In situ high spatial resolution 87 Sr/ 86 Sr ratio determination of two Middle Pleistocene (c.a. 580 ka) Stephanorhinus hundsheimensis teeth by LA–MC–ICP–MS

Bone and tooth tissues are important biological archives to study eating habits and provenance of ancient humans and animals. By taking advantage of the high spatial resolution offered by the Laser Ablation Multi Collector Inductively Coupled Plasma Mass Spectrometry (LA–MC–ICPMS) technique, we investigated the 87 Sr/ 86 Sr intra-tooth variability of two Stephanorhinus hundsheimensis tooth fragments from the Middle Pleistocene site of Isernia La Pineta. We detected significant Sr isotopes variability within the rhinos tooth enamel (enamel average ±2σ: sample RH–IS30 0.70951 ± 0.00014; sample RH–IS26 0.70976 ± 0.00015) with values higher than the “local” 87 Sr/ 86 Sr ratio (dentine average ±2σ: sample RH–IS30 0.70918 ± 0.00013; sample RH–IS26 0.70934 ± 0.00009). This is likely linked to a different water and food intake with a Sr isotopic signature similar to volcanic soils nearby (Roccamonfina: from 0.7093 to 0.7100; Colli Albani: > 0.7100) and supports the idea that the Stephanorhinus hundsheimensis species moved around seasonally. The improvement of non-destructive, accurate and precise analytical methods to decrypt the information hidden within bone and tooth hard tissues of archeological material is crucial to-unravel critical questions about evolution, migration and ecology of human and animals. We have successfully took upon this challenge using three matrix-matched standards, with variable Sr concentration (c.a. 100 to 1000 ppm), to correct unresolved interferences arising from LA analyses.


Introduction
Strontium isotopes have become an important tool in palaeontological research to track ancient human and animal migratory patterns and provenance (e.g.Price et al., 2002;Bentley, 2006;Evans et al., 2006;Montgomery et al., 2007;Copeland et al., 2011;Bentley et al., 2012).Tooth enamel, if available, is the best material for geochemical investigations because it is made of a ~ 96% mineralized hydroxyapatite more resistant to diagenesis (Bentley, 2006) and is an elemental and Sr isotope archive of the individual life.
Strontium substitutes calcium in the hydroxyapatite of bones and teeth (Burton et al., 1999) and its 87 Sr/ 86 Sr ratio is directly related to the ratio of plants and water ingested by humans and animals (e.g.Pate, 1994;Burton and Wright, 1995;Bentley, 2006).
Strontium ions enter the ecosystem through the weathering of the local bedrock, which usually retains a peculiar 87 Sr/ 86 Sr ratio related to the petrogenetic process, the age and the initial rubidium content of the rock itself.Thus, analysing the 87 Sr/ 86 Sr ratio of dental and bone tissues can help identify the place where the tissue has formed or remodelled.Given that enamel forms during the individual growth and preserves incremental features of its formation, small scale isotopic investigations allow to detect temporal variations in the Sr isotopic composition of the animal life even within one tooth (e.g.Balasse, 2003;Tafforeau et al., 2007;Martin et al., 2008;Britton et al., 2011).In such sense, the growth lines or bands visible by microscopic observations of teeth, the so-called Retzius lines (or striae of Retzius), are the bestknown long-term period incremental features of the dental enamel and represent the successive apposition of forming enamel (Smith and Tafforeau, 2008).The time between the formation of two striae of Retzius is between 2 and 12 days in primate teeth and about 7 days in rhino teeth (e.g.Tafforeau et al., 2007;Smith and Tafforeau, 2008).This time-span, namely periodicity, can be estimated counting the visible laminations (i.e.daily isochrones of the enamel deposition) per Retzius band (Tafforeau et al., 2007).
Traditionally, Sr isotopes have been analyzed by Thermal Ionization Mass Spectrometry (TIMS) and MC-ICPMS requiring total dissolution of a portion of the sample and chemical separation of strontium (e.g.Li et al., 2013).The recent development of the Laser Ablation system (LA) coupled with a Multi Collector Inductively Coupled Plasma Mass Spectrometer (MC-ICPMS) has allowed high spatial resolution analyses with a micro-destructive approach, with the opportunity to analyze even small tooth fragments and unravel possible isotopic heterogeneity at the small scale within the tooth itself (Horstwood and Nowell, 2005;Copeland et al., 2008;Horstwood et al., 2008;Richards et al., 2008;Simonetti et al., 2008;Copeland et al., 2010;Copeland et al., 2011;Balter et al., 2012).This technique can be particularly helpful when dealing with small ancient enamel fragments of migratory animals or precious remains that cannot be destroyed by canonical dissolution methods.
Despite the clear advantage of this technique, LA-MC-ICPMS analyses have some issues mainly related to analytical artefacts.In particular, during analysis several isobaric interferences are known to form; whilst some are routinely and properly corrected by many laboratories (e.g.Rb and Kr), others may hinder the achievement of precise and accurate results (e.g.REE 2+ , 40 Ca 46 Ca, 46 Ca 40 Ar, 40 Ca 31 P 16 O and 40 Ar 31 P 16 O; Copeland et al., 2008;Horstwood et al., 2008;Vroon et al., 2008;Willmes et al., 2016).
In this work, we investigated the 87 Sr/ 86 Sr ratio intra-tooth variability of two Stephanorhinus hundsheimensis teeth (LA samples n = 82) from the Middle Pleistocene site of Isernia La Pineta (Molise, Italy), to unravel possible variations within the tooth enamel and to try to reconstruct a migratory pattern for the Stephanorhinus hundsheimensis species in the area.Beside the Bison schoetensacki, this is the second most represented taxa in the area (Ballatore and Breda, 2013;Peretto et al., 2015).We analyzed our two samples in order to detect if incremental fluctuations of the 87 Sr/ 86 Sr ratio, i.e. representing temporal variations in the Sr isotopic composition, were preserved in the enamel and if they correlated with the structural growth features of the tooth.

