New Insights on Micro‐Scale Variations of Geochemical and Oxygen Isotope Compositions in Conodont and Shark Tooth Bioapatite

To help understand bioapatite microstructures and related chemical variations, their impact on O‐isotope compositions measured and give insights on sample preparation, this study analysed conodonts and shark teeth prepared in different orientations through microanalytical and bulk sampling techniques: scanning electron microscopy (SEM); electron probe microanalysis (EPMA); continuous‐flow and high‐temperature reduction – isotope ratio mass spectrometry; and secondary ion mass spectrometry (SIMS). The SEM and EPMA measurements in conodonts allowed to distinguish the tissues commonly analysed by SIMS, which included albid and hyaline crowns but given their often small‐scale intergrowth, mixtures of these are difficult to avoid. In situ SIMS O‐isotope analyses provided different δ18O values: lower values with higher variance (16 ± 1‰ n = 13, 15.7 ± 1.9‰ n = 11) for mixed albid‐hyaline tissues, and higher, homogeneous values (17.1 ± 0.2‰, n = 13) for mainly hyaline tissues. Recent shark teeth δ18OSIMS value for dentine of the same tooth was 10‰ lower than the mean δ18OSIMS value for enameloid whereas the δ18OPO4 values measured for enameloid and dentine using the HTR method were identical. The variation of δ18O seems sensitive to analytical artefacts related to sample textures, caused during the sample preparation over more porous biomineral surfaces.

The phosphate mineralised in bones and teeth, herein also referred to as 'bioapatite' (e.g., Skinner and Jahren 2007, Wudarska et al. 2022), is an important proxy archive used in palaeontological research.The general formula for biogenic phosphate minerals with its most common anion substitutions is Ca 5 (PO 4 , CO 3 , SO 4 ) 3 (OH, F, CO 3 ), and the PO 4 3-and CO 3 2-groups can have their oxygen isotope compositions measured individually (e.g., Vennemann et al. 2001).Oxygen isotope compositions of phosphate bound oxygen (d 18 O PO4 ) are an alternative to that of carbonates to help understand the palaeoenvironment of the Palaeozoic era (e.g., Kolodny et al. 1983, Wenzel et al. 2000, Joachimski et al. 2009, Trotter et al. 2008, 2016a), including studies of mass extinction events (Joachimski et al. 2012, Sun et al. 2012, Romano et al. 2013, Goudemand et al. 2019).However, the application of oxygen isotope techniques to Palaeozoic bioapatites is often limited by the small size of one of the major fossils used for this era: conodonts.These are eel-like jawless vertebrates that were ubiquitous from the Cambrian to the Triassic (Luz et al. 1984).In the fossil record, conodonts are represented by dental elements that are characterised as fluorapatite (Pietzner et al. 1968, Sweet 1988), the bioapatite variant with most resistance against diagenesis (Posner et al. 1984), and each specimen averages 500 lm in size (e.g., Wenzel et al. 2000).Since the classical geochemical methods for O isotope determinations require a minimum amount of 1 to 5 mg of phosphate (e.g., fluorination, silver phosphate microprecipitation, for more details see Vennemann et al. 2002, Mine et al. 2017), the use of conodonts was limited given their small size and mass (\ 100 lg).While sampling of several conodonts per analysis (bulk sampling) of the same and of mixed species is a common approach for palaeoclimatic/ecological interpretations (e.g., Wenzel et al. 2000, Joachimski et al. 2009, 2012, Sun et al. 2012, Romano et al. 2013), the measured d 18 O PO4 values will average possible intra-tissue differences as well as intra-and inter-specific taxon variations.
Over the last decade, the d 18 O values of individual conodonts were increasingly measured by secondary ion mass spectrometry (SIMS) (Rigo et al. 2012, Trotter et al. 2008, 2015, 2016a, Narkiewicz et al. 2017, Zhang et al. 2017, Wheeley et al. 2012, 2018, Goudemand et al. 2019, Chen et al. 2016, 2020, Albanesi et al. 2020, Edwards et al. 2022, Liu et al. 2022).This in situ technique uses a finely focused primary ion beam for sputtering a sample surface, producing secondary ions that are extracted and analysed, to determine the oxygen isotope composition of the selected sample.The high spatial resolution of SIMS measurements can measure isotopic compositions with a depth of ablation of about 1 to 6 lm, at a diameter of 10 to 30 lm (e.g., Wheeley et al. 2012).The high intensity ablation could be a disadvantage, as the sample sputtering releases oxygen from all oxygenbearing sites in the bioapatite (e.g., CO 3 2-, PO 4 3-, OH -), and these different sites not only have different O isotope compositions, but also have a distinct susceptibility to diagenetic changes compared with the oxygen bound only to the phosphate ion (Iacumin et al. 1996(Iacumin et al. , L ecuyer et al. 2010)).Because oxygen derived from the phosphate ion dominates in fossilised bioapatite with about 90 to 95% of total oxygen contribution, compared with the 3 to 6% of oxygen from carbonate and 0 to 3% hydroxyl oxygen, it is often assumed that the d 18 O values measured by SIMS are close to those of the phosphate molecule (e.g., L ecuyer et al. 2010, Wheeley et al. 2012, Zhang et al. 2017).
Another constraint to obtain precise results on such small scales is related to the surface topography of the sample that must be less than 5 lm (Marin-Carbonne et al. 2022).Surface topography is known to affect the d 18 O values measured by SIMS by creating tilted surfaces compared with the perpendicular incidence of the secondary ion beam (e.g., Kita et al. 2009).The analysis of distinct tilted surfaces can effectively deform the electrostatic field and modify the trajectory of the secondary ions (Kita et al. 2009), thus the differential sputtering of irregular sample textures can cause instrumental mass fractionations (IMF).This is an important factor to verify for bioapatites as they have different biomineralogical structures (e.g., Skinner and Jahren 2007), which may result in different behaviour during polishing.
Knowledge on the chemical composition of bioapatites is also important, both for palaeo-environmental interpretations and to verify whether a "compositional" instrumental fractionation effect exists for bioapatite.For example, studies of calcite have noted such compositional 'matrix effects' for Mg, Fe, and Mn contents that can affect the IMF (Rollion-Bard and Marin-Carbonne 2011, Sliwi nski et al. 2018).For Mg variations in carbonates a standard deviation for d 18 O of about 0.3‰ per % m/m was noted only, but for Fe as FeCO 3 only about 1 to 2 mol% of Fe can bias the d 18 O values by 2 to 3‰ (Rollion-Bard and Marin-Carbonne 2011).While the Mg in bioapatite typically has mass fractions of less than 1.5% m/m (e.g., Elliott 2002, Skinner andJahren 2007) and hence its effect is expected to be minor, Fe can certainly be enriched by diagenetic processes (Raiswell 1997, Trotter andEggins 2006).Although some quantitative and qualitative chemical analysis were monitored for conodonts and shark teeth (e.g., Trotter and Eggins 2006, Kocsis et al. 2015, Zhang et al. 2017, Shirley et al. 2018, Zigait _ e et al. 2020, Leonhard et al. 2021, Terrill et al. 2022), showing different compositions between structures like the enameloid vs. dentine in shark teeth or the albid vs. hyaline conodont crowns, few contributions illustrate their chemical distributions in more detail (Katvala and Henderson 2012, Enax et al. 2014, Terrill et al. 2018).Also, it is possible that some structures preferentially occur in a certain taxon (Leonhard et al. 2021), as is the case of the conodont albid crown, suggested to be a further adaptation to dental function by allowing the conodonts to withstand greater tensile stresses (Jones et al. 2012).Shark enameloid structures have also been noted to contain chemical distributions slightly distinct (Enax et al. 2014), with the outer shiny enameloid layer having higher mass fractions of minor elements (Mg, Na) at the cost of major components (Ca, P, F) when compared with the underlying enameloid.This compositional heterogeneity within the enameloid layers is similar to larger differences noted between the dentine vs. enameloid in shark teeth and may influence the choice of the tissue to be used for in situ investigations.
Another possible source of bias could be that data from bioapatites measured by SIMS is normalised and calibrated using the magmatic fluorapatite of Durango (cf., Trotter et al. 2008, Wheeley et al. 2012, Sun et al. 2016, Zhang et al. 2017, Wudarska et al. 2022).Durango apatite is a distinctive fluorapatite from an open-pit iron ore mine at Cerro de Mercado, Durango City, Mexico (Chew et al. 2016).Besides having a high inter-crystal oxygen isotope variation (up to 4.4‰, Sun et al. 2016), the Durango apatite is compositionally different from conodont bioapatite and may have specific sources of isotopic bias when sputtered by an ion beam during in situ analyses (e.g., Trotter et al. 2015, 2016a, Sun et al. 2016, Wudarska et al. 2022).To evaluate such matrix related effects (Eiler et al. 1997), previous studies employed a second fluorapatite reference material (RM), for example a great white shark tooth, to be analysed in parallel with the Durango RM (Trotter et al. 2008, Rigo et al. 2012).The enameloid structure in shark teeth would be an optimal choice as a secondary RM, given the relative abundance of both modern and fossil specimens (Cappetta et al. 2012).They are also frequently used as proxies for palaeoclimate, and numerous contributions were done using the bulk method for the enameloid (e.g., Kocsis et al. 2014, Leuzinger et al. 2015, Carrilo-Briceno et al. 2019).Earlier investigations using bulk sampling noted that modern shark teeth were relatively homogeneous in their d 18 O (e.g., Vennemann et al. 2001, Kocsis et al. 2015), with minor variations only noted in inter-and intra-tooth tissue (dentine vs. enameloid) analyses.In addition, the larger size of a shark tooth compared with conodonts allows a higher number of analyses such that is possible to use the same sample for different SIMS sessions.Despite the potential to serve as reference materials, a more recent SIMS study on shark teeth, obtained from sharks living under controlled environmental conditions, measured d 18 O variations of up to 4‰ in the same enameloid substructure ( Zigait _ e and Whitehouse 2014).This highlights the necessity of further studying shark teeth to test their feasibility to normalise d 18 O values.
Ideally, a conodont fluorapatite RM could be used for the measurements of conodonts.Some studies specifically addressed in situ d 18 O variations in conodonts (Wheeley et al. 2012, Rigo et al. 2012, Trotter et al. 2008, 2015, Zhang et al. 2017), but there are contrasting conclusions about the best biomineralogical structure to be used (e.g., Trotter and Eggins 2006, Trotter et al. 2007, Zigait _ e et al. 2020).It can also be challenging to characterise conodonts for in situ investigations.The conodont crown is composed of hyaline and albid tissue (Donoghue 1998), and a clear distinction between the two structures is often difficult, since their boundaries can be gradual (Donoghue 1998, Trotter et al. 2007).Furthermore, SIMS analysis was selected using a light microscope, which does not allow for an identification of conodont crown tissues.While not all previous in situ oxygen isotope measurements note a systematic difference between hyaline and albid crown, overall O-isotopic variation may be higher if different tissues are sampled (e.g., Wheeley et al. 2012, Zhang et al. 2017).A strict, unified sampling protocol may ease the visualisation and hence use of one or another crown tissue, which may be necessary should rare conodont specimens in any one sample make SIMS measurements the choice for in situ geochemical analysis.
To test for factors influencing possible instrumental mass fractionations (IMF) of bioapatites more rigorously, the oxygen isotope compositions of biogenic hard tissues of conodonts and shark teeth were analysed systematically together with their microstructure and chemical compositions.For calibration and comparison purposes, O isotope measurements were made through a classical method of silver phosphate microprecipitation and analysis in bulk sampling by high-temperature reduction (HTR).
A final objective of this study is to develop a secondary reference material for in situ isotopic measurements of bioapatites such as conodonts, to be applied to future oxygen isotope analyses in phosphate for sites where single conodont analyses are required in order to gain a better understanding of past climatic oscillations.Ultimately, improvements in the accuracy of measuring d 18 O values from conodonts will refine the interpretation and understanding of palaeoclimatic changes, such as for example the climatic oscillations following the Permian-Triassic mass extinction (e.g., Goudemand et al. 2019).

