Reconstructing Middle and Upper Paleolithic human mobility in Portuguese Estremadura through laser ablation strontium isotope analysis

Significance Laser ablation MC-ICP-MS allows in situ strontium isotope data to be obtained for incrementally formed bioapatites such as enamel with extremely high spatial resolution. Here, we provide a large-scale application of the method comparing the mobility and subsistence behavior of Middle and Upper Paleolithic humans in the same landscape. These remains and the fauna analyzed alongside come from the Almonda karst system (Portuguese Estremadura). Data suggest that regional Middle Paleolithic individuals roamed across a subsistence territory of approximately 600 km2, while Upper Paleolithic individuals moved seasonally and exploited a smaller territory of approximately 300 km2.

Over the course of these analyses, two known-value bioapatite standards were repeatedly analysed.An ashed and pressed bovine bone pellet known as BP (produced at the University of Bristol as part of a previous study; [1]) was utilised, along with several isotopically homogenous pig teeth produced during a feeding study at the University of Bristol (4).Across analysis days, repeated measurement of BP (n=29) gave a mean offset and standard error of 124 ± 95 ppm over the reference TIMS value, whilst the repeat analysis of the enamel from PT 232 (n=32) and PT 238 (n=78) gave a mean offset and standard error of 65 ± 60 and 103 ± 31 ppm over the TIMS values, respectively.
Prior to isotopic analysis, human enamel samples were surface cleaned by rapidly ablating the plotted laser path to remove any surface contaminants.For this purpose, a spot size of 150μm, a repetition rate of 10Hz and a 50μm/s -1 translation rate was utilised for human teeth.For surface cleaning of the animal teeth, which are generally larger, the translation rate was increased to 100μm/s -1 .For human enamel samples, the subsequent analysis along the pre-ablated laser track used a 150μm spot size, a repetition rate of 20Hz and a translation rate of 2μm/s -1 ; for fauna the translation rate was increased, depending on the specimen, to between 30μm/s -1 and 50μm/s -1 to account for the greater length of the samples.Gas blank data were collected for 60 seconds prior to each analysis with the same parameters, except with the laser shutter closed. 89Y was monitored as a proxy for rare earth elements (REEs) which can be taken up diagenetically (5) and present numerous isobaric interferences in the mass range of strontium.In one sample (N15-17), the data was trimmed in a region where higher levels of 89 Y covaried with 87 Sr/ 86 Sr.
Sediment samples were collected at locations representative of the different geological and soil types within a c. 50km radius of the sites.Subsoil samples were taken away from locations of obvious agriculture.Soils and sediments were leached for two weeks in RO 18Ω water at room temperature.The leach solution was filtered and divided into aliquots for the measurement of strontium concentration using a ThermoFisher Scientific XSeries2 ICP-MS, and for isotopic analysis using a ThermoFisher Scientific Triton Plus TIMS.Both aliquots were processed on ~50ml Sr-Spec resin (Triskem, France) columns using 2ml of 3M sub-boiled nitric acid to elute and the strontium collected in 1.5ml of MQ water.The samples were dried and loaded in 1.5ml of sub-boiled 1M hydrochloric acid onto an outgassed tantalum filament using a tantalum activator solution for subsequent TIMS analysis, using a multi dynamic procedure with an 88 Sr beam of 2V.Fractionation was corrected using an exponential correction normalized to 86 Sr/ 87 Sr = 0.1194.
For oxygen isotope analysis, a second enamel section parallel to the first was taken from the tooth crown.Surface dirt and any dentine was removed using a dental bur.The sections were then subdivided into sequential samples, which measured between 2-5mm long depending on the length of the tooth.Each of these sub-samples was ground into a powder using an agate pestle and mortar, and subjected to a 0.1M acetic acid wash for 15 minutes in order to remove exogenous carbonates.Each sample was then washed five times in ultrapure Milli-Q water and freeze-dried to return them to a powdered state.
Analysis was carried out at the Stable Isotope Mass Spectrometry Laboratory at the National Oceanography Centre, Southampton.Approximately 500μg of each sample was transferred to a Thermo KEIL IV carbonate device and automatically reacted with 106.7% phosphoric acid at 90°C in order to evolve CO2.The CO2 was then dried and cryogenically transferred to a Thermo Finnegan MAT 253 isotope ratio mass spectrometer.The oxygen isotope data are presented as δ-values in permil (‰) relative to the international standard Vienna Standard Mean Ocean Water (VSMOW).The typical measurement uncertainty, based on the repeated analysis of international and in-house standards, is 0.2‰.