Materials and Methods
Two fragments (a longitudinal and a transversal section) of Stephanorhinus hundsheimensis molars (RH-IS30 and RH-IS26) were collected among the nonmusealized samples found at the Isernia La Pineta site.Sample RH-IS30 was collected from the archaeological layer called 3s10 and sample RH-IS26 from the 3c layer (for more details about the stratigraphy see Peretto et al., 2015).Argon-argon dating ( 40 Ar/ 39 Ar) suggests an age of about 580 ka for both layers (Peretto et al., 2015).The surface of both samples was analyzed in detail, with 80 LA short linear scans composing four profiles (Fig. 1 and Fig. 2; Profiles A, B, C and D).We noticed that short linear scans gave better results than spots in terms of signal stability.Of these short scans, 60 were performed on the enamel surface and 20 on the dentine surface.
Dentine is usually not a good indicator of provenance because of its high susceptibility to diagenetic contamination (e.g.Bentley et al., 2006) whereas enamel is generally less affected by diagenesis and represents the best target.In the case of samples that due to their uniqueness cannot be cut to expose their inner parts, scientists have to rely on analyses of the outer enamel (e.g.Copeland et al., 2011).
Sample RH-IS26 was also tested with a slightly different method using continuous long scan profiles for the entire length of the enamel portion.We obtained two profiles: one (profile E) on the inner and one (profile F) on the outer surface of the enamel as shown in Fig. 3.
Following Horstwood et al. (2008), we have also selected and analyzed four bioapatites with different Sr concentration to establish our matrix-matched standard materials given the unavailability of solid bioapatite standards on the market.We selected a shark tooth with a Sr concentration of ~1100 ppm (average 1101 ± 47 ppm), a bovine tooth with a Sr concentration of ~350 ppm (average 347 ± 18 ppm), a swine tooth with a Sr concentration of ~250 ppm (average 253 ± 31 ppm) and a human tooth with a Sr concentration of ~100 ppm (average 104 ± 25 ppm).The shark tooth (Carcharhinus leucas, Müller and Henle, 1839) was purchased from an online specialized shop.Shark tooth enameloid is composed largely of fluorapatite (Ca5(PO4)F) and minor hydroxyapatite (Ca5(PO4)OH) (Vennemann et al., 2001), however, their Ca/P molar ratio and the relative content of Ca 2+ and PO4 3-are similar to mammal enamel (Enax et al., 2012).The swine tooth (SUS-Sr1) and the bovine tooth (BOS-Sr1) were collected from a roman archaeological site in Ravenna (Italy), while the human tooth (ROCht421) was selected among the best-preserved mummies teeth buried in the archaeological site of Roccapelago (16 th and the 18 th century AD, Modena, Italy; Lugli et al., accepted).Prior to LA-analyses, portions of our in-house bio-standards were analysed in solution by MC-ICP-MS to obtain the reference 87 Sr/ 86 Sr value.

