High-Precision In Situ 87Sr/86Sr Analyses through Microsampling on Solid Samples: Applications to Earth and Life Sciences

An analytical protocol for high-precision, in situ microscale isotopic investigations is presented here, which combines the use of a high-performing mechanical microsampling device and high-precision TIMS measurements on micro-Sr samples, allowing for excellent results both in accuracy and precision. The present paper is a detailed methodological description of the whole analytical procedure from sampling to elemental purification and Sr-isotope measurements. The method offers the potential to attain isotope data at the microscale on a wide range of solid materials with the use of minimally invasive sampling. In addition, we present three significant case studies for geological and life sciences, as examples of the various applications of microscale 87Sr/86Sr isotope ratios, concerning (i) the pre-eruptive mechanisms triggering recent eruptions at Nisyros volcano (Greece), (ii) the dynamics involved with the initial magma ascent during Eyjafjallajökull volcano's (Iceland) 2010 eruption, which are usually related to the precursory signals of the eruption, and (iii) the environmental context of a MIS 3 cave bear, Ursus spelaeus. The studied cases show the robustness of the methods, which can be also be applied in other areas, such as cultural heritage, archaeology, petrology, and forensic sciences.


Introduction
In situ radiogenic isotope determinations with microscale resolution, especially of Sr, can represent a powerful tool in different fields of geological and life sciences. In particular, this technique is nowadays one of the most important methods for the investigation and interpretation of magmatic processes, as well as of environmentally-induced responses of terrestrial mammals; it has the potential to greatly enhance our understanding of not only volcanic systems and the related magma genesis and evolution, but also of the physiological mechanisms behind specific organic adaptations.
We present a detailed protocol for in situ sampling through microdrilling, Sr purification, and thermal ionisation mass spectrometer (TIMS) high-precision determinations of small amounts of Sr (<10 ng) in both biological and geological materials at the same error levels. In situ 87 Sr/ 86 Sr analysis is presented in three case studies which deal with the following: (i) plagioclase crystals from Nisyros volcano (Greece), (ii) glassy matrix of single glassy clasts from the 2010 eruption of the Eyjafjallajökull volcano (Iceland), and (iii) bones and teeth from fossil remains of the Ursus spelaeus. In these three examples, the in situ Sr-isotope approach permits to constrain petrological and volcanological processes and to effectively outline the life habits and habitat exploitation of extinct living species. e three presented cases aim to show the high potential of the methodology, validating its wide-scale applicability in many other areas, such as cultural heritage, archaeology, and forensic sciences.

Materials and Methods
Technological improvements on mass spectrometer and microsampling devices allow researchers to collect and analyse small amounts of sample (few micrograms of sample, containing nanograms of Sr) with no loss of precision in the isotopic determination. In situ analyses have many important advantages over more typical ones on bulk samples. In geological applications, it allows to preserve the textural information and thus to combine it with the isotopic and geochemical composition of specific portions of the samples. In archaeology and palaeontology, this method has the advantage of minimising the damage and/or destruction of samples, thereby leaving significant amounts available for further applications. e procedure consists in three main stages: (i) in situ sampling through microdrilling, (ii) sample digestion and purification of the element of interest, in our case Sr, and (iii) measurement of the isotope ratios (i.e., 87 Sr/ 86 Sr) through thermal ionisation mass spectrometry (TIMS). Our Radiogenic Isotope Laboratory at the Department of Earth Sciences of the Florence University is equipped with a modern MicroMill ™ grinder device, an ultraclean laboratory ("Class 1000") for microsamples digestion and elemental separation and a thermal ionisation mass spectrometer ( ermoFinningan ™ Triton-Ti ® ) for isotopic measurements.
Compared to classic, bulk sample analyses (generally measured on 100-150 ng of Sr [55]), small-sample analysis (typically 5 to 10 ng of Sr) has the drawback of being more exposed to contamination from Sr alien to the sample. In situ micro-Sr measurements therefore require continued testing of laboratory blanks during the whole analytical procedure. esi.com/products/laser-processing/milling/micromill/). e MicroMill (Figure 1(a)) combines a binocular microscope (with 6.7x-40x magnifying power) with remotely controlled submicron stage resolution and positional accuracy and a realtime video observation (at 3x digital magnification). It includes a low-eccentricity high-torque milling chuck, with variable speed (1,200-35,000 rpm), wherein a tungsten carbide or diamond-tipped bit is fixed, and an automated highprecision sliding stage on which the sample is loaded (Figure 1). e open stage architecture can accommodate thin sections or mounts and also larger solid samples such as bones, shells, ceramics, and plastics. e plug where the bit is hosted moves with adjustable speed for both spin and vertical movement, along the Z direction. e stage moves along the X-Y direction with a precision of 1 μm and maximum shift of 5 cm, allowing a high spatial resolution to the micron scale. is yields high spatial resolution, to the micron scale, and allows small-size sampling (i.e., a few μg of powder). e digital camera, placed next to the milling chunk ( Figure 1(c)), shows a live image of the stage and of the bit position ( Figure 2(a)). Using the microscope position mode (scope position), the optical image of the sample can be monitored on the PC screen prior to and after the sampling (Figure 2(b)). e device allows in situ microsampling on several types of solid materials such as rocks, minerals, glasses, plastics, bone tissues, ceramics, metals, and alloys. e designed software package also allows performing different milling patterns such as holes arranged randomly, lines, or rasters ( Figure 2(b)). Fine adjustment of milling velocity helps the microsampling of solid materials with different hardness.

