The preservation of francolite-associated sulfate in lingulid brachiopods

Abstract

Lingulid brachiopods – the namesake “living fossils” of Darwin – have a continuous record from the Cambrian to the present and possess phosphatic valves, rendering them largely unique among extant invertebrates. Due to their poorly understood vital effects and susceptibility to taphonomic breakdown, they have not been widely adopted as geochemical archives. However, inorganic sulfate is abundant in lingulid bioapatite, which may be more resistant to diagenetic alteration than carbonate-based minerals. As such, it is possible that lingulid shells may provide an alternative, independent archive of Phanerozoic sulfate seawater chemistry, provided that the distribution and preservation of sulfur within lingulid shells can be identified. Here, we leverage sulfur K-edge micro X-ray Fluorescence (μ-XRF) imaging, X-ray Absorption Near Edge Structure (XANES) spectroscopy, and petrography to characterize sulfur speciation and distribution within extant and fossil lingulid brachiopods. In modern Lingula anatina, inorganic sulfate is dominant along francolite-rich laminae, whereas organic sulfur species are mostly observed in non-mineralized laminae and the periostracum. In the Cambrian fossil Ungula ingrica, inorganic sulfate is generally dominant in regions with the lowest sulfur abundances, with higher sulfate abundances along compact laminae than on baculate laminae. In contrast, the distribution of diagenetic pyrite is heterogenous, although largely concentrated along shell edges and particular laminae. Notably, the XANES spectrum of francolite-associated sulfate is characterized by a unique post-edge feature at 2494.0 eV. Collectively, these results suggest that carefully selected lingulid specimens – and from intra-valve laminae with minimal diagenetic phases – have the potential to be insightful for reconstructing Phanerozoic sulfate seawater chemistry.

1. Introduction

Changes in the sulfur isotopic composition of seawater sulfate (δ³⁴S_seawater) have been reconstructed from evaporites, barites, and carbonate-associated sulfate (CAS), from both bulk carbonate material and selected fossil phases (Fike et al., 2015; Present et al., 2020). There is a growing appreciation that many CAS archives – especially bulk CAS – can be dramatically influenced by source mixing, local variations in seawater chemistry, and post-depositional alteration, driving a search for the most ‘pristine’ carriers of seawater sulfate (Present et al., 2015, 2020; Richardson et al., 2019a). Of CAS-bearing archives, monospecific foraminifera samples (Cenozoic; Paris et al., 2014; Rennie et al., 2018; Thaler et al., 2023) and calcitic (rhynchonelliform) brachiopods (Mesozoic – Paleozoic) are widely considered to be the most faithful archive of seawater sulfate throughout the Phanerozoic. In particular, rhynchonelliform brachiopods have proven especially popular as geochemical archives, as they precipitate in near-equilibrium with ambient seawater with limited – although discernible – vital effects, especially when sampling the secondary calcite layer (Bajnai et al., 2018; Brand et al., 2003; Carpenter and Lohmann, 1995; Rollion-Bard et al., 2024). Furthermore, they are generally considered resistant to geochemical recrystallization and/or extensive taphonomic breakdown, at least relative to other CAS-bearing archives (Brand et al., 2003; Johnson et al., 2020; Kampschulte et al., 2001; Ullmann and Korte, 2015).

However, rhynchonelliform brachiopods do have some disadvantages. First, as with most CAS archives, they have relatively low sulfate abundances (100–1,000 ppm), necessitating either the combination of considerable material (5–10 g) for conventional analyses (i.e., EA-IRMS; Kampschulte et al., 2001; Richardson et al., 2019a) or the use of novel, high-efficiency mass spectrometric techniques (i.e., Johnson et al., 2020; Present et al., 2015). Second, the fossil record of rhynchonelliform brachiopods is not consistent throughout the Phanerozoic: they first appear in sufficient abundance for geochemical analysis in the Early-Middle Ordovician, are not particularly common in post-Jurassic strata, and suffer sharp declines in diversity and abundance in the aftermath of mass extinctions, rendering it challenging to develop high-fidelity δ³⁴S_CAS records in precisely the intervals where δ³⁴S_CAS would be most interesting to study (Fike et al., 2015; Vörös et al., 2019). Lastly, while modern rhynchonelliform brachiopods are restricted to normal-marine salinity environments, Paleozoic examples show tolerance for brackish and/or hypersaline conditions, which may imprint a distinctive local signature on any resultant geochemical records (Fürsich and Hurst, 1980).

In contrast, phosphatic brachiopods (lingulids; Order Lingulida) are composed of the carbonate-fluoroapatite mineral francolite [(Ca,Mg,Sr,Na)₁₀(PO₄,SO₄,CO₃)₆F_2–3], which has comparatively high sulfur abundances (10,000–25,000 ppm) relative to most biogenic carbonates (McArthur, 1985), and so may be more resistant to certain diagenetic styles with respect to sulfur than calcitic brachiopods. Traditionally, lingulid brachiopods have been bypassed as paleoenvironmental archives due to the poorly understood extent to which they record environmental signals or biological fractionations (vital effects) during precipitation, the susceptibility of bioapatite to diagenetic alteration, and the rapidity in which they are mechanically degraded in modern environments (Brand et al., 2011). However, they host considerable potential as geochemical archives provided that these issues can be mitigated. For instance, fossil and molecular records indicate that lingulid brachiopods originated near the basal Cambrian, with a near-continuous Phanerozoic record (Harper et al., 2017; Sperling et al., 2011; Zhang et al., 2008). Their biomineralization genes have been highly conserved, suggesting vital effect calibrations based on modern organisms are potentially robust in the geologic record (Goto et al., 2022; Luo et al., 2015). While typical modern lingulid brachiopods are found in intertidal and infralittoral zones (with potentially large environmental fluctuations in oxygenation, temperature, salinity, and pH), in vivo shell biomineralization is more favorable under fully marine conditions (Emig, 1997), which may ameliorate the potential influence of salinity and/or meteoric input on primary geochemistry (Lécuyer et al., 1996). Finally, lingulid brachiopods are particularly common constituents of marine biota in the Cambrian and in the aftermath of mass extinction, potentially providing a high-resolution archive to explore sulfur cycling during the Cambrian Explosion and major biotic crises, such as the end-Permian mass extinction (Posenato et al., 2014; Rodland and Bottjer, 2001).

