviews by Holt et al. [17] and Bataille et al. [18]. In January 2024, the authors of this paper held a five-day meeting in Cape Town, South Africa, to workshop these issues, discuss best practices in the construction and interpretation of isoscapes, and flag aspects requiring further research. This article summarizes the outcomes of those discussions. We hope it will be useful to the growing community of researchers using strontium isotope provenancing worldwide.
The application of 87Sr/86Sr as a provenance tool is based on the ratios of strontium and rubidium isotopes in the local environment, which in most cases derive from the underlying local geology. 87Sr is the stable radiogenic daughter isotope of 87Rb (h/2 = 48.8 x 109 years), while 86Sr is stable and largely invariant over time. Older rocks and those rich in 87Rb show relatively high 87Sr/86Sr, while younger (e.g. recent volcanic) substrates have lower values. Although bioavailable 87Sr/86Sr (i.e. that taken up by plants and animals) can differ from that of bedrock, these ratios are persistent in soils derived from parent substrates (albeit with other inputs, and outputs-see below), in the tissues of the plants growing in those soils, the herbivores feeding on those plants, and the predators preying on them. The single mass unit difference between 87Sr and 86Sr is small in relation to total mass, so there is little mass-dependent fractionation of these isotopes along the food chain, though see [19,20]. The small amount that does occur is corrected for by adjusting all measured 87Sr/86Sr to a constant value of 0.1194 for 86Sr/88Sr [21]. The 87Sr/86Sr of consumers should therefore track the mixing of ingested food (and in some cases water) with distinct 87Sr/86Sr, taking into account differences in strontium concentrations of each Sr source, and possible differential digestion and absorption of each source within the digestive tract of species with differing gut pH [22,23]. 87Sr/86Sr in the environment may differ from bedrock ratios if the geology is heterogeneous and contains minerals with differing 87Sr/ 86Sr and varying susceptibilities to weathering, such as granite and sandstone [24-26]. In addition, soils (and therefore plants and animals) can incorporate Sr from sources other than bedrock, such as atmospheric, fluvial and anthropogenic sources. Atmospheric inputs include precipitation and wind-transported dust, marine aerosols and volcanic ash (e.g [26-29]). The importance of these inputs depends on the magnitude of the fluxes, and the levels of endogenous bedrock-derived strontium in the soils. Coastal regions may have 87Sr/86Sr dominated by Sr from seawater-derived aerosols and precipitation or geologically recent carbonate-rich sands draped over underlying geology [4,26,3032]. This influence can extend hundreds of kilometres inland [33]. Rivers flowing through different geological formations may contain dissolved Sr and/or contribute sediments with 87Sr/86Sr that differs from local bedrock [24,34-36]. Anthropogenic activities, such as quarrying, mining, soil transport and the addition of agricultural lime, can substantially alter soil 87Sr/86Sr [37-39] and introduce greater variation through the soil profile [40]. These and similar processes may cause bioavailable 87Sr/86Sr to be decoupled from that of local bedrock [41,42].
Multiple possible Sr inputs at different scales make isoscapes of 87Sr/86Sr more challenging to develop and use than (for example) isoscapes of 18O/16O (ô180) and 2H/XH (b2H) in precipitation, in which variation driven by large-scale atmospheric processes leads to more consistent patterning [43]. For these reasons, 87Sr/86Sr provenancing in many regions cannot be undertaken with reference to geological maps alone and requires the development of isoscapes, or predictive models of geospatial variation in 87Sr/86Sr [44]. In other words, isoscapes are not simply maps of empirical data, they are sets of expectations (mathematical predictions) derived from both first principles and a limited amount of empirical data. They tend to preferentially emphasize overall patterning over the degree of variation at individual localities. Isoscapes enable us to visualize and model 87Sr/86Sr across a landscape, incorporating multiple different Sr sources. Strontium isoscapes are usually generated from baseline reference material (e.g. plants) and then used to track the mobility of animals or humans by matching 87Sr/86Sr from their tissues to the isoscape. This approach works well due to negligible trophic fractionation of 87Sr/86Sr, and the stability of these ratios over time scales relevant to ecology, forensics, archaeology and related disciplines. There are, however, several challenges, which we examine below.
First, it is important to note that 87Sr/86Sr will not address all questions about mobility and provenance. Modelling the distribution of 87Sr/86Sr requires considerable laboratory and computing effort and is not to be undertaken lightly. Researchers should first carefully consider whether there is a reasonable expectation that the spatial patterning in 87Sr/86Sr is likely to be capable of addressing the particular research question, given the nature of the landscape and its geological formations.
Below, we explore key questions that formed the main topics of discussion in our 2024 workshop. Many of us have found these difficult to resolve in our own work. Holt et al. [17] provide an excellent overview of the field. Based on their review and other recent papers, we selected the following topics: the relative merits of different approaches to isoscape development, sample collection strategies, choice of sample type and the requisite preparation and analysis, complications with using isoscapes to assign provenance, including the influence of consumer biology and metabolism, whether it is necessary to tailor isoscapes for specific applications, uncertainties and limitations and future directions for the field. These topics span both fundamental principles and laboratory practice. The choices researchers make may differ depending on the field of study and the specific research questions, but we hope that the application of these principles will guide practitioners towards better outcomes.
2. Approaches to isoscape development: are machine-learning isoscapes better than other types?
The framing of this question, and some of the section headings below, may appear to be leading the reader towards particular conclusions. These are, however, informed by careful reading of the current literature, which guided our discussions. We believe that this more targeted approach helps discriminate between various possible options and leads to a clearer understanding of current best practice.
There are currently three main approaches to developing isoscapes of bioavailable 87Sr/86Sr, each with advantages and disadvantages. The first, domain mapping, is based on clustering empirically measured 87Sr/86Sr (normally bounded or grouped by lithology), while the second and third, geostatistical and machine learning approaches, respectively, use empirical 87Sr/86Sr measurements alongside other environmental datasets to model and predict 87Sr/86Sr variation. These approaches have been described [17,45] and compared elsewhere [45-47], but for completeness, we outline the basics of each below.
Domain mapping is a clustering technique that groups empirical 87Sr/86Sr data derived from different user-defined domains, generally geological units [17]. Each unit is then assigned a median and range. This is an easy, conservative and fast approach to visualizing patterns of bioavailable 87Sr/86Sr, but it depends heavily on good sample coverage and the assignment of the domains [17]. If these are geological units, the assumption is that geology is the dominant driver of 87Sr/86Sr, and the quality of the isoscape will depend on the quality and resolution of geological maps used to delineate domains, as well as careful field observations determining the context of each sampled site. If there are other influences on 87Sr/86Sr, these are captured in the form of large uncertainties per domain. This method is, therefore, most useful in regions where lithology plays the dominant role and is well-mapped, and other Sr inputs are limited. Domains can be defined based not only on lithological units but, for example, on proximity to the coast, where 87Sr/86Sr is likely to be similar across different geological substrates because of marine-derived Sr inputs [48,49]. Similarly, specific domains can be created to encompass (for example) old-growth forests, where the more soluble (e.g. carbonate-based) Sr components of the soil may be leached out and/or sequestered in the vegetation, so the bioavailable 87Sr/86Sr signal becomes dominated by the less-soluble soil component [50]. The strontium isoscape of Great Britain, available through the British Geological Survey, was created using domain mapping (biosphere isotope domains). The isoscape, the data on which it is based and various tools, including a useful query function, are all freely available at https://www.bgs.ac.uk/datasets/biosphere-isotopedomains-gb/.
The second approach uses various geostatistical methods, such as inverse distance weighting and kriging, to develop an output that can be visualized as a contour map. This enables continuous predictions of 87Sr/86Sr variation derived from measured known-origin samples [17]. Such maps are based on the principle of spatial autocorrelation, in which points closer together are considered to be more closely related (i.e. ratios will be more similar) than those further away. The simplest such method, inverse distance weighting (IDW), simply uses the distance between a locality with unknown 87Sr/86Sr and each point with a known value to weight their influences on the prediction. The scaling between distance and weight of influence in ID W is either assumed a priori or optimized by comparing predictions against testing data. Alternatively, kriging methods predict 87Sr/86Sr at a particular location using a variogram model to account for trends in the spatial autocorrelation between pairs of points in the dataset, and assign weights to the observations. In contrast to IDW, the variogram model (and thus weights used in kriging prediction) is fitted to the known-origin data. There are several types of kriging. Simple kriging assumes a known constant mean autocorrelation, while ordinary kriging assumes an unknown constant mean, which is locally estimated. Universal kriging (or kriging with external drift) incorporates second-order effects by estimating a trend based on an auxiliary predictor and uses this instead of the local mean used in ordinary kriging [47]. Empirical Bayesian kriging automates parameter selection and accounts for the error of the semivariogram. Со-kriging incorporates one or more secondary datasets spatially correlated with 87Sr/86Sr distributions (e.g. soil, plant and water 87Sr/86Sr datasets [46]) to try to increase prediction precision. However, this also introduces more variability due to the additional estimations of the autocorrelation for the secondary variable, as well as the cross-correlation between variables. These geostatistical approaches include uncertainty assessments that allow one to assess the probability of a correct prediction [45]. One major limitation of these methods is that they assume that patterns of variation in 87Sr/86Sr are continuous across the landscape. Although geostatistics-based isoscapes are relatively easy and quick to create and can be readily extrapolated to adjacent regions, they cannot accurately represent discrete, nonlinear patterns of bioavailable 87Sr/86Sr (e.g. across heterogeneous geology), even when со-kriging with a geological map.
The third approach, the use of machine learning models, particularly random forest regression, is the most recent development. In this method, single decision trees play a central role. Each decision tree uses predictor variables to split the empirical dataset of 87Sr/86Sr into groups of similar values. The tree does this by making a series of decisions based on the variability within each predictor variable [18,45,51]. The potential predictor variables can be anything that may affect 87Sr/86Sr, such as geology, distance from the coast, dust, volcanic ash deposition, precipitation and others. Although single decision trees can theoretically be used in isolation, they are prone to overfitting and instability, where even slight changes in the training data can result in significant changes to the model. Random forest algorithms address these issues by creating an ensemble of multiple decision trees. Each tree is trained in parallel using a random subset of the training data and predictor variables, a process known as 'bagging' [52]. The performance of the model is validated through и-fold cross-validation with the possibility for spatial partitioning of the testing set to avoid geographic biases. This means that n random subsets of the data are removed prior to model development and used to test the created model. The results are averaged to produce a final model that is less prone to overfitting, more stable and more robust. This can be done multiple times with different, random test sets; however, there is a trade-off as fewer data are then available for model development. This testing process need not be uniform or truly random: it is possible to target 'problem' areas with known low accuracy based on prior knowledge, areas with low sampling density or specific areas of interest based on the research question. Initially, random forest models make no assumptions about which predictor variables are the main drivers of 87Sr/86Sr variation, and all variables are incorporated into the model. However, one can use variable selection algorithms such as variable selection using random forests (VSURF) [53] to select the most suitable predictors for a given study area (i.e. the predictors with the strongest effect on the model's predictive power). VSURF uses an iterative simulation process, constructing many random forest models but selectively removing specific variables to quantify their effect on the model performance. Through this process, the algorithm iteratively removes the least relevant predictor variables based on their lack of impact on the model performance, and finally refines the variable selection by removing any redundant predictor variables, see [53]. The relationship between predictors and 87Sr/86Sr can be conveniently visualized through partial dependence plots. Using the most relevant selected variables, the random forest algorithm is then tuned to build the most parsimonious and best-performing random forest model integrating selected geospatial predictors.
Random forest modelling is an effective method for mapping bioavailable 87Sr/86Sr, in many situations outperforming other modelling approaches [45]. Compared to domain mapping and geostatistical approaches, random forest has several advantages for predicting 87Sr/86Sr. It allows seamless integration of multidimensional data (e.g. points, categorical geological maps and continuous rasters of climate variables), and it copes well with outlying 87Sr/86Sr values. This approach accurately predicts the multiscale patterns of 87Sr/86Sr, resolving some of the limitations of contour mapping and domain mapping techniques. Although still reliant on empirical 87Sr/86Sr data, since random forest models incorporate additional correlated predictor variables, they can perform adequately in areas with low sample coverage [17]. However, as random forest models become more complex, they incorporate more predictors and require more empirical 87Sr/86Sr data, which can be difficult to obtain and can reach the limits of computing power on personal computers. At this point, they require cluster computing resources.
Additionally, the random forest framework does not account for spatial autocorrelation within the calibration dataset [54], limiting its ability to handle geographic sampling biases. Predictions and spatial uncertainty are not influenced by the location of the sampling sites, and regional bias might be introduced by sampling some areas more heavily than others. Recent studies using random forest models to map soil properties have proposed several approaches to overcome this issue. Sampling distribution can be accounted for by introducing geographic features as covariates in the model [55,56], and local variation can be integrated by combining multi-scale random forest models [57] or by weighting the prediction according to the distance from the nearest sampling sites [58]. There are also advantages in integrating multiple algorithms, including random forest, through ensemble machine learning to improve the accuracy and reduce the errors of the predictions [55,59]. While random forest models rely on a single algorithm, ensemble approaches use a set of methods that combine the predictions of multiple individual models, improving the performance and robustness of the final model [60]. The landmap package in R [61] provides methods for spatial prediction using ensemble machine learning and accounts for spatial auto-correlation by using oblique geographic coordinates as covariates [56]. Applied to bioavailable 87Sr/86Sr mapping in Eastern Canada, the ensemble approach provided similar predictions to the classic random forest model but with reduced spatial uncertainty, particularly in highly radiogenic areas with high uncertainty [54]. Although powerful, ensemble approaches that integrate local variation require a good spatial distribution of samples and more computing power than traditional random forest methods, which may limit their application to large-scale 87Sr/86Sr isoscapes.