In Situ
Microsampling on geological (e.g., rock, minerals, and glasses) and archaeological materials (glasses, ceramics, teeth, and bones) is performed using either thick polished sections (some 100 µm thick) or sample mounts (e.g., small chunks, chips, fragments of bones, and teeth). In our case studies, we used thick, polished sections for mineral and volcanic glasses and mounts for the teeth; both were fixed on the stage using either a double-sided adhesive tape or hot glue, to avoid sample displacement during milling. ick, polished sections are preferable for geological samples because they permit to characterise the petrographic features of the samples and thus to perform the microsampling according to their textural properties.

Milling Procedure and Sample Collection. A droplet of
Milli-Q ® water is placed with a micrometric pipette on the selected area prior to milling; this is performed by sticking a small punched square of warmed-up Parafilm ™ on the sample surface (Figure 1(c)). e water droplet retains the powder produced by the milling, which can then be easily collected by pipetting; it has also the effect of cooling the microdrill bit while milling. Before each drilling session, the drill bit is ultrasonically cleaned with pure ethanol and then rinsed with Milli-Q water.   e instrument software provides different drilling patterns and depths: single or multiple independent holes, spot lines, grids, line scans, or rasters ( Figure 2(b)). Milling spot lines (or grids) are more accurate, but more timeconsuming; they were used for the geological material (crystals and volcanic ashes) which requires more precise spatial resolution between the different zones of the same crystal or between the thin films of glass. Milling failure, such as crystal breaking, was prevented by setting a slow scan speed and splitting the milling into two or more steps. Line scans, which are performed faster but less accurately, were used for drilling the teeth. e number of points, lines, or rasters to be milled (which accounts for the amount of Sr to be collected) need to obtain a sufficient quantity of Sr for the TIMS measurements and can be calculated based on (i) the Sr content of the sample (independently determined by LA-ICPMS), (ii) the geometry of the drill bit, and (iii) the drilling pattern and the depth. Tips of different size and shape can be used for the drilling; therefore, the volume of material actually removed from different depths of a single hole or at different depths and lengths of a single line needs to be carefully calculated. e tungsten carbide mill bits supplied with the microdrill device (Komet-Brassler), have conical shape with an angle of 30°( Figure 3). e volume removed during each drilling is equivalent to that of the conical tip and dependent on the geometry of the drilling pattern, as well as on the specific depth ( Figure 3). e minimum amount of sample that needs to be drilled depends also on the total procedural blank, which should be at least two orders of magnitude lower than the total amount of Sr collected from the sample.
After milling was completed, the sample slurry was collected with a micropipette in a PFA beaker and then transferred in the clean lab for sample digestion and elemental purification. e blanks of the milling procedure were determined by keeping the drill bit tip into a Milli-Q water droplet on the sample surface (accurately cleaned before use) for as long as the average sampling time; the droplet was then processed as an ordinary sample. e amount of Sr in the blank was then determined through isotope dilution, by adding a single-spike solution (enriched in 84 Sr).