Thin section preparation
Incremental growth features of the transversal section of tooth RH-IS26 were studied under a transmitted-light petrographic microscope (Fig. 5).A petrographic thin section of the sample was prepared by embedding the tooth in polyester resin, gluing onto a microscope glass section, cutting in a ~1 mm section with a diamond saw and manually polishing to a thickness of 100 µm with decreasing-size polishing paper and aluminium oxide (Tafforeau et al., 2007).Observation and counting of Retzius lines was performed in polarized transmitted light with an Olympus BH-2 petrographic microscope.

Dissolution method MC-ICPMS
Samples were prepared at the Department of Chemical and Geological Sciences of the University of Modena and Reggio Emilia.The tooth enamel was manually separated from the dentine using a dentist drill.After a macroscopic cleaning with small brushes, each enamel chunk was boiled in MilliQ ® for 1h to remove any possible superficial residue of organic matter, rinsed repeatedly with MilliQ ® and then sonicated in an ultrasonic bath for 30 minutes.Before dry down, samples were rinsed 3 times with MilliQ ® and once with high purity acetone.Samples were digested in 1ml of suprapure 14N HNO3, evaporated to dryness and dissolved in 3ml of 3N HNO3.
After centrifuging, samples were loaded into 300 µl volume columns filled with Eichrom Sr spec-resin and washed with 3N HNO3.Sr was finally eluted with several reservoirs of MilliQ ® water (Deniel and Pin, 2001).
The 87 Sr/ 86 Sr isotopic composition of the samples was determined using a double focusing MC-ICPMS with a forward Nier-Johnson geometry (Thermo Fisher Masses 82 and 83 were collected to monitor the presence of Kr in the argon.Data were corrected using a 86 Kr/ 83 Kr ratio of 1.505657.Mass 85 was used to correct the signal on mass 87 for the presence of isobaric Rb, using a 87 Rb/ 85 Rb ratio of 0.3856656.Mass bias normalization, through exponential law, was carried out using a 88 Sr/ 86 Sr ratio of 8.375209.Mass fractionation for both Kr and Rb has been assumed equal to that one of Sr.The 87 Sr/ 86 Sr ratios were corrected for instrumental bias to an NBS-987 value of 0.710260 ± 0.00002 (Ehrlich et al., 2004;Durante et al., 2015).Repeated analyses of the NBS-987 yielded a 87 Sr/ 86 Sr ratio of 0.710272 ± 0.000013 (2σ; n = 8, corresponding to an external reproducibility of 18 ppm).The long-term reproducibility (c.a.three years of analyses; n > 2000) of the instrument yielded a 87 Sr/ 86 Sr ratio of 0.71027 ± 0.00002 (2σ; Durante et al. 2015).