3. What is the best sample collection methodology for creating isoscapes?
Sample collection strategies should be shaped by the chosen isoscape approach, the research questions to be answered, and the scale, resolution and budget of the project. Although systematic sampling (e.g. on a grid) may be desirable (e.g. [62]), this is not always practical or possible. Topography may limit access to some areas, and there may be restricted entry to privately owned land or reserves. Agricultural and other land use practices may make some areas unsuitable for sampling. Collaborative sampling with other projects, or obtaining samples from existing museum collections with sufficient metadata [11,51,63-65], can be cost-effective and efficient alternatives to field collection, specifically for the purpose of isoscape development.
Issues of scale and resolution are of central concern in landscape-level studies. Any interpolation method can be used to create isoscapes at any scale. Global or continental-scale isoscapes (e.g. [18,65]), that rely heavily on geospatial statistics and global covariate data are valuable as preliminary tools, including assessing whether the Sr isotope system is likely to help answer a given research question at any one locale. Many broad-scale isoscapes are, however, limited by the availability and geographically biased distribution of bioavailable 87Sr/86Sr data globally, as well as the relatively low resolution of global covariate data. As such, global models will usually require recalibration using local, potentially higher-resolution empirical and co variate information for application at regional and local scales. The required isoscape resolution is challenging to determine. For example, trying to provenance a butterfly which developed from a caterpillar that lived on a single plant is very different from tracking large mammals feeding over much bigger areas.
Ideally, one might wish for isoscapes based on covariate data and samples collected at a level of resolution relevant to the research questions about mobility or migration. Areas with few bioavailable 87Sr/86Sr data points or substantial geological complexity will require more intensive sampling. Heterogeneous felsic or metamorphic lithologies often tend to show greater 87Sr/86Sr variability, compared with more homogeneous carbonate rock types such as chalk and limestones [18]. Evans et al. [66] (as cited in [17]) found radiogenic igneous rock to have approximately 0.5% (0.0012-0.0036 per mil) 1 s.d. reproducibility, compared with approximately 0.05% (0.0004-0.0008 per mil) for carbonates.
When collecting samples, it is recommended to think beyond the current project, considering possible future use of the data by researchers interested in different questions. For each sample, it is important to record the GPS coordinates, elevation and geology or lithology, both from the published geological map of the area and actual field observations from the site. Due to limitations of scale and resolution, bedrock shown on a geological map may be draped by a thick layer of sediment or capped by a secondary deposit such as calcrete or ferricrete, and these observations are important to note. Apps in development to aid the collection of sample metadata will automatically fill out a number of these fields.
4. Are plants the best samples for isoscape creation?
The choice of material for empirical field measurements of 87Sr/86Sr depends on what is available in the area of interest and on the temporal and geographical scale of the research question. Most isoscapes are built using soil, plant, faunal or water samples, or a combination of these sample types. Terrestrial isoscapes differ from river isoscapes.
Some isoscapes are based on 87Sr/86Sr measurements of soil samples, but it can be difficult to estimate bioavailable 87Sr/86Sr from soils. Soils usually include resistant (less soluble) and more soluble components, so complete dissolution of bulk soils in strong acid yields total 87Sr/86Sr, which may be substantially different from the more soluble bioavailable signal taken up by plants and ultimately animals [34]. For the production of isoscapes, therefore, researchers typically leach soil samples to extract the more mobile (approximately equivalent to bioavailable) strontium fraction. The 87Sr/86Sr obtained is influenced by the leaching protocol (i.e. the reagent used, its concentration, duration of leaching, etc. [39,67,68]). These are examined in more detail in the sample preparation section. Some researchers report close correlations between 87Sr/86Sr of soil leachates and plants on a large scale [46,69], but several participants in the workshop have found soil leachates to yield such inconsistent results (e.g. [70]) that they now prefer other sample materials for estimation of bioavailable 87Sr/86Sr. In addition, 87Sr/86Sr in soils is highly susceptible to the effects of recent agricultural practices, such as the addition of lime [37,38,40,69], bone meal or basalt rock dust [71] used to supply minerals. Some organic fertilizers contain negligible quantities of calcium and strontium, so will not affect soil 87Sr/86Sr. Organic-rich fertilizers tend to increase soil ô15N, thus providing a marker of their presence [72-75]. Soil amendments will need to be taken into account if the intention is to provenance agricultural products [13] or apply 87Sr/86Sr in modern forensics [51].
Plants are generally considered a good material for estimating bioavailable 87Sr/86Sr [28,31,36,42,63], although they may be scarce in arid environments. As plants are primary producers at the base of the food web, patterning in plant 87Sr/86Sr will likely be a reliable proxy for ratios in consumers, although as organisms rooted in one spot, they may show more small-scale inter-individual variation than mobile animals. Any type of plant material (i.e. leaf, stem or bark, etc.) should be suitable and have similar 87Sr/86Sr [70]. Opinions differ on the importance of identifying sampled plants to species or at least genus level. There is no logical reason why 87Sr/86Sr should со-vary with taxonomy (as demonstrated by O'Regan et al. [76]), yet some researchers recommend recording plant species sampled [67]. In highly speciose biomes, identifying plants to species (and sometimes even genus) level can require the presence of flowers or fruit, which severely constrains sampling options. However, the depth and extent of root systems can vary substantially between species and between small and large specimens of the same species. It is therefore recommended to provide an estimate of plant type and rooting depth (shallow, medium, deep), as plants with different rooting depths may access different Sr sources in the soil [28,36,39,77-80], although this does not apply in all environments [47]. Attempts to characterize the bioavailable 87Sr/86Sr of the ecosystem as a whole should factor this into the collecting strategy and include both shallow- (e.g. grasses), medium- (e.g. shrubs) and deep-rooted plants (e.g. trees). Pooled samples from multiple plant species will generate an averaged signal [4,63,81] and reduce point bias [17]. A more targeted sampling strategy may be preferred for some research questions (e.g. if investigating organisms that live on specific plant species, like monarch butterflies, or that consume either deep- or shallow-rooted plants, such as browsing versus grazing herbivores [11,36]). Soil pH and cation exchange capacity can covary with 87Sr/86Sr [18,45]. O'Regan et al. [76] investigated whether mycorrhizal associations influence plant 87Sr/86Sr because mycorrhizae can change the pH of soils and thus influence the dissolution of minerals [82,83]. They found no significant differences between plants with and without associated mycorrhizae, although this may vary in different geological substrates, growing conditions and plant taxa sampled.
In some parts of the world, exotic dust can make a substantial contribution to the strontium budgets of plants [28,84]. Plant 87Sr/86Sr can also be impacted by anthropogenic alterations to the soil. If these are not relevant to the research question (i.e. the isoscape is not designed for a modern or forensic use-case), careful choice of sampling locality-avoiding areas currently or recently cultivated-can mitigate the problem. In general, the importance of fertilizer use for plant 87Sr/86Sr will need to be assessed on a case-by-case basis, including the possible influence on adjacent uncultivated areas [37,38,40].
Faunal samples (e.g. animal bones or teeth) have the advantage (compared with plants) of integrating 87Sr/86Sr over the animal's feeding range, rather than providing spot values. Small terrestrial wild or commensal species such as rodents, with relatively small territories and short life-spans, have been used to estimate 'local' 87Sr/86Sr values, with both modern [1,64,85,86] and archaeological fauna [87-89]. However, one must ensure that such animals are local and not deposited a long distance away from their home range by an owl, for example [90]. Animals with large home ranges, correlated with larger body size [91], may incorporate Sr from multiple regions with different 87Sr/86Sr, making them unsuitable for constructing isoscapes, but see [92]. Some studies have shown overall consistency between species with small foraging ranges and plants (e.g. [29]), while others have not (e.g. [93]), highlighting the context-dependent nature of strontium propagation. In regions that have been intensively farmed for a long time, with a commensurate degree of soil modification, palaeo-studies may require isoscapes based on historical or archaeological materials [87].
A number of studies have analysed snail shells, reasoning that these contain significant amounts of strontium, are straightforward to collect, will yield reliably localized signals and are easy to process in the lab. It appears, however, that the 87Sr/86Sr of snails are strongly influenced by rainwater, probably because snails are heavily dependent on rainwater to generate slime [31]. If the 87Sr/86Sr of rainwater differs from the local geological substrate, snails may not be a reliable proxy for local bioavailable values. Other studies have found snails to track 87Sr/86Sr of shallow-rooted rather than deep-rooted plants [28], to be biased towards values for soil carbonates [39], and to yield reduced ranges of variation compared with local plants [36]. They may also prove difficult to find in certain areas (acidic soils, montane regions, etc.). All in all, snail shells appear sub-optimal for constructing isoscapes of bioavailable 87Sr/86Sr.
Water samples are easy to collect. Near a river's source at a high elevation, water 87Sr/86Sr will resemble that of bedrock (at least, the fraction of bedrock susceptible to weathering). Downstream at lower elevations, the river may incorporate Sr from precipitation runoff through multiple geological formations with variable Sr contributions [24,34]. Although uncommon, freshwater 87Sr/86Sr can exhibit substantial temporal variation in environments with marked wet or dry seasons due to the alternating relative importance of surface runoff versus aquifer groundwater input [94]. Plants growing on floodplains near rivers have been found to have river-derived 87Sr/86Sr signals that differ from surrounding geological substrates [34,36]. Developing isoscapes of river systems requires the use of dendritic network models to adequately account for their complexity. These are discussed in detail elsewhere; see [95]. While the importance of Sr inputs from drinking water in consumer Sr budgets will vary with the relative concentration of Sr in the water versus that in the food, it is typically less impactful than food [96], suggesting that the collection of waters can be complementary but not necessarily a core requisite in bioavailability sampling studies.
5. How should samples be prepared and analysed?
The different sample types discussed above require different preparation for analysis. How samples are prepared will also depend on how Sr will be introduced into the mass spectrometer: typically, either in solution or via laser ablation with a multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) or in solid form with a thermal ionization mass spectrometer (TIMS). Both TIMS and solution-based MC-ICP-MS offer high precision but involve a significant amount of sample preparation, typically acid digestion of ashed, powdered or bulk samples in a cleanroom laboratory, followed by Sr isolation using ion exchange chromatography (e.g. [97,98]). Although Sr is a trace metal, it is ubiquitous in the environment, including in unfiltered air. In the US, the concentration of strontium in urban air was on average 20 ng m"3 in the 1970s and has increased with growing air pollution [99,100]. The isolation of small amounts of Sr therefore requires a cleanroom laboratory to avoid atmospheric dust and avoidance of unnecessary exposure to metal surfaces, frequently a source of Sr contamination. Class 1000 or ISO 6 certified laboratories should indicate a sufficiently uncontaminated space. However, for a more thorough assessment of potential contaminants in the lab environment and in reagents, procedural blanks should be included in each batch of samples or at least measured at regular intervals. Sr concentrations in blanks below 1% of the values in samples being analysed indicate a clean work environment and process.
5.1. Sample measurement metadata
Metadata recorded along with analytical results should include the laboratory where the measurements were conducted and the protocols used during sample preparation. Laboratories should report the long-term, 2u external reproducibility of their repeated analyses of a matrix-matched standard with Sr concentration similar to the samples (i.e. the 2u standard deviation of the facility). Along with the number of analyses, this enables the calculation of the standard error if desired. Suitable standards include US Geological Survey BCR-2 [101] and BHVO-2 [102] for rocks and soil samples, NIST SRM 1400 and 1486 for bone and tooth samples [103], and NIST SRM 1515 for plant samples [104,105]. Alongside this, the standard error of the reference materials run during a particular project should also be reported. The reference value used in the analytical facility for NIST SRM 987 to which 87Sr/86Sr measurements are referenced should also be quoted (e.g. 0.710255 from Waight et al. [106]) to enable different datasets to be renormalized to the same value for direct comparability. Note that the value of 0.71034 ± 0.00036 originally reported for SRM 987 [107] is outdated, and a large, improved set of values from many facilities is available [108]. Values for laboratory procedural blanks, the magnitude of blank elemental Sr, results for known unknowns analysed alongside project samples, and the numbers of repeat analyses should also be reported. Finally, it is helpful to include strontium concentrations of samples, if available.
5.2. Soil samples
For isoscapes of bioavailable 87Sr/86Sr, soils should be leached to isolate the bioavailable and thus biologically meaningful Sr fraction, which is water soluble and exchangeable. As mentioned above, different leaching protocols yield different 87Sr/86Sr, and there is little consensus on which are best [68]. Published studies variously report leaching with water alone [39], nitric, acetic [46,69,109] or hydrochloric acid [110] of varying concentrations (0.1-1 M) or 1 M ammonium nitrate solution (NH4NO3) [111] following the protocol DIN International Organization for Standardization (ISO) 19730 [112]. Some studies have compared multiple methods [13,67,68,113,114]. The leachates are then separated using centrifugation [115] or by filtering through a syringe or pipette equipped with a membrane [51], dried down, and Sr is isolated using ion exchange column chromatography. For the sake of data reproducibility and metadata analysis, authors should publish details of the protocol used for leaching.
5.3. Plant samples
Once collected, plant samples should be stored in unbleached, lightly closed paper bags and allowed to dry out completely to avoid the formation of mould. If long-distance (international) transport is necessary before analysis, it may be advisable to first char them (if they will ultimately be ashed), to avoid the possible need for plant importation permits. In the field, plants can be wrapped in aluminium foil and left near a fire, allowing them to turn into charcoal [116]. Alternatively, plants can be frozen at -80°C for 72 h [28] or gamma-irradiated [115] to destroy seeds and pests.
Some studies have rinsed plants with deionized or ultrapure water before analysis to remove surface dust [11,29,36,117], while others have not, presumably (although not necessarily explicitly) reasoning that atmospheric Sr deposition contributes to bioavailable 87Sr/86Sr ingested by consumers feeding on plants [4,63]. In a study in the UK, Warham [70] found that unwashed leaves and inner woody material from the same trees, Warham yielded the same 87Sr/86Sr.