S2. Consideration of diagenetic uptake of strontium
The uptake of Sr by teeth has the potential to alter or overprint the biogenic signal, confounding the interpretation of movement from Sr isotope values.While it has been observed that dentine and bone are generally very susceptible to Sr uptake, enamel is usually considered more immune because of its lower porosity and lower organic content (e.g. ( 6)).However, although diagenetic Sr signals in tooth enamel are rarely reported in the literature, the vast majority of archaeological Sr isotope studies are on fauna and humans from later prehistory to the recent past (e.g. ( 7)), and there are insufficient studies from older periods to assess the susceptibility of tooth enamel to Sr uptake over longer periods of time.
Estimates of the rate of uptake of Sr by tooth enamel from faunal teeth in contexts inundated by the North Sea (i.e.waterlogged conditions with known diagenetic Sr isotope ratio) suggest that biogenic Sr signals in enamel can become overprinted in 15-150 ka (8), although these timescales will be longer in drier conditions where the burial matrix pores and tooth enamel pores will be less coupled.Nevertheless, one should exercise caution when assuming measured Sr isotopic values are biogenic in enamel older than a few tens of thousands of years.
We compare distributions of Sr isotopic values between the enamel and dentine as a check for diagenetic Sr uptake in one of the oldest teeth in this study (OLV9, c. 93 ka; Fig. S10).The similarities in the variation of Sr isotopes between the enamel and the dentine show that even the more susceptible dentine has not suffered significant diagenetic Sr uptake.The effect of such uptake on the dentine (assuming it is uniform) would be to move the apparent Sr isotopic values towards the diagenetic value and to reduce the magnitude of the variation of the Sr isotopes in the measured profile.If the enamel were also affected by Sr uptake, it is likely to be less affected than the dentine (Lewis estimates 100-fold less, based on porosity differences (8)), and the apparent Sr isotope values of the enamel and dentine would diverge.It is significant therefore that the Sr isotope values in the enamel and dentine are broadly similar, both have a similar variation, and neither overlap with the likely diagenetic Sr isotope ratio, assumed to reflect Mesozoic limestone, i.e. in the range of 0.7068 -0.7077 (9).We therefore conclude that since the dentine has not been significantly affected by Sr uptake then we can discount Sr uptake in the enamel.
To have a biogenic, or near-biogenic Sr isotopic signal preserved in dentine is exceptional.Sr isotopic measurements on dentine (or bone) are usually made as part of the process of investigating the 'local' isotopic range for Sr, on the assumption that uptake will mean that the Sr values will reflect the Sr in the immediate burial environment rather than anything biogenic.There are cases where suites of teeth from the same context have been used to show partial overprinting of the biogenic signal in dentine -e.g.((10) Fig. 5) and ( 11) -but there are very few cases reported where the dentine is not affected by significant Sr uptake.In Oliveira, It is likely that the carbonate cementation of some of the cave sediments, and the formation of carbonate crusts (12) limited the water percolation through the deposits containing the Neanderthal teeth.This would reduce the coupling of the pore water in the burial matrix with the pores of the enamel and dentine, limiting or halting the diffusion of Sr into the teeth.

S3. Sediment sampling rationale
Sediment samples were collected at locations representative of the different geological and soil types within a c.50 km radius of the sites, away from locations of obvious agriculture.The field trips to acquire the samples took place on July 14-15, 2014.Selection of sampling locations aimed at characterizing how soils' Sr content varied as a function of the different combinations of bedrock and soil profiles present in the region.
In terms of rock types and ages, the geology of this sector of the western Iberian Peninsula is very diverse and includes Proterozoic and Paleozoic metamorphic rocks belonging to the geological basement, Paleozoic igneous lithotypes, Mesozoic sedimentary rocks (often carbonate ones), as well as Cenozoic and Quaternary sedimentary rocks and sediments (mostly terrigenous).Geological data were gathered from the different sheets of the Geological Map of Portugal (1:50,000 and 1:100,000), available on the Geoportal of LNEG (Laboratório Nacional de Energia e Geologia; https://geoportal.lneg.pt/mapa/).Due to Pleistocene uplifting and subsequent erosion (e.g. ( 15)), the soil profiles of this sector of Portugal are often derived from Tardiglacial or Holocene soil formation.Ancient soils are preserved only within karst features or structural depressions.For this reason, soil properties are often strongly controlled by the physico-chemical characteristics of their parent material.Most soil samples were collected from cambisols, fluvisols or regosols, and only a few specimens come from podzols or luvisols.Soil data were taken from (16).

S4. Stable isotope analysis
Stable carbon ( 13 C/ 12 C) and nitrogen ( 15 N/ 14 N) isotope analysis of collagen was carried out in order to directly investigate the diet of the Upper Palaeolithic individual from Galeria da Cisterna.Collagen preservation within the Almonda karst system is highly variable, limiting the range of comparative faunal data available with which to generate an isotopic baseline.A total of three Upper Palaeolithic red deer from Lapa dos Coelhos are available.Collagen was extracted from bone (fauna) and dentine (human) samples following a modified Longin (17) method at the University of Southampton.
Aliquots of collagen extracted at the University of Southampton were run in duplicate using a Thermo EA1110 elemental analyser coupled to a dual-pumped Sercon 20-20 stable isotope mass spectrometer (OEA Laboratories).Results were corrected to certified standards (δ 13 C; vPDB; δ 15 N; AIR) and are presented in Fig S11 and Table S5 .
The average δ 13 C and δ 15 N values for the Lapa dos Coelhos red deer are -20.22 ± 0.8‰ and 3.82 ± 0.45‰ respectively; consistent with terrestrial C 3 herbivores.The Upper Palaeolithic human from Galeria da Cisterna exhibits a collagen δ 13 C value of -18.65‰ and a δ 15 N value of 10.89‰.The latter is 7.07‰ more enriched than the average δ 15 N values of the red deer; two trophic positions higher based on an assumed 15 N trophic enrichment factor of 3.4‰ (18).This suggests that species occupying higher trophic positions likely contributed to the dietary protein intake of the individual.Given the presence of contemporaneous anadromous (Salmo sp. and Alosa sp.) and freshwater (Cyprinidae sp. and Barbus sp.) fish remains (along with fishhooks) in the deposits of Lapa dos Coelhos, it is likely that aquatic protein contributed to the elevated δ 15 N value.The vertebrae of four fish specimens from Lapa dos Coelhos were selected for δ 13 C and δ 15 N analysis, but unfortunately no collagen could be recovered.