Laser ablation method MC-ICPMS
The teeth portions chosen for laser analysis were cleaned with brushes, washed with MilliQ ® water and carefully dried down before introduction into the LA sample holder.
Sr data were acquired with the Neptune MC-ICPMS, coupled to a 213 nm Nd:YAG laser ablation system (New Wave Research ™ ) housed at CIGS-UNIMORE (Table 1).
Before LA analyses, the instrument was tuned using the NBS-987 standard solution, monitoring both the signals and the significant isotopic ratios.
All nine Faraday detectors were used to acquire peaks of the following masses 82 Kr, 83 Kr, 84 Sr, 85 Rb, 86 Sr, 87 Sr, 88 Sr and half masses 85.5 and 86.5.Half masses 86.5 and 85.5 were monitored to check for the contribution of doubly charged Rare Earth Elements ( 171 Yb 2+ and 173 Yb 2+ ).Usually these signals are below the detection limit of the instrument (LOD), therefore the correspondent detectors were equipped with 10 12 Ω resistors.We used the "on peak zero" method to correct for Kr interferences (Davidson et al., 2001;Vroon et al., 2008) by measuring a 60 s gas background for each mass before each analysis and subtracting the signal of each detector to the measured value of the sample.After background subtraction, the remaining signal of mass 82, was used to check the formation of Ca dimers and argides, isobars to masses 84 Sr, 86 Sr, 87 Sr and 88 Sr (Woodhead et al., 2005).Typically, Ca dimer and argide signals on mass 82 are around 0.1 mV.Likewise, the remaining signal of mass 83 (after Ca dimers argides subtraction) was used to correct for the presence of 166 Er 2+ (Copeland et al., 2010).Rb interferences were corrected using the same procedure followed for the dissolution method.We note here that the 85 Rb signal during our analytical sessions has always been lower than 1 mV.
We employed a linear ablation pattern (100 x 500 µm) to sample the tooth enamel surfaces in order to limit laser-induced elemental fractionations and to improve the beam size stability (Vroon et al., 2008;Guillong and Günther, 2002).
Our daily LA-MC-ICPMS tuning is based on a 3-step protocol using our ST1 in house-reference material.During the first step, ST1 is analysed to optimize the machine parameters tuning for the maximum signal intensity on 88 Sr.Second, background signal intensities on 83 Kr and 82 Kr are monitored.Ideally they have to be as low as possible and, as also observed by Jackson and Hart (2006), the 83 Kr/ 82 Kr ratio has to be maintained near ~1 to obtain the best results in term of precision and accuracy.Finally, the 87 Sr/ 86 Sr ratio of the ST1 standard is measured and machine parameters are tuned until the desired ratio (~0.7092) is obtained.Laser data were acquired with a block of 200 cycles, an integration time of 0.5 s and a laser spot size of 100 µm.The surface of the tooth was pre-ablated before each analysis.Summary of the typical LA parameters used during our sessions are presented in Table 1.
The two continuous long LA scan profiles obtained for tooth RH-IS26 were acquired with a block of 2000 cycles and an integration time of 0.5 s.We used a laser spot size of 100 µm and a line length of about 6 mm.We calculated a moving average of the cycles with a period of 100.
As reported by several authors, during LA analyses is difficult to address the specific interferences that contribute to the fractionation of the 87 Sr/ 86 Sr ratio (e.g.Whoodhead et al., 2005;Balter et al., 2008;Horstwood et al., 2008;Simonetti et al., 2008;Vroon et al., 2008;Nowell and Horstwood, 2009;Lewis et al., 2014).The proposed 40 Ca 31 P 16 O interference on mass 87 is still a matter of debate.In fact, while several authors claim its existence (e.g.Horstwood et al., 2008;Simonetti et al., 2008;Lewis et al., 2014), others never experienced it (e.g.Copeland et al., 2008;Müller and Anczkiewicz, 2016).More recently, Willmes et al. (2016) (Vroon et al., 2008), but with the obvious consequence of a more destructive analysis.They also suggest that a valid approach to correct for the bias arising from mass 87-interferences is a calibration correction with a series of known standards (Willmes et al., 2016;Horstwood et al., 2008).
In addition to mass-87 interference, LA isotope analyses often endure the so-called laser-induced elemental/isotope fractionation, as observed for other isotopic systematics (i.e.U-Pb), which may bias the final isotope ratio (see Horn 2008;Vroon et al., 2008).Moreover, many unknown interferences may form in the MC-ICP-MS atmosphere, mainly because of the bonds of oxygen and argon with elements of the sample matrix ionised within the plasma (Vroon et al., 2008).No matter how or why interferences arise during Sr LA analysis of a particular material, the concurrent analysis of a set of matrix-matched standards is a proper approach to monitor and correct the Sr fractionation issues, as demonstrated by Horstwood et al. (2008) and reiterated by Willmes et al. (2016).Standards with matrix equal to the samples are ideal because their behaviour during analysis is supposed to be the same of the sample (Sylvester, 2008).Thus, each of our bio-apatite in-house standard was analyzed several times (at least 5) during the whole session to observe the external reproducibility on different Sr-concentration materials.Following the method proposed by Horstwood et al. (2008), we built with our bio-standards a calibration curve based on the accuracy of the 87 Sr/ 86 Sr ratio (measured value/true value) vs. the inverse of the 88 Sr signal (1/ 88 Sr).Three of the apatites (the shark tooth, the bovine tooth and the swine tooth) were used to build the daily calibration curve, while the human tooth was used only to check the precision and accuracy of the analysis on low-Sr samples.This daily LA calibration line (r 2 = 0.82; p < 0.00001; n = 18) results in a slope of 0.0005954 and an intercept of 1.0000077 (Fig. 4).We also built a calibration curve (r 2 = 0.99) based on the 2σ of the standard materials vs. the inverse of the Sr concentration (1/Sr) to obtain the external reproducibility of an unknown sample with a given Sr concentration.Sr concentrations of our standards have been previously measured by LA-ICP-MS for comparison.
The daily 84 Sr/ 86 Sr ratio is 0.05609 ± 0.00088 (2σ), which is, within error, equal to the natural 0.0565 ratio (for a detailed review of the significance of this ratio see Horstwood et al., 2008;Vroon et al., 2008).
All LA results of our standard materials are reported in Table 2.The LA analysis of the human tooth (Sr c.a. 100 ppm) yielded an average value of 0.70917 ± 0.00050, which is identical to the true value by dissolution MC-ICP-MS (0.70917 ± 0.00001; 2SE).

