Bioavailable soil and rock strontium isotope data from Israel

Strontium isotope ratios (87Sr / 86Sr) of biogenic material such as bones and teeth reflect the local sources of strontium ingested as food and drink during their formation. This has led to the use of strontium isotope ratios as a geochemical tracer in a wide range of fields including archaeology, ecology, food studies and forensic sciences. In order to utilise strontium as a geochemical tracer, baseline data of bioavailable 87Sr / 86Sr in the region of interest are required, and a growing number of studies have developed reference maps for this purpose in various geographic regions, and over varying scales. This study presents a new data set of bioavailable strontium isotope ratios from rock and soil samples across Israel, as well as from sediment layers from seven key archaeological sites. This data set may be viewed and accessed both in an Open Science Framework repository (https://doi.org/10.17605/OSF.IO/XKJ5Y, Moffat et al., 2020) or via the IRHUM (Isotopic Reconstruction of Human Migration) database.


Introduction
Strontium (Sr) isotope geochemistry has applications in many fields of research including archaeology (Bentley, 2006;Slovak and Paytan, 2012), ecology (Barnett-Johnson et al., 2008;Hobson et al., 2010), food traceability (Voerkelius et al., 2010) and forensic sciences (Beard and Johnson, 2000). Strontium is widely distributed within geological and biological materials, and the strontium isotope ratios ( 87 Sr / 86 Sr) of these materials reflect the sources of strontium in the environment during their formation (Dasch, 1969). The use of traditional isotope systems such as hydrogen and oxygen systems in provenance studies is limited by their broad gradients across the Earth's surface, whereas 87 Sr / 86 Sr ratios show predictable geographic variability determined by lithology, with some limited temporal variability (Aggarwal et al., 2008;Willmes et al., 2018). To utilise the potential of strontium isotopes as a geochemical tracer, prior knowledge of baseline data in the region of interest is required. Bedrock strontium isotope ratios are a product of the age, mineralogy and origin of source material Faure and Mensing, 2005). The amount of 87 Sr within a geological unit typically increases over time through the radioactive decay of 87 Rb to 87 Sr, while the amounts of 84 Sr, 86 Sr and 88 Sr remain constant . In general, older rocks such as granite in continental crust have higher 87 Sr / 86 Sr ratios than younger rocks with a lower Rb / Sr ratio, such as oceanic basalt . The 87 Sr / 86 Sr ratios of carbonates reflect the water from which they precipitated, either seawater for marine carbonates or a local water body for non-marine carbonates (Faure, 1986;Neat et al., 1979). Clastic sedimentary rocks typically have 87 Sr / 86 Sr ratios which reflect the lithology of their source material, tempered by the relative ease with which minerals containing strontium are removed by weathering (McDermott and Hawkesworth, 1990). The 87 Sr / 86 Sr ratios observed for geological units of varying compositions and ages are discussed further by Capo et al. (1998), Faure and Mensing (2005), Faure and Powell (1972), and others.
The strontium isotope composition of regolith is principally derived from local weathering of bedrock  but may be augmented by windblown marine, mineral and anthropogenic aerosols; different mineral weathering rates; transportation of regolith; and pore water chemistry (Frumkin and Stein, 2004;Goede et al., 1998). Bioavailable strontium is the component of strontium present in an environment which is available for incorporation into biological systems. Regolith, in combination with water, is the principal source of bioavailable strontium, although processes including precipitation, sea spray and fertiliser use may also affect the isotopic composition of bioavailable strontium (Bentley, 2006;Frei and Frei, 2011;Price et al., 2002;Slovak and Paytan, 2012).