To this end, Xiang et al. (2024) generated δ³⁴S records derived from modern and Devonian lingulid brachiopods, and – by comparison with co-eval Devonian evaporite and CAS compilations – suggested that lingulid phosphate may serve as a reasonable δ³⁴S_seawater archive. While these results serve as a promising case study for lingulid δ³⁴S geochemistry, the role of microstructural variation or post-depositional alteration in modulating lingulid-bound δ³⁴S records remains poorly constrained (Rodland et al., 2003). This is a crucial consideration, as complex archives require careful imaging, sub-sampling, chemical pre-treatment, and analysis to calibrate their geochemical reliability. A full range of micro-scale and diagenetic factors must be constrained to adequately assess the fidelity and usefulness of geochemical data recovered from lingulid bioapatite (Lécuyer and Flandrois, 2023).

Lingulid shells are composed of multiple, alternating phosphatic and organic-rich laminae, with variable microstructual complexity (Cusack et al., 1999). Apatite is incorporated into the shell as organically coated spherules, which are further aggregated into various forms within the skeletal laminae (Agbaje et al., 2020; Cusack and Williams, 1997; Merkel et al., 2007; Schmahl et al., 2008; Williams et al., 1998). The organic component of lingulid shells is largely composed of sulphated glycosaminoglycans, fibrillar chitin, proteins, and collagen (Agbaje et al., 2021, 2020; Williams et al., 1998, 1994). There are considerable differences between compact laminae - which are composed of tightly packed apatite crystals - and baculate laminae, which are composed of trellised apatite rods (bacula) that were enmeshed into an organic matrix during the life history of the lingulid brachiopod (Williams and Cusack, 1999). In modern Lingula, the baculate framework is poorly developed; as a result, its shell structure is best referred to as botyoidal or virgose (Williams and Cusack, 2007). In particular, the preferential resistance of compact laminae to post-depositional recrystallization and authigenic apatite precipitation (Lang et al., 2016) may support the preservation of primary sulfur isotopic signatures, in a comparable manner to how the secondary shell layer tends to be the most geochemically stable phase in rhynchonelliform brachiopods (Brand et al., 2003).

Despite the poor fossilization potential of living lingulids (Emig, 1990; Kowalewski, 1996), the structural nuances of living and fossil shells can be exquisitely-preserved, often pointing towards considerable intra-valve microstructural variability (Lang et al., 2011; Lang and Puura, 2013). Moreover, microstructural style, complexity, and diversity in the lingulid clade appear to vary over the Phanerozoic, possibly imposing temporal and phylogenetic dimensions on their geochemical potential (Cusack et al., 1999; Popov et al., 2013; Zhang et al., 2024). Due to this microstructural heterogeneity, lingulid shells host multiple micro-environments for biomineralization, diagenetic precipitation, and recrystallization processes, which all likely influence the robustness and reliability of preserved geochemical signatures (e.g., Lang et al., 2016). Previous studies have suggested that at least 12% (by weight) of apatite could be added authigenically to lingulid shells during post-mortem degradation of organic matter, and skeletal apatite has been distinguished from non-skeletal apatite in bulk fossiliferous material (Lécuyer et al., 1998; Puura and Nemlihar, 2001). However, Lang et al. (2016) observed that different apatite phases can be spatially – and predicatively – distinguished within an individual lingulid shell: original skeletal apatite is largely concentrated along compact laminae, whereas non-skeletal apatite is more prominent along baculate laminae. These factors suggest that lingulid brachiopods might serve as robust paleoenvironmental archives for seawater sulfate if their microstructure and diagenetic history can be carefully characterized and accounted for in geochemical interpretations.

In this study, we take a first step towards understanding lingulid sulfur biogeochemistry by examining intra-valve and microstructural variations of sulfur abundances and speciation. We report spatial variation in sulfur speciation and abundance within individual modern Lingula anatina and Cambrian Ungula ingrica using high resolution microscale X-ray Fluorescence (μ-XRF) imaging and X-ray Absorption Near Edge Structure (XANES) spectroscopy. This data is coupled with bulk XANES spectroscopy from a suite of modern and Cambrian lingulid powders, which collectively provide insight on the structural stability of lingulid-bound sulfur through time.

2. Materials and Methods

2.1. Sampling Localities (Table 1, Fig. 1)

Table 1.

Studied samples.

Sample	Taxonomy	Collection number	Locality
Modern
Aus-1	Lingula anatina		Australia
Aus-2	Lingula anatina		Australia
H-1	Lingula anatina		Hawaii
H-1	Lingula anatina		Hawaii
JP-1	Lingula anatina		Japan
JP-3	Lingula anatina		Japan
OCSD-1	Glottidia		southern California
CSD-1	Glottidia		southern California
GIT-587-32	Lingula anatina	GIT-587	Cebu Island, Philippines
Fossil
I-Ü-3-v	Ungula ingrica	TUG 1323-5	Ülgase-3, Ülgase
I-I5-v2	Ungula ingrica	TUG 1323-5	Iru5, Iru
I-I7-v3	Ungula ingrica	TUG 1323-5	Iru7, Iru
Iru7	Ungula ingrica	TUG 1323-5	Iru7, Iru
Iru5	Ungula ingrica	TUG-1323-6	Iru5, Iru

Figure 1.

(A) Sampling locations of modern lingulid brachiopods. (B) Generalized map of Estonia, with Iru and Ülgase localities marked. (C) Stratigraphy of the Iru and Ülgase sections, with sampling levels marked (after Lang et al., 2016).

Modern:

The modern brachiopod thick section (GIT 587–32) is a Lingula anatina Lamarck 1801 from Cebu Islands (Philippines) that was implanted into polymethyl methacrylate (PMMA) resin and polished; this specimen was previously examined by Lang et al. (2016). The modern lingulid brachiopod powders are derived from L. anatina specimens previously collected from Australia, Hawaii, and Japan, and Glottida Dall 1870 specimens from southern California (Fig. 1a). The Australia, Hawaii, and Japan specimens were collected alive and flash-frozen, while the southern California specimens were collected from seafloor death assemblages. Prior to powdering, these samples were treated with a 5% NaOCl solution for 48–72 hours to remove non-mineralized phases from the shell, rinsed a minimum of three times with MilliQ water to remove any residual solution, and dried under ambient atmospheric conditions. Once dried, samples were hand-ground with an agate mortar and pestle.