Study area and material
Conodonts from five Early Triassic sites were used (Table S1): Noe Tobe (West Timor, Indonesia), Wadi Musjah (Oman), Stensi€ ofjellet (Spitsbergen), Crittenden Springs and Georgetown (United States of America).Each locality has its specific depositional features, a description of which is summarised in the online supporting information File S1.Rock samples were disaggregated in acetic acid (10%), sieved and screen washed (0.5 mm, 1 mm, 2 mm) for subsequent picking of conodonts using a binocular microscope.Taxonomic identification is based on references of Early Triassic species (Orchard 2007, Leu et al. 2019).Among the multi-elements known from Early Triassic conodont dental apparatus, platform-like elements (P 1 elements) were preferred as they are taxonomically more diagnostic.Scythogondolella ex gr.milleri were selected as our reference taxon as their segmiplanate P 1 elements were available from all localities (Table S1).This species is widely used in the conodont biostratigraphic zonation and is commonly related to the Wasatchites distractus/Anasibirites multiformis ammonoid zone (Jattiot et al. 2015, Leu et al. 2019, Goudemand et al. 2019).Scythogondolella ex gr.milleri has a relatively short fossil record (about 200 ky) and can be found in late Smithian deposits around the world (e.g., Orchard andZonneveld 2009, Leu et al. 2019).
The basal body of the conodonts, the structure more susceptible to diagenesis (Wheeley et al. 2012, Zhang et al. 2017), has not been preserved in the samples chosen.
The maximum burial temperatures of the conodonts can be assessed by the Conodont Alteration Index (CAI; Epstein 1977), which is about 1 for the set of samples chosen (\ 80 °C; Table S1).In contrast, USA specimens from Crittenden Springs have a CAI of 2.5 to 3 (\ 200 °C).One CAI 5 specimen (\ 300 °C) from Guryul Ravine, Kashmir (Leu et al. 2019) was used for comparison purposes of its chemical distribution (Table S1).
Modern shark teeth that have been studied previously (Vennemann et al. 2001) and fossil shark teeth donated to T. Vennemann completed the sample set.Shark tooth samples are all from the Lamniformes order (also known as mackerel sharks, Ebert et al. 2013): the modern representatives are Great White sharks from waters of the east coast of South Africa (KwaZulu-Natal, Vennemann et al. 2001) and fossil teeth are from Odontaspis denticulata, Anotodus sp. and Carcharias cuspidata, recovered from Oligocene-Miocene marginal marine basin settings of the Molasse basin (Vennemann et al. 1998).
While ideally all samples would have been used for all geochemical techniques mentioned in the introduction, the complex preparation techniques for the small conodonts allowed us to use some specimens for one method only.Nonetheless, these samples help illustrate the structural and chemical variations found on lm-scales within conodont bioapatite.

Methods
Conodonts and shark teeth were prepared in different views (longitudinal and transversal cross sections), to verify which sample orientation best exposes the different tissues and hence is most advantageous for in situ d 18 O measurement.The microstructure was generally investigated, both before and after a measurement session for the SIMS, by high-resolution scanning electron microscopy (SEM).Chemical compositions were verified through qualitative and quantitative measurements using electron microprobe analyses (EPMA), and for carbonate content also the Gasbench-II acid extraction method of CO 2 and subsequent isotope ratio mass spectrometry measurements (IRMS).Given the difficulty to distinguish the conodont crown tissues and their structures with the SIMS, the predominant tissue (albid, hyaline or a mixture) associated with the different sample orientations was determined through comparisons of the combined microstructural, chemical and isotopic results of this study with data from literature (Pietzner et al. 1968, Trotter and Eggins 2006, Wheeley et al. 2012, Trotter et al. 2016a, Zhang et al. 2017, Zigait _ e et al. 2020, Terrill et al. 2022).Our combined approach may hence provide additional insights about the structure, texture, and odontogenesis on the studied specimens (e.g., Katvala and Henderson 2012, Enax et al. 2012, 2014, Shirley et al. 2018, 2020), as well as shed light on analysesspecific IMF for SIMS O-isotope measurements of different biogenic phosphate tissues.
A detailed list the different techniques used is summarised in Table S1.The samples were prepared for analyses using the following methods available at the University of Lausanne (UNIL): scanning electron microscopy (SEM); qualitative and quantitative electron microprobe analyses (EPMA); continuous-flow and high-temperature reduction (HTR) and analysis by isotope ratio mass spectrometry (Gasbench-II-IRMS and HTR-IRMS); and secondary ion mass spectrometry (SIMS).
First preparation steps: At the University of Lausanne, the samples were cleaned in distilled, de-ionised water using an ultrasonic bath to reduce surface contamination prior to drying overnight at 70 °C.Prior to these laboratorial steps, the modern shark teeth had been cleaned in boiling salt water to remove organic matter (Vennemann et al. 2001).The complete jaw of the great white shark used was immersed in a large recipient of boiling water (100 °C), saturated with salt to simulate seawater (i.e., about 35 psu), and left for about an hour or until the organic matter was visually removed.The shark teeth were then individually collected from the jaw.