Strontium isotope mapping
There are a range of approaches to mapping bioavailable strontium, as summarised below and in greater detail in Bentley (2006), Maurer et al. (2012), Slovak and Paytan (2012) and Bataille et al. (2020). Archaeological human or faunal samples are considered to be one of the best indicators of the local bioavailable range, although their geographic origin may be uncertain (Maurer et al., 2012;Price et al., 2002). Modern faunal samples from known localities may also be used, although non-local food sources and fertiliser use need to be considered (Bentley, 2006;Maurer et al., 2012). Soils, plants and water may also be measured for local bioavailable strontium isotope ratios (Evans et al., 2010;Maurer et al., 2012;Price et al., 2002). While soils and plants average strontium over a smaller area, ground water and surface waters may provide an estimate of bioavailable strontium over a wider area, depending on catchment size (Evans et al., 2010;Willmes et al., 2014). As discussed previously, the strontium isotope ratios of plants, soils and water may be affected by numerous factors including precipitation, sea spray and fertiliser use, which may vary over seasonal and annual timescales (Bentley, 2006;Hoogewerff et al., 2019;Price et al., 2002). Local bioavailable strontium isotope ratios may also be modelled from bedrock taking into account age, mineralogy and weathering rates . In the case of soil and rock samples, an ammonium acetate or ammonium chloride digestion method, which removes cations from electrostatically bound pore water in addition to exchange sites on minerals and organic matter, can be used as a means of extracting the bioavailable component of these samples . This ap-proach has been found to extract between 0.1 % and 62 % of whole soil strontium (Chadwick et al., 2009).
At a country scale, strontium isoscapes have been developed for Mesoamerica (Hodell et al., 2004), the United Kingdom (Evans et al., 2010), France (Willmes et al., 2014 and Denmark (Frei and Frei, 2011) from a combination of bedrock, soil, plant and water samples. Isoscapes of bioavailable strontium have also been modelled for the contiguous USA , western Europe Hoogewerff et al., 2019) and the Caribbean region . At a larger scale, Voerkelius et al. (2010) sampled mineral waters across western Europe to predict bedrock 87 Sr / 86 Sr but found local processes to result in unexpected values.

Strontium isotope studies and mapping in Israel
Previous studies in Israel and across the Levant region have utilised strontium isotope analyses in a variety of applications in addition to undertaking small-scale mapping of bioavailable strontium. Herut et al. (1993) measured the 87 Sr / 86 Sr ratios of rainwater samples across Israel and used these results in conjunction with the chemical composition of the sample to determine the sources of Sr and soluble salts. Shewan (2004) developed a map of bioavailable strontium throughout Israel using modern faunal bones and grass samples, for comparison to bones from archaeological sites in the region. The work of Shewan (2004) and Perry et al. (2008), who measured 87 Sr / 86 Sr from archaeological faunal dental samples in western Jordan, was collated to create a bioavailable strontium map for Israel and Jordan, combining modern flora and fauna with material from archaeological sites (Perry et al., 2009). These studies highlight several distinct provinces in the region, with samples along the Jordan Valley distinguished from those in the eastern highlands of Israel and the Western Highlands in Jordan, as well as several smaller areas with varying isotopic signatures in the north of the study area (Perry et al., 2009). Rosenthal et al. (1989) report the strontium isotope composition of gastropod shells to reconstruct water sources in the Dead Sea Rift area of the Jordan Valley. Spiro et al. (2011) measured 87 Sr / 86 Sr ratios of water and mollusc shells in the Hula catchment in the Upper Jordan Valley, adjacent to the Golan Heights, and found distinct water sources and aquifers. Stein et al. (1997) sampled water and sediments around the Dead Sea to determine water sources and the evolution of the Dead Sea and its precursor, Lake Lisan, identifying two distinct periods of lake evolution. Hartman and Richards (2014) sampled plants and invertebrates in northern Israel and the Golan Heights to produce a map of modern bioavailable strontium isotope ratios, and to investigate potential sources of variability, including intersite variability and the influence of precipitation. Arnold et al. (2016) sampled plants in the vicinity of Tell es-Safi/Gath to create a local bioavailable strontium map as a baseline for use in interpreting the mobility of domestic animals from ar-chaeological sites in the region. Moffat et al. (2012) undertook spatially resolved strontium isotope analysis using laser ablation techniques to demonstrate annual migration using a bovid tooth from the archaeological site of Skhul. These studies have highlighted the value and potential of bioavailable strontium mapping in this region, and the data presented in this study aim to add to this growing knowledge base.