Fossil:

The fossil lingulid brachiopods studied here are a subset of those examined by Lang et al. (2016) from the late Cambrian - early Ordovician Kallavere Formation of Estonia (Fig. 1b). The fossil brachiopod thick section (TUG-1323) consists of several Ungula ingrica (Eichwald 1829) valves from a single locality and bed (Iru5; uppermost Cambrian) that were implanted into PMMA resin and polished. The fossil brachiopod powders analyzed were Ungula ingrica valves that were previously hand-ground with an agate mortar and pestle by Lang et al. (2016), stratigraphically spanning the Kallevere Formation in the Iru and Ülgase sections (Fig. 1c). Unlike the modern samples, the Kallevere Formation powders were not rinsed to remove organic matter.

2.2. Scanning Electron Microscopy - Energy-dispersive X-ray spectroscopy (SEM-EDS)

SEM-EDS analyses of the thick sections were performed using a TESCAN LYRA3 field-emission scanning electron microscope equipped with two Oxford X-Max-80 detectors at the FIB-SEM Facility, Department of the Geophysical Sciences, University of Chicago. Prior to analysis, samples were coated with 10 nm of carbon to reduce charging effects. Secondary electron (SE) and back scatter electron (BSE) imaging was conducted with a voltage of 7 kV and current of 7 nA under high vacuum mode, while EDS imaging was performed with a voltage of 10 kV and current of 10 nA under high vacuum mode; low energies and currents were required to avoid melting the epoxy.

2.3. μ-XRF and XANES

Sulfur XANES spectroscopy and μ-XRF imaging were performed at beamline 14–3 at the Stanford Synchrotron Radiation Lightsource (SSRL), SLAC National Accelerator Laboratory, under standard ring conditions (500 MA and 3 GeV). This beamline is equipped with a double-cooled Si (111) monochromator, which was calibrated to the sulfur K-edge using the first pre-edge feature of a sodium thiosulfate powder (2472.02 eV). A Hitachi Vortex 7-element Si drift detector and Xspress3 Quantum detectors are the detection system at this beamline. The beam size can be seamlessly changed between bulk (mm) and micro (μm) scale using a Newport hexapod positioning system mounted with Sigray axially symmetric focusing optics (for 1 or 5 μm beam spot sizes). The hexapod can be moved between pre-determined settings with the focusing optic shifted out of the beam path for bulk (mm) measurements. For bulk XANES analyses, silts were used to set the beam size to approximately 1×2 mm, while for μ-XRF imaging and μ-XANES spectroscopy a 5 μm beam spot size was used.

For bulk, homogenized powders, a small amount of powder was placed onto S-free tape (Saint Gobain) and covered with 5 μm-thick polypropylene thin film. Each sample packet was cut to approximately 5×5 mm and placed on top of a larger S-free tape strip that was adhered to a sample holder, with an array of 3×2 samples per holder. Sample holders were loaded onto a multi-sample, fully rotating sample stage within a He-purged atmosphere, which reduces X-ray attenuation in air. Since bulk XANES spectroscopy was obtained in BL 14–3 imaging mode, a camera video feed enabled us to locate each sample within our pre-arranged grid.

For thick sections, samples were mounted into the sample stage wheel, and selected specimens were μ-XRF mapped (2500 eV) to obtain elemental maps of P and S. Initially, we anticipated that since these brachiopods are phosphatic (35–40% P), tailing of the large P XRF signal into the S energy range may have potentially swamped the comparatively small S spectra. Therefore, we confirmed that the P XRF spectra did not interfere with sulfur spectra by visualizing the total S and P peaks in the full XRF spectrum of both powder samples and thick sections. Subsequently, following mapping and determination of the prominent sulfur K-edge energies, we explicitly confirmed that the limited amount of S in the epoxy did not interfere with the S distribution and speciation in the mounted samples by checking both the XRF and XANES spectra of the epoxy relative to representative points on the mounted samples. Subsequently, μ-XRF maps were collected at various energies (2470.9, 2473.1, 2473.3, 2473.8, 2481.1, 2482.4, and 2482.6 eV) corresponding to features noted in preliminary XANES spectra to detect differences in the distribution of reduced, mixed, and oxidized sulfur species. In the Microanalysis Toolkit (Webb, 2011), simplex Volume Maximization (SiVM) analysis of multi-energy S maps was used to identify endmember S speciation components and automatically identify the locations at which these components are best expressed. A representative XANES spectrum was collected for each endmember location (see numbers in Fig. 4, 5).

Figure 4.

Sulfur speciation and spatial distribution in Lingula anatina GIT-587. On top, secondary electron (SE) image of specimen, with studied region marked by blue dashed rectangles. On left, μ-XRF maps correspond to total S, computed fits for thiols, sulfonate, and sulfate, and a tri-color plot for relative thiols, sulfonate, and sulfate abundance, respectively. In middle, XANES spectra correspond to the spots marked by numbered circles on the maps, with the relevant spectral features labeled. On right, associated LCF fitting results. For ease of readability, sulfur abundance is scaled linearly and relative to the highest abundance in individual maps.

Figure 5.

Sulfur speciation and spatial distribution in Ungula ingrica TUG-1323. Insert on top right shows secondary electron (SE) image of specimen, with studied regions corresponding to the left, middle, and right columns, respectively, marked by blue dashed rectangles. In each column, μ-XRF maps correspond to total S, computed fits for thiols, sulfonate, and sulfate, and tri-color plot of thiols, sulfonate, and sulfate species scaled to their absolute abundance. In top middle insert, XANES spectra correspond to the spots marked by numbered circles on the left, with the relevant spectral features labeled. For ease of readability, sulfur abundance is scaled linearly and relative to the highest abundance in individual maps.

Subsequently, XANES spectra were processed using a combination of SIXpack (Webb, 2005) and Larix (Ravel and Newville, 2005), following previous workers (Richardson et al., 2019a). Background subtraction and normalization were performed in Larix by fitting a linearized pre-edge (at ~2465 eV) and 2nd order post-edge (~2510 eV). A linear combination fitting (LCF) was performed between a range of 2465 eV to 2520 eV to identify the relative abundance of S species in the experimental XANES spectra, using the pyrite, cysteine (representing a generalized thiol-bearing compound), thiophene (representing a generalized heterocyclic reduced sulfur compound), sulfite, sulfoxide, sulfonate, and inorganic sulfate XANES reference standards (Manceau and Nagy, 2012). We chose gypsum as a sulfate standard, rather than a phosphate-associated sulfate or any alternative inorganic sulfate, because to the best of our knowledge there is no reliable phosphate-associated sulfate XANES reference, and the gypsum spectra is well-characterized and similar in appearance to our phosphate-associated sulfate. We exclude organic sulfate esters and barite from the LCF because the observed post-edge features and energies were not characteristic of either sulfur species (e.g., Debret et al., 2017; Wilke et al., 2008), and furthermore there was no textural, petrographic, or chemical evidence for barite enrichment. All maps were processed using the Microanalysis Toolkit (Webb, 2011), by extrapolating the XANES LCF spectra using a least-squares fitting approach to generate a suite of maps corresponding to selected sulfur species (Bryant et al., 2023).

