The effects of meteoric diagenesis on the geochemical composition and microstructure of Pliocene fossil Terebratalia coreanica and Laqueus rubellus brachiopod shells from northeastern Japan

Abstract

Stable carbon (δ¹³C) and oxygen isotope (δ¹⁸O) compositions of fossil brachiopod shells can be used to interpret paleoclimatic and paleoceanographic conditions. However, the initial isotopic composition of the living shells might be modified by diagenetic alteration. To assess the degree of this modification, we analyzed δ¹³C and δ¹⁸O and three common indicators of alteration in shells: minor element (manganese (Mn), iron (Fe), and strontium (Sr)) concentrations; cathodoluminescence (CL)/non-luminescence; and the shell microstructure of fossil Terebratalia coreanica and Laqueus rubellus (rhynchonellate brachiopod) shells from an exposure of conglomerates of the Tentokuji Formation (Pliocene, 3.85–2.75 Ma) in northeastern Japan. Two indices were used to quantify the preservation state of shell microstructure: the altered fiber ratio (AFR) and the altered fiber and puncta-filling cement ratio (AF-PCR). We qualified the degree of luminescence by measuring the mean cathodoluminescence index (MCLI), defined as the mean R values in a particular area of a CL image with RGB colors. The δ¹³C and δ¹⁸O values were negatively correlated with AFR or AF-PCR and MCLI. The sampling spots with high Mn and Fe concentrations also exhibited relatively low δ¹³C and δ¹⁸O. The δ¹³C and δ¹⁸O were more strongly correlated with Mn concentration than with Fe concentration. Therefore, of the three minor elements examined in this study, Mn concentration is the most reliable indicator of meteoric diagenesis in the studied shells. Unlike Mn and Fe concentrations, Sr concentrations in the modern and fossil brachiopod shells were similar, with overlapping ranges. This indicates that Sr concentration is not likely to be a good indicator of diagenetic alteration in the studied shells. Our study provides further evidence that multiple criteria should be applied and cross-checked when assessing diagenetic alteration of brachiopod shells.

Introduction

Stable carbon isotope (δ¹³C) and oxygen isotope (δ¹⁸O) compositions of biogenic carbonates are useful for paleoenvironmental analysis. The δ¹³C and δ¹⁸O values of shells from brachiopods belonging to the subphylum Rhynchonelliformea (Williams et al. 1996; hereafter, simply called brachiopods) have been widely used for reconstructing paleoclimatic and paleoceanographic conditions during the Phanerozoic, and especially during the Paleozoic (e.g., Veizer et al. 1999), because of (1) their widespread occurrence in marine sediment and their continuous geologic record; (2) their low-magnesium calcite mineralogy, which is less susceptible to diagenetic alteration than high-magnesium calcite and aragonite; and (3) shell secretion in isotopic equilibrium with ambient seawater (Lowenstam 1961). The disequilibrium effects of shell fractionation have been very well investigated since Lowenstam (1961), and it is understood in detail which taxa and shell portions reliably record past ocean environments with the highest fidelity (Auclair et al. 2003; Yamamoto et al. 2010a, 2010b, 2011, 2013; Cusack et al. 2012; Takayanagi et al. 2012, 2013, 2015).

To evaluate the preservation of brachiopod shells, the following three screening criteria have been commonly applied (Ullmann and Korte 2015): (1) minor element (manganese (Mn), iron (Fe), and strontium (Sr)) concentrations, which vary with diagenetic alteration; (2) luminescence/non-luminescence with “cold cathode”; and (3) shell microstructures observed via scanning electron microscope (SEM). Well-preserved shells or shell portions are non-luminescent, having high Sr and low Mn and Fe concentrations and well-preserved microstructure.

However, these techniques do not work effectively in all cases. It has been shown that there might be no significant difference in the isotopic composition between well-preserved and diagenetically altered shells determined using the minor element concentrations (e.g., Bruckschen et al. 1999; Korte et al. 2005). Some unaltered modern brachiopod shells show bright cathodoluminescence (CL) images (e.g., Barbin and Gaspard 1995; Griesshaber et al. 2007), which is usually taken to be an indication of diagenetic alteration. The δ¹³C and δ¹⁸O values of luminescent and non-luminescent shells might overlap or be similar in range (Rush and Chafetz 1990; Mii et al. 1997, 2013; Yamamoto et al. 2017). Therefore, applying multiple criteria to assess the diagenetic alteration and cross-checking them are required to distinguish between diagenetically altered and unaltered brachiopod shells (Yamamoto et al. 2017).

The isotopic and minor element signatures in carbonate minerals vary with the isotopic and chemical composition of the ambient pore water (Wenzel 2000; Ullmann et al. 2014), which can be marine, meteoric, or hydrothermal water, or various mixtures of these. This variation implies that diagenetic modification of the chemical and isotopic composition in brachiopod shells embedded in carbonate deposits is likely to be different from that in siliciclastic and volcaniclastic deposits. Cathodoluminescence images of brachiopod carbonates might not be useful for detecting diagenetic alteration where sources of Mn (the activator for CL) are very limited. There are some instances showing that the δ¹³C and δ¹⁸O values or the preservation of shell microstructure are unrelated with CL intensity (Yamamoto et al. 2017).

To better understand the impact of diagenetic modification on the isotopic and chemical composition of brachiopod shells in siliciclastic deposits, we examined the δ¹³C and δ¹⁸O values, minor element concentrations, CL images, and shell microstructure of fossil brachiopod shells of two species, Terebratalia coreanica (Adams and Reeve 1850) (Family Terebrataliidae Richardson 1975) and Laqueus rubellus (Sowerby 1846) (Family Laqueidae Thomson 1927) collected from the conglomerates of the Pliocene Tentokuji Formation in Minehama, northeastern Japan. We compared these shells with modern shells from Otsuchi Bay (northeastern Japan) and Sagami Bay (central Japan). Terebratalia coreanica is generally found at water depths of 13–287 m in the northwestern Pacific Ocean (Hokkaido, Tsugaru Strait to Boso Peninsula on the Pacific side, Korean Strait north to Tsugaru Strait on the Japan Sea side, Japan; Tsingtao, China; Korea; and the Gulf of Tartary, Russia) (Hatai 1936a, 1936b, 1938; Logan 2007). Laqueus rubellus inhabits water depths of 44–907 m in the Pacific Ocean (the seas around Japan; the Japan Sea; Pailolo channel, Hawaii) (Dall 1920; Hatai 1936a; Logan 2007). The wide spatiotemporal distribution of these species means that their δ¹³C and δ¹⁸O values are potentially useful for reconstructing the paleoceanographic environments of lower sublittoral to upper bathyal zones during the late Mesozoic to Cenozoic.

Materials and sample sites

Modern specimens

Terebratalia coreanica from Otsuchi Bay

Modern T. coreanica (OTTCo; Fig. 1a) were collected by dredging at 70 m water depth in Otsuchi Bay (39° 21.86′ N, 141° 59.97′ E) along the Sanriku Ria Coast in northeastern Japan, using the research vessel Yayoi of the International Coastal Research Center, Atmosphere and Ocean Research Institute (ICRS/AORI), The University of Tokyo, on 12 September 2005. High-resolution, three-dimensional δ¹³C and δ¹⁸O values of this species have been published by Yamamoto et al. (2011, 2013). The δ¹³C and δ¹⁸O values of calcite precipitated in isotopic equilibrium with ambient seawater (equilibrium calcite) at the brachiopod collection site ranged from 1.4 to 2.6‰ (δ¹³C_EC) and from − 1.4 to 2.4‰ (δ¹⁸O_EC) (Yamamoto et al. 2011, 2013). The δ¹³C_EC values were calculated by using the δ¹³C values of inorganic carbon (DIC) at the brachiopod collection site (Otsuchi Bay), the difference between δ¹³C_HCO3− and ¹³C_DIC values (0.2‰, Grossman 1984, Romanek et al. 1992, Zhang et al. 1995), and the calcite-HCO₃⁻ enrichment factor (1.0 ± 0.20‰, Romanek et al. 1992) (Additional file 7: Doc. S1). The δ¹⁸O_EC values were estimated based on the equation given by Friedman and O’Neil (1977) (Additional file 7: Doc. S1).