Sample collection
Soil and rock samples were collected throughout Israel in September and October 2008. Sample locations were chosen opportunistically, with reference to a digital 1 : 200 000 geological map of Israel using the Old Israel Grid co-ordinate system (Sneh et al., 1998), to provide the greatest representation of the stratigraphic units present in Israel that were accessible via road. A single rock and soil sample was taken at each site, with no replicates. Soil samples were collected from the surface, at a single point, with no attempt made to sample multiple soil horizons where present. Rock samples were taken for the principal geological unit present at each site and a brief description of the lithology made in the field, which can be viewed in the data file. Sediments from stratigraphic layers in seven archaeological sites, Amud, the Atlit Railway Bridge site, Neve David, Qafzeh, Sefunim Cave, Skhul and Tabun, were also sampled for analysis via fieldwork or from archival collections. Where possible, sediments from archaeological sites were collected for multiple stratigraphic units. Sample locations and archaeological sites are illustrated in Fig. 1.

Sample treatment
Rock and soil samples were heated to 60 • C for a minimum of 24 h prior to sample preparation to comply with Australian quarantine procedures. After heating, rock samples were crushed to a powder using a hand piston. Rock and soil samples were passed through a 2 mm sieve, and 1 g aliquot of the sieved sample was leached by adding 2.5 mL of 1 M ammonium nitrate (NH 4 NO 3 ), following DIN (German Industrial Standard) ISO (International Organization for Standardization) 197310 (2009). Samples were shaken for 24 h and subsequently centrifuged at 3000 rpm for 5-10 min. 1 mL of supernatant was extracted and evaporated until dry, before being dissolved in 2 mL of 2 M high-purity nitric acid (HNO 3 ).
The concentration of strontium was determined by ICP-AES (inductively coupled plasma atomic emission spectrometry). Ion exchange chromatography was then used to isolate strontium from other elements, particularly 87 Rb to prevent isobaric interference with 87 Sr (Dickin, 2000), using two columns filled with Eichrom strontium-specific resin (prefilter resin and strontium specific resin). The strontium concentration determined by ICP-AES was used to determine the amount of sample added to the ion exchange column. Samples were diluted prior to MC-ICP-MS (multi-collector inductively coupled plasma mass spectrometer) analysis to provide sufficient additional samples to allow for reanalysis if necessary.

Neptune MC-ICP-MS analysis
A Neptune MC-ICP-MS was used to measure strontium isotope ratios in the Environmental Geochemistry and Geochronology Laboratory at the Research School of Earth Sciences, Australian National University (ANU). The isotopes measured and Faraday cup configuration used for the analysis are shown in Table 1. Data reduction was performed offline and includes a blank, 87 Rb isobar and an exponential mass bias correction.
To ensure precision, accuracy and reproducibility in the data produced, the samples were run in a sequence, which included blank and standard samples. The strontium carbonate reference material SRM987 (National Institute of Standards and Technology) was measured on the Neptune MC-ICP-MS during the sample sequence to quantify instrument drift. During the analysis period, measurements of SRM987 ranged from 0.71012 ± 0.00001 to 0.71028 ± 0.00001, with a mean of 0.71022 ± 0.00003 (n = 32). The soil and rock results were not calibrated based on these SRM987 measurements.

