Finding Mycenaeans in Minoan Crete? Isotope and DNA analysis of human mobility in Bronze Age Crete

We undertook a large-scale study of Neolithic and Bronze Age human mobility on Crete using biomolecular methods (isotope analysis, DNA), with a particular focus on sites dating to the Late Bronze Age (‘Late Minoan’) period. We measured the strontium and sulphur isotope values of animal remains from archaeological sites around the island of Crete to determine the local baseline values. We then measured the strontium and sulphur values of humans from Late Neolithic and Bronze Age sites. Our results indicate that most of the humans have sulphur and strontium isotope values consistent with being local to Crete, showing no evidence for a wide-scale movement of people from the Greek mainland or other areas away from Crete in these time periods. However, we found four individuals from the late Bronze Age (Late Minoan III) cemetery of Armenoi with sulphur isotope values not typically found in Crete and are instead consistent with an origin elsewhere. This cemetery at Armenoi also has one of only a few examples of the newly adopted Mycenaean Linear B script on Crete found outside of the palace sites, pointing to an influence (trade and possible migration) from the mainland, which may then be the place of origin of these four individuals. DNA (mtDNA) studies of eight Late Bronze Age individuals from Armenoi have results consistent with people living in Aegean region at this time and cannot be used to distinguish between individuals from Crete (‘Minoans’) and the Greek mainland [‘Mycenaeans’]).


Isotope analysis of faunal remains
We measured the strontium isotope values of 93 zooarchaeological specimens from 10 sites across Crete dating from the Neolithic to Late Bronze Age ( Table 1). The average faunal    The strontium and sulphur isotope values of the animals were very similar across the island, despite there being a variety of underlying geologies (Fig 1). We attribute this to the contribution of strontium and sulphur into soils from both sea spray (the 'sea-spray effect') and rainfall, which resulted in a more homogenous range of values than we may have predicted. Of note is that the majority of the fauna (90 of 93) and humans (49 of 52) had strontium values lower than the marine strontium isotope ratio of 0.7092 [21]. This indicates that the sea-spray effect where marine strontium was deposited in soils was not significant enough to overwhelm the contributions to soil strontium or sulphur from underlying bedrock (modern-day marine sulphate has a value of δ 34 S = 20.3‰ [22]). And instead most of the bioavailable baseline strontium and sulphur isotope values on Crete are combination of seaspray/rainfall input and the dominant geological substrate, limestones [23].
It is clear in the faunal data (and other data sets [24]) that in general the underlying geologies of Crete impart lower (compared to marine) sulphur and strontium isotope values to fauna on the island and that the sea-spray effect would generate some increases in the values of bioavailable elements. In this context high values (at or above marine) can be considered unusual values. Three faunal samples have strontium values at, or higher than, the marine strontium average values ( Table 2). These are from a cattle tooth from Zoniana in central Crete, a pig tooth from Nerokourou and a sheep/goat tooth from Kastelli, which are both sites in West Crete (Fig 1). It is possible that these were imported to the island from a region with higher strontium values, such as the Greek mainland or other areas of the Mediterranean, such as Egypt, where we know there was trade with Bronze Age Minoan Crete [25]. It is also possible that these were from a region or regions of Crete with higher strontium values, such as has recently been reported for a region of East Crete [26], which were not sampled in our study.
Similarly, the average sulphur isotope values of fauna did not show significant input from higher marine sulphur values (i.e. values at or near to present-day marine sulphate with a value of δ 34 S = 20.3‰ [22]). One pig from the site of Chamalevri in western Crete had a much higher sulphur value than the other fauna, with a value of 16.7‰ ( Table 2). This value likely indicates that this was in imported animal, although the strontium value for this animal (0.709030) is well within the range of the average ('local') faunal values for Crete. And as with the strontium results, it is possible that this pig came from a region of Crete with high sulphur values that we did not sample.