Field-collected plant samples are usually sub-sampled for processing in the laboratory. To ensure the recovery of sufficient amounts of Sr, we recommend preparing at least 1.5 g of dry plant matter or approximately 20 mg of ash. In the laboratory, plants are either ashed or treated with strong oxidizing agents (e.g. hot nitric acid) to remove organic compounds that can interfere with the separation of Sr on resins used in column chromatography. Increasingly, many laboratories use microwave digestion to achieve this more efficiently (e.g. [62]). A typical ashing protocol is as follows: the dry material is cut or crumbled into small pieces (combining different specimens if desired) to partially fill pre-cleaned crucibles made of porcelain, ceramic or pure silica or quartz, leaving space for air to circulate. The crucibles are then placed in a muffle furnace with temperatures between 500°C and 800°C for about 10 h, depending on the laboratory. The resultant ash is homogenized in the crucible, transferred to a clean storage vial and is then ready for acid digestion and separation of Sr in a clean lab setting. If plants are to be analysed for other isotope systems such as carbon, nitrogen or sulphur, sub-samples must be removed before ashing. Another method is to dry the plants overnight in an oven at approximately 40°C, then homogenize into a fine powder before acid digestion, possibly using a microwave digestor [31].
5.4. Bone and enamel samples
Like plants, organic-rich bones are either ashed or treated with hot nitric acid. For calcium-rich materials such as bone, 10-30 mg will contain sufficient Sr for analysis, with similar preparation and Sr separation methods to those employed for plants as described above. Alternatively, hard, mostly inorganic sample materials, such as well-preserved tooth enamel, are suitable for analysis by laser ablation MC-ICP-MS.
Laser ablation needs minimal sample preparation and has high spatial resolution but relatively low precision. Corrections for isobaric interferences may be required (e.g. [118-120]). It is quicker and less destructive than solution ICP-MS or TIMS since it does not require separation of Sr [121]. The method is minimally destructive for small samples (e.g. individual human teeth) that can be placed in the laser chamber; larger ones (e.g. large animal teeth) may have to be sub-sampled. Laser ablation requires samples to have relatively high concentrations of Sr and low Rb (since Rb cannot be removed before analysis and the 87Rb isobaric interference has to be corrected, although future technological developments may help to mitigate this issue). The 2o external reproducibility (two s.d. values of the mean value of multiple repeated analyses of the same sample) of 87Sr/86Sr measurements by laser ablation is typically 0.0002-0.0003 [122], larger than for solution analyses (where the uncertainty is typically in the fifth decimal place). Uncertainty in the fourth decimal place may be perfectly adequate given the numerous sources of variation in plants and animals. Laser ablation allows spatially highly resolved analyses of targeted areas of a sample, enabling the investigation of (for example) changes throughout tooth formation, or focus on areas with the best preservation. This is more difficult with solution methods, which require targeted enamel removal before digestion. If laser ablation is to be used, care should be taken that any cleaning of the sample will not compromise the isotopic measurements. Experience in the MC-ICP-MS Facility in Geological Sciences at the University of Cape Town indicates that wiping the surface of the tooth with acetone before analysis is problematic, since acetone leaves behind an organic residue. Ethanol or methanol does not, so is preferred for this purpose. This lab has also found that juvenile teeth are often not suitable for laser ablation, likely due to less mineralized enamel. While undertaking measurements on flat surfaces (sectioned or sliced samples, or even polished thin sections) is generally preferred, analyses of slightly curved surfaces can produce reliable data combined with mass correction [123].
Ancient (archaeological, palaeontological, some historical) samples may have undergone diagenetic changes that compromise our ability to measure biogenic 87Sr/86Sr. Understanding diagenesis comprises a substantial field of study in its own right [124,125], which is beyond the scope of this article, although there is one near-consensus point: bone is far more vulnerable to diagenesis than tooth enamel [126]. Unless calcined [127], or modern without prolonged contact with the soil [128], bone is best avoided as a sample material for 87Sr/86Sr. With regards to possible pre-treatment protocols, we note that methods designed to remove one diagenetic component (e.g. carbon or oxygen) from calcified tissues do not necessarily remove others (e.g. strontium). Oxygen occurs in calcified tissues in various chemical forms (carbonate, phosphate, in the hydration layer), some structural and others only loosely bound and hence readily exchangeable. Strontium, on the other hand, generally substitutes for calcium in the crystal lattice of bioapatite. Crowley et al. [129] found measurable differences in 87Sr/86Sr between samples of powdered tooth enamel prepared using different chemical pre-treatments. However, they cautioned that possible contamination introduced in the laboratory could not be ruled out [129]. Removing the outer layers of tooth enamel before analysis can help eliminate diagenetically altered material. Protocols for measurement of 87Sr/86Sr by laser ablation typically do this using a rapid pre-ablation cleaning laser sweep along the intended path of analysis with a slightly larger spot size to remove the outer few microns of enamel (typically 2-5 pm) [123].
6. What are the complications with assigning provenance?
Once an isoscape has been created, there are multiple ways to use it for geographically assigning samples of unknown origin. The simplest is visual comparison to match the 87Sr/86Sr of the unknown to the different zones on the isoscape. There are also GIS-based tools [130] and probabilistic approaches such as the assignR package in R [131], which also allows for the incorporation of additional constraints from other isotopic or non-isotopic data. The term 'assignment' is somewhat misleading, as the identified, assigned location simply represents the area of most probable origin. Bayesian probabilistic approaches like that implemented in assignR do not generate areas of zero probability (i.e. there is always some possibility of any particular locality being the origin).
The isotopic composition of consumer tissues reflects combined Sr inputs from various sources over a period of time. The more mobile the individual, and the more heterogeneous the landscape, the more sources will be incorporated. In complex cases, the consumer may match none of the contributing sources, or may yield a value compatible with more than one combination of sources, termed equifinality.
In regions of relatively homogeneous 87Sr/86Sr, such as many carbonate-based geological substrates, variations between consumers in the fourth decimal place can be meaningful [66]. In areas of more heterogeneous lithology and mineralogy, variations in the fourth place may merely be noise. Assignment can be refined with the use of contextual knowledge such as the topography of the area, rates of erosion and reworking of a landscape, known migration patterns of the species studied and/or combining with other isotopic tracers. However, one should be conscious of the influence of this prior information on the assignment. Incorrect assumptions may substantially bias or compromise the interpretations drawn or result in circular or confirmatory logic.
7. What are the influences of biology and metabolism on consumer tissue Sr?
Sr isotope provenancing of organic materials lies at the intersection between biology and geology. The geological aspects are relatively well understood, but much less attention has been paid to the biology. It has long been known that strontium is incorporated into vertebrate calcified tissues because it substitutes for calcium, replacing it in bioapatite [132,133]. Absorption of alkaline earth elements in the mammalian digestive tract favours calcium and discriminates against Sr [132,134], so the extent of Sr incorporation is proportional to [Sr]/[Ca]. Foods with high [Sr]/[Ca] (e.g. plants and dairy foods) contribute more Sr to consumer tissues than those with low [Sr]/[Ca] (mainly non-dairy animal foods) [42,135,136]. Vitamin D-rich foods promote dietary Ca absorption; this may apply also to Sr [133]. The biopurification process results in reduced variance in [Sr]/[Ca] and 87Sr/86Sr as food sources are averaged up the food chain [42]. In Britain, the interquartile range of variation seen in 87Sr/86Sr in archaeological human tooth enamel is approximately half that of plants, although the means and medians are similar [137]. Relatively little attention has been paid to [Sr]/[Ca] in isoscapes (although see [15]), but if these vary systematically across the landscape, we might expect to see differential contributions of foods from different regions (i.e. unequal visibility of various regions) in consumer 87Sr/86Sr. Lewis et al. [19] found that [Sr]/[Ca] in pig tissues increased with a greater proportion of marine food (fishmeal, which likely included small bones). We note also that the concentration of Sr in seawater is typically 8 ppm [26], compared with <1 ppm in rainwater [32]. The proportion of marine-derived Sr incorporated into the diet in the form of seafood or salt can influence 87Sr/86Sr in tissues and potentially skew provenance studies if unaccounted for [133,138]. Ingested dust influences consumer tissue 87Sr/86Sr [139], although gritstone, sandstone and granite grindstones used in grain processing appear to have little effect [140].
Details of the rates and processes of Sr uptake in different species, and mobilization and recycling of Sr already in the skeleton (especially in relation to nutritional status and pregnancy and lactation) are not well understood. Only a few, relatively small studies have analysed animals fed controlled diets. The patterns reported thus far are inconsistent: inter-individual variation in 87Sr/86Sr among animals fed isotopically monotonous diets ranged from the third decimal place (0.001) for guinea pigs [139] to the fourth decimal place (0.00062) for cattle [141] and the fifth place (0.00002-0.00008) for pigs [19] (but see [142]). Weber et al. [139] found larger inter-individual differences in guinea pigs compared with rats fed on the same diets. Inter-species differences may derive from different digestive efficiencies, perhaps linked to different pH in the digestive tract. It is critically important to know how much of the
87Sr/86Sr variation we see in our unknowns might result from within-animal biological factors alone. We can then estimate the threshold above which differences in 87Sr/86Sr are attributable to different diets and/or places of residence. Further multigenerational controlled feeding studies with larger numbers of animals are urgently needed to improve our understanding of this issue.
87Sr/86Sr in body tissues reflects food and drink consumed at the time of tissue formation, buffered by 87Sr/86Sr already circulating within the body. Even rapid movement to very different lithologies, resulting in a sudden change in dietary 87Sr/86Sr, will lead to a relatively gradual change in tissue 87Sr/86Sr due to the dilution of newly ingested Sr by the available body pool [143,144]. This effect is well-known for stable light isotopes (e.g. [145]), and additional studies on 87Sr/86Sr turnover may lead to the development of models to correct for this process in specific taxa (e.g. [144]). In mammals, bone mineral from the mother's skeleton is resorbed and incorporated into the metabolic pool during lactation, so 87Sr/86Sr in milk, and calcified tissues forming in suckling infants, reflect a mixture of the mother's current food and drink and that dating from earlier in her life. In humans, mothers typically lose 2-8% bone density during lactation [146], and some studies report that maternal skeletal turnover may be as high as 10-30% during pregnancy and lactation [147,148]. The implication is that 87Sr/86Sr of calcified tissues formed in early life will be influenced by diet and place of residence over more than one generation. Although the contribution of maternal Sr to infants is well known, the implications for provenance have not been adequately addressed.
8. Do specific applications require custom isoscapes?
The degree to which specific applications (such as provenancing animal species with different behaviours or various types of materials like glass, ceramics or eggshells) require different isoscapes probably needs to be decided on a case-by-case basis. For example, isoscapes of bioavailable 87Sr/86Sr may be suitable for assessing the origins of materials such as glass in which Sr comes mostly from wood ash [149,150], and for ceramics if geology is a major driver of 87Sr/86Sr variation in a particular study area [151]. One view is that researchers should not re-use existing isoscapes for different purposes, but rather create new ones tailored to specific research questions using the same freely available data. It is valuable to create a flexible, adjustable isoscape in which different types of samples can be included or excluded to tailor the output to a particular research question. For example, if 87Sr/ 86Sr varies significantly and consistently with rooting depth, it might be desirable to develop separate isoscapes for grazing and browsing animals, based on shallow- and deep-rooted plants, respectively. The calRaster function in the assignR package is one potentially useful tool in this regard: it can convert isoscapes developed with one type of material (e.g. plants) into versions suitable for other applications by re-calibrating with a small number of known origin samples of the second type (e.g. butterfly wings, ratite eggshell or dental enamel). The function uses linear modelling and assumes that a systematic, linear relationship exists between the existing isoscape and the target sample material (although this may not always be the case for 87Sr/86Sr). If the relationship is chaotic or not a linear function of the existing isoscape values (visualized by the generated plots and statistics), calRaster will not improve the accuracy of the isoscape. However, it will still propagate the uncertainty inherent in the comparison between the sample substrate and the isoscape, giving a more realistic estimate of uncertainty for use in subsequent assignment calculations. It is important to recognize when this is the case and adjust your analysis accordingly. For example, a very weak relationship between the isoscape and the target sample values could mean that the selected isoscape is a poor representation of the study system. One approach would be to evaluate other isoscapes (e.g. generated using different data or methods) to figure out whether they perform better. Alternatively, a weak relationship may also demonstrate more fundamental limitations of the method. For example, in the case of contemporary North Americans studied by Verostick et al. [152], the low slope and noisy relationship shown in their fig. 3B probably reflect (in large part) the isotopic homogenization of Sr sources in contemporary humans through globalization of diet and individual travel. In these cases, it is important to recognize the aspatial variation and factor it into assignment analyses.
In a comparison of 87Sr/86Sr isoscapes of Western Europe based on soils, plants, and surface waters, Bataille et al. [45] reported offsets in the third decimal place. This is not surprising given that there are difficulties with assessing bioavailable 87Sr/86Sr in soils, as discussed above, and the patterning of 87Sr/86Sr in aquatic systems differs from that of terrestrial systems. In the southern Andes, Barberena et al. [93] reported a higher correlation between measured and predicted 87Sr/86Sr for an isoscape based on rodents compared with one based on both rodents and plants collected from the same landscape. Variation in rodent ratios was driven by a smaller number of variables, mostly related to geology, compared with variation in plants, for which several bioclimatic variables played a role. As the authors point out, in this case, the faunal isoscape is likely to be better suited to provenancing archaeological fauna and humans than the plant isoscape. However, the sample sizes used in this study were relatively small, given the large study area.