S5. Consideration of dietary strontium concentration
It is important to note that strontium isotope profiles may be skewed towards the 87 Sr/ 86 Sr values of resources with higher strontium concentrations.Vegetables and fish, for example, contain an order of magnitude more strontium per kilogram than red meat (19).Certain nuts that were likely available to Middle and Upper Palaeolithic humans can exhibit even greater strontium concentrations than meat, vegetables, and fish.Molluscs in particular have been found to hyperaccumulate strontium by three orders of magnitude in freshwater communities (20,21).These variations in strontium concentration can complicate the interpretation of strontium isotope profiles in tooth enamel -for example, in a diet made up predominantly of meat, it is feasible that the majority of the strontium could be derived from plant foods (22).Likewise, an omnivorous diet may be swamped by strontium derived from a comparatively small contribution of marine molluscs.
The carbon and nitrogen isotope values of the Galeria da Cisterna human, combined with the presence of the remains of freshwater and anadromous fish (along with fishing implements) at Lapa dos Coelhos suggest that these freshwater resources did contribute to the diet of the Upper Palaeolithic individual, and therefore the enamel 87 Sr/ 86 Sr values may be weighted towards those resources.In theory, we therefore have to consider the possibility that the Galeria da Cisterna individual ranged further than the banks of the Tagus, consuming foods from the more radiogenic catchments beyond the modeled subsistence range, and that these more radiogenic values may have been overpowered by the 87 Sr/ 86 Sr values of the fish.
However, if such were the case, we would expect a correlation between the Sr concentration and isotope values, where peaks in Sr should be associated with the tending of the 87 Sr/ 86 Sr towards the values of the high strontium foods (0.7092 in the case of anadromous fish and 0.7100 in the case of freshwater molluscs).We see no obvious covariance in the 88 Sr and 87 Sr/ 86 Sr (Fig. S12) for the Magdalenian individual.Furthermore, marine mollusc food taxa were found neither at Lapa dos Coelhos (the mollusc remains retrieved in layer 3 are all ornamental species, namely, >30 specimens of Theodoxus fluviatilis, a freshwater taxon, 15 of which perforated, and two specimens of the marine taxon Littorina obtusata, both perforated) nor at Galeria da Cisterna (those retrieved in layer 3, in association with the human remains, are also perforated ornamental taxa, namely, three Theodoxus fluviatilis and one Hynia reticulata) (23)(24).The testimonial presence of marine taxa in these bead assemblages needs not imply that Upper Magdalenian Almondans regularly (e.g., seasonally) visited the coast, which is >50 km away (as the crow flies) across the limestone massif (plateau of Santo António and Candeeiros mountain range; Fig. S2), and would have been even more distant in the Tardiglacial, given the extent of the submerged continental platform (c.40 km).Such a presence is parsimoniously explained via exchange and in any case contrasts markedly with the pattern displayed by the shell bead assemblage from Gruta do Caldeirão, 25 km north-east of the Almonda spring (25)(26).Indeed, despite the greater distance from the seaside, only two of the 46 perforated shells retrieved in the Solutrean and Early Upper Palaeolithic levels of Gruta do Caldeirão are of freshwater species (Theodoxus fluviatilis and Unio pictorum); the other 44 are of marine taxa (Antalis vulgare, Aporrhais pespelecani, Littorina obtusata, and Semicassis saburon), and food taxa are entirely absent.The ornamental shell data are therefore entirely consistent with the notion that, for the people living to the east of the Central Limestone Massif, temporary residence in the Atlantic coast and subsistence exploitation of its resources (namely, shellfish) were off-limits.If anything, the bead data would in fact suggest that, by comparison with Last Glacial Maximum times, contact and exchange with littoral groups decreased markedly in the Tardiglacial, even as separation by distance decreased as a result of the rise in sea level.This pattern is in good agreement with a demographic interpretation of the strontium data.
To conclude: The 87 Sr/ 86 Sr-derived home ranges of the ibex and red deer (which would have provided much greater dietary biomass but less strontium), the consistency of the Sr-modeled human range with raw material procurement, and the lack of archaeological evidence for the exploitation of marine fauna lead us to the parsimonious explanation that the majority of the nutritional resources consumed by the Galeria da Cisterna Magdalenian individual were available and likely collected between the spring and mouth of the Almonda River.

Table S1 .
Samples analysed in this study.