Fig. 1

Photographs of the Terebratalia coreanica and Laqueus rubellus shells examined in this study. The modern T. coreanica shells were collected from Otsuchi Bay, northeastern Japan, and L. rubellus shells were collected from Sagami Bay, central Japan. The fossil shells are from the Pliocene Tentokuji Formation in Minehama, northeastern Japan. a Modern T. coreanica. b–f Fossil T. coreanica. g Modern L. rubellus. h–l Fossil L. rubellus. Only the ventral valves were examined in this study

Laqueus rubellus from Sagami Bay

Modern L. rubellus (SGLR; Fig. 1g) were collected by dredging at 80 m water depth at the eastern margin of Sagami Bay, central Japan (35° 08.17′ N, 139° 34.86′ E), during a cruise of the fishing vessel Rinkai-maru, based at the Misaki Marine Biological Station, The University of Tokyo, on 16 June 2005. Yamamoto et al. (2010a) have presented high-resolution, three-dimensional δ¹³C and δ¹⁸O values of this species. The δ¹³C_EC values ranged from 1.6 to 2.6‰, and δ¹⁸O_EC ranged from − 0.6 to 0.8‰ (Yamamoto et al. 2010a). These values were calculated by the same method as described above.

Fossil shells

Fossil T. coreanica (MNTCo1–MNTCo5; Fig. 1b–f) and L. rubellus (MNLR1–MNLR5; Fig. 1h–l) were collected from sandy conglomerates of the upper part of the Pliocene Tentokuji Formation exposed at Minehama (MH1, Amano et al. 2011), Happo-cho, Akita Prefecture, northeastern Japan (Fig. 2; 40° 17.18′ N, 140° 4.52′ E). The horizon with the fossil brachiopod shells is dated 3.85–2.75 Ma, owing to its location between the Upper Pliocene datum planes (datum planes 21 and A) (Sato et al. 2003), which represent ages of 3.85 Ma and 2.75 Ma (Sato and Kameo 1996).

Fig. 2

Location and geologic map of the sample site. a Akita Prefecture, northeastern Japan. b Enlarged view of the square in a. c Geologic map of the Minehama area (modified from Sato et al. 2003). The red star indicates the site at which the fossil Terebratalia coreanica and Laqueus rubellus shells were collected

The Upper Miocene to Pleistocene succession around Minehama consists of the Onnagawa, Funakawa, Tentokuji, and Sasaoka formations in ascending order (Fig. 2c; Sato et al. 2003). The Onnagawa Formation is composed mainly of hard mudstone (generally termed “hard shale”) conformably overlain by the Funakawa Formation, which comprises mostly mudstone and pumice tuff. The Tentokuji Formation, conformably overlying the Funakawa Formation, consists mainly of dark bluish-gray siltstone alternating with conglomerates in the middle-upper intervals of the formation. The alternating beds are regarded as a turbidite sequence (Sato et al. 2003). The conglomerates yielding the brachiopod shells analyzed in the present study are 5–10 cm thick and consist of rounded to sub-rounded, granule to pebble-sized gravel of andesite and, to a lesser extent, pumice, with a matrix of coarse-grained volcaniclastic sand. The Sasaoka Formation conformably overlies the Tentokuji Formation and consists of poorly sorted, very coarse to coarse-grained sandstone. Bivalves that co-occur with the brachiopods indicate a lower sublittoral zone depositional environment (Higo et al. 1999; Amano et al. 2011).

Methods/Experimental

High-resolution microsampling

The δ¹³C and δ¹⁸O values of the modern brachiopod shells display significant intra-shell, intra-species, and inter-species variations, and the isotope composition is not in all cases within the range of values of equilibrium calcite (Auclair et al. 2003; Yamamoto et al. 2010a, 2010b, 2011, 2013; Cusack et al. 2012; Takayanagi et al. 2012, 2013, 2015). Therefore, specific shell portions, for which the δ¹³C and δ¹⁸O values are identical to the δ¹³C_EC and δ¹⁸O_EC values, respectively, must be selected when using the isotopic signatures as paleoenvironmental proxies. Therefore, in recent studies, samples have been extracted from spot locations on the inner shell surfaces in vertical sections along the maximum growth (longitudinal) axis (e.g., Angiolini et al. 2007; Garbelli et al. 2014, 2016; Takizawa et al. 2017). This procedure yields a relatively narrow range of δ¹³C and δ¹⁸O values because of the minimal fractionation effect in this region of the shell, arising from the very low growth (shell thickening) rate, except in and around specialized shell portions (e.g., muscle scars and the foramen).

Ventral valves of the brachiopod shells were embedded in a resin epoxy matrix and vertically cut along the maximum growth axis (Fig. 3) with a 0.4 mm thick IsoMet low-speed saw (Buehler, Lake Bluff). The samples for isotopic and minor element analyses (~ 0.1 mg for each analysis) were obtained from locations within 0.5 mm of the inner shell surface at 1 mm intervals (red arrows in Fig. 3) in a thick (> 1 mm) shell portion using a 0.2 mm diameter drill and at 2 mm intervals (blue arrows in Fig. 3) in a thin (< 1 mm) shell portion with a 0.10–0.12 mm diameter drill. The sampled spots were ~ 0.6–2.3 mm in length and less than 0.5 mm wide. We used a diamond-coated drill bit for extracting samples to prevent contamination by iron present in the drill bits.

Fig. 3

Schematic illustration of the sampling procedure for geochemical analyses of the modern and fossil shells. The red and blue arrows indicate sampling intervals of 1 and 2 mm, respectively. Shells were cut along the maximum growth line and then along a line perpendicular to the maximum growth axis. Thin sections along the maximum growth axis were examined in CL images. SEM observations were carried out on the cross-section along the line perpendicular to the maximum growth axis. Individual shells were divided into three intervals (intervals 1–3) based on the shell morphology

Individual shells were divided into three intervals based on the shell morphology: interval 1 corresponding to a thick posterior shell region, interval 2 located between intervals 1 and 3, and interval 3 denoting the thin anterior shell region (Fig. 3). This division is identical to that proposed for Kikaithyris hanzawai shells (Takizawa et al. 2017). The boundary between intervals 1 and 2 in both species corresponded to a convex portion located in the center of ventral valves, which was identical to the “ventral median ridge (VMR)” reported from the ventral valves of K. hanzawai (Endo 1989), and hereafter, we use this term. The VMR was located 12.5–16.3 mm from the posterior shell edge in T. coreanica and 13.2–14.6 mm in L. rubellus. The boundary between intervals 2 and 3 was a region where the shell thickness became less than 1 mm. This region was located 4.6–12.8 and 3.8–10.2 mm from the VMR toward the anterior shell edge in T. coreanica shells and L. rubellus shells, respectively.

Powder samples were extracted from calcite cement in an internal cavity (7 mm × 4 mm in diameter) of a fossil brachiopod shell using the same method as described above. This brachiopod shell was small (a distance from the posterior shell edge to the anterior shell edge of < 1 cm) and was not used for geochemical analyses in this study. The inner shell surface of the fossil brachiopod was covered thinly with blocky equant calcites with undulated protuberances (~ 2 mm long and ~ 1 mm in diameter) showing a helictite-like appearance. This is the largest cement found in this study. Based on its location and morphology, this calcite cement is considered to have been formed by meteoric diagenesis. The intergranular space of the sandstone matrix was filled with sub-millimeter-sized, blocky cements. No sign of hydrothermal activity was found around the brachiopod sampling site (Minehama) or within the surrounding region.