SEM, EPMA and SIMS
Sample selection and analytical protocol: For the conodonts, only the Oman and Timor sites provided enough specimens to use for the bulk sampling method.Sites from the USA and Spitsbergen had few specimens for in situ measurements only.A large number of well-preserved (CAI 1) conodonts of Scythogondolella ex gr.milleri were collected from the exotic carbonate block from Noe Tobe (West Timor, Indonesia) (Jattiot et al. 2015).Their preservation state and abundance represented a good opportunity to establish our analytical protocol for SIMS measurements.Conodonts from Timor were cross-sectioned to note their structurally different tissues (albid vs. hyaline tissue).Initially, two main cross-sectioning strategies were adopted: a transversal cross section (Figure 1b, c); and a longitudinal cross section at the lower part of the crown (Figure 1d).
Conodonts transversally cross-sectioned were positioned to expose the posterior side, where the basal body was located; the basal body is often considered analogous to the vertebrate dentine root zones for the teeth (Donoghue 1998).
To verify that the samples were not influenced by remains of a broken basal body, some samples were transversally cross-sectioned at the anterior part (Figure 1c).After some experiments, samples cross-sectioned longitudinally were re- polished after a first analysis of oxygen isotopes (labelled as 'b' in Table S1).However, these cross sections where the exposed portion was at the interior of the conodonts had unique characteristics and deserved more attention.Two more samples were prepared for this approach (TM19 and 20).In total, three main conodont groups are presented in the results and discussion: conodonts with a low longitudinal cross section; with a middle longitudinal cross section; and conodonts transversally cross sectioned.As SIMS measurement spots on conodonts are chosen using light microscopy, we tried to verify the tissue type (hyaline or albid) that was sputtered using the SEM after the SIMS measurements.However, even for SEM measurements, the identification of the tissue type in conodonts is difficult to assess (e.g., Donoghue 1998, Trotter et al. 2007) and hence we also added further constraints from EPMA and carbonate content measurements (cf., Pietzner et al. 1968, Trotter and Eggins 2006, Wheeley et al. 2012, Trotter et al. 2016a, Zhang et al. 2017, Zigait _ e et al. 2020, Terrill et al. 2022).Despite our combined use of methods, the tissues could not always be clearly divided into hyaline or albid tissues, hence a large number of analytical spots are classified as "mixed tissues" for cases where the distinction was not clear (see section Bioapatite microstructures for more detail).Original output images and additional cross-sectioned specimens are presented in File S2.Based on preliminary d 18 O values (sections Oxygen isotope measurements of phosphate and Conodont chemical and oxygen isotope compositions) in the three main conodont groups from Timor, specimens from Oman, USA and Spitsbergen were prepared with a low longitudinal cross section only.
For the shark teeth, the analyses focused on the tip of the teeth.This is a zone of convergence of the outer enameloid layer (Figure 2), offering a larger surface area for the analysis.Individual tooth tips were cross sectioned to expose two views: a transversal cross section (Figure 2b, c) and a longitudinal cross section to expose both enameloid and dentine in different orientations (Figure 2a).An influence of the different c-axis orientations for minerals of the apatite group was noted for fluorine mass fractions measured by EPMA (e.g., Goldoff et al. 2012).However, no differences were reported for oxygen isotope analysis of crystallographically different orientations of analyses (e.g., Li et al. 2021, Wudarska et al. 2022).Assuming a similar microstructure as other lamniform specimens previously imaged (cf., Enax et al. 2014, Wilmers et al. 2021), the fluorine content was compared in shark teeth "parallel-bundled" enameloid tissue (PBE, Figure 2, Enault et al. 2015).Given that in lamniform, the PBE has its crystallite bundles arranged in parallel to the long growth axis of a tooth, in the transversal cross section, the PBE has its c-axis parallel to the EPMA and SIMS beams, while in the longitudinal cross section, the c-axis of the PBE is perpendicular.
Sample preparation: Samples for individual analysis were mounted into an epoxy resin (EpoFix), polished, cleaned and dried overnight (60 °C) before coating with carbon (SEM, EPMA) or gold (SIMS).For SIMS and parallel SEM, EPMA, and GasBench measurements, eight shark teeth and conodont elements (GW1, GW5, TM1-to TM6) were pre-treated with Ca-buffered acetic acid (2 mol l -1 , pH = 4.5, 2 h) to remove any exogenous carbonates, followed by several rinse cycles with distilled, de-ionised water (Koch et al. 1997).For some sessions the samples were analysed in the original resin where they were prepared, but for other sessions the resins were cut and placed into an indium mount (details in https://swisssims.com/sample-preparation/).This was necessary to have samples previously analysed by SEM or EPMA in the same sample mount, together with the Durango RM for the SIMS analysis.Also, conodonts appeared to have a slightly different resistance during polishing depending on their size.Smaller conodonts generally are 'ready' to analyse after a shorter polishing time than larger conodonts; smaller conodonts tend to be lost when polished too long.To avoid material loss, conodonts were polished separately and then mounted in an indium mount.New and used indium beads were melted into the mounts and pressed several times using a hydraulic press.The pressure was gradually increased (1000 to 5000 psi, 1-5 min per step) until a satisfactory surface topography was achieved.To avoid analytical artefacts related to sample geometry and topography, samples were prepared to be within 10 mm from the centre of the mount and to have less than 5 lm of surface topography relative to the RM in the same mount (cf., Kita et al. 2009).A good sample topography for analysis was readily obtained for the smaller conodonts (about 3 lm), but for the larger bioapatites of sharks this was more difficult (File S3).The topography varied within the enameloid zones (Figure S1) and in some samples, layers had more than 5 lm of relief compared with the RM in the mount (Durango).Although most parts of the dentine of modern sharks were within the 5 lm relief, dentine has a more irregular surface texture in modern compared with the fossilised teeth, given that the void space in dentine is filled by newly formed and larger crystals of bioapatite during fossilization.Relief was particularly poor for recent samples of shark teeth pre-treated with acetic acid (GW1 in Figure 2C, GW5 in File S2).Details about how this textural variation could have influenced the results are discussed below (Major element compositions and mineralogy and Shark teeth chemical and oxygen isotope compositions).
All reported uncertainties in the results are calculated in 1 standard deviation (1s).The raw data obtained from all analyses are available at a Data Repository [https://doi.org/10.5281/zenodo.8195665](Luz et al. 2023).We provide additional data, data treatment and normalisation in the online supporting information.
Scanning electron microscopy (SEM): Analyses were made using a Tescan Mira II LMU in backscattered electron emission (BSE) mode.Conodonts were examined for growth layers and albid/hyaline tissues in different orientations (Figure 1, File S2).Only transversally cross sectioned shark teeth were studied but it was enough to observe the enameloid components and the dentine, as well as to estimate the structural arrangement of samples longitudinally cross sectioned (Figure 2, File S2).We used the terms and definitions of biomineralogical structures presented by Trotter and Eggins (2006), Orchard (2007) and Zhang et al. (2017) for conodonts and by Enault et al. (2015) and Wilmers et al. (2021) for sharks in the Results and Discussion.
Electron probe microanalyses (EPMA): Quantitative analyses were made on seven bioapatites (three conodonts, four sharks, Table S1) and qualitative chemical distribution maps were prepared for fourteen bioapatites (eight conodonts and six sharks, Table S1) in order to shed light on likely systematic zoning in conodont elements and shark teeth (Figures 4, 5, Tables 1, S2, File S4).Chemical concentrations and distributions were determined using a JEOL JXA-8530F HyperProbe.For the quantitative analysis (spatial resolution: 5 lm), longitudinal and transversal cross sections of shark and conodonts (Table S2) were selected.For the qualitative maps of sharks, the same bioapatite orientations were used (Table S2, File S4), and for conodonts we included a sample with the anterior part transversally cross sectioned (Figure 1b, Table S2, TM28 in File S4).
Measurements were produced in four sessions (October 2017, May, November and December 2019).The accelerating voltage was between of 15 kV in the first session (only qualitative maps done, samples: GW5, MS2, TM28 in File S4), and of 10 kV in the three latter sessions (quantitative and qualitative analyses), all with a current of 20 nA.The reason for using a lower voltage in the latter sessions was to reduce the sample damage and to be able to analyse the same specimens by SIMS in the same layers investigated by EPMA.For all sessions, the diameter of the beam was 5 lm and was slightly defocused, also for reducing sample damage.Each sample point was analysed during 50 ms.For quantitative analyses, element selection includes major elements present in the conodont bioapatite (Ca, P, F) and minor and trace elements common in bioapatite (Na, S, Mg, Fe, Sr, Cl, Y, Al, K) (e.g., Trotter and Eggins 2006, Katvala and Henderson 2012, Enax et al. 2014, Zhang et al. 2017, Shirley et al. 2018).For the mapping, ten elements were analysed per counting cycles in a total of two cycles.Among the selected elements, the Cl mass fractions were only analysed after the first quantitative analysis results (session of May 2019), hence it is absent from some analyses given in the Table S2.
The analytical criteria for data normalisation used were based on those described by Katvala and Table 1.Measurement results for major and minor element oxides (and elements) by EPMA of the bioapatite materials studied (results are means expressed as % m/m with 1s) The "unmeasurable portion" is considered to be largely organic matter (now void space) and/or CO 32-ions.
The analytical spots of each measurement are available in Table S2.
The detection limits (in lg g -1 ) for each element were: Na 80, K 120, Mg 80, Ca 135, Sr 425, Fe 580, Al 65, Y 330, P 300, S 145, F 235, Cl 185.The asterisk symbol within the squares of the structure results (e.g., FeO results for the modern shark) means that at least one of the measured values of the referred structure or element were below the detection limit (DL).
Henderson (2012).The "unmeasurable" portion of our mass fractions (% m/m) refers to the balance of the total mass fraction of the elements up to 100%, which are portions that are not measurable by EPMA, interpreted to represent carbonate, organic matter, or simply hydrocarbons of the resin in void space.To minimise uncertainties related to the carbonate component, Gas-Bench II and mass spectrometric CO 2 analyses of concentration and isotopic composition were made on some biogenic phosphates (bioapatite).Details are given below.
Data were selected based on an evaluation of edge effects, instrument measurement uncertainties and analytical artefacts from sample preparation.Calcium and P mass fractions were constant throughout the bioapatite, but some analytical points near the edges had unusually low Ca and P concentrations and a lower total compared with adjacent sample spots.This edge effect suggests that the beam was not centred on a flat part of the sample.A total of two data points were excluded from our quantitative analysis calculations and details of this evaluation are given in Table S2.Dark areas and areas depleted in major element concentrations (e.g., Ca, P) are due to traces of indium in void spaces that remained in the samples (verified using a reflected light microscope; see File S4).

Carbonate content estimates (GasBench-IRMS) and oxygen isotope analyses (HTR, SIMS)
GasBench isotope ratio mass spectrometry (-GasBench-IRMS): Stable isotope measurements were done of the carbonate component of four shark teeth (two modern and two fossil, Table S1) with the objective to estimate their carbonate content.The enameloid and dentine surfaces were abraded with a micro drill to collect sample powders.About 10-15 lg of a pure in-house carbonate RM (CM -Carrara Marble) was weighed into glass vials.For the shark enameloid, this quantity corresponded to 800-1000 lg, while for the dentine 500-600 lg was used.Analyses were made following the procedures of Sp€ otl and Vennemann (2003), adapted to use smaller sample quantities.The oxygen and carbon isotope compositions were analysed using a GasBench II coupled to a Thermo Fisher Scientific Delta V mass spectrometer.In the GasBench II, the vials were flushed using helium and reacted with 99% orthophosphoric acid to produce a CO 2 gas to be analysed by the mass spectrometer.The measured isotopic ratios were normalised to an in-house Carrara marble calcite RM calibrated against NBS-19.
High-temperature reduction isotope ratio mass spectrometry (HTR-IRMS): Oxygen isotopes were measured in twelve bioapatites (two conodonts, eight enameloid vs. Secondary-ion mass spectrometry (SIMS): Sixty bioapatite samples (fifty conodonts and ten enameloid vs. dentine of sharks, Table S1) were measured one to ten times (typically four sample spots) in situ for the oxygen isotope composition (d 18 O).Longitudinal and transversal cross sections of shark teeth (Figure 2) and conodonts (Figure 1) were selected.About fifty sample spots were measured per mount (File S5).Oxygen isotope measurements were made using the IMS CAMECA 1280HR of the SwissSIMS national facility in three different sessions (December 2018, June 2019, November 2019).A 1.5nA Cs + ion beam (10 kV voltage) was focused into a spot of about 15 lm and with a depth resolution of about 1.5 lm. 16O -and 18 O -secondary ions were detected on multicollector Faraday cups, equipped with 10 10 (L'2) and 10 11 (H'2) Ohms resistors, respectively.The analytical routine consists of calibration of the Faraday cups at the beginning of each day.Each analysis consists of 30 s of pre-sputtering, automated centring of secondary ion beam, followed by sixteen cycles of 4 s of secondary ion detection.In order to calibrate the measurements, two crystals of Durango RM were used for correction of d 18 O values.Their d 18 O PO4 values were determined by laser fluorination (first crystal: 8.7 AE 0.1‰, n = 2; second crystal: 9.0 AE 0.1‰, n = 4, details in File S1) (Vennemann et al. 2002).Each run was normalised with their respective Durango batch (File S5).Typical count rate on Durango was about 1.3 x 10 9 counts per second, similar to the unknowns.
An in-house quartz RM (UNIL-Q1) was measured repeatedly during each session, in order to monitor the instrument stability.Typical reproducibility (in 1s) was 0.15-0.2‰.Durango RMs were also measured repeatedly, although less often as they reproduce less well (e.g., Sun et al. 2016, Wudarska et al. 2022, File S5)

Bioapatite microstructures
High-resolution images were prepared in back-scattered electron (BSE) mode for fourteen bioapatites (nine conodonts and five sharks, Table S1) and they are given in Figures 1, 2 and 3, as well as in File S2.Both albid and hyaline crown growth laminae measure less than 10 lm across (Figures 1,  3), as such the analyses by SIMS will represent mean d 18 O values of at least two growth layers.Measurements of longitudinal cross sections for the lower part of the crown, were judged to be dominated by hyaline crown textural features, and there is a restricted degree of porosity near cracked portions (Figures 1d, 3b).In the transversally crosssectioned conodonts, the hyaline crown also appears to be dominant, from the central axis extending laterally outwards to the platforms.The upper parts of the denticle generally had a higher degree of porosity suggestive of albid tissues, often however mixed with layers of hyaline tissue (Figures 1b,  c, 3a, TM29 in File S2).Longitudinal cross sections of the lateral conodont view also have a high porosity at the denticles when compared with the lower parts (Figure 1e, f, g).While this distinction is suggestive  ).These structures are also distinguishable using reflected light microscopy (Figure 2a).The single crystallite enameloid (SCE) parts cover the outer surface of the crown; and the random orientation of crystallites occur at the edges of transversal cross sections and at the tip of the teeth in longitudinal sections.Chemical compositions may help to distinguish different layers too (see below).From the BCE unit, two of the three components can be seen: the parallel bundled enameloid (PBE) just adjacent to the SCE unit, with crystallites arranged parallel to the crown surface; and the tangled bundled enameloid (TBE) that is observed between the PBE and the dentine from fossil teeth, with crystallites of more random orientation but with a 'smooth' surface similar to the PBE.The dentine of recent shark teeth is less commonly exposed as it is readily damaged and even removed during sample preparation.In contrast, the lower degree of porosity for the recrystallised dentine parts in fossil shark teeth compared with the recent shark teeth (Figure 2b, c, GW5 in File S2), allows these parts to be preserved and then also not to be damaged by the polishing procedure.