9. What are the remaining uncertainties and limitations of strontium isoscapes?
Regardless of the details of isoscape creation, robust and honest quantification of the uncertainties in the isoscape predictions is critical when evaluating the origin of unknown samples using probabilistic methods [153,154]. In these approaches, the predicted isoscape value (mean prediction and uncertainty) at any given location is used to estimate the distribution of 87Sr/86Sr at each location of the isoscape. By comparing the 87Sr/86Sr of a sample of unknown origin to the estimated 87Sr/86Sr distribution at each site, one can assess the likelihood that this sample might have come from any given location on the isoscape. Overly optimistic uncertainty estimates will give the impression that the sample origin is more tightly constrained than it really is, while pessimistic estimates will underrepresent the strength of the isotopic assignment. A common error is to confuse uncertainty estimates of the mean isoscape predictions with the variance expected among individuals originating from a common location. As we are almost always seeking to constrain the origin of individuals, isoscape uncertainty estimates should ideally incorporate the variability between individuals within a local population. This is currently not standard practice for 87Sr/86Sr, although it is for 18O/16O and 2H/XH isoscapes in which researchers frequently analyse multiple known-origin specimens from the same location (e.g. [155,156]). Different methods exist for generating robust uncertainty estimates, ranging from analytical propagation of component uncertainties to estimating 'bulk' uncertainty by comparing system-specific, known-origin data with isoscape predictions (as is done in the calRaster function discussed above, for example [153]). Each method has benefits and limitations, which are beyond the scope of this review, but isoscape developers should report how uncertainty was estimated, and users should assess the adequacy of the estimates for their application.
To illustrate the uncertainties associated with the predictive modelling approach, isoscapes should be accompanied by standard error maps. Higher predictive standard error in some regions may be the result of low sample size, skewed sample distribution, spatially heterogeneous predictor variables or a combination of these factors. These prediction error maps are particularly important when using isoscapes to assign provenance of a given specimen (e.g. in assignR), allowing a degree of assignment uncertainty integration. Areas with high prediction errors can also help identify regions with poorer sample coverage for targeted future sampling efforts. However, standard predictive error maps may not be the only tool to identify areas that require local re-sampling or cautious use; in their strontium isoscape of sub-Saharan Africa, Wang et al. [65] employed multivariate Mahalanobis distance as an indicator of environmental dissimilarity, to identify regions with unusual environmental (predictor variable) conditions not well represented by the training dataset [157].
A number of uncertainties and limitations associated with Sr isoscapes remain. They include uncertainties associated with the covariates, such as the accuracy of geological maps and model calibration for the past (e.g. changes in the position of shorelines, shifts in vegetation cover, changes in surface geology like alluvium deposition and other significant natural or human modification of the landscape). There are also uncertainties in the development of isoscapes, including how to manage regions without training data. Should one extrapolate into unsampled areas, or should they be left blank to be filled in by later studies? Extrapolation will produce predictions with increased uncertainty and runs the risk of an end-user being overly confident in the results. The decisions surrounding the risks of extrapolation can, of course, be case-specific, but leaving areas blank is advisable given that random forest cannot reliably extrapolate 87Sr/86Sr predictions into unsampled areas [158]. This is particularly true for continental-scale isoscapes with large spatial gaps between datapoints and highly heterogeneous environmental conditions, but possibly less relevant for comparatively focused isoscapes in homogeneous landscapes. Some areas may be less likely candidates for activities such as ancient human land use, based on the historical or archaeological record or even the topography. A large sedimentary basin or area of loess may have significantly less variability in 87Sr/86Sr across a large area than a mountain range, but may represent a more likely area of use by agriculturalists.
9.1. Sources of error
There are many possible sources of error during the process of sample collection and analysis. Errors can occur at the point of initial sample collection (e.g. mistakes in animal species identification, plants growing in unusual microhabitats) or during subsequent preparation (e.g. inadequate homogenization). Contamination of the sample, such as with solvents used as cleaning agents, can occur before or after the sample arrives in the lab and this can be challenging to detect. Instrument errors can result from (for example) low Sr concentrations, or interferences. It is important to analyse matrix-matched known unknowns (i.e. test samples of materials similar to the samples) and choose appropriate international and in-house standards to monitor, assess and minimize such errors.
10. Where to from here?
10.1. Adding other isotopes
Including other isotope systems alongside 87Sr/86Sr is useful for provenance studies and should increase the accuracy of assignments [159] and may expand the current boundaries of the field by making visible previously unforeseen connections. Oxygen (18O/16O, expressed as ő18O) and hydrogen (2H/1H, expressed as ŐD or ő2H) show gradients with precipitation [153,160,161]. Patterning in sulphur (34S/32S, expressed as ő34S) is more complex, but applications include proximity to the coast and identifying wetland inhabitants [162-164]. Lead isotopes can help to refine likely origins when 87Sr/86Sr is not definitive e.g. [165]). Pb derives from both natural geological sources and anthropogenic ones, such as lead pipes and leaded petrol; the extensive use of Pb-rich materials in the Roman Empire is wellknown [166,167]. It may be easy to identify contamination by anthropogenic sources. Millard et al. [168] found that if samples have Pb concentrations below 0.87 ppm, there is little risk of anthropogenic contamination, so isotope analysis is likely to provide meaningful clues to provenance. Both Pb and Sr can be isolated from the same sample, although the very low concentration of Pb in human teeth means that reliable analyses require relatively large sample sizes. The precise relationship between geological and bioavailable plant Pb has not been closely investigated, but because animals take up Pb mostly via accidental ingestion of soil or dust [169,170], rather than from food, isotope measurements for Pb-containing minerals are more usable as a tracing tool, and there is less need for mapping of bioavailable Pb isotope ratios than there is for Sr. Neodymium isotopes (143Nd/144Nd) show an inverse relationship with Sr isotopes. The mammalian gut discriminates strongly against the absorption of rare earth elements, so there are very low concentrations of Nd in mammal teeth [171]. Neodymium isotopes may therefore be less helpful for provenancing humans or other mammals, but potentially useful for materials such as ratite eggshells.
In addition to the use of other isotope systems to bolster interpretations based on 87Sr/86Sr, strontium isotopes are also important in reinforcing or challenging inferences drawn from other isotope datasets in archaeology and palaeoecology. 87Sr/86Sr can be used to trace or exclude non-local provenance and past mobility. For example, studies reconstructing regional palaeotemperatures based on ő18O of ungulate teeth have also analysed 87Sr/86Sr on intra-tooth oxygen isotope peaks and troughs to exclude the possibility that the animal was migratory; we can thus be confident that the ő18O values reflect 'local' seasonal climatic variation (e.g. [172]). Similarly, ó66Zn analysis, a relatively new method of reconstructing trophic level in ancient food webs from tooth enamel, may increasingly be paired with 87Sr/86Sr to ensure that variability in zinc isotopes reflects trophic level rather than variations in bedrock composition [173,174].
10.2. Isoscapes for forensics
The application of isotopes for provenance in forensics has garnered substantial interest, including the use of multiple isotope systems to assist in reconstructing residence and travel history for the identification or repatriation of human remains [9,175], and sourcing of illegally traded wild plants [176] and animals [43,177,178]. The value of 87Sr/86Sr is probably limited in urban human populations that buy and consume internationally sourced foods from supermarkets (although see [179]); it is likely to be more useful in the case of rural subsistence farmers. Strontium isotopes also play a role in 'food forensics' [180] to demonstrate the geographical origins of food products, sometimes required for food authentication. For example, 87Sr/86Sr has been shown to be useful in differentiating the origins of European wines (e.g. [181]), allowing the detection of counterfeit wines [182]. Along with other isotope systems, strontium isotopes are, however, used mainly as a forensic investigative tool (rather than as evidence in court) because of the limitations of their inherent uncertainty. Isotope approaches can only identify areas of probable origin or exclude other unlikely regions. Definitive proof of provenance requires additional types of forensic evidence. Nevertheless, forensic studies incorporating isotopes are a growing focus. It is worth noting that for forensic purposes, agricultural and industrial alteration of landscape 87Sr/86Sr will need to be explicitly incorporated into isoscape design, instead of being excluded as is usually the case in archaeological or other palaeo-studies.
10.3. Data repositories
As the quantity of isotopic data in the literature grows, long-term open-access storage, making data available and accessible for multiple different uses, is a goal for the field as a whole. There are multiple repositories for isotope data (e.g. [183,184]); we highlight two. IsoBank is a multi-disciplinary repository founded in 2017 with funding from the US National Science Foundation [185]. It is USAand Canada-focused but is expanding to other regions [186]. There is a strong focus on metadata requirements to promote usage in multiple, diverse applications, and there are plans to create database interfaces that will allow data to be accessed and used directly in scientific software packages that enable isoscape development and analysis. The second, IsoArcH, is an archaeology-focused, community-driven platform that stores isotopic data for academic purposes with standardized metadata definitions and language, shared using the FAIR and CARE principles [187,188]. It was originally focused on Europe and, to a lesser extent, Asia, but has expanded to incorporate over 65 000 measurements from all over the world. An excellent map function shows the geographical distribution of data. It is an open-access repository where one can create private projects with a DOI and multiple collaborators that can instantly go public upon manuscript acceptance. It requires donations to keep it running, which can be built into grant budgets. Initiatives are underway to compile and harmonize existing published bioavailable 87Sr/86Sr data for incorporation into these repositories, to provide researchers across many disciplines with a foundation for their individual research projects. Researchers are encouraged to engage with these platforms, making use of currently available data and uploading newly acquired data directly. These repositories also include valuable lists of references to published studies. Since these data repositories are open-access, there is a chance that data may be misused to produce research of questionable scientific value or that contributors to the repository may not be given proper credit. Therefore, robust peer review remains critical to achieve quality publications and ensure correct citation of the papers where the data were originally published, not just the repository itself.
10.4. Ethics and accessibility
The use of isotope analyses for investigating diet, provenance, mobility and migration can raise ethical concerns. These require consideration to ensure we work mindfully in the academic and non-academic spheres in which our research is relevant [189]. Additionally, there are ethical challenges with publishing forensic data, such as the locations of endangered plant or animal species, at risk of being exploited.
Isoscape maps should be developed using colour palettes that, where possible, are legible for people with red-green colour vision deficiency, and we should use different shapes or sizes for data points in graphics, rather than relying on colour alone to distinguish them. Various tools and palettes have been developed to support the visualization of different types of data, and many of these can be accessed directly in scientific software environments like R and Python. The colorbrewer tool and libraries (colorbrewer2.org) have several useful colour palettes, and COBLIS (color-blindness.com/coblis-colorblindness-simulator) is a useful website that simulates how a colour-blind person would see an uploaded figure.
It is a concern that many researchers around the world have little or unequal access to the expertise and facilities required to carry out 87Sr/86Sr and other isotopic analyses. Many regions have significant economic challenges that hinder exposure to and training in laboratory-based sciences. Nevertheless, there are opportunities to obtain funding for training and collaboration with established laboratories.
11. Conclusions
This article highlights the enormous potential of 87Sr/86Sr isoscapes to help address questions about the origin and movement of goods, people and animals in many different fields, including archaeology, ecology, forensics, wildlife conservation, food science and others. Major concerns for practitioners include the choice of approach to creating an isoscape. In many cases, domain mapping performs well, yielding user-friendly isoscapes. In complex situations, however, especially where there are multiple important strontium inputs, machine learning approaches (especially random forest) are probably best. These require significant developer skills and access to adequate computing capacity. The choice of samples to be collected will depend on what is available, but most researchers favour plants because they form the base of the food web and are widely available (except in extreme environments). For landscape-level isoscape construction, the most widely applicable approach to bioavailable 87Sr/86Sr determination is by sampling plants within defined geological contexts. Faunal remains (modern and/or ancient) can, where present, supplement and complement plants. We survey strategies and processes for sample collection, preparation and analysis. An important contribution of this paper is that it highlights the relatively understudied influence of biology and strontium metabolism on 87Sr/86Sr in consumer tissues, particularly with regards to dietary choices and gut absorption, and also maternal recycling of skeletal Sr during pregnancy and lactation and how this may influence the chemistry of the growing tissues of offspring. We discuss the limits of applicability of isoscapes, in terms of materials and uncertainties, and the estimation of those uncertainties and the role this should have in assignment and data interpretation. Researchers need to make their own choices about the issues explored here, depending on their field of study and research questions, but we hope that the insights offered here will guide practitioners towards better outcomes.
Ethics. This work did not require ethical approval from a human subject or animal welfare committee.
Data accessibility. Supplementary material is available online [190].
Declaration of Al use. We have not used AI-assisted technologies in creating this article.
Authors7 contributions. M.J.S.: conceptualization, funding acquisition, project administration, writing-original draft, writing-review and editing; A.A.: writing-review and editing; S.H.A.: writing-review and editing; S.B.: writing-review and editing; R.B.: writing-review and editing; C.B.: writing-review and editing; G.J.B.: writing- review and editing; K.B.: writing-review and editing; H.C.: writing-review and editing; R.D.: writing-review and editing; A.D.: writing-review and editing; J.A.E.: writing-review and editing; E.F.: writing-review and editing; K.G.: conceptualization, funding acquisition, project administration, writing-review and editing; P.H.-S.: writing-review and editing; E.H.: writing-review and editing; H.F.J.: writing-review and editing; A.J.: writing- review and editing; M.L.C.: writing-review and editing; P.I.R.: conceptualization, funding acquisition, project administration, writing-review and editing; J.L.-T.: writing-review and editing; A.M.: writing-review and editing; P.J.M.: writing-review and editing; J.M.: writing-review and editing; B.M.: writing-review and editing; V.M.O.: writing-review and editing; M.P.: writing-review and editing; M.P.R.: writing-review and editing; C.T.S.: writing-review and editing; F.S.-S.: writing-review and editing; A.S.: writing-review and editing; C.Sta.: writing-review and editing; C.Sn.: writing-review and editing; B.S.: writing-review and editing; C.Stu.: writing-review and editing; D.T.: writing-review and editing; A.G.W.: writing-review and editing; C.W.-S.: writing-review and editing; J.S.: conceptualization, funding acquisition, project administration, writing-original draft, writing-review and editing.