Sample Dissolution and Sr Purification.
e purification of the element of interest, in our case Sr, is crucial to obtain high-precision isotopic measurements for at least two reasons. First, it avoids isobaric interferences on the masses that will be analysed by mass spectrometry; in the specific case of Sr isotope measurements, even a small amount of 87 Rb will add to 87 Sr, yielding an overestimate of the 87 Sr/ 86 Sr ratio. Secondly, the presence of other elements of the matrix will compete with Sr during the thermal ionisation process, reducing the Sr signal and thus yielding less accurate measurements. e possibility of collecting and processing the sample for the purification of the element of interest is one of the major advantages of the method presented here over other methodologies, which do not achieve the same degree of accuracy and precision. e LA-MC-ICPMS methods allow faster data acquisition and higher sample throughput than mechanical microdrilling plus TIMS procedures, thanks to the possibility of introducing the samples directly into the mass spectrometer without chemical separation. On the other hand, LA-MC-ICPMS measurements require careful monitoring and corrections to minimize isobaric interferences in order to achieve suitable analytical accuracy and precision (e.g., [49,53,54,56]).
Powder digestion and Sr purification were carried out in our ultraclean laboratory ("Class 1000") aiming at the following: (i) optimising the separation of Sr and Rb to avoid interference of 87 Rb with 87 Sr, (ii) purifying the Sr collection form all the matrix analytes, (iii) maximising the yield of the columns during the chromatographic purification, and (iv) preserving low procedural blanks. Sample digestion was performed by sequential HF-HNO 3 -HCl as described in [55]. Chromatographic Sr purification was performed using Eichrom ® Sr-Spec ™ resins (100-150 µm) in quartz microcolumns (0.14 ml volume; Figure 4). Matrix elements were flushed out through elution with 14 column volumes of 3 N HNO 3 . Sr was then collected in Milli-Q (13 column volumes). e collected Sr fractions were further treated with concentrated HNO 3 and H 2 O 2 (fluxing at 150°C on a hotplate) to remove any organic residue. After this final step, samples were diluted in HNO 3 (10 vol.%) and were finally ready for loading on filaments for mass measurements. e  Figure 3: (a) SEM imaging of a tungsten carbide drill bit used for milling: the tip angle is 30°and geometrical shape of (b) a single conical hole and single line scan.
whole analytical procedure was performed with acids of ultra-pure quality. In order to thoroughly assess the contamination levels, we measured two types of blanks, one considering only the amount of Sr deriving from the chemical digestion and Sr separation, the other accounting for the whole procedure, including the drilling process, as described in the previous section. e results were 17 ± 6 (1 SD, n � 12) and 38 ± 19 pg (1 SD, n � 16), respectively, over a 14-month period (Table 1), thus allowing sampling as low as 4 ng of Sr for isotope analysis.

87 Sr/ 86 Sr Measurements on TIMS.
Sr isotope ratios were determined using a multicollector, thermal ionization mass spectrometer (TIMS: ermoFinningan Triton-Ti ™ ) ( Figure 5), equipped with nine moveable collectors, which allow to simultaneously detect all the natural masses of Sr ( 84 Sr, 86 Sr, 87 Sr, and 88 Sr). e mass of 85 Rb was also measured to monitor possible 87 Rb interference, but it was always lower than the detection limit of the instrument, confirming the quality of the separation procedure described above. A detailed description of instrumental characteristics and performances are given in [55], along with standardised routine, measuring conditions, and setting for normal-sized samples (100-150 ng of Sr). Instrumental mass bias (e.g., [57][58][59]) was corrected to the natural value of 86 Sr/ 88 Sr � 0.1194 using an exponential law (e.g., [55,59]). e most critical aspects of measuring small-size samples are related to (i) the procedure of sample loading onto the filaments and (ii) the measurement mode (i.e., static versus multidynamic, e.g., [55,59]). Both are very important to maximise the Sr signal during the measurements, to balance the analysis time and the analytical errors that are a function of sample size. e measurement protocol was tested by replicate analyses of an international certified standard (NIST-SRM987), properly diluted to attain sample sizes (5 to 10 ng Sr) comparable to those of the microsamples. en, we tested the whole procedure, from in situ sampling (<10 ng of Sr) to isotope measurement, on the international glass reference sample BHVO-2G. BHVO-2G reference sample is a synthetic basaltic glass (provided by USGS) obtained by melting the BHVO-2 powder collected from a Hawaiian lava flow. e glassy slices are supplied in epoxy resin mounts (https://crustal. usgs.gov/geochemical_reference_standards/microanalytical_ RM.html).