3. Results

3.1. Bulk powder XANES and LCF (Fig. 2)

Figure 2.

XANES spectra and LCF fitting results for bulk powders. Blue spectra are from sulfate-bearing mineral standards; black spectra are from modern lingulid powders; red spectra are from fossil lingulid powders. Note the gradual increase in the size of the sulfate peak – and the proportional contribution of sulfate to the total sulfur inventory – with increasing stratigraphic height.

The modern lingulid specimens are dominated by a single excitation peak at 2842.6 eV (sulfate). No reduced or mixed-valence sulfur species were observed in any of the modern collected brachiopods (Japan, Hawaii, Australia) – which, as noted in Section 2.1, had been previously treated to remove organic matter. Some reduced or mixed-valence sulfur species were detected in the bulk California samples, which have undergone early taphonomic breakdown. These spectra are also characterized by broad post-edge features at 2486.0 eV, 2494.0 eV, and 2500.0 eV. While the peaks at 2486.0 eV and 2500.0 eV broadly resemble post-edge features in other calcium sulfates, such as gypsum or brachiopod-bound CAS, the feature at 2494.0 eV appears to be largely unique to francolite (Majzlan et al., 2011; Paris et al., 2001; Xu et al., 2020).

In the powdered Cambrian U. ingrica specimens, peaks at 2472.2 eV (pyrite) and 2482.6 eV (sulfate) are dominant. The proportion of pyrite increases throughout the stratigraphic section (from Ü3 to Iru7), in conjunction with a decrease in the proportion of sulfate. As in the L. anatina powders, there are additional broad post-edge features at 2486 eV, 2494 eV, and 2500 eV in spectra dominated by the sulfate peak energy.

3.2. Spatial analysis of modern Lingula anatina (Fig. 3, 4)

Figure 3.

SEM-EDS images and spectra for two locations on Lingula anatina GIT-587 (see Fig. 4). The middle images are from near the edge of the shell, while the upper and lower images are from the center of the shell. Note that the number, style, and elemental distribution within laminae differ between the sites, and that the top image is vertically exaggerated by a factor of 2 for visual clarity. See also Fig. 4, Lang et al. (2016) for corresponding SEM-EDS images from a Ungula ingrica specimen.

The L. anatina specimen (GIT-587) is narrow, with a length of ~18 mm and a thickness of <150 μm. The shell is composed of 5–8 discrete laminae of variable thickness (~5–50 μm), interspersed by somewhat thinner (~5–20 μm) interlaminae spaces with less biomineralized material (Fig. 3). These two alternating features likely correspond to the compact and baculate laminae in fossil lingulid specimens, respectively. For clarity, L. anatina do not have true ‘baculate laminae’, as do fossil specimens, but rather possesses a virgose fabric in which the bacula rods have not developed. We refer to this virgose fabric as ‘baculate laminae’ for ease of comparison. The ‘compact laminae’ feature elevated Ca and P abundances compared to the ‘baculate laminae’ which is consistent with previous observations from Cambrian U. ingrica specimens (Lang et al., 2016). Fluorine, Na, and Mg abundances are generally higher in the ‘compact laminae’ than in the ‘baculate laminae’ (Fig. 3; see also Lang et al., 2016, Fig. 4). In some, but not all, cases, the lower boundary of the ‘compact laminae’ is diffusive into the next ‘baculate laminae’.

On the L. anatina polished section, XANES spectra have several spectral features, with the main characteristic energies identified at 2473.5 eV (thiol), 2476.4 eV (sulfoxide), 2481.0 eV (sulfonate), and 2482.6 eV (sulfate) (Fig. 4). In spectra with dominant sulfate peaks, additional post-edge features are observed at 2486 eV, 2494 eV, and 2500 eV, although these are less prominent than in the bulk powders.

To observe how sulfur speciation and abundance patterns vary across the L. anatina polished section, we plot fitted μ-XRF maps for thiols, sulfonate, sulfate, and total sulfur (Fig. 4). As sulfoxide generally scales with sulfonate in XANES spectra, we do not plot its spatial distribution. Total sulfur abundances are banded within GIT 587–32, with the highest relative abundances observed in association with phosphatic laminae, intermediate abundances in the inter-laminae intervals, and the lowest abundances in the periostracum. In turn, relative sulfate abundance is generally correlated with total sulfur abundances – the highest sulfate abundances are in the phosphatic laminae, whereas the lesser-mineralized inter-laminae have little-to-no sulfate. While XANES spectra confirm that thiols are present throughout the specimen, it has the highest relative abundance in the inter-laminae intervals and on the external periostracum margin. Finally, sulfonate species are mostly co-located with thiols and sulfoxides throughout GIT 587–32, they are the dominant sulfur species on the internal valve margin (e.g., Spots 2, 4, and 6 in Fig. 4) and the inter-laminae closest to the internal valve margin (Fig. 7).

Figure 7.

A conceptual illustration of biomineralization, sulfur speciation, and elemental distribution in both modern (left) and fossil (right) lingulid brachiopods. In the modern Lingula anatina, the dominant sulfur species in compact laminae is sulfate, the dominant sulfur species in virgose laminae are sulfonates, and the dominant species in the periostracum and individual layers within virgose laminae are thiols. While sulfate is the dominant non-pyrite species in the Cambrian Ungula ingrica, sulfate concentrations are generally higher in compact laminae than in baculate laminae, while thiol concentrations are relatively constant throughout. Accordingly, accordingly, the color banding in the baculate laminae represents their higher thiol-to-sulfate ratios relative to the compact laminae. Pyrite is heterogenously disseminated throughout the shell, but is more concentrated near the shell margin and in the baculate laminae. The Elemental concentration data on Ungula ingrica from Lang et al. (2016).