Results
A total of 60 soil samples and 48 rock samples were analysed for 87 Sr / 86 Sr. The 87 Sr / 86 Sr ratios of the soil samples range from 0.70577±0.00001 to 0.71020±0.00003. The 87 Sr / 86 Sr ratios of the rock samples range from 0.70529 ± 0.00001 to 0.74072 ± 0.00001. These results are illustrated on satellite images of Israel in Fig. 2 (rock samples) and Fig. 3 (soil samples). Gross lithologies, defined by observations in the field and geological maps of the region (Sneh et al., 1998), are used to group the results of Sr isotope analysis. The soil and rock samples analysed show some differences in 87 Sr / 86 Sr ratios between the different lithologies sampled, as illustrated in Fig. 4, although there is some over-  lap between most lithologies. Figure 4a displays all samples analysed, while Fig. 4b displays an inset removing a high 87 Sr / 86 Sr granite sample (site IS044) and a high 87 Sr / 86 Sr rhyolite sample (site IS047), to display the rest of the data set more clearly. Basalt samples are generally less radiogenic in 87 Sr / 86 Sr ratio than the other lithologies sampled for both soil and rock samples. Soil samples from carbonate (limestone, dolostone, chalk and marl), granite, kurkar (aeolian quartz sandstone with carbonate cement), rhyolite, siliciclastic lithologies and areas with no bedrock have comparable 87 Sr / 86 Sr ratios (Fig. 4b). The median 87 Sr / 86 Sr ratios from rock samples are slightly lower than those from soils for basalt, carbonate and kurkar samples (Fig. 4b). The 87 Sr / 86 Sr ratios measured from granite rock samples (two samples) and the rhyolite rock sample are substantially elevated compared to other units, and compared to the soil samples from the same lithologies. At 43 of the sample locations, both soil and rock samples were collected, and the 87 Sr / 86 Sr ratios of these samples are compared in Fig. 5. As with Fig. 4, Fig. 5a illustrates the entire data set, while Fig. 5b has a granite rock sample (site IS044) and a rhyolite sample (site IS047) removed to better display the variation in the rest of the data set. The variation between soil and rock 87 Sr / 86 Sr ratios for samples collected from the same site ranges from 0.00001 to 0.03130. This offset between rock and soil samples is also visualised in Fig. 6, in which the reference line indicates where rock and soil sam-  ples have the same 87 Sr / 86 Sr ratios for both rock and soil samples (y = x). Points which lie above this reference line have higher 87 Sr / 86 Sr ratios for soil samples than for rock samples, which in this study is the case for most basalt, carbonate, siliciclastic and kurkar samples. Points which lie be-low the y = x reference line have higher 87 Sr / 86 Sr ratios for rock samples compared to soil samples; in this study this is the case for granite and rhyolite samples. The mean offsets from rock to soil 87 Sr / 86 Sr ratios of samples collected from the same sites are shown in Table 2, grouped by lithology. Sediments from seven archaeological sites were also analysed for 87 Sr / 86 Sr: 4 from Amud, 1 from the Atlit Railway Bridge site, 2 from Neve David, 5 from Qafzeh, 3 from Sefunim Cave, 3 from Skhul and 16 from Tabun. These results are illustrated in Fig. 7, with samples R_1024 and R_1026 from Amud highly elevated compared to the rest of the data set (Fig. 7a). Figure 7b displays a subset of the data excluding these elevated samples to better display the variability in the rest of the samples. For the sites other than Amud, the samples from Qafzeh show the largest range in 87 Sr / 86 Sr ratios between layers, while for Neve David, Sefunim Cave, Skhul and Tabun, most samples have similar 87 Sr / 86 Sr ratios.