SEM, energy dispersive spectrometer (EDS), and CL observations

After the shells had been micro-sampled for isotopic and minor element analyses, they were cut parallel to the maximum growth axis and made into thin sections (Fig. 3). The spatial offset between the geochemical sampling and the CL sections were less than 0.5 mm. CL images of the thin sections were examined using a Relion Industries Reliotron VII cold cathodoluminescence microscope mounted on a polarizing microscope at the Department of Earth Sciences, Environmental and Resources Engineering, Waseda University, Japan. The beam current was ~ 0.06–0.2 mA, and the voltage was 10–13 kV. The beam diameters were ~ 7 mm for the observation of modern shells and ~ 1.5 cm for the fossil shells. We measured the mean CL intensity (MCLI) in the corresponding areas (~ 400 μm × 1000 μm and 200 μm × 2000 μm in thick and thin shell portions, respectively) of the isotopic and minor element sampling locations to quantify the degree of luminescence. The CL images represent color information for each dot as a mixture of R (red), G (green), and B (blue), each represented in 256 gradations from 0 to 255. The MCLI was defined as the mean R values of a CL image. The R values were measured using ImageJ software developed at the National Institutes of Health, USA (ver. 1.59 m9 with Java 1.8.0_101).

After making the thin sections for CL observations, the shells were cut transversely (perpendicular to the maximum growth axis; Fig. 3) along a transect located ~ 13–35 mm from the posterior shell edge. The cut surface was etched with 0.2 mol/L hydrochloric acid (HCl) for 30 s and then coated with carbon. The surface was then observed using a Keyence VE-8800 SEM at the Institute of Geology and Paleontology, Graduate School of Science, Tohoku University (IGPS), with an accelerating voltage of 10 kV. The secondary shell layer of brachiopods is composed of calcite fibers that extend along the maximum growth axis. During diagenesis, these calcite fibers are dissolved and replaced with blocky calcite. The altered fibers become amalgamated, and their outlines become indistinct (Samtleben et al. 2001; Yamamoto et al. 2017). Traditionally, longitudinal sections are used to examine shell microstructure (e.g., Bates and Brand 1991; Qing and Veizer 1994; Azmy et al. 1998; Korte et al. 2005), but this diagenetic alteration of fibers can be observed more clearly in transverse sections, which enables a better determination to be made of whether or not the fibers retained their initial forms (e.g., Samtleben et al. 2001; Garbelli et al. 2014; Takizawa et al. 2017; Yamamoto et al. 2017).

Chemical microanalyses were performed using an Oxford Instruments energy dispersive spectrometer (EDS) installed in a JEOL JSM-7001F field emission SEM (FE-SEM) at the IGPS. The constant accelerating voltage for the EDS analysis was 15 kV and the probe current 1.47 A.

We compared the microstructures of the modern and fossil brachiopod shells and defined two divisions of preservation state: well-preserved (Fig. 4e, f) and diagenetically altered (Fig. 4g, h). In the well-preserved shell portions, the outlines of the calcite fibers were well defined, and each fiber was plainly discernible. In the altered portions, the individual fibers were unobservable owing to their amalgamation. We used two indices to quantify the preservation state of shell microstructure: altered fiber ratio (AFR) and altered fiber and puncta-filling cement ratio (AF-PCR). The former is defined as a ratio (%) of the area of altered fibers to the total area of fibers observed in a particular area (200 μm × 250 μm (magnification = × 400) or 212 μm × 266 μm (magnification = × 600)). The AF-PCR, which corresponds to the AFR of Yamamoto et al. (2017), represents a ratio (%) of the area occupied by altered fibers plus puncta-filling cement to the total area of the fibers where we measured the AFR. We used the two indices because the impact of puncta-filling cement on the diagenetic modification of the isotopic compositions and minor element concentrations of the fossil shells is not negligible and needs to be properly evaluated. We did not measure these indices for interval 1, as the isotopic compositions for this interval cannot be used for paleoenvironmental analysis, as discussed below.

Fig. 4

SEM and CL images of shell microstructure observed in the studied brachiopod shells. a SEM image of modern Terebratalia coreanica (OTTCo). b CL image of modern T. coreanica (OTTCo)*. c SEM image of modern Laqueus rubellus shell (SGLR). d CL image of modern L. rubellus shell (SGLR). e SEM image of well-preserved shell microstructure of fossil T. coreanica (MNTCo3). f CL image in the shell portion adjacent to e, with luminescent punctae. g SEM image of altered shell microstructure (surrounded by dotted lines) in a fossil L. rubellus (MNLR2) shell. h CL image of the shell portion adjacent to g, showing bright luminescence. i SEM image of cement precipitated in the inter-fiber spaces of fossil T. coreanica (MNTCo1). Image brightness and contrast were adjusted with Adobe Photoshop software (Adobe Systems Incorporated, San Jose, USA) so that the luminescent bands become clearly visible

Isotopic analysis

Stable carbon and oxygen isotope analyses of carbonate powder samples were performed using a Thermo Fisher Delta V Advantage isotope ratio mass spectrometer coupled to a ThermoQuest Kiel-III automated carbonate device, at the IGPS. The samples (~ 0.1 mg) were reacted with 100% phosphoric acid at ~ 72 °C. The isotope ratios were expressed in conventional (δ‰) notation and calibrated to the NBS-19 international standard relative to the Vienna Pee Dee Belemnite (VPDB). The external precision (1σ) based on replicate measurements (n = 91) of the laboratory reference material (JCt-1; Okai et al. 2004) was 0.02‰ for the carbon isotope analysis and 0.05‰ for oxygen isotope analysis.

It is known that the δ¹⁸O values for magnesian calcite (Mg-calcite) are higher than those for pure calcite formed under the same conditions (e.g., Tarutani et al. 1969). However, if a fractionation correction for Mg-calcite relative to pure calcite (0.06‰ per mol% MgCO₃, according to Tarutani et al. 1969, 0.17‰ per mol% MgCO₃, according to Jiménez-López et al. 2004) is applied to δ¹⁸O values of the studied shells, this effect is negligible or limited due to their low Mg contents (< 1.18 mol% MgCO₃). Therefore, the Mg effect on the δ¹⁸O values of brachiopod shells (Brand et al. 2013) does not need to be accounted for in this study.

Minor element analysis

Powdered brachiopod shell samples (20.5–189.8 μg) were dissolved in ~ 5.0–10.0 mL of volume-specific 2% (v/v) nitric acid (HNO₃). The volume of the acid was varied to adjust the Ca concentration in the solutions to ~ 3–4 ppm for most analyses. Concentrations of the three minor elements (Mn, Fe, and Sr) were analyzed using an Agilent 7700x inductively coupled plasma mass spectrometer at the Department of Earth and Environmental Sciences, Nagoya University, Japan. Minor elemental concentrations are expressed as ppm (= μg/g). The precision of the analytical method, expressed as the relative standard deviation (σ) of repeated analyses of the laboratory reference material (JCt-1; Okai et al. 2004), was 1.8–4.6% for Mn, 10.8–11.9% for Fe, and 2.2–2.4% for Sr.

Results

All analytical data are given in Additional file 1: Table S1, Additional file 2: Table S2, Additional file 3: Table S3, and Additional file 4: Table S4.