Organosulfur compounds form a considerable – perhaps even dominant – component of total sulfur in the L. anatina polished section. In the phosphatic laminae, LCF fitting results indicate that only 50% of total sulfur is derived from sulfate – interpreted here to represent francolite-bound inorganic sulfate, rather than organosulfate species. The majority of organosulfur in these laminae are thiol-bearing (24%) and sulfonate (23%) species. In the non-biomineralized laminae, the dominant sulfur species are again thiol-bearing and sulfonate species, with modest contributions from sulfoxide species.

As with total sulfur abundance, the spatial distribution of each of these constituent sulfur species is dominated by laminae-parallel banding. There is minimal lateral heterogeneity in the observed excitation maps; where there is lateral variability, it is largely associated with post-mortem physical damage to the L. anatina shell (likely arising from the mounting and/or polishing process).

3.3. Spatial analysis of Cambrian Ungula ingrica (Fig. 5, 6)

Figure 6.

Sulfur speciation and spatial distribution in Ungula ingrica TUG-1323. On left, XANES spectra for spots on Fig. 5. On right, associated LCF fitting results.

The examined U. ingrica specimen (TUG-1323) is shorter (~6 mm) and thicker (~2 mm) than the L. anatina specimen. The shell is composed of alternating compact and baculate laminae, with the compact laminae slightly thinner than the baculate counterparts (Lang et al., 2016). Pyritization is focused along the outer margin of the shell and along baculate laminae. P, Na, and Mg abundances are elevated along the compact laminae, while Ca and F abundances are generally higher along the baculate laminae (see also analyses of the same specimen in Lang et al., 2016).

On the U. ingrica polished section, the main XANES features have energies of 2472.2 eV (pyrite), 2473.5 eV (thiol), and 2482.6 eV (sulfate) (Fig. 5). As in the L. anatina specimen, there are additional broad post-edge features at 2486.0 eV, 2494.0 eV, and 2500.0 eV in spectra dominated by sulfate.

To observe sulfur speciation and abundance patterns within TUG-1323, we plot μ-XRF maps for thiols, sulfonate, sulfate, and total sulfur (Fig. 5). To first order, the spatial distribution of total sulfur abundance is patchy, especially in comparison to the modern L. anatina specimen. However, some spatial patterns can be recognized. For instance, total sulfur abundances tend to be highest in laminae-parallel bands, in particular on (i) the outer commissure and (ii) the outer margins of baculate laminae. In many cases, sulfur abundance ‘hotspots’ coincide with pyrite framboids that can be readily recognized either visually or via SEM, and are confirmed by XANES spectroscopy (Fig. 6). In other cases, however, sulfur abundance ‘hotspots’ do not necessarily coincide with the growth orientation of the shell.

Relative pyrite abundance is almost universally associated with the highest total sulfur abundances. As such, it is extremely variable throughout the shell, although appears to be more common at the edges of the specimens and along certain laminae, as noted above. By contrast, sulfate is the most dominant sulfur species (>82%) in intervals with lower average sulfur abundances – as would be expected based on francolite stoichiometry [1.0–2.5% S by mass] vs. pyrite stoichiometry [53.3% S by mass]. While there is spatial variability in relative sulfate abundances, they tend to be much more homogenous than corresponding sulfide abundances. Notably, sulfate is demonstrably elevated in some – but not all – total sulfur abundance hotspots. Moreover, sulfate abundance trends along the original fabric, with sulfate abundances somewhat higher in the compact laminae than in the baculate laminae. Finally, XANES spectra and fits to multi-energy μ-XRF maps confirm that thiol species are disseminated throughout the shell, albeit at considerably lower relative abundances than in L. anatina. While thiol is generally a minor component of shell-bound sulfur species, in some isolated locations it can compose up to 42% of the proportional sulfur abundance. Many isolated thiol hotspots are co-located with pyrite. In general, sulfate-to-thiol ratios tend to be higher within baculate laminae, and lower within compact laminae.

4. Discussion

4.1. Sulfur speciation in modern lingulids

The compounds represented in XANES spectra for the modern Lingula anatina are assigned to, in order of increasing oxidation state, reduced organic sulfur (likely thiol-bearing), sulfoxide, sulfonate, and inorganic sulfate (Fig. 3). While organic sulfate esters have similar spectral features to a combination of sulfonate and sulfate, the spectral feature at ~2481.0 eV is lower than the leading shoulder of a sulfate ester (Almkvist et al., 2010). In addition, the relative amplitudes of the peaks at 2481.0 eV and 2482.6 eV are spatially variable, indicating that the features are unlikely to represent a single compound.

The vertical distribution of sulfur is heterogeneous in the L. anatina polished section, with bands of relatively high sulfur abundance corresponding to mineral-rich regions of the shell, and slightly lower sulfur abundance in more organic-rich regions of the shell (Fig. 4, 7). These discrete compositional bands are 5–20 μm in thickness, laterally continuous, and oriented parallel to the valve margin. Perhaps unsurprisingly, organic sulfur species are dominant in the non-biomineralized (e.g., phosphate-poor) laminae, and inorganic sulfate dominates in the mineral-rich bands. Based on the unique sulfate post-edge feature at ~2494 eV, we suggest that the sulfate in the shells is likely lattice-bound in apatite, in solid solution with phosphate and carbonate ions, rather than present as micro-inclusions of other sulfate-bearing minerals, such as gypsum or barite (Anné et al., 2019; Konecke et al., 2017; Xu et al., 2020). This interpretation is supported by the largely homogeneous nature of the sulfate within mineralized bands.

The absence of organic sulfur species in powdered L. anatina specimens (Fig. 2) is likely due to sample preparation - namely, the 5% NaOCl solution treatment prior to powdering. The small presence of reduced and mixed-valence sulfur species in the Glottida specimens is potentially due to early taphonomic remineralization of shell-bound organic sulfur species. Ideally, to fully capture a picture of sulfur speciation, the powdered specimens would not have had organic sulfur species removed prior to XANES analysis. However, the powders were originally prepared for future δ^44/40Ca analysis beyond the scope of this study, for which organic matter must be removed prior to column chromatography (Todes and Blättler, 2022). Moreover, the minimal presence of detectable organic sulfur species in the powders suggests that treatment with 5% NaOCl solution is generally sufficient to remove all organic-bound sulfur species, which may have implications for measuring the isotopic composition of FAS in lingulid brachiopods (see section 4.4).