Sample Loading onto Filament.
e Sr fraction collected from the columns was dissolved into 1 µl HNO 3 10 vol.% and loaded on single Re filaments under a horizontal laminar flow hood. Due to the small amount of Sr available, it is important to confine the sample on the smallest possible area on the filament, so that the whole loaded sample can be ionised at the same time from a single spot. To attain this, a thin layer of Parafilm ™ was melted at both sides of the filament surface, to prevent any spreading of the solution, leaving a small gap (1 mm) for the droplet at the center on the filament ( Figure 5(b)). e loading was performed by sandwiching the sample between two 0.5 µl   ) were slowly dried by passing a current on the filament, which was increased at the end of the procedure until the filament starts glowing. e loaded filaments were then placed on the filament-holder turret and then inserted into the mass spectrometer (Figures 5(c) and 5(d)).

Measurement Procedure Reproducibility and Accuracy.
e measurement routine was established to obtain the best internal and external precisions, and the accuracy, on 87 Sr/ 86 Sr, was achieved by experimentally comparing runs performed in static versus dynamic conditions at a variable number of cycles and integration times. A detailed description of the static and dynamic methods is provided in [55,59]. In brief, the static mode consists of simultaneous measurements of all isotopes in a single "jump," so that the magnetic field remains static and the masses always hit the same detectors (Table 2). Static measurements have the advantage of considerably reducing the acquisition time in comparison with the dynamic mode, which becomes important when little amount of Sr is available, as dealing with small samples. e main limitation of this method is related to the uncertainty on the Faraday cup efficiency and on the drift of the electronics (i.e., the amplifiers).
e Triton-Ti is equipped with a virtual amplifier, which enables a variable connection between amplifiers and Faraday cups and allows a complete switching between amplifiers and cups during a single measurement. However, the virtual amplifier is not able to correct for the different Faraday cup efficiency and its variation with time.
In contrast, the dynamic (or multidynamic) mode is a peak-jumping procedure where a number of different cup configurations are employed for determining a single isotopic ratio ( e best configuration for static mode measurements was found by measuring 300 cycles with an integration time of 8 s, which corresponds to a total measuring time of about 35 minutes for each sample. For dynamic mode measurements, we performed 120 cycles (each including 3 magnetic jumps), with 8 s of integration time and an idle time of 3 s between the different jumps, for a total of 70 minutes for each sample.
In both static and dynamic methods, the filament was slowly warmed up, for a total of about 45 minutes, to stabilise the ion emission until the suitable intensity is achieved. During the heating, the beam was accurately optimised by peak-centering and focusing. e optimal beam intensity for the measurement varied from static to dynamic mode, with higher intensity allowed by the shorter static (3-3.5 V on 88 Sr) mode with respect to dynamic (1.5-2 V on 88 Sr) mode, which instead requires maintaining a stable beam, owing to the longer duration of the measurement.
e results are shown in Figure 6 and Table 3. Static and dynamic mode measurements on NIST-SRM987 reference samples (10 ng of Sr measured) yielded 87 Sr/ 86 Sr average values of 0.710247 ± 0.000026 (2 SD, n � 30) and 0.710251 ± 0.000018 (2 SD, n � 51), respectively, with internal precisions of 13 ppm (2 SE) and 16 ppm (2 SE), respectively. Both values are within the recommended reference value for NIST-SRM987 ( 87 Sr/ 86 Sr � 0.710248 ± 0.000011; Figures 6 (a) and 6(b) and Table 3 [59]). Static measurement reduced the experimental time but showed a worse external reproducibility than that obtained in dynamic mode ( Figures  6(a) and 6(b)), yet maintaining similar internal precision. erefore, the dynamic mode was chosen for the experimental work both on the international glass standard BHVO-2G and on the unknown samples. Further measurements of SRM987 in the dynamic mode were performed after the initial testing, along with the studied samples, yielding consistent results ( 87 Sr/ 86 Sr � 0.710252 ± 0.000018, 2 SD, n � 47; Figure 6(c)).
Results on the BHVO-2G are reported in Table 3. e results were also compared to standard measurements (150 ng of Sr) on the BHVO-2 powder reference sample. BHVO-2G versus BHVO-2 results found for micro-and normal-size samples, respectively, are well within the internal analytical error (Table 3) and in agreement with the reference values reported in [60,61] for bulk powder (i.e., standard BHVO-2). e significantly larger standard deviation of the microdrilled BHVO-2G measurements, with respect to both micro-Sr SRM987 and BHVO-2 powder data, is likely partly related to small isotopic heterogeneities of the glass standard. e few available micro-Sr data on the same sample provide similar averages and reproducibility of our data [54,62] ( Table 3).
Comparing our results with LA-MC-ICPMS data is more difficult; in fact, the latter vary largely depending on the material used for the analyses. External reproducibility obtained with LA-MC-ICPMS on material with high Sr contents and low Rb/Sr (e.g., apatite [49,53], marine shells, and synthetic plagioclase [16]) is comparable or slightly worse than that attained with our method; yet, small but significant differences in accuracy have been reported [53]. On the other hand, the internal and external reproducibility worsen significantly (e.g., by a factor 5 to 10 in [16]) in materials with low Sr and high Rb/Sr.
In summary, the method presented here generally provides more accurate and precise results than LA-MC-ICPMS, independently on the nature of the analysed material, despite being more time-consuming. It is therefore suitable for a wider range of applications.