SEM, EDS, and CL observations

Modern shells

The shells of modern T. coreanica (OTTCo) and L. rubellus (SGLR) consisted of two calcite layers: the thinner outer primary shell layer and a thicker inner secondary shell layer (Fig. 4a, c). The primary shell layers of both brachiopod species displayed a fine granular structure measuring up to 30 μm thick in T. coreanica and up to 25 μm in L. rubellus. The secondary shell layer accounted for > 90% of the shell thickness and was composed of thin elongated calcite fibers oriented parallel or slightly oblique to the maximum shell growth axis. The two shell layers were penetrated with punctae perpendicular to the outer and inner shell surfaces. The calcite fibers within the secondary shell layer were clearly identifiable. We regarded this state as the initial fiber structure (Fig. 4a, c).

The modern shells of the two species exhibited intrinsic blue luminescence (Fig. 4b, d; Additional file 5: Figure S1-1b; Additional file 6: Figure S2–1b,) caused by a lack of Mn-activated visible luminescence and by crystal imperfections (Sippel and Glover 1965; Machel et al. 1991; Habermann et al. 1998). Here, we term this blue luminescence as “non-luminescence.”

Fossil shells

The primary shell layer, which is commonly lost in fossil shells (Angiolini et al. 2007; Garbelli et al. 2014, 2017), was preserved in all fossil specimens (MNTCo1–MNTCo5 and MNLR1–MNLR5). The secondary shell layer of the fossil brachiopods consisted of calcite fibers (Fig. 4e) similar to those of the modern shells. However, traces of the dissolution and amalgamation of calcite fibers (Fig. 4g) were identified in many shell portions. The punctae (< 10 μm in diameter), perforations penetrating brachiopod shells, were scattered, commonly infilled with calcite cements. The abundance of altered structures was highly variable among the shell portions, with AFR values ranging from 0 to 92% and AF-PCR values from 1 to 95%.

Some parts of the fossil shells exhibited bright luminescence in CL images (Fig. 4f, h), with MCLI values ranging from 18 to 213 in T. coreanica and from 12 to 216 in L. rubellus. The strongly luminescent shell portions were also diagenetically altered, as indicated by the traces of dissolution and amalgamation of calcite fibers observed via SEM and the high AFR and AF-PCR values. The fossil shells had luminescent dots and lines (< 15 μm in diameter), each corresponding to a calcite cement filling a puncta (Fig. 5).

Fig. 5

CL images of puncta-filling cements. CL images of the vertical sections through the studied brachiopod shells (a Terebratalia coreanica, MNTCo2; b Laqueus rubellus, MNLR2). Note that strongly luminescent dots and lines each corresponding to a calcite cement filling a puncta

Submicron-sized, subhedral, blocky, isopachous fringing calcite cements were found in the inter-fiber spaces of some fossil brachiopod shells with SEM (Fig. 4i). These cements formed in a narrow space between the calcite fibers. The remnants of organic sheaths surrounding rarely occurred. EDS analysis indicated that the Mn concentrations of these cements were as high as 22,000 ppm and their Fe concentrations up to 11,000 ppm, although the concentrations were highly variable. In contrast, Sr was not detected in such cements.

Isotopic composition

Modern shells

The δ¹³C values of samples from the inner shell surfaces of modern T. coreanica from Otsuchi Bay (OTTCo) ranged from − 0.15 to 1.38‰ and from − 0.12 to 1.75‰ for δ¹⁸O (Table 1, Fig. 6a, and Additional file 5: Figure S1-1). Interval 1 exhibited wider ranges of isotopic values than did intervals 2 and 3, containing two local δ¹³C minima and four local δ¹⁸O minima (Fig. 6a, and Additional file 5: Figure S1-1). The δ¹³C profile in this interval consisted of lower values near the posterior shell edge and higher values near the VMR, which were separated by a local minimum δ¹³C value between the two sections. The δ¹³C values in the lower-value section varied within a narrow range of 0.16 to 0.51‰. In contrast, the δ¹³C values in the higher-value section exhibited large-amplitude fluctuations within a range of − 0.15 to 1.38‰. The δ¹⁸O profile exhibited large-amplitude fluctuations throughout interval 1. The δ¹³C and δ¹⁸O profiles in intervals 2 and 3 showed small-amplitude fluctuations with values from 0.65 to 1.36‰ for the δ¹³C values and from 0.93 to 1.75‰ for the δ¹⁸O values. Although the δ¹⁸O values in intervals 1–3 were within the range of equilibrium calcite, the δ¹³C values were not (Additional file 5: Figure S1-1).

Table 1 Modern and fossil Terebratalia coreanica and Laqueus rubellus δ¹³C and δ¹⁸O values and minor-element concentrations

Fig. 6

Profiles of geochemical, microstructural, and optical signals in the studied brachiopod. These panels compare the δ¹³C and δ¹⁸O values, minor element concentrations, mean CL intensity (MCLI), altered fiber ratio (AFR), and altered fiber and puncta-filling cement ratio (AF-PCR) between the modern (OTTCo and SGLR) and fossil (MNTCo1–MNTCo5 and MNLR1–MNLR5) shells. a Terebratalia coreanica shells. b Laqueus rubellus shells. The geochemical profiles of each sample were fitted to the average length of interval 2 (7.74 mm for T. coreanica and 5.27 mm for L. rubellus)

The δ¹³C and δ¹⁸O values of samples from modern L. rubellus from Sagami Bay (SGLR) covaried (r = 0.57, n = 27, p < 0.01), ranging from 0.52 to 1.96‰ and from 0.26 to 0.94‰, respectively (Table 1, Fig. 6b, and Additional file 6: Figure S2-1). The δ¹³C profile showed a gradual increase from 0.52 to 1.57‰ toward the VMR in interval 1, followed by a plateau ranging from 1.49 to 1.94‰ with an average value of 1.68‰ in interval 2 and a range of 1.28 to 1.96‰ (average 1.56‰) in interval 3 (Fig. 6b). The δ¹⁸O values varied within a narrow range from 0.26 to 0.94‰ (average 0.62‰) throughout intervals 1–3. The δ¹⁸O profile showed a gradually increasing trend in interval 1 and a decreasing trend in interval 3. The δ¹³C values in interval 1 were lower than those of equilibrium calcite, and the δ¹³C values in intervals 2 and 3 were slightly lower than or identical to the lowest values of equilibrium calcite (Additional file 6: Figure S2-1). The δ¹⁸O values fell mostly in the upper half of the δ¹⁸O range of equilibrium calcite.

Fossil shells

The δ¹³C and δ¹⁸O values of the five fossil T. coreanica (MNTCo1–MNTCo5) shells fell in wide ranges from − 3.31 to 2.43‰ for δ¹³C and from − 0.42 to 2.18‰ for δ¹⁸O (Table 1, Fig. 6a, and Additional file 5: Figures S1-2–6). The δ¹³C profiles from interval 1 of the fossil shells had a wider range (2.41–4.37‰) than that of the modern shell (1.53‰). These δ¹³C profiles for interval 1 generally followed the pattern of a lower-value section near the posterior shell edge and a higher-value section near the VMR. These two sections were separated by a local minimum δ¹³C value, with an overall increasing trend toward the VMR. This was the same pattern seen in the modern shell. However, the higher-value section near the VMR of interval 1 from two shells (MNTCo2 and MNTCo5; Additional file 5: Figure S1-3, 6) was associated with a local minimum in δ¹³C values. The δ¹³C profiles from intervals 2 and 3 of the fossil shells exhibited a different trend from the same intervals of the modern shell. Shells MNTCo1 and MNTCo2 both showed a high-value plateau extending over interval 2. The δ¹³C profile of MNTCo3 was similar to those of MNTCo1 and MNTCo2 but exhibited a local minimum in this interval (Additional file 5: Figures S1-2–4). Interval 2 of shells MNTCo4 and MNTCo5 exhibited a short high-value plateau near the VMR, followed by a marked decrease toward the anterior shell edge. The δ¹³C values in interval 3 were lower than those of the high-value plateau in interval 2. Shell MNTCo1 had a local δ¹³C minimum in this interval, and shell MNTCo3 had two local minima (Additional file 5: Figures S1-2, 4).