Retzius line counting
Retzius lines were counted only in sample RH-IS26 because it was the best available tooth fragment with the proper orientation for this type of observations (Fig. 5).
Relying on visible striae, we have estimated the presence of c.a. 30-35 Retzius lines along the entire length of the enamel and of about 10-15 lines along the OES.The number of laminations per Retzius band (i.e.periodicity) is 7, in agreement with the observations of Tafforeau et al. (2007).Thus, sample RH-IS26 should cover a period of about 200-250 days along the EDJ and of about 80-100 days along the OES.In the upper part of the sample (nearest to the apex), we approximately calculated an outward extension rate of the enamel of about 0.5 mm/month (~ 15 µm/d) in agreement with literature data (e.g.Tafforeau et al., 2007;Bendrey et al., 2015).
All dissolution results, including those of the standard materials, are reported in Table 2.

In situ LA Sr isotopes
The whole dataset of RH-IS30 and RH-IS26 87 Sr/ 86 Sr ratios is reported in Table 3.
The 87 Sr/ 86 Sr ratios, however, vary significantly within the RH-IS30 enamel ranging from 0.70966 to 0.70938.The mean 87 Sr/ 86 Sr value of LA dentine analyses is 0.70918 ± 0.00013 (2σ, n = 10), a ratio quite similar to modern seawater (~0.70917;Dia et al., 1992).Using our standard materials with known Sr concentrations, we have also calculated a Sr concentration of 361 ± 49 ppm (2σ) in the enamel and 447 ± 76 ppm (2σ) in the dentine for sample RH-IS30.

Diagenesis
Both teeth show significant isotopic variability along the enamel surface (Fig. 1 and 2).Enamel, unlike dentine (e.g.Lee-Thorp and Sponheimer, 2003;Bentley, 2006), is rarely affected by diagenetic contamination even in prehistoric samples (Budd et al., 2000), we can, therefore, consider these variations as real.The 87 Sr/ 86 Sr ratio of dentine, however, is still an important information because is used as the reference value for the "local" substrate isotopic signature (e.g.Bentley, 2006;Britton et al, 2011).This is due to the large pores (1µm tubules) of dentine that make this tissue highly susceptibility to diagenetic alteration (Kohn et al., 1999) especially in prehistoric samples.Any significant Sr enrichment in dentine can be due to the effect of local diagenetic contamination and/or animal physiology (see Evans et al., 2006).
Sample RH-IS30 shows a dentine-enamel difference of 86 ppm, while sample RH-IS26 of 35 ppm.The smaller Sr concentration difference of sample RH-IS26 might be interpreted as a minor diagenetic alteration for this sample that in fact retains a 87 Sr/ 86 Sr ratio (0.70934) higher than sample RH-IS30 (0.70918) and perhaps less affected by the modern seawater signal of 0.7092, that locally dominates the Sr rainwater isotopic composition (e.g.Montgomery, 2010).Therefore, we can infer that the local Sr isotope signature should be lower than ~0.7093, probably close to modern seawater.

External reproducibility
The external reproducibility is the error associated with repeated measurements of reference standard materials, generally expressed as a multiple of the standard deviation.
We calculated an external reproducibility of 0.00008 (2σ) for sample RH-IS30, whereas the interpolated external reproducibility is 0.00005 (2σ) for sample RH-IS26 (see section 2.3 of this paper for more details).Therefore, we consider all values outside this range as significantly different Sr isotopic compositions that can be related to morphological growth features of the teeth as discussed in the following section.