4.2. Sulfur speciation in fossil lingulids

In contrast to the modern L. anatina specimen, the major compounds represented in the U. ingrica XANES spectra are pyrite and inorganic sulfate (Fig. 5–7). While organic sulfur species (primarily thiols) can be detected throughout the lingulid matrix, they are present at much lower abundances than within the L. anatina thick section.

The abundance and distribution of pyrite throughout U. ingrica is presumably linked to the reactivity and abundance of organic matter within the lingulid during their lifetime. Modern lingulid shells contain a notably high organic matter contribution, especially within baculate laminae (50–70%; Williams et al., 1994), in contrast to the relatively low abundance of organics within rhynchonelliform brachiopod shells (<3%; Cusack et al., 2008). Moreover, organic matter within modern lingulids (e.g., Discinisca tenuis (Sowerby)) contains a significant proportion of highly-labile biopolymers, such as sulfonated glycosaminoglycans (Williams and Cusack, 1999). Under suitable conditions, the post-mortem degradation of these compounds could have promoted pyrite formation, which in turn would preferentially occur within baculate laminae. To this end, Lang et al. (2016) speculated that the authigenic pyrite within these brachiopods suggested that sulfate-reducing bacteria were taphonomically active during very early diagenesis. The same bacterially-mediated sulfate-reduction process would promote the precipitation of secondary, authigenic apatite (Lécuyer et al., 1998), as has been previously observed in the baculate laminae of this specimen (Lang et al., 2016). Additionally, pyrite tends to be more concentrated along the marginal surfaces of the shell and in the baculate layers (Fig. 5), which is suggestive of the presence of post-mortem degrading agents (e.g., bacterial mats) both around the edge of the shell and in the baculate layers.

The relative variations in sulfate abundances within U. ingrica are attributed to the differing primary vs. authigenic nature of francolite within compact and baculate laminae, respectively (see also Lang et al., 2016). In general, relative sulfate abundances are lower within the baculate laminae relative to the compact laminae, which would suggest sulfate depletion within the post-depositional apatite phase (Fig. 5, 7). Notably, the authigenic apatite phase associated with porous baculate laminae is characterized by higher Ca, F, and CO₃²⁻ abundances than that associated with the primary apatite phase in compact laminae (Lang et al., 2016). Conversely, primary (‘biotic’) apatites have higher Na and Mg abundances than the authigenic (‘diagenetic’) apatite phase (Lang et al., 2016). As sulfate in modern Lingula is distributed in mineralized laminae rather than organic-rich laminae, it is likely that the higher sulfate concentrations in the compact laminae of the fossilized specimen are also a relic of the original composition of the shell. Furthermore, sulfate incorporation can – although does not necessarily – result in increases in the a-axis and c-axis lattice parameters of apatite (Alshemary et al., 2013; Tran et al., 2017), consistent with the fact that (i) the a-axis lattice parameter in modern Lingula is the same as that in the compact laminae, and (ii) is larger than the a-axis lattice parameter in the baculate laminae (Lang et al., 2016). In experimental abiotic apatite precipitation, sulfate substitution into an apatite structure is mediated by the charge balance of the counter ion (e.g., Na⁺, Mg²⁺), such that the incorporation of sulfate scales with Na and Mg abundances, and is inversely proportional to Ca abundance (Tran et al., 2017) – consistent with the observations presented here. While there is no direct experimental evidence, the observed negative correlation of sulfate and carbonate abundances could conceivably be driven by either (i) competition for lattice space between the comparably-sized sulfate and carbonate anions, (ii) differences in the solubility of sulfur vs. carbonate species associated with organic matter degradation and authigenic apatite precipitation processes, and/or (iii) tight in vivo constraints on carbonate chemistry within a living lingulid.

In isolated cases, elevated sulfate abundances are associated with total sulfur and sulfide abundance ‘hotspots’ (Fig. 5). In some – but not all – cases, these are associated with pyrite nodules, confirmed petrographically. We tentatively attribute these pyrite-associated elevated sulfate abundances to sulfide oxidation, which may be potentially associated with exposure to oxidating fluids post-pyritization, surficial oxidative weathering, and/or sample preparation (Chander and Briceno, 1987; Richardson et al., 2019a).

No sulfonates (e.g., resulting from glycosaminoglycan degradation) were detected in the fossil shell, which is surprising given that they (i) constitute a substantial proportion of modern organophosphatic lingulid shells, including the L. anatina specimen studied here, and (ii) their existence has been inferred in fossil lingulids based on ultramorphologic considerations (Williams et al., 1994). We infer that any sulfonate moieties were preferentially degraded during early taphonomic stages, as polysaccharides are particularly reactive organic compounds, especially when bearing multiple functional groups (Cowie et al., 1992). In contrast, we attribute the more pronounced presence of thiol-bearing organic sulfur compounds (Fig. 5) to either proteins within the organic fibrous framework present throughout the matrix, or original matrix-based organic matter that was diagenetically sulfurized by reducing fluids and subsequently incorporated into the authigenic apatite, perhaps contemporaneous with pyrite formation. Thiol-bearing organic sulfur compounds are known to have a strong inhibitory effect on hydroxyapatite crystallization (Koutsopoulos and Dalas, 2001), which may help explain why the crystallites in both the compact laminae and bacula have maintained their original size, while secondary apatite developed large, blocky crystals (Lang et al., 2016; Fig. 7). While thiol-bearing organic sulfur compounds are present in the modern L. anatina, the spatial distribution is different than in U. ingrica, suggesting either (i) physiological differences between the two species or, more likely (ii) a pronounced diagenetic pathway. In lingulid brachiopods, such frameworks encase trellised apatite rods, appear to be relatively resilient during early diagenesis, and may form a template upon which authigenic apatite can precipitate (Lang et al., 2016). In the modern rhynchonelliform brachiopod Terebratulina retusa (Linnaeus, 1758), thiols are associated with organic sheaths encasing calcitic fibers, which is comparable to their inferred purpose in the bacula of U. ingrica (Cusack et al., 2008).

Sulfur speciation in the powdered U. ingrica specimens reflect a mixture of inorganic sulfur species, with no observed organic sulfur contribution. In the Iru section, the relative abundance of sulfide modestly increases with stratigraphic height, concomitant with a decrease in the relative abundance of sulfate (Fig. 2). While these are largely qualitative observations, they do suggest that the intensity of authigenic pyrite formation increased upwards in the Kallavere Formation, which may result from secular evolution and/or spatial heterogeneity in the subsurface redox profile (Graul et al., 2023a, 2023b; Lumiste et al., 2021).