Isotope results from humans
We measured the strontium values of 52 humans from 9 Neolithic and Bronze Age sites across Crete. We also measured the sulphur isotope values of 72 humans from 4 Neolithic and Bronze Age sites. These results were then compared to the baseline isotope map of Crete developed using strontium and sulphur isotope measurements of animals from these and other archaeological sites across Crete.
The sulphur and strontium values of humans from these sites are plotted with the values for sulphur and strontium that we obtained for animals from Crete in Figs 4 and 5. The majority of humans had strontium isotope values lower than the marine average strontium value of 0.7092, and most have values that fall within the range of the average strontium value for all fauna from Crete. The exception to this is a number of individuals from a single site (Armenoi) which show values above 0.7092. Likewise, most of these humans also had sulphur values within the average range of the fauna we measured from Crete, but as with the strontium values, some of the individuals from the site of Armenoi had higher sulphur isotope values than the faunal average. We then interpret this result as indicating that the majority of the humans we measured had strontium and sulphur isotope values consistent with their originating from Crete.
The isotope evidence indicates then that the majority of these humans can be considered 'local' to Crete. This may indicate a somewhat insular Neolithic and especially Late Minoan society where there was limited migration of people to the Island, and the cemeteries contained mostly local people. This is, of course, is based on the background isotope maps we have produced using faunal enamel strontium values and bone collagen sulphur values, and these baseline values may change with additional future sampling. Another important point is that we have very little data for the strontium and sulphur values from these time periods from the surrounding regions of the eastern Mediterranean, meaning that migrants from regions with similar isotope values to Crete would be invisible as migrants to Crete using these isotope methods.
One important site of note, however, is the Late Minoan (LMIII) cemetery (necropolis) site of Armenoi [19,27,28] where four individuals showed sulphur isotope values that were higher (>2 s.d.) than any other measured sulphur values (humans and fauna) from Armenoi itself or other sites across Crete (Table 3). Similarly, two of these same individuals (ARM 501, ARM 504) also showed the highest strontium isotope ratios for humans and higher than all but two of the zooarchaeological samples in our study-higher even than the marine strontium value. Indeed, these two individuals, and a third individual also from Armenoi, are the only humans we measured with strontium values higher than the marine average strontium value. As strontium values record location during childhood tooth enamel formation [29], it could be argued that two of these individuals then may have lived elsewhere than Crete during their childhoods. Sulphur isotope values are from bone collagen, which can reflect dietary intake of sulphur over many years, but specifically in adulthood, so is a useful companion isotope to strontium for mobility studies [22]. Four individuals from Armenoi had sulphur values above 2.s.d from the mean value observed for animals and humans from Crete, they likely lived elsewhere in the years before their death. There was no clear evidence of marine food consumption by these individuals, which may have had an influence on their sulphur or strontium isotope values. The carbon and nitrogen data show a terrestrial-based diet for these humans ( Table 4), something commonly observed for Bronze Age Greece [30] with the notable exception of Mycenae [31].
The carbon isotope evidence also shows there was no significant consumption of C 4 foods by humans or animals. It should be noted that the carbon and nitrogen data presented here are from new sample preparations and measurements in Leipzig so the results may differ slightly from the earlier measurements on some of the same samples measured at Oxford [31].
The two individuals with both outlier strontium and sulphur values likely spent their childhood and much of their adult lives away from Crete and perhaps then were recent arrivals in Crete before their death and internment in Armenoi.

DNA results for humans
The DNA preservation was overall rather poor, with the percentage of endogenous human DNA in every sample between 0.01 and 0.18 (Table 6). Due to the low yield of human shotgun reads, the genetic sex could not be determined. Nevertheless, the mitochondrial capture resulted in 10 complete or nearly complete mtDNA genomes, with reads showing patterns of C-to-T and G-to-A misincorporations at the 5' and 3' ends, respectively, which are typical of   authentic ancient DNA. We estimate the mitochondrial contamination as 3% or lower for all 10 genomes. The mtDNA evidence shows that most of the people we analysed had sequences typical for this region and time period [19]. For the ten individuals analysed from Late Bronze Age Armenoi and Neolithic Gerani, we did not find any individuals with DNA haplogroups that are unexpected for this region. At Armenoi we likely detect several maternal related individuals, who all carry the identical haplogroup H59.