Rhino tooth growth and Sr isotopic composition
The estimated crown formation time of rhino species ranges between about 1.5 and 3 years (Tafforeau et al., 2007;Hillman-Smith et al., 1986), a long enough period during which the animal may have migrated through the landscape.For all the profiles we studied at the microscope and analysed by LA, we approximately followed the growth direction of the enamel (e.g.Tafforeau et al., 2007;Bentley, 2006;Khon et al., 2002), trying to observe possible seasonal variations both in the growth features and in the Sr isotopic composition.Because dentine growth direction is inward from the EDJ, those portions of the profiles that include dentine samples are inverted in terms of tissue growth.However, because of the observed diagenetic contamination of dentine, we considered this tissue as a homogeneous matrix, ignoring any possible incremental feature.

Tooth sample RH-IS30
The RH-IS30 sample is a horizontal section of a rhino molar fragment, but given that enamel nucleates at EDJ and grows downward from the tooth crown and outward from the EDJ itself, it is difficult to determine the precise time-span that this tooth section covers.In large herbivores, namely horses from Kohn et al. (2002), the outward growth rate is about 1 mm/month (Kohn et al., 2002).Given that the enamel thickness of RH-IS30 sample is about 2 mm, the time recorded by this sample should be at least 2 months, or even longer considering that rhino enamel growth seems slightly slower than the one of a horse (Bendrey et al., 2015;Tafforeau et al., 2007;Kohn et al., 2002).RH-IS26 sample, based on the observation of laminations, revealed an outward growth rate of about 0.5 mm/month in agreement with the growth rate of rhino teeth (Tafforeau et al. 2007).This suggests a recorded time span of about 3-4 months for sample RH-IS30.Both Sr isotope profiles obtained for this sample follow the outward direction (from the EDJ) of the enamel growth.Profile A, composed by 32 small linear analyses, covers the entire length of the sample with the first line near the EDJ and the last line close to the external surface (Fig. 1).This profile includes relatively regular cyclic variations of the 87 Sr/ 86 Sr ratio, from highly radiogenic ratios (>0.70960) to less radiogenic ratios (<0.70945).A simple t-test (>0.70960, single sample one-tailed t-test p = 0.02; <0.70945, p = 0.04) shows that these ratios are statistically different from the mean of the profile (0.70953).Profile B, measured half within the enamel and half within the dentine, also shows cyclic variations of the Sr isotopic ratios but with higher periodicity and clearly displays the chemical boundary between the enamel and the dentine.The enamel part of the profile includes only one value (0.70957) statistically different from the mean value of the profile itself.

Tooth sample RH-IS26
Sample RH-IS26 is a vertical section of a rhino molar fragment.For this sample, we analyzed both the exposed inner surface (Fig. 2) and the external surface (OES, Fig. 3) of the enamel.We observe cyclic variations of the Sr isotopic composition in both profiles with different periodicity.Theoretically, the OES and the exposed internal surface should encrypt similar incremental information.However, Tafforeau et al. (2007) showed how the external and the internal surface (along the EDJ) of the rhino enamel have different growth rate.When Retzius lines reach the OES of the enamel, they form ridges called perikymata (e.g.Smith and Tafforeau, 2008).Our sample represents the last ~ 8 mm (distance from the cervix) of the crown.Hence, the internal exposed surface of the enamel based on rates by Tafforeau et al. (2007) should represents a time-span of about 300-350 days, while the external surface a time-span of about 100 days.Our counting of Retzius lines agrees well with these estimates, revealing at least a time-interval of about 200-250 days along the EDJ and of 80 days along the OES.
According to the different growth rate between OES and inner surface, we observe that the second half of the Sr isotopic composition profile E seems to reflect the same fluctuations of the entire length of the F profile.In Figure 3c we compared the two profiles and shifted them to obtain the strongest possible correlation (r 2 = 0.56; p < 0.00001), confirming the fact that the two profiles show equivalent temporal variations in Sr isotopic composition.In addition, the other two profiles (C and D), which should encrypt fewer (about 2) months of the rhino life, display similar cyclical variations.