Applications
In this section, we report three case studies, two of which were previously published [8,10], as examples of possible applications of the presented methodology in different fields of science. Indeed, in the last decade, the in situ isotope microsampling approach has been used and applied in many pilot studies in a wide range of research fields, including, among others, palaeoenvironmental and palaeoecologic reconstructions (i.e., [53,63]) and climate changes (i.e., [64]).
In the three presented case studies, the samples were thoroughly characterised both texturally (optical microscope and SEM) and chemically (electron micro probe analyses) before drilling. e strontium element concentrations (in ppm) in all the samples was then determined through LA-ICPMS.

Micro-Sr Isotope in Minerals.
Rock-forming minerals in igneous rocks display variable chemical composition depending on several processes and parameters such as (i) the physicochemical conditions of the magmas, (ii) open system processes (e.g., magma mixing and mingling), and (iii) recycling of cumulated crystals triggered by new arrivals of magma within crustal reservoirs. Radiogenic isotope ratios in minerals, or portions of them, can be used as a petrogenetic "DNA" to record the history of the magma reservoir (crystal isotope stratigraphy, e.g., [7]) and their evolution within the crust (e.g., [1,5,7,8,[65][66][67]). Combining in situ Sr-isotope fingerprints with other approaches, such as textural evidences and crystal size distribution, offers the opportunity to understand the processes and timescales through which magmas are stored, differentiated, and delivered prior to eruption (e.g., [5][6][7]13  e case study presented here is related to the active Nisyros volcano, the easternmost volcanic island of the South Aegean Active Volcanic Arc (Figure 7) [5,[70][71][72][73][74][75][76]. Nisyros volcanic products are typically porphyritic rocks, with clear petrographic evidence of recurrent mixing and mingling of different magmas during the whole volcano's history, which is likely interpreted as the triggering mechanism for its eruptions (e.g., [74,77,78]). Sr-isotope determinations at the subcrystal scale, along with detailed petrographic microscopic textural evidence, provided significant data for better defining the interaction of different magmas, concerning pre-eruptive mechanisms. e  (Figures 7(b) and 7(c)) (e.g., [8,78]). In situ Sr-isotope ratios were determined on plagioclase phenocrysts (Figures 7(c)-7(e)), from both domes and enclaves, which preserve evidence of the complex history of interaction between the mafic (i.e., enclaves) and felsic (i.e., rhyodacitic domes) magmas in their growing zones [8]. e 87 Sr/ 86 Sr values determined on micromilled samples from the different growth zones of the plagioclase phenocrysts show clear Sr-isotope disequilibria between (i) cores and rims of single crystals (Figures 7(c)-7(e)) and (ii) the crystals and the host magmas (Figure 7(e)). is suggests that some of the phenocrysts that had formed in the rhyolitic magmas were later enclosed (as xenocrysts) within the more mafic one (i.e., as enclaves). Whereas the rim of the phenocrysts is isotopically intermediate between the rhyolitic and mafic magmas ( Figure  7(f)), their cores show higher 87 Sr/ 86 Sr values, quite close to that of the previously erupted upper pumice magma. is clearly indicates that the phenocrysts originated in a different, older system. In this light, the large plagioclase phenocrysts found inside the dome lavas and enclaves can be interpreted as recycled from previously cumulated crystals (called "antecrysts" by Davidson et al. [7]). ese results have also demonstrated that the dome lavas are multicomponent magmas formed by progressive mingling/mixing processes between (i) a rhyolitic, and more Sr-radiogenic melt derived from the original upper pumice magmas, and (ii) the enclave-forming mafic, and less Srradiogenic, melts refilling the felsic magma chamber. e constraints involved in interpreting in situ isotope data have further implications for the timing and style of eruption. e inferred delay between the mafic input (i.e., enclaves) and the relative dome eruption allows time for reheating and consequent drop in magma viscosity, thus favouring dome extrusion rather than explosive activity [8,78].