δ¹⁸O profiles of intervals 1 and 2 of the fossil T. coreanica were similar to those of the modern shell. Interval 1 had larger δ¹⁸O amplitudes than did intervals 2 and 3 as a result of a series of erratic, negative δ¹⁸O spikes. The δ¹⁸O values were more consistent in interval 2 than in intervals 1 and 3. Each profile from interval 2 fell in a narrow range (0.07–1.22‰), and the total range of the δ¹⁸O values from the five shells was only 1.44‰. The δ¹⁸O values in interval 3 of the three fossil shells (MNTCo2, MNTCo4, and MNTCo5) fell in a narrow range (0.28–0.57‰), as did the modern shell (0.82‰). The other two fossil shells (MNTCo1 and MNTCo3) exhibited distinct negative spikes in the δ¹⁸O profiles in interval 3 (one in MNTCo1 and two in MNTCo3; Additional file 5: Figure S1-2, 4).

The δ¹³C values of samples from fossil L. rubellus shells (MNLR1–MNLR5) ranged from − 3.58 to 2.42‰, and δ¹⁸O ranged from 0.03 to 2.37‰ (Table 1, Fig. 6b, and Additional file 6: Figures S2-2–6). The δ¹³C values in interval 1 were highly variable, and each profile had a wider range (3.07–5.47‰, average 3.71‰) compared with the modern shell (1.05‰, Fig. 6b). The gradually increasing trend in δ¹³C toward the VMR in this interval was recognized in the fossil shells, but it was not as obvious as in the modern shell. The δ¹³C profiles showed three trends in interval 2: relatively constant (MNLR1 and MNLR3), increasing toward the anterior (MNLR4), and decreasing toward the anterior (MNLR2 and MNLR5). The δ¹³C values were again highly variable in interval 3 without any common trend among the fossil shells, and almost all values were lower than those of the modern shell except for one (MNLR1).

The δ¹⁸O values of fossil L. rubellus shells were less varied than the δ¹³C values throughout the three intervals. Each profile exhibited a narrow range (1.23–1.81‰), and the total range of the δ¹⁸O values from the five shells was < 2.34‰.

Calcite cement

The δ¹³C and δ¹⁸O values of the calcite cement sample were much lower than those of the fossil shells. The mean values from eight repeated measurements were − 17.87‰ (σ = 0.12‰) for δ¹³C and − 6.62‰ (σ = 0.09‰) for δ¹⁸O (Fig. 7).

Fig. 7

Cross-plots of δ¹³C versus δ¹⁸O values of the studied brachiopod shells. The samples were extracted from the modern and fossil T. coreanica and L. rubellus shells. The cement δ¹³C versus δ¹⁸O values are also plotted

Minor elements

Modern shells

The minor element concentrations (n = 30) in the modern T. coreanica shell fell within the ranges of 2–10 ppm (mean = 5 ppm, σ = 3 ppm, median = 4 ppm) for Mn, 30–1100 ppm (mean = 220 ppm, σ = 260 ppm, median = 100 ppm) for Fe, and 810–1100 ppm (mean = 980 ppm, σ = 80 ppm, median = 980 ppm) for Sr (Table 1, Fig. 6a, and Additional file 5: Figure S1-1). The profiles of these elements showed no systematic variations.

In the modern L. rubellus shell, minor element concentrations (n = 27) ranged from 4 to 30 ppm (mean = 10 ppm, σ = 6 ppm, median = 8 ppm) for Mn, 70 to 10,100 ppm (mean = 1000 ppm, σ = 2000 ppm, median = 510 ppm) for Fe, and 800–1100 ppm (mean = 940 ppm, σ = 90 ppm, median = 920 ppm) for Sr (Table 1, Fig. 6b, and Additional file 6: Figure S2-1) and also showed no systematic variations.

Fossil shells

The minor element concentrations (n = 127) in the fossil T. coreanica (MNTCo1–MNTCo5) shells varied between 80 and 3300 ppm (mean = 580–1100 ppm, σ = 470–800 ppm, median = 280–740 ppm) for Mn, 200 and 6900 ppm (mean = 1000–2600 ppm, σ = 930–1700 ppm, median = 600–2200 ppm) for Fe, and 730 and 1100 ppm (mean = 870–930 ppm, σ = 50–80 ppm, median = 860–920 ppm) for Sr (Table 1, Fig. 6a, and Additional file 5: Figures S1-2–6).

In the fossil L. rubellus shells, minor element concentrations (n = 103) of (MNLR1–MNLR5) fell in the ranges of 120–2700 ppm (mean = 400–1000 ppm, σ = 200–830 ppm, median = 370–910 ppm) for Mn, 180–5800 ppm (mean = 1300–1900 ppm, σ = 830–1800 ppm, median = 670–1500 ppm) for Fe, and 730–1000 ppm (mean = 840–960 ppm, σ = 40–70 ppm, median = 830–970 ppm) for Sr (Table 1, Fig. 6b, and Additional file 6: Figures S2-2–6).

Roughly speaking, the Mn and Fe concentrations in the fossil shells of both species decreased from the posterior shell edge toward the VMR and increased toward the anterior shell edge. In contrast, the Sr concentrations in the fossil shells (T. coreanica, 730–1100 ppm; L. rubellus, 730–1000 ppm) were comparable with those of the modern shells (T. coreanica, 810–1100 ppm; L. rubellus, 800–1100 ppm). As with the modern shells, no discernable patterns were seen in the Sr profiles.

Calcite cement

The minor element concentrations of the calcite cement sample were 6200 ppm for Mn, 14,400 ppm for Fe, and 440 ppm for Sr. Mn concentration was 6 times higher than the median concentrations of the fossil shells, and the Fe concentration was 7 times higher. The Sr concentration of the cement was lower than that of the fossil shells.

Discussion

The diagenetic modification of isotope compositions

The δ¹³C and δ¹⁸O profiles in interval 1 of the modern T. coreanica shell are characterized by the occurrence of two local δ¹³C minima and four local δ¹⁸O minima. Similar depletions are found in the same interval of the modern L. rubellus shell, although they are not so distinct as those in the T. coreanica shell. Interval 1 includes the region of the shell to which the diductor and adjustor muscles are attached (Shiino and Kitazawa 2010) and a shell cavity where the digestive and reproductive organs are housed (Williams and Rowell 1965). This suggests that the observed δ¹⁸O minima are potentially related to the biological activities of the brachiopods, although we cannot specify the exact mechanism. The δ¹³C values of the muscle scar are up to 2‰ lower than those of the secondary shell layer (Parkinson et al. 2005; Yamamoto et al. 2010a, 2013) due to secondary calcification, which has been reported in the muscle scars, loop, and teeth (Williams 1968). Therefore, it is highly likely that the depletion of the δ¹³C values in interval 1 is caused by precipitation and/or dissolution (Hughes et al. 1988) around the specialized portions.

The δ¹³C amplitude in intervals 2 and 3 of the fossil brachiopods is ~ 7.0–7.5 times greater than those of modern shells, and the δ¹⁸O amplitude is ~ 3 times greater. The δ¹³C and δ¹⁸O values of these intervals in three T. coreanica and four L. rubellus shells display significant positive correlations (T. coreanica: r = 0.89–0.98, p < 0.01, excluding MNTCo2 and MNTCo4; L. rubellus: r = 0.88–0.99, p < 0.01, excluding MNLR1) (Fig. 7). Both δ¹³C and δ¹⁸O trend negatively toward the δ¹³C and δ¹⁸O values of the calcite cement. The δ¹⁸O value of calcite precipitated in oxygen isotope equilibrium with meteoric water in northeastern Japan, including the Minehama region (δ¹⁸O of meteoric water = − 7.7 to − 10‰ VSMOW; Mizota and Kusakabe 1994) at the mean annual temperature (11.2 °C, Japan Meteorological Agency 2018) is estimated to be − 6.62 to − 8.92‰ following Friedman and O’Neil (1977). The δ¹⁸O of the calcite cement (− 6.62‰) falls within this range, indicating that the δ¹³C and δ¹⁸O values of the highest end-members on the cross-plots are close to their initial values and that the isotopic values of brachiopod shells at Minehama decreased toward those of the cement during meteoric diagenesis.