Discussion
The biomolecular evidence presented here strongly points to little movement of people between Crete and the surrounding regions during the Neolithic and Bronze Age periods. However, at the end of the Bronze Age, in the postpalatial Late Minoan period (LMIII) we found four individuals with non-local isotope values suggesting they may have originated from outside of Crete and were then buried in the Late Minoan cemetery of Armenoi. The site of Armenoi dates to the Late Minoan IIIB (LMIIIB) period and contains over 230 burial features. It has been excavated by Y. Tzedakis over a period of more than 40 years, first starting in 1969 and has many rock-cut burial features (chamber tombs) of different sizes [19,27,28]. Some of the larger tombs contained grave goods including pottery and burials in large free-standing ceramic larnakes. Many of the other tombs also contained pottery, and some of the smaller ones had no grave goods at all. These chamber tombs usually contained either multiple or single burials. Most of the pottery dates to the LMIIIB period, and there is the remarkable find of a pottery vessel (stirrup jar) from tomb 146 (which has DNA results presented in this paper, but we were unable to measure isotope values of humans from this tomb) that is inscribed with Linear B script (the inscription is 'wina-jo'). The Linear B writing script was not in use in Early or Middle Bronze Age Crete (which instead used Linear A script), but it was in widespread use at Late Bronze age Mycenaean sites on the Greek mainland. Therefore, Armenoi (and some of the palace sites, such as Knossos, where this new Linear B script is the main script used on tablets in the Late Bronze Age) may have been settled by people from the mainland who used this Mycenaean script or may have been imported from the mainland through trade or other contacts and was then locally adopted as the main script. It is possible then that the four individuals with 'non-local' sulphur isotope values (two of these individuals also had higher strontium isotope values than the marine strontium value and our bioavailable faunal baseline average value) were new Mycenaean settlers or traders (especially the two individuals with both strontium and sulphur outlier values) from the mainland that settled at or near to Armenoi and the existing cemetery was used for burials of these newcomers.
While the mtDNA sequences do not offer conclusive evidence for the origins of the studied individuals, previously published nuclear DNA data from Armenoi specimen 503 showed that this individual was genetically distinct from preceding Cretans, and in her genetic profile more similar to contemporaneous Myceneans from the mainland in that she also harbored ancestry derived from Bronze Age steppe pastoralists that Minoans pre-LMIII lacked [19].