Micro-Sr Isotope in Natural Glasses.
Glasses are found in nature generated by rapid quenching of molten material. ey represent a volumetrically small component of crustal rocks and can have different genesis (i.e., volcanic, lightning strikes, meteorite impact, and anthropogenic). In this light, the radiogenic isotopic compositions (i.e., Sr, Nd, and Pb) can provide fundamental information to discriminate among the processes involved in their formation. Glasses constitute the main component of ash and pyroclastic deposits, and their composition and Sr-isotope signature can provide important information in defining the triggering mechanisms of explosive eruptions (e.g., [10,69,[79][80][81][82][83][84]). Glasses may also be found in ceramics, as well as in other artefacts. Sr-isotope data are therefore also important to define the possible source of raw materials for pottery, which is particularly relevant for cultural heritage, or to track trade routes in archaeology (e.g., [85][86][87][88]). Microscopic scale Sr-isotopic measurements on glasses would help in minimising the amount of sample that needs to be milled, which is of crucial importance both dealing with small-sized volcanic ashes and ejecta and with human artefact of archaeological interest. e present study is a case study of submicroscopic scale Sr-isotopic determination on ashes of the 2010 Eyjafjallajökull (Iceland) volcano's explosive eruption, which caused enormous disruption to air travel across the northern hemisphere (Figure 8(a)) (e.g., [89][90][91][92][93][94]). is eruption represents a unique opportunity to test the potential of microscale Srisotope determinations on ash glasses from tephra deposits that were well preserved within the ice/snow pack [10]. e population of tephra is composed by four different types of ash fragments: (i) fluidal, (ii) coarsely vesicular, (iii) spongy fine vesicular, and (iv) blocky (Figure 8(b); [93]), generated by different fragmentation processes during the eruption. A detailed microanalytical geochemical and in situ Sr-isotope study performed on glassy groundmass of single ash clast showed unusually high 87 Sr/ 86 Sr values (up Table 3: Accuracy and reproducibility on reference standard material (NIST-SRM987 and BHVO-2).

Standard
Sr content 87 (Figure 8(c)); these high isotopic values were also associated to atypical elemental compositions compared to most of the juvenile ash fragments of the eruption. e anomalous, high Sr-radiogenic clasts belong to the blocky type (Figure 8 (b)) and are concentrated in the first, thin ash level emplaced during the initial phase of eruptive activity. ese clasts originated from the magma quenched from the contact with the ice cap filling the summit caldera of the volcano [10]. ese anomalous findings in the Icelandic magmatic environment can be explained supposing that during its rise and before intruding into the ice cover, the erupting magma selectively assimilated hydrothermal minerals (i.e., zeolites, silica phases, and anhydrite) with seawater- related high-Sr isotopic ratios, hosted in altered volcanic/epiclastic rocks. is selective assimilation took place at the tip edge of the first rising magma body, resulting in a high degree of contamination restricted to the rather small amount of melt directly in contact with the hydrothermal veins. Indeed, evidence for this process is recorded only by the very first erupted juveniles (i.e., the blocky clasts). e results obtained through submicroscopic-scale micromilling and relative Sr-isotope determination revealed the dynamics of the processes involved in the initial stages of magma ascent to the surface; this provides significant insights into the interpretation of the precursory signals of the eruption (mostly consisting of ground deformation or increased seismicity) [10]. ese transient processes, which interested only a small, well-confined part of the magma, cannot be detected using traditional, Sr-isotope determination on whole-rock samples but can be revealed only analysing single-glassy clasts separately.