Chemical compositions
Quantitative analyses (mass fractions in % m/m): One dental element from each of the three conodont groups was analysed: the TM18 for TM low longitudinal (sample spots n = 27), TM20 for TM mid longitudinal (n = 29) and TM21 for TM transversal (n = 30) (Tables 1, S2).Conodonts longitudinally cross-sectioned (TM low longitudinal and TM mid longitudinal ) are similar in their totals for the major components (Ca, P and F), but with slightly lower mean totals measured for the TM low longitudinal samples.As outlined in the Material and methods section, the "unmeasurable" portions of our mass fractions (% m/m) are interpreted to represent carbonate, organic matter, or simply hydrocarbons of the resin void space.These "unmeasurable" mass fractions are more contrasting between the conodonts sectioned longitudinally, with 2.6 AE 1 (TM low longitudinal ) and 1 AE 0.8% (TM mid longitudinal , Table 1), respectively, albeit nevertheless within the measurement uncertainty.Samples transversally cross-sectioned (TM transversal ) clearly had a higher "unmeasurable" content at the lowest part of the crown (5.2 AE 1.2%) compared with the upper part (1.6 AE 0.7%).Magnesium, S and Na mass fractions are also higher in the lower part, compensated for by lower Ca, P, and F in the same parts (Table 1).When excluding data points from the lower part of the crown (n = 4), the Ca, P and F mass fractions of the TM transversal are within the measurement uncertainty the same as to those of the TM mid longitudinal sample.
Four shark teeth were examined by quantitative analysis: two fossil and two modern specimens prepared in transversal and longitudinal cross-sections (Table 1, Table S2).The enameloid and dentine clearly have different compositions.In general, the fossil and modern enameloid are similar in their Ca, P and F mass fractions (Table 1), with only minor differences for mass fractions of Fe and S.However, larger differences are noted for their dentine parts (Table 1).The most notable is in the "unmeasurable portion" that is high in the modern dentine (31.6 AE 10%), lower in the fossil dentine tissues (7 AE 1.9%), and lowest in the enameloid parts (fossil: 1.8 AE 0.8%; modern: 1.3 AE 1%).The F mass fraction is also similar in the enameloid of modern and fossil teeth (about 3.8%), while it is lower in the dentine parts (2.8 AE 0.1% in fossil and 0.2 AE 0.2% only in the modern parts; Table 1).No significant difference is measurable for the F content of the PBE, which has its c-axis distinctly oriented parallel to the EPMA beam in the transversal sections (OD2 and GW4), but perpendicular to the beam in the longitudinal sections (OD1 and GW3).However, for the samples transversally cross-sectioned (OD2, GW4), the F mass fraction is higher in the PBE than in the TBE (3.9 AE 0.2 vs. 3.3 AE 0.3%, Table S2).Not surprisingly, minor elements in bioapatite such as Na, Mg, Sr, Fe and S, have higher mass fractions in the dentine than in the enameloid (Table 1), with the fossilised dentine being enriched in some particular elements.For example, the Mg mass fraction of the dentine in modern specimens is the highest amongst all samples measured in this study (1 AE 0.2%), while Fe mass fraction can only be found in trace amounts except for the dentine of fossil shark teeth (1.7 AE 0.8%).Similarly to the latter, the S mass fraction in the fossil dentine is the highest among the bioapatites (2.7 AE 1.4%).
Zoning of major and trace elements: Qualitative chemical maps were produced of the same specimens presented above (Figures 4, 5), and for additional specimens shown in File S4.The chemical map for the sample TM19 (TM mid longidutinal ) is only presented in the supporting information (File S4) since it had a indium surface contamination that could not be removed prior to chemical mapping.The TM low longitudinal conodont (TM18) shows slight compositional gradients in its elemental distribution (Figure 4b), with detectable zoning in both Ca, P and F contents, which have higher mass fractions in the anterior and posterior parts of the conodont, decreasing towards the platform structures.For Mg and S, and to some degree Na, this zoning is reversed with higher mass fractions at the platform structures, decreasing towards the inner parts (Figure 4b).Comparison with individual quantitative data points corroborates the chemical heterogeneity (Table S2).The TM mid longidutinal conodont (TM20) has a less exposed area and is more homogeneous in elemental distribution (Figure 4b), with only a minor change in the Mg distribution, increasing towards the platform portions.The TM transversal sample also has a heterogeneous elemental distribution (TM21) (Figure 4b), also corroborating the quantitative measurements (Tables 1, S2).Calcium and F distributions are homogeneous in the upper part of the conodont until the middle part/growth centre, but in the lower part there is a decrease in the mass fraction of these elements.Again, Mg, S, and Na chemical distributions compensate for Ca and F and the former have higher mass fractions at the lower portion of the crown.
For the shark teeth the enameloid has higher Ca, P and F contents with lower Mg and S mass fractions than the dentine (Figure 5b), while the SCE has higher mass fraction of Mg, comparable in amount to the dentine part (transversal sectioned specimens in Figure 5).In the BCE unit, the PBE unit has slightly higher Ca and F mass fractions and lower Mg and Na mass fractions than the TBE unit.The dentine chemical compositions and distributions vary between modern and fossil specimens, but in the latter group, Fe and S contents are higher (Figure 5b).

Carbonate content
The carbonate content measured for four shark teeth has a range from 1 to 6% m/m (Table S3).Shark enameloid has lower carbonate contents than the dentine.In the enameloid, the results were 0.6% m/m (n = 2) for the modern shark and 1.0% m/m (n = 2) for the fossil shark.In the dentine, the values were 2.3 AE 0.6% m/m (n = 4) and 5.2 AE 0.9% m/m (n = 2) for the modern and fossil shark, respectively.

Oxygen isotope measurements of phosphate
The d 18 O values of seventy-two bioapatites measured by HTR as well as SIMS have a total range from 10.8‰ to 23.6‰ (Tables 2, 3, File S5).Fifty-two analyses are from conodonts with a range from 10.8‰ to 17.4‰, and twenty are from shark teeth with a range from 10.8‰ to 23.6‰.
The d 18 O PO4 values from six shark teeth (four modern and two fossil) range between 21 and 22.7‰ (Table 3, Figure 6).Two modern teeth had only their enameloid analysed, while the other four had their enameloid and dentine analysed (Table 3).All enameloid structures were homogenized and the d 18 O PO4 values should represent their mean.Samples with both tissues analysed have within the measurement uncertainty the same d 18 O PO4 values, with the exception of one fossil shark tooth where the enameloid has somewhat higher values compared with dentine (OD2en: 21.8 AE 0.2‰ n = 2 vs. OD2de: 21.0 AE 0.0‰, n = 2).

Secondary ion mass spectrometry (SIMS):
The d 18 O values from samples analysed by SIMS (n = 60) have a range between 10.8‰ and 23.6‰ (Table 2).Fifty conodonts have d 18 O values in a range from 10.8‰ to 17.4‰ with the 1s for most of them being of 0.3‰ (n = 26), and specimens with a variation [ 0.3‰ d 18 O (n = 22) are not necessarily samples where more analysis spots were done (cf.Wheeley et al. 2012, Figure 7, Table 2).Pre-treated conodonts (TM1-TM6) had d 18 O values similar to those of non-pre-treated samples of the same group (Table 2).Statistical tests (e.g., Student's t-test, One-way ANOVA, Tukey's pairwise) were performed in the datasets using the PAST3 software (Hammer et al. 2001, Table S4).Among the tests, the one-way ANOVA was particularly important as it compares means of three or more groups that are mutually independent and satisfy the normality and equal variance assumptions (cf., Kim 2017).The three groups of Timor conodonts from different cuts have significant differences in their d 18 O values (ANOVA: F = 5.07, p = 0.01183, Table S4).For samples where the hyaline crown is mixed with the albid crown, variations in d 18 O values are larger (TM transversal : 16 AE 1‰, n = 13; TMmid longitudinal: 15.7 AE 1.9‰, n = 11) compared with only having hyaline crown analysed (TM low longitudinal : 17.1 AE 0.2‰, n = 13).Furthermore, the d 18 O values from the hyaline crown are within 1s the same as the bulk samples from Timor analysed by HTR (17.3 AE 0.4‰, n = 7).
Hyaline crown samples of Sc. ex gr.milleri from USA have a mean d 18 O value of 15.4 AE 0.3‰ (n = 6), higher than the mean d 18 O value of Scythogondolella from Spitsbergen (14.4 AE 0.1‰, n = 7) (Table 2, Figure 7).For the TM low longitudinal results, statistical tests including one-way ANOVA and Tukey's pairwise for the three datasets of Timor, USA and Spitsbergen confirm that this same species has different mean d 18 O values in the three localities (Table S4).
Ten shark teeth (six recent and four fossil) analysed by SIMS have d 18 O values within a range from 10.8‰ to 23.6‰, and most of the individual enameloid d 18 O values are higher than dentine (Table 2, Figures 6, S1).Averaging the results of the enameloid structures (i.e., SCE + PBE + TBE to total enameloid), the recent shark d 18 O values are between 10.8‰ and 22.9‰, and fossil shark d 18 O values are between 20.2‰ and 23.6‰.Among the enameloid layers, the edges of the SCE are shown separately (SCED, Figures 2, 6), in order to discuss the influence of a higher surface sample area for this component.The d 18 O values between enameloid zones in the same tooth can differ by up to 1.5‰, while the single tissue d 18 O differences are lower (up to 1‰ only in SCED of OD1 in Figures 6, S1).The mean d 18 O values of each enameloid component (SCED, SCE, PBE, TBE) were individually compared with the d 18 O PO4 values for the same tooth analysed by HTR (Figure 6c).d 18 O values from PBE, SCED and TBE are the same as measured for HTR, while values from the SCE may show some differences.All shark dentine results reproduced within AE0.3‰ (1s), with the exception of one sample of a recent tooth (GW4de, Table 2, Figure 6c), and d 18 O values from modern specimens are substantially lower (about 10‰) compared with those measured by HTR (GW3de: 10.7 AE 0.2‰ and GW4de: 10.9 AE 0.8‰ lower for the SIMS analyses).To examine the contribution of the carbonate group or also the organic matter to the total d 18 O values, mass balance calculations were made (Table 4), assuming their proportions within dental biogenic apatites to be a maximum of 10% (% m/m).To the carbonate group, an 18 O enrichment of carbonate compared with phosphate of between +7 to +9‰ (Δ carbonate-phosphate ; Iacumin et al. 1996, Vennemann et al. 2001, L ecuyer et al. 2010, Aguilera et al. 2017) is hypothesised.For organic matter, the estimates are based on collagen d 18 O values measured from herbivorous mammals (Kirsanow et al. 2008, Drewicz et al. 2020), as shark collagen has