Previous work
There have been two previous studies of strontium isotopes applied to Bronze Age archaeological sites in Crete. The first, Nafplioti [32] presented strontium data from the site of Knossos. As with our study, most of the humans measured showed enamel strontium values below the marine strontium value of 0.7092. There were, however, a number of exceptions, with a few individuals having values higher than the marine strontium value, and with strontium isotope values that were very similar to the two individuals with these high values from Armenoi reported here. Interestingly, Knossos is the main palace site on Crete where Linear B script has been found in the Late Bronze Age (LMIII) period, and Linear B was also found at Armenoi.
A second, more recent study [26] applied strontium isotopes to the early and middle Bronze Age site of Sissi in east Crete. Here, strontium values of four snails from near to the site were used as being indicative of the 'local' strontium value. Three of these modern snail shells showed an average value below 0.7092 and within the range of values we observed for zooarchaeological enamel values from the island (0.70830 +/-0.00062). The fourth snail shell showed a much higher strontium isotope value of 0.70974, higher than all of the zooarchaeological enamel strontium values we measured in our study across Crete. Interestingly at this site, the human enamel samples almost all had strontium values higher than the marine value of 0.7092, which is much higher than the majority of humans and animal data presented here and previously, as well as in three of the four snails from the site. Is this perhaps one localized region of the island that does have higher strontium values than elsewhere? Clearly, further work is needed to better establish the strontium baselines across Crete at higher resolution. With an increased dataset from Crete perhaps we will be able to provide additional evidence that this area of Crete did have high strontium baseline values in the Bronze Age, and also if there are other areas of Crete that also may have had these higher strontium values at this time. With more strontium baseline values from Crete (especially from the surrounding areas of the eastern Mediterranean) we will be better able to indicate possible locations where the individuals we and others have identified with higher strontium values (greater than the marine value of 0.7092) may have originated.
One study attempted to produce a widescale map of the bioavailable strontium values from across all of Greece, and included data from six sites on Crete [24] (and these results are also included in a more recent paper by Frank et al. [33]) The strontium data from these six sites derive from archaeological bone samples (not tooth enamel), which is not a reliable substrate for strontium measurements as it is often contaminated by post-depositional uptake of strontium from soil [29,34]. Although these results should be used with caution, which was also acknowledged by Nafplioti [24], the results from these six sites (that include some of the same sites we reported in our study) all show strontium values lower than the marine strontium value and are in line with the strontium values we obtained from animal teeth from the same, or nearby, sites.

Conclusions
The archaeology of Neolithic and Bronze Age Crete is among the longest and most studied regions and time periods in the world. This is particularly the case for the Late Bronze Age-Late Minoan periods. The end of the Bronze Age and the associated first appearance of Mycenaean material culture and architectural styles on Crete is particularly intriguing. Here we sought to add to the body of evidence for human mobility in these important time periods using biomolecular isotope and DNA methods. Our results point to a largely insular Neolithic and Bronze Age Crete, with little evidence of the movement of people to the island. We did find the exciting possibility of identifying newcomers to Crete at the end of the Minoan period, perhaps from mainland Mycenean Greece, at the LMIII necropolis/cemetery site of Armenoi, a site which also has evidence of Mycenaean Linear B script on a pottery vessel from this site at this time. Of course, this could have been an adoption of the written language and material culture from the mainland, or part of a trade network with Mycenaeans, rather than a movement of people. However, if our isotope evidence does indeed show that there were some individuals buried at Armenoi at this transition period at the end of the Minoan era that did not originate from Crete, we may then have evidence that this new writing style was brought to Crete by people moving from mainland Mycenaean Greece who were then buried at the cemetery in Armenoi.

Study sites
We sampled Neolithic and Bronze faunal and humans for isotope and DNA analysis from across Crete for this study. We sought to sample as many sites as possible across a wide geographic range of Crete. We sampled as many sites as we were able to, depending on the availability and accessibility of the material, and the generous collaborations with our many collaborators and the Greek Ministry of Culture. The list of sites and their chronology is given in Table 1 and their locations are plotted in Fig 1. Samples were taken by MPR, HM and KD, with zooarchaeological identifications by KD. Permits for sample export and analysis for all samples were obtained from the Greek Ministry of Culture. All necessary permits were obtained for the described study, which complied with all relevant regulations.