All authors gave final approval for publication and agreed to be held accountable for the work performed therein.
Conflict of interest declaration. We declare we have no competing interests.
Funding. We are grateful to the following organizations for supporting the workshop at which these ideas were developed: the Wenner-Gren Foundation (wennergren.org; conference and workshop grant no. 943 to J.S.), the South African Research Chairs initiative of the National Research Foundation of South Africa (nrf.ac.za; grant no. 84407 to J.S.), the MC-ICP-MS Facility, Department of Geological Sciences, University of Cape Town (UCT), the UCT Research Committee Hosting a Conference Grant (uct.ac.za), the South African Biogeochemistry Research Infrastructure Platform (BIOGRIP; biogrip.ac.za). Thanks also to the following funding bodies for supporting attendance: European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement no. 101022577 to A.S. (marie-sklodowska-curie-actions.ec.europa.eu); and Fonds Wetenschappelijk Onderzoek (FWO), Belgium (fwo.be) for a workshop travel grant to H.F.J.
Cite this article: Spies MJ etai. 2025 Strontium isoscapes for provenance, mobility and migration: the way forward. R. Soc. Open Sei. 12:250283.
Received: 9 February 2025
Accepted: 22 May 2025
Subject Category:
Earth and environmental science
Subject Areas:
ecology, biogeochemistry
Author for correspondence:
Maximilian J. Spies
e-mail: [email protected]
References
1. Price TD, Burton JH, Bentley RA. 2002 The characterization of biologically available strontium isotope ratios for the study of prehistoric migration. Archaeometry 44,117-135. (doi:10.1111/1475-4754.00047)
2. Richards M, Harvati K, Grimes V, Smith C, Smith T, Küblin JJ, Karkanas P, Panagopoulou E. 2008 Strontium isotope evidence of Neanderthal mobility at the site of Lakonis, Greece using laser-ablation PIMMS J. Archaeol. Sci. 35,1251-1256. (doi:10.1016/j.jas.2007.08.018)
3. Copeland SR, Sponheimer M, de Ruiter DJ, Lee-Thorp JA, (odron D, le Roux PJ, Grimes V, Richards MP. 2011 Strontium isotope evidence for landscape use by early hominins. Nature 474,76-78. (doi:10.1038/nature10149)
4. Copeland SR, Cawthra HC, Fisher EC, Lee-Thorp JA, Cowling RM, le Roux PJ, Hodgkins J, Marean CW. 2016 Strontium isotope investigation of ungulate movement patterns on the Pleistocene Paleo-Agulhas Plain of the Greater Cape Floristic Region, South Africa. Quat. Sei. Rev. 141,6584. (doi: 10.1016/j.quascirev.2O16.04.002)
5. Alt KW etai. 2014 Lombards on the move-an integrative study of the migration period cemetery at Szólád, Hungary. PLoS One 9, e110793. (doi:10.1371/journal.pone.0110793)
6. Song BY, Ryu JS, Shin HS, Lee KS. 2014 Determination of the source of bioavailable Sr using 87Sr/86Sr tracers: a case study of hot pepper and rice. J. Agric. Food Chern. 62,9232-9238. (doi:10.1021/jf503498r)
7. Brennan SR, Zimmerman CE, Fernandez DP, Cerling ТЕ, McPhee MV, Woolier MJ. 2015 Strontium isotopes delineate fine-scale natal origins and migration histories of Pacific salmon. Sei. Adv. 1,e1400124. (doi:10.1126/sciadv.1400124)
8. Crowley BE, Miller JH, Bataille CP. 2017 Strontium isotopes (87Sr/86Sr) in terrestrial ecological and palaeoecological research: empirical efforts and recent advances in continental-scale models. Biol. Rev. 92,43-59. (doi:10.1111/brv.12217)
9. Bartelink EJ, Chesson LA. 2019 Recent applications of isotope analysis to forensic anthropology. Forensic Sci. Res. 4, 29-44. (doi:10.1080/ 20961790.2018.1549527)
10. Stewart BA, Zhao Y, Mitchell PJ, Dewar G, Gleason JD, Blum JD. 2020 Ostrich eggshell bead strontium isotopes reveal persistent macroscale social networking across late Quaternary southern Africa. Proc. Natl Acad. Sci. USA 117,6453-6462. (doi:10.1073/pnas.1921037117)
11. Reich MS, Flockhart DTT, Norris DR, Hu L, Bataille CP. 2021 Continuous-surface geographic assignment of migratory animals using strontium isotopes: a case study with monarch butterflies. Methods Ecol. Evol. 12,2445-2457. (doi:10.1111/2041 -21 Ox.13707)
12. Barakat S, Le Corre M, Will mes M, Cohen J, Vuillien M, Desclaux E, Britton K. 2023 Laser ablation strontium isotopes and spatial assignment show seasonal mobility in red deer (Cervus elaphus) at Lazaret Cave, France (MIS 6). Front. Ecol. Evol.VL (doi:10.3389/fevo.2023.988837)
13. Saar de Almeida B, Fedele L, D'Antonio M, Morra V, Mercurio M, Stevenson R, Widory D. 2023 Characterizing wine terroir using strontium isotope ratios: a review. Isotopes Environ. Health Stud. 59,1-22. (doi:10.1080/10256016.2023.2245122)
14. Heddell-Stevens P, Jöris 0, Britton K, Matthies T, Lucas M, Scott E. 2024 Multi-isotope reconstruction of Late Pleistocene large-herbivore biogeography and mobility patterns in Central Europe. Commun. Biol. 7,1-14. (doi:10.1038/s42003-024-06233-2)
15. Montgomery J. 2010 Passports from the past: investigating human dispersals using strontium isotope analysis of tooth enamel. Ann. Hum. Biol. 37,325-346. (doi:10.3109/03014461003649297)
16. Britton K, Grimes V, Dau J, Richards MP. 2009 Reconstructing faunal migrations using intra-tooth sampling and strontium and oxygen isotope analyses: a case study of modern caribou (Rangifertarandus granti). J. Archaeol. Sei. 36,1163-1172. (doi:10.1016/j.jas.2009.01.003)
17. Holt E, Evans JA, Madgwick R. 2021 Strontium (87Sr/86Sr) mapping: a critical review of methods and approaches. Earth Sei. Rev. 216,103593. (doi: 10.1016/j.ea rsci rev.2021.103593)
18. Bataille CP, Crowley BE, Woolier MJ, Bowen GJ. 2020 Advances in global bioavailable strontium isoscapes. Palaeogeogr. Palaeoclimatol. Palaeoecol. 555,109849. (doi: 10.1016/j.palaeo.2020.109849)
19. Lewis J, Pike AWG, Coath CD, Evershed RP. 2017 Strontium concentration, radiogenic (87Sr/86Sr) and stable (688Sr) strontium isotope systematics in a controlled feeding study. Star 3,45-57. (doi:10.1080/20548923.2017.1303124)
20. Guiserix D, Albalat E, Ueckermann H, Davechand P, laccheri LM, Bybee G, Badenhorst S, Balter V. 2022 Simultaneous analysis of stable and radiogenic strontium isotopes in reference materials, plants and modern tooth enamel. Chern. Geol. 606,121000. (doi:10.1016/j.chemgeo. 2022.121000)
21. Nier АО. 1938 The isotopic constitution of strontium, barium, bismuth, thallium and mercury. Phys. Rev. 54,275-278. (doi:10.1103/physrev. 54.275)
22. Thyvalikakath P. 1965 Studies on the intestinal absorption of radioactive strontium. PhD, McGill University.
23. Merchant HA, McConnell EL, Liu F, Ramaswamy C, Kulkarni RP, Basit AW, Murdan S. 2011 Assessment of gastrointestinal pH, fluid and lymphoid tissue in the guinea pig, rabbit and pig, and implications for their use in drug development. Eur. J. Pharm. Sei. 42,3-10. (doi:10. 1016/j.ejps.2O10.09.019)
24. Blum J, Erei Y, Brown К. 1993 87Sr/86Sr ratios of Sierra Nevada stream waters: implications for relative mineral weathering rates. Geochim. Cosmochim. Acta 57,5019-5025. (doi:10.1016/S0016-7037(05)80014-6)
25. Åberg G. 1995 The use of natural strontium isotopes as tracers in environmental studies. Water Air Soil Pollut. 79, 309-322. (doi:10.1007/ BF01100444)
26. Capo RC, Stewart BW, Chadwick 0A. 1998 Strontium isotopes as tracers of ecosystem processes: theory and methods. Geoderma 82,197-225. (doi: 10.1016/s0016-7061 (97)00102-x)
27. Vitousek PM, Kennedy MJ, Derry LA, Chadwick 0A. 1999 Weathering versus atmospheric sources of strontium in ecosystems on young volcanic soils. Oecologia 121,255-259. (doi:10.1007/s004420050927)
28. Hartman G, Richards M. 2014 Mapping and defining sources of variability in bioavailable strontium isotope ratios in the Eastern Mediterranean. Geochim. Cosmochim. Acta 126,250-264. (doi: 10.1016/j.gca.2O13.11.015)
29. Serna A, Prates L, Mange E, Salazar-Garcia DC, Bataille CP. 2020 Implications for paleomobility studies of the effects of quaternary volcanism on bioavailable strontium: a test case in North Patagonia (Argentina). J. Archaeol. Sei. 121,105198. (doi: 10.1016/j.jas.2020.105198)
30. Whipkey CE, Capo RC, Chadwick 0A, Stewart BW. 2000 The importance of sea spray to the cation budget of a coastal Hawaiian soil: a strontium isotope approach. Chern. Geol. 168,37-48. (doi:l0.10l6/s0009-2541 (00)00187-x)
31. Evans JA, Montgomery J, Wildman G, Boulton N. 2010 Spatial variations in biosphere 87Sr/86Sr in Britain J. Geol. Soc. London 167,1-4. (doi:10. 1144/0016-76492009-090)
32. Alonzi E, Pacheco-Forés SI, Gordon GW, Kuijt I, Knudson KJ. 2020 New understandings of the sea spray effect and its impact on bioavailable radiogenic strontium isotope ratios in coastal environments. J. Archaeol. Sei. 33,102462. (doi:10.1016/j.jasrep.2020.102462)
33. Vet R, Artz RS, Carou S, Shaw M, Ro CU, Aas W. 2014 A global assessment of precipitation chemistry and deposition of sulfur, nitrogen, sea salt, base cations, organic acids, acidity and pH, and phosphorus. Atmos. Environ. 93,3-100. (doi:10.1016/j.atmosenv.2013.11.013)
34. Sillen A, Hall G, Richardson S, Armstrong R. 1998 87Sr/86Sr ratios in modern and fossil food-webs of the Sterkfontein Valley: implications for early hominid habitat preference. Geochim. Cosmochim. Acta 62,2463-2473. (doi:10.1016/s0016-7037(98)00182-3)
35. Brennan SR, Schindler DE, Cline TJ, Walsworth TE, Buck G, Fernandez DP. 2019 Shifting habitat mosaics and fish production across river basins. Science 364,783-786. (doi:10.1126/science.aav4313)
36. Britton K, Le Corre M, Wil imes M, Moffat I, Grün R, Mannino MA, Woodward S, Jaouen К. 2020 Sampling plants and malacofauna in 87Sr/86Sr bioavailability studies: implications for isoscape mapping and reconstructing of past mobility patterns. Front. Ecol. EvoL 8, 579473. (doi:10. 3389/fevo.2020.579473)
37. Thomsen E, Andreasen R. 2019 Agricultural lime disturbs natural strontium isotope variations: implications for provenance and migration studies. Sei. Adv. 5, eaav8083. (doi:10.1126/sci a d v.a a v8083)
38. Andreasen R, Thomsen E. 2021 Strontium is released rapidly from agricultural lime-implications for provenance and migration studies. Front. Ecol. Evol. 8,588422. (doi:10.3389/fevo.2020.588422)
39. Maurer AF, Galer SJG, Knipper C, Beierlein L, Nunn EV, Peters D, Tütken T, Alt KW, Schöne BR. 2012 Bioavailable 87Sr/86Sr in different environmental samples - effects of anthropogenic contamination and implications for isoscapes in past migration studies. Sei. Total Environ. 433,216-229. (doi:10.1016/j.scitotenv.2012.06.046)
40. Frei R, Frei KM, Jessen S. 2020 Shallow retardation of the strontium isotope signal of agricultural liming - implications for isoscapes used in provenance studies. Sei. Total Environ. 706,135710. (doi:10.1016/j.scitotenv.2019.135710)
41. Kennedy MJ, Hedin LO, Derry LA. 2002 Decoupling of unpolluted temperate forests from rock nutrient sources revealed by natural 87Sr/86Sr and 84Sr tracer addition. Proc. Natl Acad. Sei. USA 99,9639-9644. (doi:10.1073/pnas.152045499)
42. Bentley R. 2006 Strontium isotopes from the Earth to the archaeological skeleton: a review. J. Archaeol. Method Theory 13,135-187. (doi:10. 1007/s10816-006-9009-x)
43. Bowen GJ, Wassenaar LI, Hobson KA. 2005 Global application of stable hydrogen and oxygen isotopes to wildlife forensics. Oecologia 143,337348. (doi:10.1007/S00442-004-1813-y)
44. West JB, Sobek A, Ehleringer JR. 2008 A simplified GIS approach to modeling global leaf water isoscapes. PLoS One 3, e2447. (doi:10.1371/ journal.pone.0002447)
45. Bataille CP, von Holstein ICC, Laffoon JE, Wil imes M, Liu XM, Davies GR. 2018 A bioavailable Strontium isoscape for Western Europe: a machine learning approach. PLoS One 13, eO197386. (doi:10.1371/journal.pone.0197386)
46. Adams S etai. 2019 A strontium isoscape of north-east Australia for human provenance and repatriation. Geoarchaeology 34,231-251. (doi: 10.1002/gea.21728)
47. Willmes M, Bataille CP, James HF, Moffat I, McMorrow L, Kinsley L, Armstrong RA, Eggins S, Grün R. 2018 Mapping of bioavailable strontium isotope ratios in Francefor archaeological provenance studies. Appl. Geochem. 90,75-86. (doi:10.1016/j.apgeochem.2017.12.025)
48. Evans JA, Montgomery J, Wildman G. 2009 Isotope domain mapping of 87Sr/86Sr biosphere variation on the Isle of Skye, Scotland. J. Geol. Soc. 166,617-631. (doi:10.1144/0016-76492008-043)
49. Laffoon JE, Davies GR, Hoogland MLP, Hofman CL. 2012 Spatial variation of biologically available strontium isotopes (87Sr/86Sr) in an archipelagic setting: a case study from the Caribbean. J. Archaeol. Sei. 39,2371-2384. (doi:10.1016/j.jas.2O12.02.002)
50. Johnson L, Evans J, Montgomery J, Chenery C. 2022 The forest effect: biosphere 87Sr/86Sr shifts due to changing land use and the implications for migration studies. Sei. Total Environ. 839,156083. (doi:10.1016/j.scitotenv.2022.156083)
51. Hoogewerff JA etai. 2019 Bioavailable 87Sr/86Sr in European soils: a baseline for provenancing studies. Sei. Total Environ. 672,1033-1044. (doi: 10.1016/j.scitotenv.2019.03.387)
52. Breiman L. 2001 Random forests. Mach. Learn. 45,5-32. (doi:10.1023/A:1010933404324)
53. Genuer R, Poggi JM, Tuleau-Malot C. 2015 VSURF: an R package for variable selection using random forests. R J. 7,19. (doi:10.32614/rj-2015018)
54. Le Corre M, Dargent F, Grimes V, Wright J, Côté S, Reich M. 2025 An ensemble machine learning bioavailable strontium isoscape for Eastern Canada. FACETS 10,1-17. (doi:10.1139/facets-2024-0180)
55. Hengl T étal. 2021 African soil properties and nutrients mapped at 30 m spatial resolution using two-scale ensemble machine learning. Sei. Rep. 11,6130. (doi:l 0.1038/S41598-021 -85639-y)
56. Møller AB, Beucher AM, Pouladi N, Greve MH. 2020 Oblique geographic coordinates as covariates for digital soil mapping. SOIL 6, 269-289. (doi:10.5194/soil-6-269-2020)
57. Georganos S, Grippa T, Niang Gadiaga A, Linard C, Lennert M, Vanhuysse S, Mboga N, Wolff E, Kalogirou S. 2021 Geographical random forests: a spatial extension of the random forest algorithm to address spatial heterogeneity in remote sensing and population modelling. Geocarto Int. 36,121-136. (doi:10.1080/10106049.2019.1595177)
58. Sekulič A, Kilibarda M, Heuvelink GBM, Nikolič M, Bajat В. 2020 Random forest spatial interpolation. Remote Sens. 12,1687. (doi:10.3390/ rs12101687)
59. Wadoux AMJC, Minasny B, McBratney AB. 2020 Machine learning for digital soil mapping: applications, challenges and suggested solutions. Earth Sei. Rev. 210,103359. (doi:10.1016/j.earscirev.2020.103359)
60. Zhang C, Ma Y (eds). 2012 Ensemble machine learning: methods and applications. New York, NY: Springer. (doi:10.1007/978-1-4419-9326-7)
61. landmap HT. 2024 An R package for automated spatial prediction using ensemble machine learning. GitHub. See https://github.com/ Envirometrix/landmap.