Discussion
Major element compositions and mineralogy Calcium and P are the major elements of all bioapatites analysed, making up about 95% m/m, with the exception of 1 6 2 the dentine of the recent shark tooth where the unmeasurable fraction makes up about 32% m/m (Table 1).The major and minor element mass fractions measured are all similar to those previously reported for bioapatite (Elliot 2002, Skinner and Jahren 2007, Katvala and Henderson 2012, Enax et al. 2012, Kocsis et al. 2015, Zhang et al. 2017).The F mass fractions in recent and fossil shark tooth enameloid as well as in the conodonts are all similar at close to 3.8% m/m, which is within the measurement uncertainty of the ideal stoichiometric proportion of F in a pure fluorapatite Ca 5 (PO 4 ) 3 F (Table 1).
During the fossilisation process of bioapatite, the pore spaces ("unmeasurable portions" in Table 1) left after decomposition of the organic matter are commonly filled by secondary phosphate or other secondary minerals, allowing for the introduction of new elements but also for cation replacements on newly formed or recrystallised biogenic phosphate (e.g., Kohn andCerling 2002, T€ utken andVennemann 2011).Common cation replacements to the Ca site are Mg, Na, Sr, Ba, K, Rare Earth Elements (REE) and even U (Kohn and Cerling 2002, Skinner and Jahren 2007, T€ utken and Vennemann 2011, Hughes and Rakovan 2015).Magnesium and Na mass fractions have a contrasting distribution compared with Ca mass fraction in all studied bioapatites, with higher values in those parts that have or have had higher unmeasurable portions (Table 1, Figures 4, 5), confirming previous findings that these elements concentrate in newly formed, secondary phosphate (Katvala andHenderson 2012, Enax et al. 2014).Sulfur distribution is correlated with that of Mg and Na, and the three additionally contrast with P and F contents, again with marginally higher mass fractions in tissues with higher unmeasurable portions.Using the data from quantitative analyses, the ionic formula for the biogenic phosphate analysed was determined (carbonate vs. hydroxyl vs. fluor -apatite) (Deer et al. 1992).Conodonts had a calculated formula of Ca 4.7 Na 0.17 (CO 3 ) 0.22 (PO 4 ) 2.88 F 1.05 (Table S2), similar to shark enameloid, hence both can be classified as carbonate-fluorapatite.

Shark teeth chemical and oxygen isotope compositions
Chemical compositions of tissues analysed: In the enameloid, the unmeasurable portion from fossil and modern specimens (about 1.5%) are similar to the carbonate content of about 1% m/m for both samples (Table S3), typical for enameloid (Table 1) (Vennemann et al. 2001, Enax et al. 2012, Kocsis et al. 2014, Aguilera et al. 2017).As expected, the dentine in the fossil shark teeth has a much lower unmeasurable portion (7 AE 1.9%, Table 1) compared with the modern shark teeth (31 AE 10%), in part because of the recrystallisation and filling of pore spaces during fossilisation (Figure 2b).However, the high unmeasurable mass fraction in the modern specimens does not represent only the collagen portion in the dentine structure but is also related to poor polishing for such porous material, notably for the material where a pre-treatment has removed much of the organic matrix in recent teeth (Figure 2c).Calcium and F have qualitatively higher mass fractions in the outer layers of the enameloid (SCE and PBE), compared with the TBE and the dentine (Figure 5).Teeth transversally cross-sectioned have differences between the SCE and PBE, notably for the F and Mg distributions.Similar patterns were measured by qualitative energy dispersive X-ray (EDX) maps in modern Isurus oxyrinchus (Enax et al. 2014), another lamniform shark.While the heterogeneity between the SCE and PBE are less noteworthy in our specimens compared with those in Enax et al. (2014), more details on structurally different tissues can be indicated when mapping the chemical compositions of the tissues.The chemical variations within the enameloid of the modern species can be explained by the microstructural differences between the components (i.e., SCE and TBE vs. PBE).Based on observations of the enameloid of modern Carcharias taurus (also a lamniform taxon, Wilmers et al. 2021), as well as of I. oxyrinchus  (Enax et al. 2014, Wilmers et al. 2021), the fluorapatite crystallite bundles forming the SCE and TBE are randomly oriented within the structures, whereas the PBE has its crystallite bundles arranged in parallel to the crown surface (e.g., Figure 6 in Wilmers et al. 2021).The entangled aspect of the SCE and TBE crystallite bundles may create pore spaces of about 1 lm, which are not seen in the single-oriented densely packed bundles of the PBE (e.g., Wilmers et al. 2021).These characteristics favour diagenetic or surficial in vivo remineralisation processes in the TBE and SCE, or even while the structures are precipitated originally in the tooth bed (Enax et al. 2014).Quantitative analyses from the OD2 and GW4 also give lower F mass fractions for the TBE compared with the PBE.This may explain why the fossil enameloid has comparable elemental distributions and mass fractions to the modern teeth, and thus may be expected to be less influenced during diagenesis and fossilisation (e.g., Trueman andTuross 2002, Kohn 2008).
Iron and S were found in higher mass fractions in fossilised shark tooth "dentine" tissues compared with all other bioapatites (Table 1).Iron is an element usually incorporated into the bioapatite structure during diagenesis and, in its common association with S, suggests the secondary mineralisation of pyrite (FeS 2 ), a mineral frequently formed in sedimentary rocks and biominerals rich in organic matter (Raiswell 1997, Van Dover 2007).The large spatial enrichment of these two elements in the dentine of our fossil shark teeth (about 50 to 100 lm wide) supports the presence of secondary pyrite (Table 1, Figure 5).Such zones of secondary mineral enrichment are clearly to be avoided for SIMS d 18 O measurements.

Oxygen isotope compositions of tissues analysed:
The oxygen isotope composition in all shark bioapatite tissues analysed by HTR reproduced to within 0.2‰ (Table 3).For any one tooth, the different tissue types also have the same oxygen isotope compositions.This contrasts with the d 18 O values measured using the SIMS (Table 2, Figure 6).While the dentine is known to be more prone to alteration and hence often excluded from analyses, this tissue, notably that of the recent teeth was analysed by SIMS to test for matrix and textural effects.In the modern shark, the dentine had about 10‰ lower d 18 O values compared with the enameloid of the same tooth when measured by the SIMS but had identical d 18 O values to enameloid for the HTR method.Relative to carbonates, the phosphate values are likely to be 7 to 9‰ (Δ carbonate-phosphate between +7 to +9‰; Iacumin et al. 1996, Vennemann et al. 2001, L ecuyer et al. 2010, Aguilera et al. 2017).Compared with organic matter (i.e., collagen), a Δ collagen-phosphate value of -5‰ is used (Kirsanow et al. 2008, Drewicz et al. 2020).
The low d 18 O value measured for the modern shark tooth using SIMS may, in part, be related to the presence of organic matter (e.g., Drewicz et al. 2020).While there are no collagen d 18 O values reported for sharks in the literature, the collagen d 18 O values of herbivore mammals are about 5‰ lower compared with that for enamel d 18 O PO4 values (Kirsanow et al. 2008, Drewicz et al. 2020).Assuming the same degree of isotopic 18 O depletion (i.e., Δ collagen-phosphate ) for shark tooth collagen, and assuming that the complete unmeasurable portion represents collagen only, such a contribution would decrease the d 18 O values by only 1.5‰ (Table 4).A speculative 18 O-depletion of 20‰ per weight percent for the collagen, hence four times the Δ collagen-phosphate , would lower the values by 7‰, again not enough to justify the low d 18 O of dentine measured here.Contributions of sputtered oxygen from organic matter are possible, but it is also unlikely that the teeth contain 30% m/m of organic remnants after the pre-treatment procedure (boiling salt water) and hence this contribution will likely be small.In a similar comparison of different biogenic apatite tissues from sandbar shark teeth (Carcharhinus plumbeus), Zigait _ e and Whitehouse (2014) noted about 2 to 3‰ lower d 18 O values compared with the enameloid.While this is a smaller difference compared with our measurements, other factors would need to be considered to explain the more than 10‰ lower SIMS d 18 O values of recent dentine compared with their HTR values (Figure 6c).
Another possibility to explain this discrepancy for this large difference in d 18 O enameloid and d 18 O dentine for the recent shark teeth may be related to sample texture, that is problems with irregular sample heights (pits) and/or tilted sample surfaces that may both be created during the complex polishing procedures (e.g., Kita et al. 2009).We found that our even mild pre-treatment of the larger great white shark tooth resulted in an irregular sample surface that was difficult to polish prior to SIMS analyses (Figure 2c).This was not noted for the smaller sample chips or the conodonts that were easier to polish leaving a flat surface without irregular heights nor tilted sample surfaces.Kita et al. (2009) noted that if samples have topographic differences of between 10 to 40 lm or more, a bias for in situ measurements of up to +4‰ may be obtained.Differences in sample height caused by pits that may have rather flat bottoms are expected to be less relevant as such height differences may be corrected for by refocussing the sample during a SIMS analysis.Similar problems were also noted for biogenic carbonates, which may have fine-grained and porous textures (Orland et al. 2015).SIMS d 18 O values measured on otoliths with high porosity were about 2‰ lower than values measured by conventional isotope ratio mass spectrometry (Helser et al. 2018).In ammonite shells, a depletion of 1‰ in the d 18 O values was noted for analyses near cracked portions of the biomineral surface (Linzmeier et al. 2020).