Rhino seasonal movements
Given that the Sr isotopic composition of the growing rhinos teeth derives from the land where the animal lived, these cyclic variations of the 87 Sr/ 86 Sr ratio likely indicate different water and food intake and, therefore, seasonal movements of the two animals to different lands, e.g. with different Sr isotopic composition.Even the shorter time range (about 100 days) observed seems to be characterized by cyclic variations of the 87 Sr/ 86 Sr ratio and consequently of the migration pattern.Moreover, both tooth samples show an enamel 87 Sr/ 86 Sr ratio considerably higher than its relative dentine ratio that represents the local water/food signal (0.70951 vs. 0.70918 for RHIS-30, one-tailed Mann-Whitney U-test p = 0.04; 0.70976 vs. 0.70934 for RH-IS26, onetailed Mann-Whitney U-test p = 0.06).
We try now to identify the Sr reservoir responsible for the chemical variations observed in the rhinos.The site of Isernia and the surrounding area are characterized by Holocene and Pleistocene fluvial deposits and limestone and dolostone ranging between the Cretaceous and the Holocene (Peretto et al., 2015).Therefore, the 87 Sr/ 86 Sr isotopic composition of these rocks should range approximately between 0.7073 and 0.7092 according to the global Sr isotopes seawater curve of McArthur et al. (2001).This is a big interval, however, our samples have even higher 87 Sr/ 86 Sr ratios.One of the possible sources of the more radiogenic ratios of our samples can be identified in the Roccamonfina volcano, located 50 km to the southeast of Isernia.
Previous geological work has shown that the lavas from the 630 ka eruption of the Roccamonfina volcano have a 87 Sr/ 86 Sr isotopic signature ranging between about 0.7093 and 0.7100 (Conticelli et al., 2009).In addition, located about 150 km to the northeast of Isernia, the Middle-Pleistocene Colli Albani volcano lavas show 87 Sr/ 86 Sr ratios higher than 0.7100 (Giaccio et al., 2013;Boari et al., 2009).Likely, our Isernia Rhinos have travelled to these volcanic grounds and fed themselves with water/food "contaminated" with highly radiogenic 87 Sr/ 86 Sr ratios.
According to palaeoenvironmental reconstructions, the Isernia La Pineta environment consisted of an association of closed forests, grasslands and semi-open woodlands (Orain et al., 2013;Thun Hohenstein et al., 2004), seemingly a good ground for rhinos grazing.However, based on our data, these rhinos were very mobile and seasonally reached the fertile substratum offered by the volcanic soils (Ugolini and Zasoski, 1979).Understanding the specific direction of the rhinos movements (either to the Roccamonfina volcano or to the Colli Albani area) remains no more than a speculation.Yet, we can certainly state that the Isernia rhinos have seasonally migrated as shown by the Sr isotopic composition temporal variations measured in their teeth, covering a distance of at least 50 km and returned cyclically to Isernia La Pineta or to a different site, but farther away from the isotopic influence of the volcanos.