Another explanation for the positive correlation of the δ¹³C and δ¹⁸O values is a kinetic isotopic fractionation effect. Kinetic isotope fractionation during CO₂ hydration and hydroxylation results in an enhanced incorporation of ¹²C and ¹⁶O during precipitation of skeletal carbonates (McConnaughey 1989a, 1989b; McConnaughey et al. 1997; Auclair et al. 2003; Yamamoto et al. 2013; Takayanagi et al. 2015). This effect is closely related to rapid growth rates and high metabolic activity and is typically characterized by a positive correlation between δ¹³C and δ¹⁸O values (or δ¹³C–δ¹³C_EC and δ¹⁸O–δ¹⁸O_EC values). However, it is reported that growth rates along the inner surface (i.e., shell thickening rates) of brachiopod shells, including T. coreanica and L. rubellus, are very slow and that within a single shell, the kinetic effect is negligible there (Yamamoto et al. 2010b, 2013). The δ¹³C and δ¹⁸O values of interval 1 are less strongly correlated than those of intervals 2 and 3 (Fig. 7). This is because the isotopic composition in interval 1 of the fossil shells is influenced by the biogenic effect mentioned above as well as by meteoric diagenetic modification.

The isotopic data indicate that the highest end-members on the cross-plots (Fig. 7) are likely to be closest to the initial shell δ¹³C and δ¹⁸O values. The highest end-members are from samples extracted from the thick portion of interval 2. This portion corresponds to “interval 2” of Pleistocene K. hanzawai shells from the Ryukyu Islands (Takizawa et al. 2017), which was regarded as a good shell portion to use for paleoenvironmental reconstructions because it exhibited relatively small intra-shell and intra-species variations. Therefore, this interval is likely to be a good region of the shells for paleoenvironmental analysis in many brachiopod taxa.

The diagenetic modification of minor element concentrations

Minor element concentrations in carbonate minerals have been used to identify diagenetically altered brachiopod shells or shell portions (e.g., Wenzel 2000; Ullmann et al. 2014). As meteoric water generally contains more Mn and Fe and less Sr compared with seawater, Fe and Mn concentrations increase and Sr concentration decreases in diagenetically altered marine carbonates (Brand and Veizer 1980; Marshall 1992).

The Mn concentrations in the studied modern shells are relatively uniform. Concentrations in T. coreanica fall within a narrow range of 2–12 ppm, and L. rubellus concentrations are between 4 and 30 ppm (Fig. 6a, b). These ranges are comparable with values reported in previous studies of modern brachiopod shells (5–500 ppm, Morrison and Brand 1986; 0–200 ppm, Brand et al. 2003; 4–30 ppm, Lee et al. 2004; 0–70 ppm, Ullmann et al. 2017) and are lower than the common Mn concentrations in modern shells (< 80 ppm) compiled by Brand et al. (2003).

In contrast, the Mn concentration in the fossil T. coreanica and L. rubellus shells is higher than that of the modern shells. The observed range is 80–3300 ppm in the T. coreanica shells and is 120–2700 ppm in the L. rubellus shells. These values exceed the common Mn concentration in modern shells (Brand et al. 2003) and indicate that there are significant intra-shell and intra-species variations in the fossil brachiopod Mn concentrations (Fig. 6). The samples with high Mn concentrations did not correspond to specific regions of the shells but exhibited low δ¹³C and δ¹⁸O values, high Fe concentrations, and strong luminescence (Fig. 6, Additional file 5: Figures S1-2–6, and Additional file 6: Figures S2-2–6). This indicates that the increased Mn concentrations are the result of meteoric diagenetic alteration.

Although some anomalously high Fe concentrations (> 600 ppm) are observed, the range of Fe concentrations in the modern T. coreanica shells ranged from 30 and 490 ppm and from 70 to 570 ppm in L. rubellus (Fig. 6a, b). These values exceed the common Fe concentrations of modern brachiopod shells (< 140 ppm) presented by Brand et al. (2003). This difference is at least partly due to the fact that the Fe concentration threshold reported in Brand et al. (2003) was derived from bulk shell samples, which represent spatially averaged values.

The Fe concentrations in the fossil T. coreanica (200 to 6900 ppm) and L. rubellus (180 to 5800 ppm) shells are much higher than those of the modern shells and the common Fe concentration range (Brand et al. 2003). The samples with high Fe concentrations also exhibited high Mn concentrations, low δ¹³C and δ¹⁸O values, and strong luminescence. As with Mn, anomalously high Fe concentrations also occur in the fossil shells. These anomalously high concentrations are not always associated with any microstructural (dissolution and amalgamation of fibers), optical (strong luminescence on CL images; see below), or geochemical (decreased δ¹³C or δ¹⁸O values or increased Mn concentration) signals of diagenetic alteration. As such, we consider that these anomalously high Fe concentrations are likely not diagenetic products but represent initial concentrations from when the brachiopods were alive. Similar anomalously high Fe concentrations without signs of diagenetic alteration in well-preserved fossil (~ 20 and ~ 70 ka) K. hanzawai shells have also been described by Takizawa et al. (2017).

Unlike Mn and Fe concentrations, Sr concentrations in the modern and fossil brachiopod shells were similar and their ranges overlap (Table 1). This indicates that Sr concentration in these brachiopod shells is not likely to be a good indicator of diagenetic alteration. The Mn and Sr concentrations show a moderate negative correlation (fossil T. coreanica: r = − 0.46, p < 0.01, Fig. 8c; fossil L. rubellus: r = − 0.52, p < 0.01, Fig. 8d). This differs from the result of Jelby et al. (2014), who found that Mn/Ca of diagenetically altered Lower Maastrichtian brachiopod shells was strongly negatively correlated with Sr/Ca (r = − 0.94).

Fig. 8

Cross-plots of minor element (Mn, Fe, and Sr) concentrations in the studied brachiopod shells. The samples were extracted from the modern and fossil T. coreanica (a, c, e) and L. rubellus (b, d, f) shells

The Mn concentrations in the calcite cement were ~ 2–3 times higher than the maximum Mn concentrations in the fossil shells, and Fe concentrations were ~ 2 times higher. This indicates that Mn and Fe concentrations in brachiopod shells at Minehama increased during meteoric diagenesis, approaching those of the cement. The Sr concentrations in the calcite cements were ~ 50% or less of mean Sr concentrations in the fossil shells. This implies that Sr concentrations in brachiopod shells at Minehama decreased during meteoric diagenesis, approaching those of the cement. However, the decrease in the Sr concentrations was relatively more gradual than the increase in the Mn and Fe concentrations. This is the reason why Sr is not sensitive to diagenesis at Minehama and shows a moderate negative correlation with the Mn concentrations. In conclusion, our study shows that of the three minor elements analyzed in this study, Mn concentration is the most reliable indicator of diagenetically altered shells or shell portions.

Diagenetic alteration detected by CL image observations

Cathodoluminescence observations are useful for determining the paragenetic sequences of diagenetic products precipitated from the solutions containing Mn. Altered shell portions are easily detectable by luminescence activated by the presence of Mn. It has been shown that the δ¹³C and δ¹⁸O values of luminescent shell portions are commonly lower than those of non-luminescent portions (e.g., Grossman et al. 1993; Mii et al. 2012).