62. Sengeløv A, Capuzzo G, Dalle S, James H, Sabaux C, Stamataki E. 2025 From plants to patterns: constructing a comprehensive online strontium isoscape for Belgium (IsoBel) using high density grid mapping. Geoderma 453,117123. (doi:10.1016/j.geoderma.2024.117123)
63. Wang X, Bocksberger G, Lautenschläger T, Finckh M, Meller P, O'Malley GE, Oelze VM. 2023 A bioavaila ble strontium isoscape of Angola with implications for the archaeology of the transatlantic slave trade J. Archaeol. Sei. 154,105775. (doi:10.1016/j.jas.2O23.105775)
64. Funck J, Bataille C, Rasie J, Woolier M. 2021 A bio-available strontium isoscape for eastern Beringia: a tool for tracking landscape use of Pleistocene megafauna J. Quat. Sei. 36,76-90. (doi:10.1002/jqs.3262)
65. Wang X etai. 2024 Strontium isoscape of sub-Saharan Africa allows tracing origins of victims of the transatlantic slave trade. Nat Commun. 15, 10891. (doi:10.1038/S41467-024-55256-0)
66. Evans JA, Mee K, Chenery CA, Cartwright CE, Lee KA, Marchant AP. User guide for the Biosphere Isotope Domains GB. See https://nora.nerc.ac. uk/id/eprint/520128/ (accessed 20 December 2024).
67. Ryan SE, Snoeck C, Crowley QG, Babechuk MG. 2018 87Sr/86Sr and trace element mapping of geosphere-hydrosphere-biosphere interactions: a case study in Ireland. Appl. Geochem. 92,209-224. (doi:10.1016/j.apgeochem.2018.01.007)
68. Frank AB, Frei R, Kristiansen K, Frei KM. 2022 A multi-proxy, bioavailable strontium isotope baseline for southern Almeria, Spain: using modern environmental samples to constrain the isotopic range of bioavailable strontium. Appl. Geochem. 144,105405. (doi:10.1016/j.apgeochem. 2022.105405)
69. James HF, Adams S, Willmes M, Mathison K, Ulrichsen A, Wood R, Valera AC, Frieman CJ, Grün R. 2022 A large-scale environmental strontium isotope baseline map of Portugal for archaeological and paleoecological provenance studies. J. Archaeol. Set 142,105595. (doi:10.1016/j.jas. 2022.105595)
70. Warham J. 2013 Mapping biosphere strontium isotope ratios across major lithological boundaries. A systematic investigation of the major influences on geographic variation in the 87Sr/86Sr composition of bioavailable strontium above the Cretaceous and Jurassic rocks of England. PhD, University of Bradford, https://bradscholars.brad.ac.uk/handle/10454/5500.
71. Conceiçao LT, Silva GN, Holsback HMS, Oliveira C de F, Marcante NC, Martins É de S, Santos FL de S, Santos EF. 2022 Potential of basalt dust to improve soil fertility and crop nutrition. J. Agri. Food Res. 10,100443. (doi:10.1016/j.jafr.2022.100443)
72. Szpak P. 2014 Complexities of nitrogen isotope biogeochemistry in plant-soil systems: implications for the study of ancient agricultural and animal management practices. Front. Plant Set 5,288. (doi:10.3389/fpls.2014.00288)
73. Styring AK etai. 2017 Isotope evidence for agricultural extensification reveals how the world's first cities were fed. Nat. Plants 3,1-11. (doi:10. 1038/nplants.2017.76)
74. Trandel MA, Vigardt A, Walters SA, Lefticariu M, Kinsel M. 2018 Nitrogen isotope composition, nitrogen amount, and fruit yield of tomato plants affected by the soil-fertilizer types. ЛС5 Omega 3,6419-6426. (doi:10.1021 /acsomega.8b00296)
75. Santana-Sagredo F etai. 2021 'White gold' guano fertilizer drove agricultural intensification in the Atacama Desert from AD 1000. Nat. Plants 7, 152-158. (doi: 10.1038/s41477-020-00835-4)
76. O'Regan HJ, Wilkinson DM, Wagner D, Evans J. 2023 'Why so high?' Examining discrepancies between the Sr biosphere map and archaeological tooth data from the Peak District, England J. Archaeol. Sei. 157,105826. (doi:10.1016/j.jas.2O23.105826)
77. Poszwa A, Ferry B, Dambrine E, Pollier B, Wickman T, Loubet M, Bishop K. 2004 Variations of bioavailable Sr concentration and 87Sr/86Sr ratio in boreal forest ecosystems: role of biocycling, mineral weathering and depth of root uptake. Biogeochemistry 67,1-20. (doi:10.1023/b:biog. 0000015162.12857.3e)
78. Oelze VM, Nehlich 0, Richards MP. 2012 'There's no place like home'-no isotopic evidence for mobility at the Early Bronze Age cemetery of Singen, Germany. Archaeometry 54,752-778. (doi:10.1111 /j. 1475-4754.2011.00644.x)
79. Aguzzoni A, Bassi M, Robatscher P, Scandellari F, Tirler W, Tagliavini M. 2019 Intra- and intertree variability of the 87Sr/86Sr ratio in apple orchards and its correlation with the soil 87Sr/86Sr ratio J. Agric. Food Chern. 67,5728-5735. (doi:10.1021 /acs.jafc.9b01082)
80. Snoeck C, Ryan S, Pouncett J, Pellegrini M, Claeys P, Wainwright AN, Mattielli N, Lee-Thorp JA, Schulting RJ. 2020 Towards a biologically available strontium isotope baseline for Ireland. Sei. Total Environ. 712,136248. (doi:10.1016/j.scitotenv.2019.136248)
81. Ventresca Miller AR, Winter-Schuh C, Usmanova ER, Logvin A, Shevnina I, Makarewicz CA. 2018 Pastoralist mobility in Bronze Age landscapes of Northern Kazakhstan: 87Sr/86Sr and 6180 analyses of 1284 human dentition from Bestamak and Lisakovsk. Environ. Archaeol 23,352-366. (doi: 10.1080/14614103.2017.1390031)
82. Koele N, Dickie IA, Blum JD, Gleason JD, de Graaf L. 2014 Ecological significance of mineral weathering in ectomycorrhizal and arbuscular mycorrhizal ecosystems from a field-based comparison. Soil Biol. Blochem. 69,63-70. (doi:10.1016/j.soilbio.2013.10.041)
83. Allen M. 2022 Mycorrhizal dynamics in ecological systems. Cambridge, UK: Cambridge University Press.
84. Guinoiseau D, Singh SP, Galer SJG, Abouchami W, Bhattacharyya R, Kandler К, Bristow C, Andreae MO. 2022 Characterization of Saharan and Sahelian dust sources based on geochemical and radiogenic isotope signatures. Quat. Sei. Rev. 293,107729. (doi:10.1016/j.quascirev.2O22. 107729)
85. Hoppe KA, Koch PL, Carlson RW, Webb SD. 1999 Tracking mammoths and mastodons: reconstruction of migratory behavior using strontium isotope ratios. Geology 27,439. (doi:10.1130/0091 -7613(1999)0272.3.co;2)
86. Radloff FGT, Muciņa L, Bond WJ, Ie Roux PJ. 2010 Strontium isotope analyses of large herbivore habitat use in the Cape Fynbos region of South Africa. Oecologia 164,567-578. (doi:10.1007/s00442-010-1731-0)
87. Kootker LM, van Lanen RJ, Kars H, Davies GR. 2016 Strontium isoscapes in The Netherlands. Spatial variations in 87Sr/86Sr as a proxy for pa\aeomobW\tyJ.Archaeol.Sci. 6,1-13. (doi:10.1016/j.jasrep.2016.01.015)
88. Patterson DB, Lehmann SB, Matthews T, Levin NE, Stynder D, Bishop LC, Braun DR. 2016 Stable isotope ecology of Cape dune mole-rats (Bathyergus suillus) from Elandsfontein, South Africa: implications for C4 vegetation and hominin paleobiology in the Cape Floral Region. Palaeogeogr. Palaeoclimatol. Palaeoecol. 457,409-421. (doi:10.1016/j.palaeo.2016.04.044)
89. Porder S, Paytan A, Hadly EA. 2003 Mapping the origin of faunal assemblages using strontium isotopes. Paleobiology 29,197-204. (doi:10. 1017/s0094837300018066)
90. Copeland S, Sponheimer M, Lee-Thorp J, de Ruiter D. 2010 Using strontium isotopes to study site accumulation processes. J. Taphon 8,115127. (doi:10.5167/uzh-38862)
91. Feranec RS, Madly EA, Paytan A. 2007 Determining landscape use of Holocene mammals using strontium isotopes. Oecologia 153,943-950. (doi:10.1007/s00442-007-0779-y)
92. Crowley BE, Slater PA, Arrigo-Nelson SJ, Baden AL, Karpanty SM. 2017 Strontium isotopes are consistent with low-elevation foraging limits for Henst's goshawk. Wlldl. Soc. Bull. 41,743-751. (doi:10.1002/wsb.840)
93. Barberena R etai. 2021 Bioavailable strontium, human paleogeography, and migrations in the Southern Andes: a machine learning and GIS approach. Front. Ecol. Evol. 9. (doi:10.3389/fevo.2021.584325)
94. Crook DA, Lacksen K, King AJ, Buckle DJ, Tickell SJ, Woodhead JD, Maas R, Townsend SA, Douglas MM. 2017 Temporal and spatial variation in strontium in a tropical river: implications for otolith chemistry analyses offish migration. Can. J. Fish. Aguat. Sei. И, 533-545. (doi:10.1139/ cjfas-2016-0153)
95. Brennan SR, Torgersen CE, Hollenbeck JP, Fernandez DP, Jensen CK, Schindler DE. 2016 Dendritic network models: improving isoscapes and quantifying influence of landscape and in-stream processes on strontium isotopes in rivers. Geophys. Res. Lett. 43,5043-5051. (doi:10.1002/ 2016gl068904)
96. World Health Organization. 2010 Strontium and strontium compounds. World Health Organization. See https://iris.who.int/handle/10665/ 44280.