The tooth tip prepared by
Zigait _ e and Whitehouse (2014) was smaller than the great white shark tooth tip used in our study.Their sample also had a surface topography of less than 1 lm, including the dentine parts, and the surface was flat with no visible irregular heights.As their d 18 O offset is also lower compared with ours, this offset may indeed be related to the organic matter content and/or carbonate content of the dentine compared with much lower contents thereof in enameloid.In contrast, our higher offset of more than 10‰ may be attributed to a combination of a contribution of sputtered oxygen from the unmeasurable portion of the dentine, in addition to an irregularly tilted surface texture that results from the small nano-meter size crystallites in modern dentine during polishing.Hence, this is an analytical artefact of the SIMS measurement.
The individual SIMS d 18 O values measured for the enameloid structures of the shark teeth (Figure 6) were also different.The larger d 18 O variations in the fossil tooth may be the result of secondary alteration processes, as also suggested by the higher mass fractions of Fe and S in these tissues (Figure 6b, c).For the modern teeth, the individual d 18 O values for the different tissues have a variation of AE 1.3‰ (Table 2, Figure 6c), similar to the inter-tissue d 18 O heterogeneity within the enameloid tissues noted for C. plumbeus teeth by Zigait _ e and Whitehouse (2014).With the exception of the GW1, the modern enameloid d 18 O values are marginally enriched in 18 O compared with the mean d 18 O PO4 measured through HTR by about 0.5 to 1‰ (Figure 6c).Comparable results were obtained for most of the SIMS investigations using bioapatites (e.g., Rigo et al. 2012, Trotter et al. 2015, 2016a, Goudemand et al. 2019), and the d 18 O values had to be corrected in order to interpret values for past environmental conditions.
An enrichment in 18 O measured using SIMS analysis and compared with the classical silver phosphate preparation could potentially be related to a contribution of carbonate-ion sourced oxygen, as this is commonly noted to be enriched in 18 O compared with the phosphate group by about 9‰ (e.g., Iacumin et al. 1996, Vennemann et al. 2001, L ecuyer et al. 2010, Aguilera et al. 2017, Sisma-Ventura et al. 2019).Given the low unmeasurable content of the shark enameloid (\ 3% m/m), mass balance estimates of the carbonate contribution to these analysed tissues would, however, be limited to about 0.3‰ for the d 18 O values (Table 4).This suggests that for the majority of our fluorapatite dataset, contributions of oxygen ions from other oxygen-bearing sites (CO 3 2-, SO 4 2-, OH -) that can bias the d 18 O values measured by SIMS are limited and hence unlikely to be related to the carbonate content only.By analogy to the study by Zigait _ e and Whitehouse (2014), it is likely that the enameloid of the PBE is the best to analyse as it has simple crystallite orientations, restricted pore spaces due to its arrangement and, as for all enameloid layers, lesser matrix interferences due to carbonate content or unmeasurable portions.Alternatively, this may well be an ontogenetic effect of the different enameloid tissues that can indeed only be detected with SIMS measurements.Analysis of more samples is required to address any patterns of specific enameloid structures that may cause IMF.Nonetheless, enameloid layers that have contrasting mass fractions of Mg and F in their structures (SCE and TBE,Figures 5,7b), had a good d 18 O reproducibility for most samples, with some of their results overlapping with the 'reference' HTR d 18 O PO4 values (Figure 6c).Hence chemical matrix effects may be excluded.

Conodont chemical and oxygen isotope compositions
Chemical compositions of tissues analysed: For conodonts, the total unmeasurable mass fraction is low (mean of 1.9 AE 1.3%) but was higher for the lower part of the TM transversal specimen (5.2 AE 1.2%).Among the conodont biomineralogical tissues, the albid tissue is known to have a lower content of CO 3 2-compared with the hyaline tissue (Trotter and Eggins 2006).A lamellar pattern characteristic of hyaline tissue is recognised for most of the crown in the TM transversal (Figure 3a), however, the unmeasurable content (including CO 3 2- ) was significantly higher only at the lowermost part of the crown (Table 1).This part is also separated by a "dark" contour and darker lower zone that can be seen in specimens analysed by SEM (Figures 1b, c,  3a).The Mg, S and Na mass fractions are also higher in this lower part of the conodont, in parallel to the unmeasurable portion (Table 1, Figure 4), and at the cost of the Ca, P, and F mass fractions.The crown has a similar pattern of chemical distributions to the enameloid, but the lowest part of the crown is comparable to the dentine of shark teeth (Table 1, Figures 4, 5).These observations are similar to those reported by Katvala and Henderson (2012), but this is the first time that the chemical distribution of segmiplanate conodont P 1 elements has been analysed.One could expect such chemical pattern at the posterior extremity of a conodont dental element, because this is where the basal body is originally located.However, one specimen with its anterior extremity cross-sectioned transversally shows a similar pattern to that of the TM transversal result (Figure 1c).
We analysed additional conodonts to test if this chemical distribution pattern is exclusive for our Sc.ex gr.milleri specimens from Timor, or if this pattern could still be detected in samples subjected to a relatively high temperature of fossilisation (300 °C, CAI 5) (specimens: WMJ1, WMJ2, GR1 in Table S1, File S4).Interestingly, a specimen of Borinella sp. from the early Spathian of Oman (WMJ1) also had a similar pattern at the lower part of the crown.The same can be said for an element of Sc. ex gr.milleri conodont from Kashmir (Guryul Ravine, Leu et al. 2019) with a CAI 5 (GR1), but the chemical zonation is less pronounced when compared with conodonts with a lower CAI index and interpreted to be well preserved.
A pristine or secondary elemental 'imprint' in the conodont bioapatite?: Major (Ca, P, F) and trace elements (Mg, S, Na) as well as the unmeasurable portion have complementary mass fractions within the crown of the conodonts.Distinct chemical mass fractions between albid and hyaline crown were expected as the latter tissue was reported to have a higher carbonate and trace elemental content (e.g., Trotter andEggins 2006, Trotter et al. 2016b).However, the relative enrichment for Mg, S, and Na mass fractions and to the unmeasurable portion was only noted for the reticulated, "darker" areas situated at the lowermost part of the crown (Figures 3a, 4b, Table 1).Sodium as well as Mg may be used to interpret if these trace elements were introduced during bioapatite diagenesis (e.g., Kohn et al. 1999, Dauphin andWilliams 2004), while the S can be indicative of soft tissue residue (e.g., Terrill et al. 2018).
One suggestion that may explain this chemically distinct zone is that the lower part of the conodont crown was covered by soft tissue at least for a period of the animal's life, thus creating the chemical heterogeneity noted.Such hypothesis was first based on microstructural studies that described regular and irregular reticulate surfaces on some dental elements, thought to be a reflection of the cell pattern of the tissue that covered the crown (von Bitter and Norby 1994, Donoghue 1998, Donoghue and Purnell 1999, Turner et al. 2010).However, recent geochemical investigations using Palaeozoic taxa have further corroborated this characteristic, as they also measured peculiar chemical compositions at the lowest part of the conodont crown (Katvala andHenderson 2012, Terrill et al. 2018).More studies utilising a higher number of samples are necessary to fully understand the nature of the various dental morphologies that these animals present between different taxa.

Oxygen isotope compositions of tissues analysed:
Segmiplanate P 1 elements of Sc. ex gr.milleri from Early Triassic strata of Timor were of sufficient abundance to allow us to prepare sample sets in order to examine different analytical protocols.The d 18 O values for the TM low longitudinal , representing largely the hyaline crown (Figure 3b), are the most homogeneous tissues analysed (17.1 AE 0.2‰, n = 13, Figure 7).Their mean d 18 O value is also within the measurement uncertainty the same as the reference value for Timor conodonts analysed by HTR (17.3 AE 0.4‰, n = 7).In contrast, the TM transversal samples that may include denticles and/or albid crown tissues (Figure 3a), have a higher d 18 O variance with also lower mean values (16 AE 1‰, n = 13, Figure 7).The conodonts from the TM mid longitudinal group, polished at a higher level and measured towards the interior of the dental elements (Figure 3c) also have a higher variance and lower d 18 O values (15.7 AE 1.9‰, n = 11, Figures 7, 8).The lower d 18 O values and higher variance in the albid tissue and in the interiors of conodonts is in line with previous SIMS measurements of conodonts (Wheeley et al. 2012, Zhang et al. 2017).The similarity of the TM mid longitudinal with the TM transversal d 18 O values, as well as the quantitative element analyses and SEM measurements for these parts suggest that these sections represent dominantly albid tissue in both conodont groups.For samples of Scythogondolella from USA and Spitsbergen, prepared also for a low longitudinal section, the measurements of d 18 O values also reproduced well with a low variance (1s of 0.3‰ or better; Figure 7), even though the absolute values are quite different and likely related to different palaeoenvironmental conditions.
Despite the TM transversal and the TM mid longitudinal groups being interpreted to have dominantly albid tissues (Figure 8a, b), some of the d 18 O values measured still overlap with the HTR reference value for these "bulk sample" conodonts.Previous studies attributed the low d 18 O measured in the albid crown and in the interiors of conodonts to a higher susceptibility of these tissues to diagenetic alteration (e.g., Wheeley et al. 2012, Zhang et al. 2017).However, if these tissues were more susceptible and their values measured are indeed lower due to a diagenetic overprint, one would expect the d 18 O values measured via HTR analysis with about 60 to 100 conodonts, to give mean values between those of the albid and the hyaline crown.For the specimens studied here, the albid tissue makes up an estimated 20 to 50% of the conodont crown and is mainly concentrated in the denticles (e.g., Figures 1, 3a, 8a, b, Donoghue 1998).If their d 18 O values are indeed systematically lower compared with the hyaline parts due to a diagenetic or even ontogenetic reason, the bulk HTR values would be expected to average at values between the albid and hyaline SIMS values.In contrast though, the HTR d 18 O values are rather similar to the d 18 O values measured in the hyaline tissue and/or for many hyaline/albid mixed tissues in the crown (i.e., TM low longitudinal , Figure 7).
In view of the above discussion for the shark teeth that illustrates a clear sensitivity of the SIMS O isotope measurements to sample topography, one possible factor that may best explain the higher variance and lower mean measured values for albid compared with hyaline tissues is a slight microstructural difference between these tissues.Detailed microstructural investigations of the albid crown have characterised these tissues as relatively porous, with pore spaces of hundreds of nanometres (about 160 nm) being dominant (Atakul-€ Ozdemir et al. 2021).In the specimens imaged here, estimates of the pore sizes of albid tissue are about 1 lm in size (Figures 1, 3a).The hyaline crown, despite its lamellar structure, has intracrystalline pores with a range of up to tens of nanometres only (Kemp 2002, Trotter et al. 2007).Such a difference in pore size relative to the analytical beam size of 15 lm used for the SIMS measurement may perhaps be sufficient to create irregular surfaces during secondary ion ablation.Depending on the exact density of the pores relative to the beam size used, this may also explain why in many cases the albid tissue or at least the albid/hyaline mixed tissues still give individual analysis similar to the hyaline tissues (e.g., Rigo et al. 2012, Trotter et al. 2008, 2015, Narkiewicz et al. 2017, Wheeley et al. 2018) and to the bulk HTR values, while other studies do not have similar d 18 O values for these two tissues neither (Wheeley et al. 2012, Zhang et al. 2017).
Given the low unmeasurable content of all the conodont tissues, by analogy to the studies on enameloid of the shark teeth (\ 3% m/m), mass balance estimates of the carbonate contribution to these tissues would be limited to less than 0.3‰ for the d 18 O values (Table 4).Marginally higher 18 O enrichments would be possible on in situ measurements at the lower part of the conodont crown only (+0.5‰).This also suggests that for most of our dataset, contributions of oxygen ions from other oxygen-bearing sites (CO 3 2-, SO 4 2-, OH -) that can bias the d 18 O values measured by SIMS are also limited (e.g., L ecuyer et al. 2010, Wheeley et al. 2012, Zhang et al. 2017).Also, oxygen isotope contributions from organic remnants still present in the conodonts is considered minimal as conodonts are not known to preserve and neither contain larger amounts of organic matter in the first place (e.g., Wheeley et al. 2012), but well-preserved dental elements (i.e., CAI 1) could contain epithelial tissue within conodont tissues (Donoghue 1998).Considering the quantitative estimate of only 2.6 AE 1% m/m of the unmeasurable portion in TM low longitudinal that is hyaline tissue (Table 1), mass balance calculations similar to those for the shark teeth above, would imply a depletion of 18 O of about 0.15‰ only, perhaps accounting for the small but consistent offset between the SIMS d 18 O (17.1‰) and the HTR d 18 O PO4 (17.3‰) means.
The elemental mass fractions and distributions measured in conodonts are similar in their tendency to those noted for the fossil shark enameloid vs. dentine: higher substituting cations and unmeasurable portions in the more porous and/or dentine parts of shark teeth, as well as in the lowest crown parts, albeit with much lower portions of unmeasurable elements in the conodonts.In addition, all conodont tissues, as well as the enameloid from shark teeth, have nearly stoichiometric Ca, P, and F mass fractions for fluorapatite in the enameloid or upper crown parts.In parallel to the shark teeth, those parts with the higher unmeasurable contents also have the highest variance for their d 18 O values and may also have lower mean d 18 O values, although this difference in d 18 O is by far not as pronounced as that for the shark teeth dentine.In addition, parts of the conodonts with marginally higher substituting trace element content and also unmeasurable portions, may still have analytical SIMS points with d 18 O values similar to those tissues of stoichiometric fluorapatite composition and lower porosity such as represented by the hyaline crown and some albid tissues too (Figure 8a, b).Finally, the analytically most homogeneous hyaline tissue of the TM low longitudinal zones that also have mean d 18 O values (17.1‰,AE 0.2‰) identical to the bulk HTR values (17.3 AE 0.3‰) may also show some systematic differences in the elemental distributions of Ca, P, F, Mg, Na, and S (e.g., Figure 8c.1, d).Also, in situ analyses in the interiors of conodonts on surfaces similar to the shark enameloid, but with near-stoichiometric Ca, P and F mass fractions and low Mg, Na and S mass fractions (i.e., TM mid longitudinal , Table 1, Figure 4) also have larger isotopic variations (Figures 7, 8c.2, e) and lower mean d 18 O values (15.7 AE 1.9‰) compared with the HTR values.Collectively this suggests that the chemical variations of well-preserved conodonts do not cause significant IMF for the SIMS O-isotope measurements.Instead, the lower and more scattered d 18 O values may be related to an irregular texture of the conodonts, as most of the low d 18 O was measured in sample sets that included albid tissue with a somewhat larger porosity (i.e., for the TM transversal and TM mid longitudinal groups) compared with hyaline tissue in the analysis.Any textural irregularities inherent to the crystallite size and orientation in the biogenic apatite of the conodonts may also be enhanced by subsequent sample preparation.Hence, care has to be taken with the sample preparation and the choice of the final biogenic apatite tissue to be analysed.
In summary, by analogy to previous suggestions (cf., Wheeley et al. 2012, Zhang et al. 2017), analysis of the albid tissue for some conodonts may be prone to give lower and more variable d 18 O values for in situ measurements using SIMS.Just as for the recent shark dentine d 18 O values, we attribute the biased d 18 O values measured in conodonts to irregular sample textures (for example a tilted analytical surface; Kita et al. 2009) for SIMS analysis.By analogy to the distinct shark teeth enameloid tissues (SCE and PBE, see above), the hyaline crown appears to be the most reliable conodont tissue for in situ SIMS O isotope measurements and longitudinal cross sections of this tissue can be readily prepared for microanalyses techniques.Sample texture is also a more relevant factor for reliable SIMS O-isotope measurements than specific tissue choices, hence ideally all analytical spots, whether in albid or hyaline crown portions should ideally be verified with subsequent SEM imaging to identify the exact analysed tissue, as well as an optical profilometer (cf.Kita et al. 2009) to confirm the final analytical texture.The relatively small variations in chemical compositions for the conodont tissues do not, however, cause IMF for the SIMS O-isotope measurements (cf.Wheeley et al. 2012, Zhang et al. 2017).Ideally also, one should concentrate the analysis spots in one area to avoid analyses of distinct growth zones and texturally different zones within the conodont tissue.