4.3. Implications for lingulid biomineralization

Despite considerable changes in the abundance of intra-crystalline organic matter, the presence of similar organic sulfur compounds (e.g., thiol-bearing species) in both extant and Cambrian lingulid brachiopods may provide insight into the processes and evolutionary history of lingulid biomineralization processes. Such intra-crystalline sulfur-bearing compounds have been previously identified in modern corals, bivalves, and rhynchonelliform brachiopods, and are thought to have significant influences on a variety of biomineralization processes, including crystal nucleation, growth and orientation (Cuif et al., 2003; Cusack et al., 2008; Dauphin et al., 2005; Richardson et al., 2019a, 2019b). Intra-crystalline thiol-bearing compounds have been previously identified in both modern and fossil rhynchonelliform brachiopods, and attributed to organic sheaths encasing calcitic fibers (Cusack et al., 2008; Richardson et al., 2019a). Specifically, it has been proposed that these thiol-bearing compounds are ultimately responsible for facilitating the growth of the fibrous secondary layer in rhynchonelliform brachiopods, which may help to explain their preservational potential and comparatively limited vital effects (Richardson et al., 2019a). It has also been observed that sulfur concentrations are lower within the fibrous secondary layer in modern Terebratulina, coinciding with an inferred change in the speciation of the sulphated organic matrix towards thiol-bearing compounds that is inferred to have a major influence on biomineralization processes (England et al., 2007).

The presence of intra-crystalline thiol-bearing compounds in the phosphatic lingulid brachiopods analyzed in this study – even at low abundance – may suggest that thiol-bearing matrix proteins are potentially homologous across the Brachiopoda, despite differing mineralogies and the early Cambrian divergence between the Linguliformea and the Rhynchonelliformea. This, in turn, suggests that brachiopod biomineralization pathways are deeply-rooted and have been highly conserved throughout the Phanerozoic, consistent with modern genomic, transcriptomic, and proteomic evidence (Jackson et al., 2015; Luo et al., 2015; Wernström et al., 2022). This observation is also consistent with the generally comparable – although diverse – assemblage of ultrastructural features, laminae texture, and mineralogical – organic matter interactions in both calcitic (Rhynchonelliformea, Craniiformea) and phosphatic (Linguliformea) brachiopods (Merkel et al., 2009, 2007; Simonet Roda et al., 2021). However, we emphasize that the specific roles played by these thiol-bearing organic sulfur species during lingulid biomineralization cannot be determined from this study alone, and may have varied between genera and throughout evolutionary time; for instance, there have been significant changes in chemio-structural, microtextural, and nannotextural features within the lingulid lineage over the Phanerozoic (Cusack et al., 1999; Lang et al., 2011).

4.4. Lingulid-bound francolite-associated sulfate (FAS) as an archive of seawater sulfate?

While bulk δ³⁴S_FAS compositions – typically from inorganic phosphorites – has been previously measured, it has not generally been accompanied by information about sulfur speciation. Bulk δ³⁴S_FAS signatures have generally been interpreted as reflecting local environmental controls, with isotopic signatures reflecting local changes in the location and shape of the redox gradient and/or microbial sulfur cycling (Benmore et al., 1983; McArthur, 1985; McArthur et al., 1986; Goldberg et al., 2005; Sanders et al., 2024; see Shields et al., 2004; Hough et al., 2006 for opposing views). At face value, this would suggest that bulk francolite with no obvious macrofaunal biogenicity would not inherently reflect global seawater sulfate δ³⁴S, given the marked local, syn-depositional influence. However, bulk francolite deposits are usually microbially-mediated and/or abiotic, and generally spatially restricted to phosphorite facies: they are likely not a reasonable comparison for lingulid-bound FAS, given that biomineralization in lingulid is tightly regulated (Cusack et al., 1999), occurs under normal marine salinity and oxic conditions, and is demonstrably dominated by inorganic sulfate.

Xiang et al. (2024) reported bulk δ³⁴S_FAS and δ¹⁸O_FAS values from modern Lingula (Guangxi, China) and Devonian lingulid brachiopods, through which they suggested that (i) there is a −6‰ offset between lingulid δ³⁴S_FAS and ambient seawater and (ii) that lingulid brachiopods could reasonably track both modern and Devonian δ³⁴S_seawater, the latter based on comparison to the δ³⁴S values of Devonian evaporites. They excluded significant organosulfur contribution based on the invariance of their δ³⁴S_FAS measurements across a variety of experimental extraction procedures, and significant diagenetic overprint based on comparable phosphorus-to-sulfur ratios between modern and Devonian lingulid brachiopods. While their observations are stimulating, given that the lingulid XANES spectra presented here indicate the presence of organosulfur compounds (up to 90% in modern L. anatina, and 30% in the Cambrian U. ingrica), the contribution of organosulfur compounds to previously reported modern bulk δ³⁴S_FAS and δ¹⁸O_FAS should be further considered.

First, even an organosulfur contribution could decrease lingulid δ³⁴S_FAS values by at least −6‰. For instance, if 20% of lingulid shell sulfur consists of organic sulfur, there is no major fractionation between δ³⁴S_FAS and δ³⁴S_SW, and the offset between modern δ³⁴S_SW (21.15‰; Johnston et al., 2014; Paris et al., 2013; Tostevin et al., 2014), and the δ³⁴S values of shell-bound organic matter (−5 to +10‰, Fichtner et al., 2021; Newton et al., 2018; Pape et al., 2018) are comparable to that of modern bivalves, than incorporation of organosulfur species could decrease measured δ³⁴S values by −5‰. This is a conservative estimate for the organic matter contribution, considering that LCF fittings (Fig. 4) indicate that >50% of modern lingulid shell sulfur is composed of organic sulfur species. This is comparable to the modern ‘vital effects’ observed by Xiang et al. (2024), which could certainly poise a major concern. Xiang et al. (2024) did not treat their samples with an oxidative-reductive cleaning protocol to remove organic matter prior to analysis, which could have inadvertently led to the incorporation of organic-derived sulfur in their reported δ³⁴S_FAS values. As such, we suggest that lingulid ‘vital effects’ should be recalibrated using an extraction protocol that diagnostically removes organosulfur material (e.g., 72 hr of NaOCl, as presented here).