The modern T. coreanica and L. rubellus shells exhibit faint to weak luminescent bands that are arranged almost parallel to the inner shell surface (Fig. 4b, d; Additional file 5: Figure S1-1b, and Additional file 6: Figure S2-1b). Similar luminescent bands are observed in shells of modern inarticulate brachiopods and in those of other marine invertebrates (Barbin and Gaspard 1995; Barbin 2013; Riechelmann et al. 2016). These bands or lines are the result of biological activity (e.g., bleeding cycles and changes in growth rates) and/or environmental stresses (e.g., input of coastal water, tidal rhythms, and changes in seawater temperature). In the present study, it is difficult to determine the main reason for the difference in the CL intensity between the modern shells of the two species, noting that CL can also be influenced by the measurement conditions (England et al. 2006).

The fossil shells display brighter luminescence than that of the modern shells. No luminescent bands are observed in the fossil shells. The MCLI profiles of the fossil brachiopod shells show statistically significant positive correlations with Mn concentration (T. coreanica: r = 0.73, n = 122, p < 0.01; L. rubellus: r = 0.51, n = 103, p < 0.01) (Figs. 6a, b and 9a–c). The luminescent regions also correspond to the samples with higher Fe concentrations and lower δ¹³C and δ¹⁸O values than those of adjacent non-luminescent regions within an individual shell (Fig. 6a, b), indicating that these shell portions have been altered by meteoric diagenesis.

Fig. 9

Double-logarithmic cross-plots of mean CL intensity versus Mn concentration in the studied fossil shells. a T. coreanica. b L. rubellus. c Combined data from both species

The punctae of the fossil brachiopods are partly to fully filled with cement and display bright luminescence (Fig. 4f, h) as has been described in other species (e.g., Takizawa et al. 2017; Casella et al. 2018). This is because the punctae are filled with organic material in live brachiopods (Williams 1997; Pérez-Huerta et al. 2009), although no such material was observed in the studied modern shells, and void spaces are created during postmortem decomposition, with subsequent precipitation of diagenetic cements (Garbelli et al. 2016). We conclude that the presence of carbonate cements with extremely high Mn and Fe concentrations in the punctae and inter-fiber spaces is one of the major causes of high concentrations of these elements in diagenetically altered shell portions, as discussed below.

The diagenetic alteration of shell microstructure

The secondary shell layer is composed of calcite fibers, each with a concave outer surface and convex inner surface (Gaspard 1991; Samtleben et al. 2001). However, altered fibers become irregularly shaped and amalgamated with an increased diagenetic alteration. The amalgamation of two or more fibers occurs locally in modern brachiopods, but the outline of such fibers remains defined, which means that pristine amalgamated fibers can be distinguished from diagenetically altered fibers (Samtleben et al. 2001).

Altered shell portions (Additional file 5: Figures S1-2–6 and Additional file 6: Figures S2-2–6) are characterized by low δ¹³C and δ¹⁸O values, high Mn and Fe concentrations, and luminescent CL images. In contrast, the well-preserved shell portions exhibit comparatively high δ¹³C and δ¹⁸O values, low Mn and Fe concentrations, and weakly luminescent or non-luminescent CL images (Fig. 4f). AFR and AF-PCR are correlated negatively with δ¹³C and δ¹⁸O but positively with Mn and Fe concentrations and MCLI in the fossil T. coreanica and L. rubellus shells (Fig. 10).

Fig. 10

Cross-plots of isotopic composition versus AFR or AF-PCR (a, b, e, and f) and of isotopic composition versus MCLI (c, d, g, and h) . The gray bar on the y-axis indicates a range of δ¹³C or δ¹⁸O values of the modern brachiopod shell (OTTCo or SGLR)

Some studies have argued that the impact of in-filling cements in punctae upon the chemical and isotopic composition of shells is negligible because their volume accounts for only a small portion of the shell (e.g., < 10% of a Silurian brachiopod, Resseralla elegantula, shells; Samtleben et al. 2001) compared with that of the shell-forming calcite fibers. However, if the isotopic compositions and chemical concentrations of the puncta-filling cements are one or more orders of magnitude greater than those of the shell-forming fibers, the impact of such cements on the geochemistry of spot samples might not be negligible. Here, we assume that the isotopic composition and Mn and Fe concentrations of the puncta-filling cements are identical to those of the cement formed by meteoric diagenesis (δ¹³C = − 18‰, δ¹⁸O = − 7‰, Mn concentration = 6000 ppm, Fe concentration = 14,000 ppm). The δ¹³C values are at least 19‰ lower than those of the fossil shells, and the δ⁸O values are 9‰ lower. The Mn concentrations are at least 5000 ppm higher than those of the fossil shells, and Fe concentrations are 12,000 ppm higher. Under this assumption, if the punctae of the studied modern shells are filled with such cements and they account for 5% of the shell volumes, then the δ¹³C and δ¹⁸O values decrease by ~ 1‰ and ~ 0.4‰, respectively (Table 2). Conversely, the Mn and Fe concentrations increase by ~ 300 and ~ 600 ppm, respectively. In fact, the punctae account for 1–9% of the fossil T. coreanica shell volumes and 1–12% of the L. rubellus shell volumes. These indicate that the isotopic composition and chemical concentration of the puncta-filling cements can potentially significantly bias the isotopic composition and minor element concentrations of spot or bulk shell samples.

Table 2 Effect of puncta-filling cements to the isotopic composition and minor element concentrations of shell samples. We assume that the isotopic composition and minor element concentrations of the puncta-filling cements are identical to those of the cement formed by meteoric diagenesis. Then, we calculate the isotopic composition and minor element concentrations of the mixtures of shell and puncta-filling cements when the punctae are filled with such cements

AF-PCR was correlated more negatively with δ¹³C and δ¹⁸O and more positively with Mn and Fe concentrations and MCLI than was AFR (Table 3, Fig. 9). This suggests that the occurrence of puncta-filling cements is one of the major causes of diagenetic modification of isotopic compositions and minor element concentrations. However, it is not clear whether the puncta-filling cements have similar isotopic compositions and minor element concentrations to cements formed by meteoric diagenesis. The punctae of some modern brachiopod shells are luminescent (e.g., Griesshaber et al. 2007; Barbin 2013), indicating that they are filled with carbonate minerals that are enriched in Mn compared with shell calcite. It is deduced that such puncta-filling cements are chemical and/or microbial precipitates formed in a closed reductive space (e.g., Kaźmierczak and Iryu 1999) or biogenic products. As such, the puncta-filling cements in this study might include carbonate minerals of different origins, with likely differences in their isotopic composition and minor element concentrations.

Table 3 Cross-correlations between geochemical, microstructural and optical signals in the studied fossil brachiopodCross-correlations are calculated between isotope composition (δ¹³C and δ¹⁸O), minor-element (Mn, Fe, and Mn) concentrations, MCLI, and AFR or ΑF-PCR of the fossil Terebratalia coreanica and Laqueus rubellus shells. Coefficient values in brackets are not statistically significant (p > 0.01)

Figure 10 shows the cross-plots of isotopic composition versus AFR/AF-PCR and isotopic composition versus MCLI. The intercepts of these cross-plots can be regarded as the isotopic composition of non-altered brachiopod shells (AFR or AF-PCR = 0, MCLI = 0), theoretically corresponding to the initial isotopic composition of the fossil brachiopod shells. The intercepts are not identical but are close to each other (the difference in δ¹³C values is < 1.1‰ and in δ¹⁸O values is < 0.6‰). The intercepts of the δ¹⁸O values of the fossil T. coreanica shells are up to 0.5‰ greater than the δ¹⁸O values of the modern shell from Otsuchi Bay. Those of L. rubellus are 2‰ greater than the δ¹⁸O values from Sagami Bay. Assuming that the brachiopods occupied similar habitats during the Pliocene to those occupied by their modern counterparts, this indicates the presence of cooler and/or more saline seawater in the Minehama region during the Pliocene compared with that of the modern Otsuchi and Sagami bays. This interpretation is supported by the occurrence of cool-water molluscan species within the outcrop from which the fossil brachiopod shells of this study were collected (MH1; Amano et al. 2011). This suggests that AFR/AF-PCR and MCLI are good indices for quantitatively estimating the extent of diagenetic alteration within brachiopod shells.