Conclusions
We exploited the LA-MC-ICP-MS technique to study ancient animal migration patterns of two Stephanorhinus hundsheimensis teeth discovered in the Middle Pleistocene site of Isernia La Pineta, where the most ancient human Italian tooth has been unearthed.The two teeth, here analysed in high-spatial resolution, revealed time-related fluctuations of the Sr isotopic composition, likely linked to seasonal movements of the rhinos, probably towards the more fertile volcanic soils of Roccamonfina and/or Colli Albani.Moreover, while dentine has been possibly altered by diagenetic processes, enamel seems to be pristine and therefore suitable for reliable Sr isotopic analysis.As expected, both the inner and the outer enamel surface of the vertically sectioned sample, RH-IS26, showed highly correlated variations of the 87 Sr/ 86 Sr ratio, but with different time spans, related to the differential growth rate of the tooth portions.Additionally, we have successfully investigated the 87 Sr/ 86 Sr ratio of low-Sr (to c.a. 100 ppm) bioapatites, using three matrix-matched standards to correct unresolved interferences arising from LA analyses.100x 500 µm) LA analyses; blue line is the average of the analyses; grey area is the 2σ-error of 0.00008 calculated from the standard materials; x-error bars are the 2-standard errors.c.Profile B starts from the dentine and proceeds outward to the OES; while for enamel the profile follows the tooth growth, for dentine the direction is inverted (see text for details); symbols are the same as in profile A. d.Example of a tooth horizontal section (redrawn from Ballatore and Breda, 2013).Profile F is a single LA analysis which cover all the length of the outer enamel, reported as a moving average with a period of 100; blue line is the average of the analyses; grey area is the 2σ-error of 0.00005 calculated from the standard materials.c.Comparison between profile F and the second half of profile E (Fig. 2d); the two profiles show a high correlation (r 2 = 0.56).For a better visual comparison, profile E, which has relatively less data, is smoothed with a Savitzky-Golay filter.
Scientific, Neptune ™ ) housed at the Centro Interdipartimentale Grandi Strumenti (CIGS) of the University of Modena and Reggio Emilia.Seven of the nine Faraday detectors, fitted with 10 11 Ω resistors, were used to collect signals of the following masses: 82 Kr, 83 Kr, 84 Sr, 85 Rb, 86 Sr, 87 Sr, 88 Sr. Solutions were diluted to 200 ppb of Sr and introduced into the Neptune via quartz spray chamber and a 100 microliter/min nebulizer.The 88 Sr signal of the shark tooth (~1100 ppm), after proper dilution to 200 ppb, was ~15 V and the blank level (0.5 HNO3) 0.02 V. Samples and standards were analyzed in a static multi-collection mode in a single block of 100 cycles, with an integration time of 8 s per cycle.A classic bracketing sequence [blank/standard/blank/sample/blank] has been used to monitor any eventual drift of the instrument.
have attributed the interference on mass 87 to the 40 Ar 31 P 16 O molecule, because of the omnipresence of Ar in the plasma atmosphere.They correct this polyatomic interference adding a flux of nitrogen gas to reduce the formation of oxides within the plasma and monitoring mass 71 ( 40 Ar 31 P) as a proxy of the 40 Ar 31 P 16 O oxide production.This method seems to solve the interference problem related to mass 87 but largely reduces (factor of 2-3) the intensity of Sr signals, a serious problem when dealing with low-Sr samples.They partially resolve this issue by employing a larger spot size (~250 µm), resulting in a more intense signal because of the increased volume of ablated material

Fig. 1
Fig. 1. a. RH-IS30 sample photo, with visible Hunter-Schreger bands.b.Profile A is composed by 32 small (100 x 500 µm) LA analyses; blue line is the average of the analyses; grey area is the 2σ-error of 0.00008 calculated from the standard materials; x-error bars are the 2-standard errors.c.Profile B starts from the dentine and proceeds outward to the OES; while for enamel the profile follows the tooth growth, for dentine the direction is inverted (see text for details); symbols are the same as in profile A. d.Example of a tooth horizontal section (redrawn fromBallatore and Breda, 2013).

Fig. 2
Fig. 2. a. RH-IS26 sample photo of the exposed inner surface.b.Profile C cover a short portion of the enamel near the OES and is composed by 11 LA analyses; blue line is the average of the analyses; grey area is the 2σerror of 0.00005 calculated from the standard materials; x-error bars are the 2-standard errors.c.Profile D starts from the dentine and proceeds outward to the OES; while for enamel the profile follows the tooth growth, for dentine the direction is inverted (see text for details); symbols are the same as in profile C. d.Profile E is a single LA analysis which cover all the length of the enamel, reported as a moving average with a period of 100; symbols are the same as in profile D. e. Example of a tooth vertical section (redrawn from Ballatore and Breda, 2013).

Fig. 3
Fig.3.a. RH-IS26 sample photo of the outer enamel surface (OES).b.Profile F is a single LA analysis which cover all the length of the outer enamel, reported as a moving average with a period of 100; blue line is the average of the analyses; grey area is the 2σ-error of 0.00005 calculated from the standard materials.c.Comparison between profile F and the second half of profile E (Fig.2d); the two profiles show a high correlation (r 2 = 0.56).For a better visual comparison, profile E, which has relatively less data, is smoothed with a Savitzky-Golay filter.

Fig. 4 .
Fig. 4. Correlation between the accuracy of the 87 Sr/ 86 Sr analysis (observed/true) and the inverse of the 88 Sr signal.This high correlation is attributed by some authors to the Ca-P-O interference which occurs during LA analysis of bioapatites (see text for details).The line equation is used to correct our LA data.

Table 1 .
Parameters and operating conditions for Sr isotope determination by laser ablation MC-ICP-MS