Numerous geochemical proxies show considerable variation between modern L. anatina populations that cannot be readily linked to temperature or palaeoceanographic conditions (Tuross, 2024). Consequently, δ³⁴S_FAS measurements from several other populations are needed to confirm that the observed vital effects are stable. While it is impossible to examine vital effects in extinct genera – such as the Ungula examined in this study – the two modern lineages of lingulid brachiopods studied here (Glottida and Lingula) diverged in the Carboniferous to Triassic according to morphological and genomic constraints (Goto et al., 2022; Williams et al., 2000). As such, if the vital effects of lingulid δ³⁴S_FAS are approximately the same in these two clades, it would seem plausible that vital effect calibrations based on modern organisms are geologically robust. Alternatively, if the vital effects vary significantly between extant Glottida and Lingula, additional work must be conducted to determine if lingulid-based δ³⁴S_FAS measurements are truly robust constraints on δ³⁴S_SW.

Third, previous studies have suggested that bulk phosphorus-to-sulfur ratios may be useful for tracing diagenetic alteration within lingulid brachiopod shells (e.g., Xiang et al., 2024). However, phosphorus-to-sulfur ratios are inherently spatially variable within both modern and fossil lingulids (Fig. 8), and so may not be a viable marker of diagenesis. There is substantial evidence that sulfur incorporation into lingulid phosphate can be influenced by both ontogenic and seasonal factors, rather than just post-depositional controls (Mazumdar, 2024). Moreover, while phosphorus concentrations are generally constant throughout the U. ingrica specimen [with the exception of places with visible pyrite], total sulfur concentrations can vary by several orders of magnitude, largely driven by the pyrite abundance. In previous studies, Devonian lingulids occupy a much smaller range of phosphorus-to-sulfur ratios (67.7–80.2) than modern lingulids (61.1–120.9); it is unclear if the phosphorus-to-sulfur ratios identified by Xiang et al. (2024) is evolutionarily conservative or results from preservational constraints (Xiang et al., 2024). Furthermore, since sulfate is the preferred target for isotopic analysis, and lingulid brachiopods can incorporate variable proportions of sulfide and organic sulfur species while the sulfate is structurally bound (see section 4.1), it may be more appropriate to consider phosphate-to-sulfate ratios when possible, especially as non-sulfate species can be easily added or lost during organic matter sulfurization and pyritization (Fig. 8). FAS abundance tends to be higher within compact laminae than within baculate laminae, and additional francolite-bound sulfate can be generated by the oxidation of pyrite during weathering or sample preparation, both of which would impact phosphate-to-sulfate ratios (Fig. 8). As such, we suggest that pure elemental ratios may not be ideal for tracking diagenesis in lingulid brachiopods, especially given observed micro-scale diagenetic variations in other proxies (e.g., Lang et al., 2016).

Figure 8.

Phosphorus:sulfur (P:S) and phosphorus:sulfate (P:SO₄²⁻) ratios in both Lingula anatina GIT-587 and Ungula ingrica TUG-1323. For ease of readability, P:S and P:SO₄²⁻ ratios are scaled linearly and relative to the highest abundance in individual maps. Note that the scaling and fitting approaches used in this study means that absolute concentrations cannot be directly compared across maps – as such, we emphasize the qualitative patterns.

Given the observed heterogeneity in sulfur speciation, spatial distribution, and abundance within individual lingulid brachiopods (Fig. 7), it is plausible that lingulid sulfur isotopic composition may be spatially variable within an individual lingulid specimen and incorporate several sulfur species. The EA-IRMS method employed by Xiang et al. (2024) necessitated the combination of multiple lingulid shells to obtain robust δ³⁴S_FAS measurements. Given the observed spatial heterogeneity in sulfur speciation within individual lingulids, such approaches may unavoidably integrate multiple sulfur phases and the geochemical aftermath of various diagenetic pathways. MC-ICP-MS approaches that require significantly less material (Bryant et al., 2023; e.g., Paris et al., 2013; Sanders et al., 2024) may be preferable for measuring lingulid δ³⁴S_FAS, at least until the spatial distribution of lingulid δ³⁴S_FAS is better constrained. Targeted micro-sampling along compact laminae, away from any organic and/or sulfidic species hotspots, should ensure that FAS is the dominant sulfur constituent. Since lingulid compact laminae are thought to reflect skeletal apatite, whereas baculate laminae are influenced by early authigenic francolite precipitation (Lang et al., 2016), we suggest that geochemical sampling from lingulid shells should primarily target compact laminae where it is possible to distinguish these differing laminae styles. While pyrite can partially oxidize to sulfate-bearing species during sample preparation and/or under ambient earth surface conditions, the fact that FAS hotspots are typically associated with pyrite suggests that it is possible to distinguish pyrite-derived sulfate from primary FAS. Beyond micro-sampling, extreme care should be taken to remove organic-bound sulfur and avoid inadvertent dissolution of reduced sulfur species, such as pyrite.

5. Conclusions

In this study, we leverage sulfur K-edge μ-XRF imaging, XANES spectroscopy, and taphonomy to characterize the speciation, distribution, and geologic stability of sulfur within lingulid brachiopods. We conclude that fracolite-bound sulfate (FAS) XANES spectra are broadly comparable to CAS, with a distinctive post-edge feature at 2494.0 eV that is present in both modern and fossil specimens and is clearly distinguishable from other sulfate species. FAS appears to be more abundant in (primary) compact laminae than in (authigenic) baculate laminae (Fig. 7). Moreover, while organic sulfur compounds are more abundant and diverse in the modern Lingula anatina specimen than in the Cambrian Ungula ingrica specimen, they are still present in the latter, often in appreciable abundance. Collectively, these results suggest that lingulid-based sulfur isotope measurements from carefully selected lingulid specimens, focusing on intra-valve laminae diagnostically free of diagenetic phases, may be useful for reconstructing Phanerozoic sulfur seawater chemistry.

Highlights:

• First sulfur K-edge XANES spectra from lingulid (phosphatic) brachiopods

• Considerable intra-specimen variability in sulfur abundance and speciation

• Differences between modern and Cambrian lingulid sulfur distribution and speciation

• Francolite-associated sulfate has a unique XANES post-edge feature at 2494.0 eV

• Lingulid fossils must be spatially characterized prior to interpreting δ³⁴S records