The submicron-sized cements in a narrow space in the inter-fiber spaces (Fig. 4i) show the morphological and geochemical features that are common in cements formed during meteoric diagenesis (Tucker and Wright 1990), and the occurrence of such cements is an additional cause of high concentrations of Mn and Fe in diagenetically altered shells.

The cross-correlation coefficients of isotopic composition, minor element concentration, and MCLI and AF-PCR indices indicating the alteration of shell portions are presented in Table 3. The δ¹³C and δ¹⁸O values of the fossil T. coreanica and L. rubellus shells show moderate to strong negative correlations with Mn concentration, MCLI, and AF-PCR. Mn concentration, MCLI, and AF-PCR were positively correlated. The negative correlation of Fe concentration with isotopic composition is weaker than that of Mn concentration because of the occurrence of localized high Fe concentrations that are more anomalous than the high Mn concentrations. Fe concentration shows a weak (although not statistically significant (p > 0.01)) positive correlation with MCLI. Mn and Fe concentrations increase with increased diagenetic alteration and act as an activator (Mn) and a quencher (Fe) for CL images (Machel et al. 1991). The weak positive correlation of Fe concentration with MCLI indicates that the activator (Mn) acts more efficiently as an activator than Fe does as a quencher. Sr concentration was not correlated significantly with any other features or measures. In summary, shell portions with high Mn concentration, high MCLI, and high AF-PCR generally exhibit low δ¹³C and δ¹⁸O values, indicating that these measures are useful for detecting shell portions altered by meteoric diagenesis.

Lithologic control on the diagenetic alteration of brachiopod shells

Our study shows that the four indices commonly used to specify diagenetically altered carbonates—δ¹³C and δ¹⁸O values, minor element concentrations, the preservation state of shell microstructure, and cathodoluminescence—are applicable to brachiopod shells. However, it is known that diagenesis alters the initial paleoenvironmental signals extracted by these indices (Ullmann and Korte 2015). Yamamoto et al. (2017) studied the four indices of fossilized Kikaithyris hanzawai shells from the Pleistocene reef complex deposits (Iryu et al. 2006) and showed that although altered fibers are commonly observed in brachiopod shells, the initial isotopic composition can be largely preserved, and luminescent shells cannot always be regarded as having undergone degradation of shell microstructure or isotopic modification by diagenesis. The findings of those authors indicate that CL images are not always sufficient for detecting diagenetic alteration in shells. This is at least partly because of the differences in the diagenetic processes modifying the chemical and isotopic compositions between brachiopod shells embedded in siliciclastic and volcaniclastic deposits and those in carbonate deposits. The geochemistry of brachiopod shells embedded in siliciclastic and volcaniclastic deposits is affected by formation water relatively rich in Mn, which originates from meteoric water. This can explain why diagenetically altered shells are luminescent in CL images, have low δ¹³C and δ¹⁸O values, and high Mn concentrations. In contrast, carbonate deposits have limited sources of Mn, except when the deposits are associated with siliciclastics (including paleosols) or volcaniclastics. As such, CL images might not necessarily be useful for the detection of diagenetic alteration of shells in carbonate deposits.

The relationship between the degree of preservation of shell microstructures and the initial isotopic composition has been controversial. It is known that some Paleozoic brachiopods, in spite of their well-preserved shell microstructure, have low δ¹³C and δ¹⁸O values, similar to those of diagenetically altered marine carbonate fossils and sparry cements (Rush and Chafetz 1990), or exhibit large variations in isotopic composition caused by a diagenetic alteration (Banner and Kaufman 1994). To explain this conflict, it has been suggested that only large-scale dissolution and subsequent recrystallization of shells can result in isotopic resetting in the per mil range (Wadleigh and Veizer 1992; Qing and Veizer 1994). However, the significantly lower δ¹³C and δ¹⁸O values in intervals 2 and 3 of the fossil brachiopod shells compared with those of the modern shells in the present study indicate that the resetting of isotopic values in the per mil range can occur without large-scale dissolution and subsequent recrystallization.

Shell portions in which microstructures are preserved but their initial isotopic composition has been modified present a significant challenge with respect to reconstructing paleoceanographic conditions. Applying and cross-checking multiple criteria to assess diagenetic alteration are required to distinguish between diagenetically altered and unaltered brachiopod shells (Yamamoto et al. 2017).

Conclusions

To estimate the extent of modification in the isotope composition of fossil brachiopod shells with increasing diagenetic alteration, we analyzed δ¹³C and δ¹⁸O and three commonly applied indicators of meteoric diagenesis: minor element (Mn, Fe, and Sr) concentrations, cathodoluminescence, and shell microstructure. The samples for these analyses were collected from the inner surfaces of fossil T. coreanica and L. rubellus (rhynchonellate brachiopod) shells collected from conglomerates of the Tentokuji Formation (Pliocene, 3.85–2.75 Ma) exposed in northeastern Japan. The results of the study can be summarized as follows:

• (1) Individual shells of T. coreanica and L. rubellus can be divided into three intervals (intervals 1–3) based on the shell morphology, each with characteristic chemical and isotopic compositions.

• (2) Interval 2 is characterized by relatively small intra-shell and intra-species variations in δ¹³C and δ¹⁸O values in both modern and fossil shells. The δ¹³C and δ¹⁸O values from this interval of the fossil shells plots as the highest end-member, and these values are therefore likely to be the closest to the initial shell δ¹³C and δ¹⁸O values.

• (3) Of the three minor elements examined in this study, Mn concentration is the most reliable indicator of diagenetically altered shells. Mn and Fe concentrations in the fossil brachiopod shells are higher than those of the modern shells, displaying large intra-shell and intra-species variations, with several anomalously high localized concentrations. The samples with high Mn and Fe concentrations do not correspond to specific regions of the shells but also exhibit low δ¹³C and δ¹⁸O values and strong luminescence. This indicates that the increased Mn concentrations are the result of meteoric diagenetic alteration. The anomalously high Fe concentrations, which are also observed in modern shells, are not necessarily associated with microstructural (dissolution and amalgamation of fibers), optical (strong luminescence on CL images), or isotopic (decreased δ¹³C or δ¹⁸O values). This indicates that these anomalously high Fe concentrations are likely not diagenetic products but represent initial concentrations from when the brachiopods were alive. Unlike Mn and Fe concentrations, Sr concentrations in the studied modern and fossil brachiopod shells are similar and their ranges overlap, indicating that Sr concentration in carbonate minerals is not always a good indicator for identifying diagenetic alteration.

• (4) We quantified the preservation state of shell microstructure using two indices: altered fiber ratio (AFR) and altered fiber and puncta-filling cement ratio (AF-PCR). To quantify the degree of luminescence, we measured the mean cathodoluminescence index (MCLI) defined as the mean R value in a particular area of a CL image. The δ¹³C and δ¹⁸O values of the fossil brachiopod shells are moderately or strongly negatively correlated with Mn concentration, MCLI, and AF-PCR. Mn concentration, MCLI, and AFR/AF-PCR are positively correlated with each other. Fe concentration shows a weaker negative correlation with isotopic composition than does Mn concentration on account of the occurrence of anomalously high localized Fe concentrations.

Our study provides further evidence that applying and cross-checking multiple criteria to assess diagenetic alteration is the optimal method for distinguishing between diagenetically altered and unaltered brachiopod shells.