A preliminary in-house SIMS reference material
A principal objective of this research was to refine the SIMS measurements and its usefulness for different tissues in bioapatite, as well as to define a possible bioapatite RM to be used for future SIMS measurements, particularly of conodonts.The Durango apatite is compositionally and structurally different compared with fossil bioapatite (Table S2), hence it is more appropriate to use a bioapatite RM, or at least to combine with the Durango RM for SIMS measurements and the data reduction.The selected bioapatite needs to be homogeneous in its oxygen isotope composition, which can be difficult if different growth lines or structures of the studied biomineral are analysed during a SIMS session (beam diameter: 15 lm).However, the bulk analysis by HTR done in parallel provides a good basis for studying the possible variations in d 18 O by SIMS (Tables 2 and 3).
For the recent shark teeth, the variation in measured d 18 O values is clearly related to the type of tissue analysed with the SIMS, as no such differences were noted for the HTR measurements on these different tissue types (dentine and enameloid).The d 18 O differences for the dentine between HTR and SIMS are likely to be related to the irregular sample texture analysed in situ, but minor contributions of sputtered oxygen released from the organic matrix, if present, cannot be excluded.The relatively high Fe and S mass fractions in fossil specimens, suggestive of diagenetic alteration, can be used to justify the intra-and inter-tissue variations in d 18 O values measured as either primary or secondary in nature.The variation in structure, texture and density of the shark teeth render them more complex than previously thought and not ideal as bioapatite RMs, notably if used in crosssection to the structurally different tissue types.To minimise such analytical artefacts that can ultimately cause bias for in situ microanalysis techniques, we recommend consulting a recent contribution by Shirley et al. (2020).It summarises common setbacks of phosphatic microfossils' sample preparation and the different solutions to solve them, which vary with the nature of the studied material.Nevertheless, given the larger size of shark teeth and the ease of exposing and hence polishing only the outer enameloid structure, there is still potential to use them as in-house RM material, or for an individual study where no alternative bioapatite RMs are available.Sharks, for which their teeth were formed under controlled environmental conditions have potential as the chemical composition of the water can be controlled (e.g., Zigait _ e and Whitehouse 2014), though research in animals under stress conditions in captivity reported disrupted bone formation (e.g., Yirmiya et al. 2006).
For the conodonts, sample spots should focus on a single tissue, and while the albid tissue can provide reliable d 18 O values, it is more prone to present irregular sample textures given the larger pore spaces compared with hyaline tissue.Based on the oxygen isotope compositions measured, the Sc.ex gr.milleri from Timor, now named 'TM-SM', was hence selected as a preliminary internal RM for oxygen isotope analysis of conodonts for the SwissSIMS laboratory at the UNIL (Table 5).About 100 specimens measuring around 300 to 600 lm are picked and ready to be prepared into epoxy resins, and additional specimens can still be recovered from the Timor sediments.A follow-up that could be done is to send about a dozen specimens mounted into epoxy resins to other institutions capable of SIMS analysis, to verify the absolute d 18 O values and the consistency of the d 18 O values.In addition, to continue testing the compositional homogeneity of the TM-SM, some individual specimens can be provided for additional geochemical investigations (e.g., studies of other stable isotopes).S4).The d 18 O mean of all sample splits used for both methods (SIMS and HTR) for the Sc.ex gr.milleri from Timor is 17.1 AE 0.3‰ (n = 20, Table 5), and this value will be considered for the TM-SM preliminary in-house RM.This conodont RM is thus a good, high d 18 O RM, which can be used together with the low d 18 O RM of the Durango Table 5. Summary of the oxygen isotope (HTR and SIMS) and EPMA measurements of the preliminary in-house reference material TM-SM fluorapatite.Both TM-SM and Durango may provide a suitable basis for scaling the instrumental fractionations of SIMS measurements of bioapatite.

Conclusions
A detailed geochemical and oxygen isotope study of conodonts and shark teeth provides new insights for SIMS O-isotope measurements of bioapatites.Systematic and/or more random variations in chemical compositions that may occur within the different bioapatite-mineralised hard tissues can be used to discriminate the importance of primary versus secondary mineralisation processes and hence help select parts of the biogenic tissue that are of interest for subsequent O-isotope measurements and palaeoenvironmental interpretations.However, no systematic relations are noted between relatively small chemical variations in conodont or enameloid shark tooth biogenic tissues and variations of their oxygen isotope compositions measured by SIMS.Conodont as well as shark tooth enameloid also have near-stoichiometric major element compositions of fluorapatite.The d 18 O variations measured by SIMS in such tissues, when compared with conventional HTR measurements of the same material, are likely to be analytical artefacts related to small differences in structures and final analytical textures of the phosphatic tissues.Adjacent tissues with different crystallite size, densities of crystallites or orientation, and with relatively large pore spaces (≥ 1 lm) are prone to amplifying textural differences during sample preparation and polishing.Evidence for this is given by a comparison of the d 18 O values measured by SIMS and the classical silver phosphate precipitation method, which also isolates the PO 4 3-ion from other potential O-bearing sources.A single biomineralogical layer that is readily prepared with a minimal final sample topography should be analysed.Outer fluorapatite enameloid layers from shark teeth or also fluorapatite of lower crown tissues from conodonts would both analytically, as well as in terms of minimisation of secondary mineralisation or remineralisation during diagenesis, be suitable tissues for SIMS analyses.While enameloid shark bioapatite is recommended to be used as a secondary reference material for in situ oxygen isotopic investigations in vertebrate dental tissues, the interpretation of its SIMS d 18 O values should be done with caution as the different "enameloid" layers can also have contrasting d 18 O values.For conodont bioapatite, the hyaline crown is considered to be the most reliable tissue for SIMS d 18 O measurements.Best results can be obtained from cross sectioning longitudinal cuts of the lower crown of the conodonts.These sections are relatively easy to be prepared and also avoid albid tissue commonly mineralised within the denticles.Based on the present results, a conodont sample set from Timor (TM-SM) was selected to be used as an internal RM for stable isotope analyses of bioapatite for the SwissSIMS laboratory at the University of Lausanne.Additional material of this sample could be made available for cross-laboratory tests and potential RM calibration.

Figure 1 .
Figure 1.(a) P 1 dental element orientation of Sc. ex gr.milleri, and specimens imaged by SEM in back-scattered electron (BSE) mode (SEM-HV: 20 kV): transversal cross sections of the posterior (b), and anterior extremities (c); longitudinal cross section of the lower crown (d), and vertical longitudinal cross sections of the conodonts for (e), (f) and (g) (lateral view).The encircled red dashed area is over more porous regions, which should indicate albid tissue presence.The blue dashed line delineates a reticulated dark contour at the lowest part of the conodont crown.