Isotope analysis
Background. The use of strontium isotope analysis to trace mobility has been widely used in archaeology since the late 1980's [29,35]. Strontium in mammal tooth enamel originates from food and (to a lesser extent) water in diets and it substitutes for calcium during tooth enamel formation. The strontium isotope ratio between amounts of two isotopes of strontium, 87 Sr and 86 Sr (reported as 87 Sr/ 86 Sr) that were present in food and water are preserved in the enamel when the strontium is deposited, so are directly related to the strontium isotope ratio ( 87 Sr/ 86 Sr) of food and water that was being consumed during enamel formation. In turn the food (i.e. plants and animals) and water strontium ratios are related to the soils and water environment where the food was grown or raised. Therefore, the strontium isotope ratio in enamel can be used to tell us the geographical source of the food that contained the strontium that was eventually deposited in teeth. To use this strontium isotope method as a mobility indicator for humans, it is essential to understand the 87 Sr/ 86 Sr ratios of the geographical areas that were potential food and water sources. This is usually done by first exploring the geological maps of a region and predicting the strontium isotope values of plants and soils based on the age and type of the underlying bedrocks. As the actual human strontium isotope ratios are not directly related to bedrock geology strontium, where possible a map of the strontium that is actually incorporated into plants and animals ('bioavailable' strontium) needs to be produced, often by measuring the strontium of animal remains from different regions, as well as sampling modern plants and molluscs (e.g. [36,37]). Once the map of strontium isotope ratios in a region is produced, the human 87 Sr/ 86 Sr ratios are compared to the map to determine if the values match the 'local' values, or not ('non-local). With this method then, the best use of it is to determine if humans are local or non-local to a region, and if they are non-local, it is not usually possible to determine definitively their origin as there simply are not enough region-specific detailed bioavailable strontium isotope maps available (however see [38] for a first attempt at a global scale-modelled map), therefore many regions of the world do not have these baseline strontium values based on empirically-derived data. Also, it is important to note that for humans, enamel is usually formed during childhood (depending on the specific tooth) so the strontium values of the enamel of permanent teeth of an adult likely indicates where that person was getting their food from while they were children.
Sulphur isotope analysis, on the other hand, is measured in bone collagen so can be used as an indicator of mobility in later life, depending on which bone was sampled [22]. Similar to strontium, sulphur is derived from diet, however sulphur is deposited in bone collagen in the amino acid methionine [22,39,40]. Sulphur is both a dietary and mobility indicator, as it is very useful to determine which ecosystem dietary protein was derived from, and can be used to determine freshwater proteins (e.g. fish) compared to terrestrial based proteins (e.g. terrestrial herbivore meat and milk) [41,42]. As plant sulphur isotope values are related to the soil sulphur isotope values these ratios in human collagen can, in a very similar way to strontium, be used to determine the location where an individual obtained most of their foods when the collagen was being synthesised and deposited within bones.
The combination of strontium and sulphur isotope analysis is then a powerful tool to help reconstruct the life history of a person, allowing the comparison of childhood location with where they were living in later life.
Sample preparation. Samples of tooth and bone were prepared and analysed at the Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany. Collagen was extracted from bone for carbon, nitrogen and sulphur isotope analysis in 0.5M HCl at 5˚C following established protocols outlined in [43] with the addition of an ultrafiltration step [44]. The %C, %N and where measured, %S were determined for each collagen extract and only those with acceptable collagen preservation criteria (% collagen yield, C:N and C:N:S ratios, [45,46]) were included here, and these values are given in Table 4. Carbon and nitrogen isotope measurements were undertaken using a Thermo Flash EA coupled to a Thermo Delta V continuous flow isotope ratio mass spectrometer (CF-IRMS). Sulphur isotope measurements of collagen were undertaken on a Heka EuroVector elemental analyser Thermo-Finnigan Delta V plus following procedures outlined in [40]. Isotope ratios were calculated in relation to international and internal standards. Specifically, methionine (Brad-001) and NBS Liver 1577b internal standards (both externally certified) and the international IAEA N1, N2, CH6 and CH7 standards for carbon and nitrogen measurements. The sulphur measurements were calculated using the international standards NBS127, IAEA S1, S2, S3, SO-5 and NIST1577b. Errors on the isotope measurements are (1σ) δ 13 C ± 0.1‰, δ 15 N ± 0.2‰, and δ 34 S ± 0.5‰, based on long-term measurements of the internal secondary standards and international standards listed above.
The strontium isotope measurements of the faunal and human teeth were made on solutions of purified and isolated strontium prepared from sampled tooth enamel. The preparation protocol is described in [47]. This involved digestion of~20mg enamel, previously mechanically abraded and ultrasonicated in Deionized (DI) H 2 O (18.3MO), in 1ml of double-distilled 14.3M nitric acid (HNO 3 ) at 120˚C for 1 hour within a closed 3ml PFA vial (Savillex). The resulting solution was dried and the sample residue transferred in 1ml of 3M HNO 3 to a 2ml chromatography column (Eichrom) containing previously cleaned Sr-spec resin (Eichrom). Following three successive 1ml washes of 3M HNO 3 , strontium was eluted from the column into a clean 3ml PFA vial (Savillex) using 2ml of DI H 2 O. The strontium containing solution was then dried and re-dissolved in 3% HNO 3 for analysis. The strontium isotope ratios were measured using a Thermo Fisher Neptune MC-ICP-MS with instrument parameters and data collection method outlined in [48]. Each sample 87 Sr/ 86 Sr value is the result of 50 analyses (i.e., 50 cycles at 2 second integrations) and involved interference (from krypton and rubidium) and mass bias ( 88 Sr/ 86 Sr = 8.375209) normalization [49] and inverse mass bias correction. Within the data collection analytical sessions, the NIST strontium carbonate standard SRM987 was also measured as a 3% HNO 3 solution at 200ppb Sr and gave 87 Sr/ 86 Sr = 0.710262 (+/-0.000017 1σ, n = 30). All 87 Sr/ 86 Sr sample data were externally adjusted such that SRM987 87 Sr/ 86 Sr = 0.710240, resulting in typical data corrections of <0.00004, and total procedural blanks were considered a negligible (e.g. <0.5%) contribution to the sample 88 Sr signal. The strontium isotope data is given in Table 5.