Micro-Sr Isotope in Teeth and
Bones. Due to their similar chemical properties, Sr can substitute for Ca in the bioapatite [Ca 5 (PO 4 ,CO 3 ) 3 (OH,F)] of mammalian bones and teeth, reaching contents of few hundreds of ppm that allows the isotope analysis by microdrilling. e Sr isotope composition of human and animal hard tissues is a function of their dietary habits (e.g., [22]) and depends on the isotopic composition of the food and water ingested during life, which in turn are related to the geological substrate [95][96][97].
A pioneering study [107] demonstrated that migrant individuals who moved between different geologic regions might be traced by comparing 87 Sr/ 86 Sr in adult tooth enamel, formed between four and twelve years of age, and in the bones, which remodel throughout life and therefore representative of adulthood [30]. Unlike bones and dentine, dental enamel formed during childhood [108] remains unaltered throughout the years. Different 87 Sr/ 86 Sr in the teeth and bones of an individual may thus reflect the fact that it moved around the landscape passing through different isotopic environments during its youth and maturity [109,110]. Teeth enamel is generally preferred to dentine and bones in the analysis of Sr concentrations and isotope ratios because it is virtually unaffected by postmortem diagenesis (e.g., [111][112][113]). e micro-scale Sr-isotope measurements of samples obtained using the submicroscopic-scale micromill technique is perfectly able to discriminate between enamel and dentine in single-tooth samples. In addition, this technique increases the accuracy of sampling and also reduces the amount of specimen to be destroyed for high-precision Srisotope analysis.
Ursus spelaeus was an endemic, widespread European Late Pleistocene species. In contrast to many other taxa, it has fairly rich fossil records, especially thanks to its recurrent use of caves or shelters for winter hibernation [114][115][116]. Caves were frequently used by bears for many generations, and numerous individuals eventually died in them, so that significant quantities of their remains accumulated over considerable periods of time.
Many studies used stable isotopes to determine the dietary habits of living and extinct bears (e.g., [117][118][119][120][121]), but until now only a small number of papers have considered employing Sr isotopes to possibly elucidate the factors influencing habitat use and gain insights into the foraging behaviour of cave bears [122]. We report here the first 87 Sr/ 86 Sr data obtained through in situ microsampling on teeth and bones of Ursus spelaeus found in Grotta all'Onda cave.
e study was aimed at defining the lifestyle and feeding behaviour of one of the most prominent European Late Pleistocene mammals [123][124][125].
Grotta all'Onda cave is located 708 m above sea level (a.s.l.) in the Apuan Alps nearby the village of Camaiore (Tuscany, Italy) (Figure 9), in a sub-Mediterranean habitat [126].
All the three teeth had well-preserved dentin and very thin enamel layer ( Figure 10). In contrast, the three bones were rather differently preserved and had different porosity. In particular, SCT3 was the most heavily mineralised and best preserved, whereas SCT1 was densely vacuolated and preserved higher amounts of organic components.
Major element analyses of these specimens revealed that the enamel bioapatite was more mineralised than that of dentin and bones. Moreover, the dentin bioapatite was found enriched in Sr and in other trace elements [128].
For the purpose of this study, dentin and enamel of the teeth and cortical sections of the bones from Grotta all'Onda were micromilled and analysed for micro-Sr isotope determination. e two soil samples were also processed for Sr-isotope determination using the traditional, large sample method [37,55]. e 87 Sr/ 86 Sr data are reported in Table 4. Tooth enamel shows higher 87 Sr/ 86 Sr and lower Sr/Ca than microsamples of dentin and bones do (Figures 10(e) and 10(f)). is strongly suggests that 87 Sr/ 86 Sr of tooth enamel is unaffected by diagenetic alteration, in contrast to the other organic-rich samples (i.e., bones and dentin). e 87 Sr/ 86 Sr composition of the dentin samples is close to that of the local soil samples (Figure 10(e)); even closer to the latter is that of the bone samples, with the only exception of SCT3 due to its high degree of mineralisation. ese results, despite the relatively recent age of the fossil specimens (ca. 40 ka [127]), indicate that bone tissues have been more exposed to postmortem diagenetic exchange processes compared to more heavily mineralized enamel. is speaks for a possible isotope reequilibration between dentin and bones (but not enamel) and soil, due to Sr exchange with percolating fluids. e soil samples have 87 Sr/ 86 Sr values comparable to those of the "Calcare a Rhaetavicula Contorta" formation, which forms the cave's bedrock [129], through which fluids filter into the cave (Figure 10(e)).
In summary, our study shows that the Grotta all'Onda bones and dentin are unsuitable to determine the characteristics of the habitat where the cave bear lived, due to their interaction with percolating water and their consequent contamination by the soils in which they had been preserved.
In contrast, the 87    enamel samples is consistent with that of the "Calcare Massiccio Formation" and in particular with the dolomite fraction [129] ( Figure 10(f)). e mismatch between the 87 Sr/ 86 Sr values of the enamel samples and those of the cave soils ( Figure 10(e)) (i.e., local substrata of the cave) indicate that the cave bear died away from its customary habitat. Bears cannot indeed find food in caves, where they find refuge as shelter for winter hibernation or for night resting. e enamel isotopic values obtained during our study indicate that Ursus spelaeus from Grotta all'Onda roamed in search for food within a confined area not far from the cave, where the "Calcare Massiccio" is largely exposed and did not move too far from the area during its whole life.