Figure 2 .
Figure 2. Schematic representation of the longitudinal and transversal cross sections prepared for modern (great white shark, Carcharodon carcharias) and fossil (Odontaspis denticulata) shark teeth.Shark biomineralised structures can be noted even in reflective light microscope images in longitudinally cross-sectioned teeth (a).SEM images in BSE mode (b), (c) of transversal cross sections have clearly shown such components: at the outer part the single crystallite enameloid (SCE); just below it the parallel bundled enameloid (PBE) can be identified, followed by the tangled bundled enameloid (TBE) surrounding the dentine.An irregular sample surface is noted in the dentine of the modern specimen (GW1).

Figure 3 .
Figure 3. Schematic representation of the sampling protocol (upper-left) employed for Sc.ex gr.milleri conodonts and scanning-electron microscopy images (right) in BSE mode for two of the three main groups classified in this research.Specimens were prepared for visualisation of longitudinal (TM18, TM19) and transversal (TM21) cross sections.A transversal cross section (TM transversal ) of the posterior extremity (a) reveals the presence of patches of albid crown/tissue (above the encircled red dashed area) in the denticle.The dark contour mentioned in the text is also indicated and remarked with a blue dashed line, starting in the ridges and extending laterally in the platforms, just over the growth axis.In the longitudinal cross section at the lower conodont crown (TM low longitudinal ), the lamellar hyaline crown is the predominant structure (b).The third group comprises specimens also cross-sectioned longitudinally (TM mid longitudinal ) and polished up to the interior of the conodont crown (c).
sections Chemical compositions and Conodont chemical and oxygen isotope compostion), we conclude that the re-polished samples also have a mixed aspect of albid and hyaline tissue in their surface.Thus, the three set of conodont samples from Timor mentioned above (section SEM, EPMA and SIMS) were classified as: dominantly hyaline tissue in a low longitudinal cross section -TM low longitudinal ; mixed tissue (i.e., hyaline and albid) in a middle longitudinal cross section -TM mid longitudinal ; and mixed tissue in a transversal cross section -TM transversal .SEM scans of shark teeth show the two types of textural/mineralogical units: the single crystallite enameloid (SCE, a.k.a.'shiny enameloid layer',Enault et al. 2015; or as 'ridge cutting edge layer',Wilmers et al. 2021) and the bundled crystallite enameloid (BCE,Enault et al. 2015) unit(Figure 2b, c, GW5 and MS1 in File S2

Figure 4 .
Figure 4. Qualitative EPMA chemical mapping of Sc. ex gr.milleri specimens showing the distribution of Ca, P, F, Mg, Na and S in the conodont crown.Scale bars represent the intensity of X-rays detected by the EPMA, correlated with the chemical distribution of each element in the sample.(a) The three conodont groups from Timor were evaluated by this technique and their respective results are presented below in (b).Significant heterogeneity can be observed in the lower part of the crown of TM transversal (TM21) (1).The TM mid longitudinal (TM20) is rather homogeneous (2), and SIMS analysis spots are scarcely visible in the F and Ca chemical maps (only sample analysed after a SIMS session).The TM low longitudinal (TM18) also presents a degree of heterogeneity on its chemical distribution (3), notable for the Mg.Scale bars: 50 lm.

Figure 5 .
Figure 5. Qualitative EPMA chemical mapping of modern (C.carcharias) and fossil (O.denticulata) shark teeth showing the distribution of Ca, P, F, Mg, Na and S. Scale bars represent the intensity of X-rays detected by the EPMA, correlated with the chemical distribution of each element in each sample.(a) Longitudinal (GW3, OD1) and transversal (GW4, OD2) cross sections were prepared from teeth tips where enameloid tissues (SCE, PBE and TBE) and dentine are observed.(b) Enameloid and dentine had expected variations in their elemental mass fractions and a relative chemical heterogeneity is observed between the PBE and TBE, layers that make up the bundled crystallite enameloid (BCE).Scale bars: 100 lm.

Figure 6 .
Figure 6.Modern and fossil shark bioapatite d 18 O values measured by SIMS and HTR (TC/EA).In (a), a schematic representation of the cross sections and where sample spots were sputtered in each enameloid layer and in the dentine (symbols) is shown.Chemical distributions maps discussed to explain some d 18 O variations noted in modern and fossil specimens are shown (b), and the qualitative scale for them is explained in Figure 5.For the modern shark tooth, we selected the F distribution map mainly to show the contrast between the TBE vs. PBE.For the fossil shark tooth, the Fe distribution map is given, which associated with the S suggests the presence of pyrite in the dentine.In (c), SIMS d 18 O values of each layer are presented individually by white/black symbols (see (a) for tissue types represented by the different symbols), while the red symbols refer to the d 18 O values obtained by HTR.All enameloid structures were homogenized for the HTR analysis.Profilometer topographic profiles are given in (d), and the irregular sample texture of the dentine can be better seen in the GW4 specimen.

Figure 7 .
Figure 7. Mean d 18 O values and standard deviations (1s) of Sc. ex.gr.milleri specimens analysed by SIMS.At the top left, (a) dashed arrows indicate the cross sections done on the three main conodont groups from Timor, and below examples of their respective qualitative maps for Mg are shown, with information regarding the predominant associated tissues.We selected the Mg for better visualisation of the relative chemical heterogeneity, but Ca, F, Na and S have similar distributions (colour and scale bars in Figure 4).Based on the more homogeneous values, conodonts from USA and Spitsbergen were prepared similarly to the sample set (3). Below the chemical maps, (b) symbols for each dataset are given.At the right, the red line and the marked area correspond respectively to the d 18 O PO4 mean value of Sc. ex.gr.milleri from Timor obtained by HTR (TC/EA) and its standard deviation (1s).Vector drawings of each sample and sample spot are illustrated at the side or close to their respective results, and blue circles represent individual d 18 O analysis.The sample order in Table 2 is plotted inversely upwards, hence, the lowest symbol in the figure is the first sample in the table (TM2).Differences in the absolute d 18 O values for USA and Spitsbergen are due to the contrasting depositional conditions between Timor (offshore) and the Panthalassa sites (inshore).

Figure 8 .
Figure 8. Example of conodonts from the three main sampled groups, with comparisons between their chemical patterns the sample spots analysed by SIMS.(a), (b) and (c) are SEM images in BSE mode (details in Figures 1, 3), while (d) and (e) are chemical distribution maps for Mg (details in Figure 4).Again, Mg was selected for better visualisation.Sample spots over the visible porosity of the albid tissue (encircled red dashed area) still reproduced within the HTR reference d 18 O value (a, b), as well as analysis on zones enriched chemically in minor components (c.1, d).Analyses of conodonts more chemically homogeneous have more negative and scattered d 18 O values (c.2, e).The chemical variations do not appear to relate with an IMF to SIMS measurements.The samples (c.2) and (e) have visible roughness over the crown surface due to a quick polishing done after the SIMS analysis were made.Scale bars: 50 lm.
Using the most reliable dataset (TM low longitudinal ), the d 18 O values measured by SIMS are statistically indistinguishable from the TM-HTR d 18 O PO4 values (t test: p [ 0.173, Table Mine et al. (2017)harks and two in the enameloid only, TableS1) to compare the results of the isolated phosphate component (d 18 O PO4 ) with the in situ d 18 O values containing all oxygen-bearing sites (SIMS).To obtain the d 18 O PO4 values by HTR, the phosphate group in apatite was separated via precipitation as silver phosphate, following a method adapted afterMine et al. (2017).For this method, about 1 to 1.5 mg of original bioapatite is required.Enameloid and dentine of shark teeth was sampled by abrasion of the crown surface using a micro-drill or by fracturing small pieces of a tooth using an agate mortar and pestle.Conodonts were sampled in bulk and several specimens (60 to 100) were necessary for each sample, but the results should be a representation of the crown (albid and hyaline tissues) since no basal body is preserved.All samples were pre-treated in 1 mol l Vennemann et al. (2002)te (pH = 4.5, 2 h) and rinsed several times inde-ionised water  (Koch et al. 1997).International RM (NIST SRM 120c) and in-house laboratory RM were prepared in parallel with each sequence of samples.Duplicates or triplicates of Ag 3 PO 4 were analysed for each sample on a TC/EA (hightemperature conversion elemental analyser) coupled to a Finnigan MAT 253 mass spectrometer according to the method described inVennemann et al. (2002), where silver phosphate is converted to CO at 1450 °C via reduction with graphite.Measurements were corrected to in-house (LK-2L: 12.1‰, LK-3L: 17.9‰) and international Ag 3 PO 4 phosphate RMs (USGS80: 12.5‰, USGS81: 34.7‰), which had AE0.3‰ (1s) during measurements (File S5).The NIST SRM 120c phosphorite reference material had a mean value of 21.8 AE 0.3‰ (n = 17).The isotope compositions are expressed in the d notation relative to Vienna Standard Mean Ocean Water (VSMOW).
. Indeed, typical standard deviation (1s) for Durango is 0.25 to 0.3‰.Results are reported as a mean of the sample spots done per sample.The shark enameloid d 18 O values are given as an overall mean as well as means of each structural tissue, to individually compare them with the d 18 O PO4 values measured by HTR.The isotope ratios are expressed in the d-notation relative to Vienna Standard Mean Ocean Water (VSMOW).

Table 2 .
Conodont and shark teeth d 18 O values measured by SIMS

Table 2 (
continued).Conodont and shark teeth d 18 O values measured by SIMS n refers to the number of sample spots analysed for each specimen.Samples labelled with a 'b' (i.e., TM mid longitudinal group) are re-analysis for specimens of the TM low longitudinal group (TM2, TM5-TM12).Shark enameloid abbreviations are SCED: single crystallite enameloid edge; SCE: single crystallite enameloid; PBE: parallel bundled enameloid; TBE: tangled bundled enameloid.

Table 3 .
Conodont and shark tooth d 18 O PO4 values measured by HTR (TC/EA) far not been analysed yet.This relationship was expressed as Δ collagen-phosphate , and an isotopic 18 O depletion of -5‰ of the d 18 O collagen relative to the d 18 O PO4 is assumed.More details related to this table are discussed below in the sections Shark teeth chemical and oxygen istope composition and Conodont chemical and oxygen isotope composition. so

Table 4 .
Mass balance calculations for carbonate and organic matter contents and their d 18 O contribution based on percent mass fraction (% m/m) to the SIMS analysis, expressed as Δ carbonate -phosphate (i.e., d 18 O CO3d 18 O PO4 ) and Δ collagen -phosphate (i.e., d 18 O collagend 18 O PO4 ), respectively