DNA analysis
We processed 16 samples from 12 individuals from the sites Armenoi and Gerani for DNA analyses (Table 6). Pre-PCR steps took place in the clean room facilities of the Institute for Archaeological Sciences at the University of Tübingen, Germany. After surface irradiation with ultraviolet light, the tooth was sawn apart transversally at the border of crown and root, and dentine powder from the inside the crown was sampled using a sterile dentistry drill. Bone surface at the sample area was removed with a dental drill, before extracting bone powder using a new drill bit. DNA extraction was carried out on around 100 mg of dentine powder per sample according to Dabney et al. [50]. DNA libraries were prepared from 10 or 20μl of extract each [51,52], and enriched for human mitochondrial DNA using a bead-based hybridization protocol [53]. Negative controls were included in the extraction and library preparation and taken along for all further processing steps. Sequencing of shotgun libraries and mitochondrial-DNA-enriched libraries took place at the facilities of the Frauenklinik of the University of Tübingen on an Illumina HiSeq 2500 for 2 × 101 + 8 cycles.
Base call files produced by the instrument's software were converted to raw sequences that were demultiplexed according to the index combinations they received during library preparation. The software pipeline EAGER [54] was used to clip adapter sequences, map reads to the reference-hg19 for shotgun data and Reconstructed Sapiens reference sequence (RSRS) for mitochondrial capture data-using Burrows-Wheeler Aligner (BWA) [55] with parameters "-l 10000 -n 0.01," and to remove duplicate reads. Mitochondrial consensus sequences were called while jointly estimating the rate of deamination damage and contamination from present-day human sources with the probabilistic iterative method applied in the software schmutzi [56] using the accompanying tool log2fasta with parameter -q 20. Poly-C regions and mutational hotspots at positions 303-315, 515-522, and 16519 were masked. Sex was assigned on the basis of the shotgun-sequencing data [57] and results are reported S1 Table, but we caution that these are extremely uncertain due to the low number of mapping shotgun reads. Mitochondrial haplogroups were assigned manually by consulting the output of Haplofind [58] and Haplogrep2.0 [59].
Supporting information S1

Author Contributions
Conceptualization: Michael Richards.