97. Pin C, Briot D, Bassin C, Poitrasson F. 1994 Concomitant separation of strontium and samarium-neodymium for isotopic analysis in silicate samples, based on specific extraction chromatography. Anal. Chim. Acta 298,209-217. (doi:10.1016/0003-2670(94)00274-6)
98. Pin C, Gannoun A, Dupont A. 2014 Rapid, simultaneous separation of Sr, Pb, and Nd by extraction chromatography prior to isotope ratios determination by TIMS and MC-ICP-MS. J. Anal. Spectrom. 29,1858-1870. (doi:10.1039/c4ja00169a)
99. Dzubay T, Stevens R. 1975 Ambient air analysis with dichotomous sampler and X-ray fluorescence spectrometer. Environ. Sei. Technoi. 9,663668. (doi:10.1021/es60105a011)
100. Agency for Toxic Substances and Disease Registry (ATSDR). 2004 Strontium Toxicol. Profile. See https://wwwn.cdc.gov/TSP/ToxProfiles/ ToxProfiles.aspx?id=656&tid=120.
101. Jweda J, Bolge L, Class C, Goldstein SL. 2016 High precision Sr-Nd-Hf-Pb isotopic compositions of USGS reference material BCR-2. Geostand. Geoanalytical Res. 40,101-115. (doi:10.1111/j.1751-908X.2015.00342.x)
102. Raczek I, Jochum KP, Hofmann AW. 2003 Neodymium and strontium isotope data for USGS reference materials BCR-1, BCR-2, BHVO-1, BH VO-2, AGV-1, AGV-2, GSP-1, GSP-2 and eight MPI-DING reference glasses. Geostand. Newsl. 27,173-179. (doi:10.1111/j.1751 -908x.2003.tb00644.x)
103. Weber M, Lugli F, Jochum KP, Cipriani A, Scholz D. 2018 Calcium carbonate and phosphate reference materials for monitoring bulk and microanalytical determination of Sr isotopes. Geostand. Geoanalytical Res. 42,77-89. (doi:10.1111/ggr.12191)
104. Liu HC, Chung CH, You CF, Chiang YH. 2016 Determination of 87Sr/86Sr and 688/86Sr ratios in plant materials using MC-ICP-MS. Anal. Bioanal. Chern. 408,387-397. (doi:10.1007/s00216-015-9070-y)
105. Gerritzen C, Goderis S, James H, Snoeck C. 2024 Optimizing Zr-doped MC-ICP-MS sample-standard bracketing to simultaneously determine 87Sr/86Srand ð88Srfor high sample-throughput.Spectrochim. ActaPartBlVJ, 106955. (doi:10.1016/j.sab.2024.106955)
106. Waight T, Baker J, Peate D. 2002 Sr isotope ratio measurements by double-focusing MC-ICPMS: techniques, observations and pitfalls. Int. J. Mass Spectrom. 221,229-244. (doi:10.1016/s1387-3806(02)01016-3)
107. Moore LJ, Murphy TJ, Barnes IL, Paulsen PJ. 1982 Absolute isotopic abundance ratios and atomic weight of a reference sample of strontium J. Res. Natl. Bur. Stand. 87,1. (doi:10.6028/jres.087.001)
108. Jochum KP, Nohl U, Herwig K, Lammel E, Stoll B, Hofmann AW. 2005 GeoReM: A new geochemical database for reference materials and isotopic standards. Geostand. GeoanalyticalRes. 29,333-338. (doi:10.1111/j.1751 -908x.2005.tb00904.x)
109. Lugli F, Cipriani A, Bruno L, Ronchetti F, Cavazzuti C, Benazzi S. 2022 A strontium isoscape of Italy for provenance studies. Chern. Geol. 587, 120624. (doi:10.1016/j.chemgeo.2021.120624)
110. Drouet T, Herbauts J, Gruber W, Demaiffe D. 2007 Natural strontium isotope composition as a tracer of weathering patterns and of exchangeable calcium sources in acid leached soils developed on loess of central Belgium. Eur. J. Soil Sei. 58, 302-319. (doi:10.1111/j.13652389.2006.00840.x)
111. Ladegaard-Pedersen P, Frei R, Frank AB, Saracino M, Zorzin R, Martinelli N, Kaul F, Kristiansen K, Frei KM. 2022 Constraining a bioavailable strontium isotope baseline for the Lake Garda region, Northern Italy: a multi-proxy approach J. Archaeol. Sei. 41,103339. (doi:10.1016/j.jasrep. 2022.103339)
112. IS019730.2008 ISO [Internet]. See https://www.iso.org/standard/41019.html (accessed 28 November 2024).
113. Pett-Ridge JC, Derry LA, Barrows JK. 2009 Ca/Sr and 87Sr/86Sr ratios as tracers of Ca and Sr cycling in the Rio Icacos watershed, Luquillo Mountains, Puerto Rico. Chern. Geol. 267,32-45. (doi:10.1016/j.chemgeo.2008.11.022)
114. Pierson-Wickmann AC, Aquilina L, Weyer C, Molénat J, Lischeid G. 2009 Acidification processes and soil leaching influenced by agricultural practices revealed by strontium isotopic ratios. Geochim. Cosmochim. Acta 73,4688-4704. (doi:10.1016/j.gca.2009.05.051 )
115. Willmes M et al. 2014 The IRHUM (Isotopic Reconstruction of Human Migration) database - bioavailable strontium isotope ratios for geochemical fingerprinting in France. Earth Syst. Sci. Data 6,117-122. (doi:10.5194/essd-6-117-2014)
116. Nikita E, Alexander M, Cox S, Radini A, Le Roux P, Chaoua I i M, Fenwick C. 2023 Isotopic evidence for human mobility in late antique Bulla Regia (Tunisia). J. Archaeol. Sei. 47,103816. (doi:10.1016/j.jasrep.2022.103816)
117. Janzen A et al. 2020 Spatial variation in bioavailable strontium isotope ratios (87Sr/86Sr) in Kenya and northern Tanzania: implications for ecology, paleoanthropology, and archaeology. Palaeogeogr. Palaeoclimatol. Palaeoecol. 560,109957. (doi:10.1016/j.palaeo.2020.109957)
118. Nowell GM, Horstwood MSA. 2009 Comments on Richards et al., Journal of Archaeological Science 35, 2008 Strontium isotope evidence of Neanderthal mobility at the site of Lakonis, Greece using laser-ablation PIMMS' J. Archaeol. Sei. 36,1334-1341. (doi:10.1016/j.jas.2009.01. 019)
119. Müller W, Anczkiewicz R. 2015 Accuracy of laser-ablation (LA)-MC-ICPMS Sr isotope analysis of (bio)apatite - a problem reassessed. J. Anal. Spectrom. 31,259-269. (doi:10.1039/c5ja00311 c)
120. Le Corre M, Grimes V, Lam R, Britton К. 2023 Comparison between strip sampling and laser ablation methods to infer seasonal movements from intra-tooth strontium isotopes profiles in migratory caribou. Sci. Rep. 13,3621. (doi:10.1038/s41598-023-30222-w)
121. Horstwood MSA, Evans JA, Montgomery J. 2008 Determination of Sr isotopes in calcium phosphates using laser ablation inductively coupled plasma mass spectrometry and their application to archaeological tooth enamel. Geochim. Cosmoehim. Acta 72, 5659-5674. (doi:10.1016/j. gca.2008.08.016)
122. Copeland SR, Sponheimer M, Ie Roux PJ, Grimes V, Lee-Thorp JA, de Ruiter DJ, Richards MP. 2008 Strontium isotope ratios (87Sr/86Sr) of tooth enamel: a comparison of solution and laser ablation multicollector inductively coupled plasma mass spectrometry methods. Rapid Commun. Mass Spectrom. 22,3187-3194. (doi:10.1002/rcm.3717)
123. Le Roux PJ, Lee-Thorp JA, Copeland SR, Sponheimer M, de Ruiter DJ. 2014 Strontium isotope analysis of curved tooth enamel surfaces by laserablation multi-collector ICP-MS. Palaeogeogr. Palaeoclimatol. Palaeoecol. 416,142-149. (doi:10.1016/j.palaeo.2014.09.007)
124. Kendall C, Eriksen АМН, Kontopoulos I, Collins MJ, Turner-Walker G. 2018 Diagenesis of archaeological bone and tooth. Palaeogeogr. Palaeoclimatol. Palaeoecol. 491,21-37. (doi:10.1016/j.palaeo.2017.11.041)
125. Reynard B, Balter V. 2014 Trace elements and their isotopes in bones and teeth: diet, environments, diagenesis, and dating of archeological and paleontological samples. Palaeogeogr. Palaeoclimatol. Palaeoecol. 416,4-16. (doi: 10.1016/j.palaeo.2014.07.038)
126. Hoppe KA, Koch PL, Furutani TT. 2003 Assessing the preservation of biogenic strontium in fossil bones and tooth enamel. Int. J. Osteoarchaeol. 13,20-28. (doi:10.1002/oa.663)
127. Snoeck C, Lee-Thorp J, Schulting R, de Jong J, Debouge W, Mattielli N. 2015 Calcined bone provides a reliable substrate for strontium isotope ratios as shown by an enrichment experiment. Rapid Commun. Mass Spectrom. 29,107-114. (doi:10.1002/rcm.7078)
128. Tuross N, Behrensmeyer A, Eanes E. 1989 Strontium increases and crystallinity changes in taphonomic and archaeological bone. J. Archaeol. Sei. 16,661-672. (doi:10.1016/0305-4403(89)90030-7)
129. Crowley BE, Bruff Simpson EM, Hammer SJ, Smith JM, Johnson TM. 2023 Comparison of powdered enamel sample pretreatment methods for strontium isotope analysis. Front. Environ. Chern. 4. (doi:10.3389/fenvc.2023.1114807)
130. Evans JA, Mee K, Chenery CA, Marchant AP. 2023 Biosphere Isotope Domains GB (V2): interactive website. Br. Geol. Surv. (doi:10.5285/ 2ce7fc22-1 b6e-4979-968f-42058c0120fb)
131. Ma C, Vander Zanden HB, Wunder MB, Bowen GJ. 2020 assignR: an r package for isotope-based geographic assignment. Methods Ecol. Evol. 11, 996-1001. (doi:10.1111/2041-210x.13426)
132. Comar C, Wasserman R. 1963 Strontium. In Mineral metabolism (eds C Comar, F Bronner), pp. 523-572. New York, NY, USA: Academic Press.