Comparison between rock and soil 87 Sr / 86 Sr ratios
The difference between the 87 Sr / 86 Sr ratios of soil and rock sampled from the same site is in most cases greater than the analytical uncertainty (Fig. 5), which may be due to inputs to the soil from sources other than bedrock, such as sea spray, irrigation or Saharan aeolian dust, or alternately due to differences in the weathering rates of minerals. Rainwater collected across Israel has been found to contain strontium associated with sea spray, marine minerals and dust from carbonate minerals, the relative proportions of which affect the 87 Sr / 86 Sr ratio (Herut et al., 1993).
The input of aeolian dust is expected to be particularly important to the 87 Sr / 86 Sr ratios of these soil samples, as it has been reported that aeolian material may make up to 50 % of soils formed on hard limestone rocks in Israel (Yaalon, 1997). The 87 Sr / 86 Sr isotope ratio of dust in the region is known to have varied over time, which is an important consideration in the use of modern isoscapes for the interpretation of archaeological studies. 87 Sr / 86 Sr ratios of dust in this region varied from 0.711-0.712 during the marine isotope stage 2 and 4, and 0.709-0.710 during MIS 1, 3 and 5 (Haliva-Cohen et al., 2012). The delivery of aeolian dust 3648 I. Moffat et al.: Bioavailable soil and rock strontium isotope data from Israel  is also affected by climate variations (Frumkin et al., 2011), with Sr isotope ratios found to increase at major climate transitions that correspond to sapropel formation in the Mediterranean (Stein et al., 2007). Speleothem analyses have shown that colder and drier conditions are associated with higher Sr isotope ratios due to an increase in sea spray and aeolian dust (Ayalon et al., 1999;Frumkin and Stein, 2004).  (Madella et al., 2002). These elevated 87 Sr / 86 Sr ratios are likely due to the ash component in these sediments, and the variable nature of the interbedded sediments at this site.  (Herut et al., 1993). The wide range of 87 Sr / 86 Sr isotope results in the Golan Heights region may be explained by the rapid depletion of Sr from rocks and soil in this region due to weathering and the large contribution of aeolian dust to soil profiles (Singer, 2007, 202-206). Carbonate units were the most sampled lithology in this study, due to their widespread distribution across the coun- try. This lithology groups a range of carbonate units (limestones, dolostones, chalk and marl), and the 87 Sr / 86 Sr ratios of rock samples range from 0.70733 to 0.70911, while soil 87 Sr / 86 Sr ratios range from 0.70803 to 0.71020. Hartman and Richards (2014) report bedrock 87 Sr / 86 Sr ratios from northern Israel for carbonate units to range from 0.7073 to 0.7078, comparable to the lower range of rock samples measured in this study. Ligneous plant samples have 87 Sr / 86 Sr ratios ranging from 0.70789 to 0.70910, nonligneous plants from 0.70791 to 0.70919 and invertebrates from 0.70807 to 0.70876 (Hartman and Richards, 2014), comparable to the results of this study from rock and soil samples. Rosenthal et al. (1989) summarise 87 Sr / 86 Sr isotopes from carbonate-derived water sources in the Jordan Valley as having 87 Sr / 86 Sr ratios ranging from 0.7070 to 0.7080. Arnold et al. (2016) sampled plants in the vicinity of Tell es-Safi/Gath to create a local bioavailable strontium map as a baseline for use in interpreting the mobility of domestic animals from archaeological sites in the region. Of the 10 plant samples collected, 4 were from kurkar soils, with 87 Sr / 86 Sr ratios ranging from 0.70863 to 0.70881. The kurkar soil 87 Sr / 86 Sr ratios reported in this study are slightly more elevated, ranging from 0.70894 to 0.70934. Hartman and Richards (2014) also measured 87 Sr / 86 Sr ratios from plants and invertebrates from kurkar units, ligneous plants ranged from 0.70899 to 0.70903, non-ligneous plants ranged from 0.70900 to 0.70903, and two mollusc shells were measured with 87 Sr / 86 Sr ratios of 0.70897 and 0.70904, comparable to the kurkar soil samples reported in this study. Coastal plain grasses were analysed by Shewan (2004) in the region of El Wad cave, south of Haifa. These 87 Sr / 86 Sr ratios ranged from 0.70886 to 0.70965, with an outlying 87 Sr / 86 Sr ratio of 0.71003 attributed to influence from the road base (Shewan, 2004). The results from the kurkar samples in this study are comparable to those of coastal plain grasses as reported by Shewan (2004).

Comparison with regional data sets
Samples from granite, rhyolite and siliciclastic lithologies, as well as samples from regions with no bedrock, are also reported in this study. Hartman and Richards measured 87 Sr / 86 Sr ratios from plants growing in alluvium, with a range of 0.70712 to 0.70830 for combined ligneous and nonligneous (n = 4) plants. From this study, samples sites where no bedrock was found had 87 Sr / 86 Sr ratios ranging from 0.70783 to 0.70952, comparable to the results of Hartman and Richards (2014). There are no comparable studies which sample bioavailable 87 Sr / 86 Sr for the granite and rhyolite samples included in this study.

Data availability
The data set can be viewed and downloaded on the IRHUM (Isotopic Reconstruction of Human Migration) database (http://www.irhumdatabase.com, last access: 23 November 2020). The IRHUM database architecture and functionality are described by Willmes et al. (2014). The data are also available at the Open Science Framework data repository "Data Associated with Bioavailable Soil and Rock Strontium isotope data from Israel" https://doi.org/10.17605/OSF.IO/XKJ5Y (Moffat et al., 2020).

Conclusion
This data set represents a substantial contribution of bioavailable soil and rock 87 Sr / 86 Sr ratios for Israel, to complement and build on previous research. The 87 Sr / 86 Sr results are shown to be principally controlled by lithology and are in broad agreement with previous smaller-scale studies in the region. Soil and rock 87 Sr / 86 Sr ratios from the same site are generally offset, and determining the mechanisms behind this offset could be the focus of future research to supplement this study.
Author contributions. RG conceived the study. IM collected sample material. IM, MW, GM, LK, LM, RA and MA undertook the analysis. IM and RR wrote the first draft and produced the figures. All authors reviewed the paper.