Summary and Conclusions
e present study show the potential of 87 Sr/ 86 Sr determination by TIMS on micro-scale samples, based on micromilling solid specimens, not only for geological applications, but also for other fields, such as archaeology, forensics, medical, and life sciences, where it has hardly, if ever, been used. Reported here is a detailed description of all the analytical protocols, including results on replicate analyses of international standards (SRM 987 and BHVO-2G), which yield good accuracy and precision. In addition, three case studies are presented, performed in our laboratory, where in situ microdrilled Sr isotopes have been used in different fields of application. e first case study on micro-Sr isotope determination at subgrain microscopic scale regards the petrogenetic processes relevant to the understanding of the plumbing system dynamics under active volcanoes. is example revealed the role played by the interaction of different magmas, which are normally characterised by distinct 87 Sr/ 86 Sr signatures, comingled in the plumbing system of the Nisyros volcano, which was capable of triggering the eruption. e second case regards volcanic glasses with extremely low total Sr content (i.e., tholeiitic). e micromilling determination of 87 Sr/ 86 Sr ratio was performed on ashes, with different shapes and nature, erupted by different phases, during the 2010 eruption of Eyjafjallajökull volcano (Iceland). e 87 Sr/ 86 Sr data provide information on the eruptive mechanism involved during the eruption, as well as on the interaction between the magma and the hydrothermallyderived minerals attained before the thawing of the ice cap; it also provides significant insights into the interpretation of the precursory signals of the eruption. e third case displays the use of 87 Sr/ 86 Sr microdrilled in enamel, dentin, and bones, to show that only enamel has more chances to preserve the original Sr-isotope signatures than bone and dentin. e analysis also revealed the close relationship existing between the radiogenic-Sr of the organic materials and that of the geologic Cretaceous substratum of the Apuan Alps, which provides valuable insights into the palaeoenvironment of the local cave bears. In this case, the Sr isotopes proved particularly useful for determining the foraging habits of extinct mammals, which substantiates the well-known statement "YOU ARE WHAT YOU EAT" (cit. Anthelme Brillat-Savarin).

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.

Authors' Contributions
Lorella Francalanci provided the input to the set-up and the development of the in situ micro-Sr isotope procedure at the Radiogenic Isotope Laboratory of Firenze, providing also continuous scientific stirring and encouraging to the other research fellows, coauthors of the present paper.