133. Lahtinen M, Arppe L, Nowell G. 2021 Source of strontium in archaeological mobility studies-marine diet contribution to the isotopic composition. Archaeol. Anthropol. Sei. 13,1. (doi:10.1007/s12520-020-01240-w)
134. Spencer H, Warren JM, Kramer L, Samachson J. Passage of calcium and strontium across the intestine in man. Clin. Orthop. Reiat. Res. 1973. (doi: 10.1097/00003086-197303000-00031 )
135. Klusek CS. î982Strontium-90intheUSdiet. New York, NY: Environmental Measurements Lab, Department of Energy. (doi:10.2172/6685059)
136. Burton JH, Wright LE. 1995 Nonlinearity in the relationship between bone Sr/Ca and diet: paleodietary implications. Am. J. Phys. Anthropol. 96, 273-282. (doi:!0.1002/ajpa.l330960305)
137. Evans JA, Chenery CA, Montgomery J. 2012 A summary of strontium and oxygen isotope variation in archaeological human tooth enamel excavated from Britain J. Anal. Spectrom. 27,754-764. (doi:10.1039/c2ja10362a)
138. Dalle S etai. 2022 Strontium isotopes and concentrations in cremated bones suggest an increased salt consumption in Gallo-Roman diet. Sei. Rep. 12,9280. (doi:l 0.1038/s41598-022-12880-4)
139. Weber M, Tacail T, Lugli F, Clauss M, Weber K, Leichliter J, Winkler DE, Mertz-Kraus R, Tütken T. 2020 Strontium uptake and intra-population 87Sr/86Sr variability of bones and teeth-controlled feeding experiments with rodents (Rattus norvegicus, Cavia porcellus). Front. Ecol. Evol. 8, 569940. (doi:10.3389/fevo.2020.569940)
140. Johnson L, Montgomery J, Evans J, Hamilton E. 2019 Contribution of strontium to the human diet from querns and millstones: an experiment in digestive strontium isotope uptake. Archaeometry 61,1366-1381. (doi:10.1111/arcm.12485)
141. Towers J, Bond J, Evans J, Mainland I, Montgomery J. 2017 An isotopic investigation into the origins and husbandry of Mid-Late Bronze Age cattle from Grimes Graves, Norfolk J. Archaeol. Sei. 15,59-72. (doi:10.1016/j.jasrep.2O17.07.007)
142. Anders D, Osmanovic A, Vohberger M. 2019 Intra- and inter-individual variability of stable strontium isotope ratios in hard and soft body tissues of pigs. Rapid Commun. Mass Spectrom. 33,281-290. (doi:10.1002/rcm.8350)
143. Montgomery J, Evans JA, Horstwood MSA. 2010 Evidence for long-term averaging of strontium in bovine enamel using TIMS and LA-MC-ICPMS strontium isotope intra-molar profiles. Environ. Archaeol. 15,32-42. (doi:10.1179/146141010x12640787648694)
144. Yang D, Bowen GJ, Uno KT, Podkovyroff K, Carpenter NA, Fernandez DP, Ceding ТЕ. 2023 BITS: a Bayesian Isotope Turnover and Sampling model for strontium isotopes in proboscideans and its potential utility in movement ecology. Methods Ecol. Evol. 14,2800-2813. (doi:10.1111/2041 - 210x.14218)
145. Ayliffe LKef al. 2004 Turnover of carbon isotopes in tail hair and breath CO2 of horses fed an isotopically varied diet. Oecologia 139,11-22. (doi: 10.1007/s00442-003-1479-x)
146. Augustine M, Boudreau R, Cauley JA, Majchel D, Nagaraj N, Roe LS, Sood P, Stewart AF, Horwitz MJ. 2023 Bone mineral density during and after lactation: a comparison of African American and Caucasian women. Calcif. Tissue Int. 113,426-436. (doi:10.1007/s00223-023-01125-9)
147. Gulson B, Pounds J, Mushak P, Thomas B, Gray B, Korsch M. 1999 Estimation of cumulative lead releases (lead flux) from the maternal skeleton during pregnancy and lactation. J. Lab. Clin. Med. 134,631-640. (doi:10.1016/S0022-2143(99)90104-X)
148. Gulson BL, Mizon KJ, Korsch MJ, Palmer JM, Donnelly JB. 2003 Mobilization of lead from human bone tissue during pregnancy and lactation-a summary of long-term research. Sei. Total Environ. 303,79-104. (doi:10.1016/s0048-9697(02)00355-8)
149. Henderson J, Ma H, Evans J. 2020 Glass production for the Silk Road? Provenance and trade of Islamic glasses using isotopic and chemical analyses in a geological context. J. Archaeol. Sei. 119,105164. (doi:10.1016/j.jas.2020.105164)
150. Lü QQ, Chen YX, Henderson J, Bayon G. 2023 A large-scale Sr and Nd isotope baseline for archaeological provenance in Silk Road regions and its application to plant-ash glass. J. Archaeol. Sd. 149,105695. (doi:10.1016/j.jas.2022.105695)
151. Frigolé C, Ie Roux P, Sanhueza L, Lucero G, Falabella F, Cardillo M. 2024 Multi-isotope approach allows tracking the circulation of ceramics across the Andes (Argentina and Chile) J. Archaeol. Sei. 57,104684. (doi:10.1016/j.jasrep.2024.104684)
152. Verostick KA, Serna A, Stantis C, Bowen GJ. 2025 Human tissue oxygen and strontium isotope values in North America: a data compilation and assessment for forensic geolocation J. ForensicSci. 70,905-920. (doi:10.1111/1556-4029.70030)
153. Bowen GJ, West JB. 2019 Chapter 3 - isoscapes for terrestrial migration research. In Tracking animal migration with stable isotopes (eds KA Hobson, LI Wassenaar), pp. 53-84,2nd edn. Cambridge, MA: Academic Press. (doi:10.1016/B978-0-12-814723-8.00003-9)
154. Pouncett J. 2020 Spatial approaches to assignment. In Archaeological spatial analysis: a methodological guide (eds M Gillings, P Hacigüzeller, G Lock). Abingdon, UK: Routledge. (doi:10.4324/9781351243858-11 )
155. Wunder MB, West JB, Bowen GJ, Dawson TE, Tu KP. 2010 Using isoscapes to model probability surfaces for determining geographic origins. In Isoscapes, pp. 251-270. Dordrecht, The Netherlands: Springer. (doi:10.1007/978-90-481 -3354-3_12)
156. Ghouri S et al. 2024 A hydrogen isoscape for tracing the migration of herbivorous lepidopterans across the Afro-Palearctic range. Rapid Commun. Mass Spectrom. 38, e9675. (doi:10.1002/rcm.9675)
157. Owens HL etai. 2013 Constraints on interpretation of ecological niche models by limited environmental ranges on calibration areas. Ecol. Model. 263,10-18. (doi:10.1016/j.ecolmodel.2013.04.011)
158. Hengi T, Nussbaum M, Wright MN, Heuvelink GBM, Gräler В. 2018 Random forest as a generic framework for predictive modeling of spatial and spatio-temporal variables. PeerJ 6, e5518. (doi:10.7717/peerj.5518)
159. Bataille CP, Jaouen K, Milano S, Trost M, Steinbrenner S, Crubézy É, Colleter R. 2021 Triple sulfur-oxygen-strontium isotopes probabilistic geographic assignment of archaeological remains using a novel sulfur isoscape of western Europe. PLoS One 16, e0250383. (doi:10.1371/ journal.pone.0250383)
160. Bowen GJ. 2010 Isoscapes: Spatial pattern in isotopic biogeochemistry. A™. Rev. Earth Planet. Sei. 38,161-187. (doi:10.1146/annurev-earth040809-152429)
161. Mancuso CJ, Ehleringer JR. 2019 Residentand nonresident fingernail isotopes reveal diet and travel patterns. J. ForensicSci. 64,77-87. (doi:10. 1111/1556-4029.13856)
162. Tarrant D, Richards M. 2024 Modern plants and sulfur isoscapes-a review, discussion, and construction of a pilot 634S isoscape for mobility and provenance studies. Rapid Commun. Mass Spectrom. 38, e9908. (doi:10.1002/rcm.9908)
163. Lamb AL, Chenery CA, Madgwick R, Evans JA. 2023 Wet feet: developing sulfur isotope provenance methods to identify wetland inhabitants. R. Soc. Open Sei. 10,230391. (doi:10.1098/rsos.230391)
164. Richards M. 2023 Sulphur isotope ratios of archaeological human collagen as a mobility indicator. In Isotopic proveniencing and ability: the current state of research (ed T Price), pp. 187-197. Cham, Switzerland: Springer International Publishing. (doi:10.1007/978-3-031 -25722-3_8)
165. Evans JA, Pashley V, Mee K, Wagner D, Pearson MP, Fremondeau D, Albarella U, Madgwick R. 2022 Applying lead (Pb) isotopes to explore mobility in humans and animals. PLoS One VI, e0274831. (doi:10.1371 /journal.pone.0274831 )
166. Montgomery J, Evans J, Chenery S, Pashley V, Killgrove K. 2010 Gleaming, white, and deadly: using lead to track human exposure and geographic origins in the Roman period in Britain J. Roman Archaeol. 78,199-226.
167. Delile H, Blichert-Toft J, Goiran JP, Keay S, Albarède F. 2014 Lead in ancient Rome's city waters. Proc. Natl Acad. Sei. USA 111,6594-6599. (doi: 10.1073/pnas.1400097111)
168. Millard AR, Montgomery J, Trickett M, Beaumont J, Evans J, Chenery S. 2014 Childhood lead exposure in the British Isles during the industrial revolution. See https://durham-repository.worktribe.com/output/1680173/childhood-lead-exposure-in-the-british-isles-during-theindustrial-revolution (accessed 12 November 2024).
169. Kameňov GD, Gulson BL. 2014 The Pb isotopic record of historical to modern human lead exposure. Sei. Total Environ. 490,861-870. (doi:10. 1016/j.scitotenv.2014.05.085)
170. Kameňov GD, Krigbaum J. 2023 Pb isotopes and human mobility: natural, cultural, or diagenetic signal? In Isotopicproveniencing and mobility: the current state of research (ed. TD Price), pp. 163-185. Cham, Switzerland: Springer International Publishing. (doi:10.1007/978-3-031-257223_7)
171. Plomp E, von Holstein ICC, Koornneef JM, Smeets RJ, Baart JA, Forouzanfar T, Davies GR. 2019 Evaluation of neodymium isotope analysis of human dental enamel as a provenance indicator using 1013 0 amplifiers (TIMS). SeiJustice 59,322-331. (doki0.1016/j.scijus.2O19.02.001)
172. Pederzani S et al. 2021 Reconstructing Late Pleistocene paleoclimate at the scale of human behavior: an example from the Neandertal occupation of La Ferrassie (France). Sci. Rep.11,1419. (doi:10.1038/s41598-020-80777-1)
173. Bourgon N etai. 2020 Zinc isotopes in Late Pleistocene fossil teeth from a Southeast Asian cave setting preserve paleodietary information. Proc. Natl Acad. Sci. USA 117,4675-4681. (doi:10.1073/pnas.1911744117)
174. Jaouen K etai. 2020 Zinc isotope variations in archeological human teeth (Lapa do Santo, Brazil) reveal dietary transitions in childhood and no contamination from gloves. PLoSOne 15, e0232379. (doi: 10.1371 /journal.pone.0232379)
175. Ubelaker DH, Francescutti C. 2020 The role of stable isotope analysis in forensic anthropology. In Forensic science and humanitarian action: interacting with the dead and the living (eds RC Parra, SC Zapico, DH Ubelaker), pp. 273-284. Hoboken, NJ: John Wiley & Sons, (doi: 10.1002/ 978lll9482062.ch19)
176. Retief K, West AG, Pfab MF. 2014 Can stable isotopes and radiocarbon dating provide a forensic solution for curbing illegal harvesting of threatened cycads? J. ForensicSci. 59,1541-1551. (doi:10.1111/1556-4029.12644)
177. van der Merwe NJ, Lee-Thorp J, Thackeray F, Hall-Martin A, Kruger F, Coetzee H, Bell RHV, Lindeque M. 1990 Source-area determination of elephant ivory by isotopic analysis. Nature 346,744-746. (doi:10.!038/346744a0)
178. Hopkins JB, Frederick CA, Yorks D, Pollock E, Chatfield MWH. 2022 Forensic application of stable isotopes to distinguish between wild and captive turtles. Biology 11,1728. (d oi : 10.3 3 90/bio logy 11121728)
179. Tipple BJ, Valenzuela LO, Ehleringer JR. 2018 Strontium isotope ratios of human hair record intra-city variations in tap water source. Sci. Rep. 8, 3334. (d o i : 10.1038/s41598-018-21359-0)
180. . Kelly S, Heaton K, Hoogewerff J. Tracing the geographical origin of food: The application of multi-element and multi-isotope analysis. Trends Food Sei. Technoi.. 2005;16:555-567. doi:W.1016/j.tifs.2005.08.008
181. Horn P, Schaaf P, Holbach В, Hölzl S, Eschnauer H. 1993 87Sr/86Sr from rock and soil into vine and wme.ZLeb. Unters Forch 196,407-409. (doi: 10.1007/BF01190802)
182. Epova EN, Bérail S, Séby F, Vacchina V, Bareille G, Médina B, Sarthou L, Donard OFX. 2019 Strontium elemental and isotopic signatures of Bordeaux wines for authenticity and geographical origin assessment. Food Chem. 294,35-45. (doki0.1016/j.foodchem.2019.04.068)
183. Pezo-Lanfranco L, Mut P, Chavez J, Fossile T, Colonese AC, Fernandes R. 2024 South American Archaeological Isotopic Database, a regional-scale multi-isotope data compendium for research. Sci. DataAV 336. (doi:10.1038/s41597-024-03148-9)
184. Billings TN, Scott E, Cocozza C, Hixon S, Boivin N, Roberts P, Spengler RN, Fernandes R. 2025 The North American Repository for Archaeological Isotopes. Sei. DataU, 50. (doi:10.1038/s41597-024-04175-2)
185. Pauli JN et al. 2017 Why we need a centralized repository for isotopic data. Proc. Natl Acad. Sei. USA 114, 2997-3001. (doi:10.1073/pnas. 1701742114)
186. Shipley 0, Dabrowski A, Bowen G, Hayden B, Pauli J, Jordan C. 2024 Design, development, and implementation of IsoBank: a centralized repository for isotopic data. PLoS One 19, e0295662. (doki 0.1371/journal.pone.0295662)
187. Salesse K, Fernandes R, de Rochefort X, Brůžek J, Castex D, Dufour É. 2018 IsoArcH.eu: an open-access and collaborative isotope database for bioarchaeological samples from the Graeco-Roman world and its margins. J. Archaeol. Sei. 19,1050-1055. (doki 0.1016/j.jasrep.2O17.07.030)
188. Plomp E etai. 2022 The IsoArcH initiative: working towards an open and collaborative isotope data culture in bioarchaeology. Data Brief 45, 108595. (doi:!0.10l6/j.dib.2022.108595)
189. Stantis letal. 2024 Ethics and applications of isotope analysis in archaeology. Am J. Biol.Anthropol. 186, e24992. (doki 0.1002/ajpa.24992)
190. Spies MJ, Alblas A, Ambrose SH, Barakat S, Barberena R, Bataille C et al. 2025 Supplementary material from: Strontium isoscapes for provenance, mobility, and migration: the way forward. Figshare, (d o i : 10.6084/m 9.f ig s h a re.c.7873 380)
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2025. This work is published under http://creativecommons.org/licenses/by/4.0/ (the "License"). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Strontium isotopes (87Sr/86Sr) are increasingly used as a provenance tool in multiple disciplines. Application to biological materials requires knowledge of the variation in bioavailable 87Sr/86Sr across the landscape, potentially in the form of an isoscape (a quantitative model of spatial isotopic variability). This paper summarizes and provides advice on our current understanding of the main concerns in creating and interpreting isoscapes of bioavailable 87Sr/86Sr. Isoscape creation approaches include domain mapping, geostatistical contour mapping and machine learning, the last becoming more readily achievable with the availability of software packages. It is critically important to develop isoscapes at a resolution appropriate for addressing the research questions. Choice of sample materials depends on the research questions and availability: plants or fauna with small ranges are favoured, with some analytes (snails, soil leachates) posing challenges. Interpreting 87Sr/ 86Sr in biological tissues requires considering Sr metabolism and the timing of tissue formation, thus far underappreciated. The numerous sources of error involved in developing and applying isoscapes must be recognized to avoid over-interpreting data and spurious provenance precision. We hope this paper will help researchers investigating provenance, mobility, landscape use and migration to develop the most appropriate isoscapes for their purposes, and possible future use by others.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details
1 Department of Archaeology, biogeochemistry Research Infrastructure Platform (BIOGRIP),University of Cape Town, Cape Town, South Africa
2 Division of Clinical Anatomy, Faculty of Medicine and Health Sciences, Stellenbosch University, Cape Town, South Africa
3 Department of Anthropology, University of Illinois, Urbana, IL, USA
4 Department of Archaeology, University of Aberdeen, Aberdeen, UK
5 Dentro de Investigación, Innovación y Creación (CIIC), Universidad Católica de Temuco Facultad de Ciencias Sociales y Humanidades, Temuco, Araucania, Chile