Figures and Tables

top

Introduction

Biogenic carbonates are important palaeoclimate archives, providing long term perspectives on climate change and contextualising major biosphere transitions throughout Earth history. The δ¹⁸O composition of skeletal carbonates such as corals, foraminifera, and bivalves are often measured to reconstruct marine palaeotemperatures (e.g., Chen and Watkins, 2025

Chen, S., Watkins, J.M. (2025) Oxygen and Carbon Isotopes in Marine Carbonates: A Biogenic Climate Archive Built Upon Disequilibria. Elements 21, 112–117. https://doi.org/10.2138/gselements.21.2.112

). However, this requires knowledge of the isotopic value of formation water (δ¹⁸O_water). Clumped isotope measurements circumvent the need for a priori δ¹⁸O_water quantities by recording palaeotemperature independently, which can subsequently be used alongside carbonate δ¹⁸O to calculate δ¹⁸O_water empirically (Huntington and Petersen, 2023

Huntington, K.W., Petersen, S.V. (2023) Frontiers of Carbonate Clumped Isotope Thermometry. Annual Review of Earth and Planetary Sciences 51, 611–641. https://doi.org/10.1146/annurev-earth-031621-085949

). In thermodynamic equilibrium, bonding between “heavy” oxygen and carbon isotopes within the carbonate lattice varies exclusively with mineralisation temperature (Ghosh et al., 2006

Ghosh, P., Adkins, J., Affek, H., Balta, B., Guo, W., Schauble, E.A., Schrag, D., Eiler, J.M. (2006) ¹³C–¹⁸O bonds in carbonate minerals: A new kind of paleothermometer. Geochimica et Cosmochimica Acta 70, 1439–1456. https://doi.org/10.1016/j.gca.2005.11.014

; Schauble et al., 2006

Schauble, E.A., Ghosh, P., Eiler, J.M. (2006) Preferential formation of ¹³C–¹⁸O bonds in carbonate minerals, estimated using first-principles lattice dynamics. Geochimica et Cosmochimica Acta 70, 2510–2529. https://doi.org/10.1016/j.gca.2006.02.011

). Therefore, the excess abundance of multiply substituted (“clumped”) isotopologues of CO₂ evolved from CaCO₃ compared to the predicted abundance based on randomly arranged isotopes (Δ₄₇) functions as a well calibrated palaeothermometer (e.g,. Kelson et al., 2017

Kelson, J.R., Huntington, K.W., Schauer, A.J., Saenger, C., Lechler, A.R. (2017) Toward a universal carbonate clumped isotope calibration: Diverse synthesis and preparatory methods suggest a single temperature relationship. Geochimica et Cosmochimica Acta 197, 104–131. https://doi.org/10.1016/j.gca.2016.10.010

; Anderson et al., 2021

Anderson, N.T., Kelson, J.R., Kele, S., Daëron, M., Bonifacie, M., Horita, J., Mackey, T.J., John, C.M., Kluge, T., Petschnig, P., Jost, A.B., Huntington, K.W., Bernasconi, S.M., Bergmann, K.D. (2021) A Unified Clumped Isotope Thermometer Calibration (0.5–1,100°C) Using Carbonate-Based Standardization. Geophysical Research Letters 48, e2020GL092069. https://doi.org/10.1029/2020GL092069

). This has improved records of palaeohydrological dynamics underpinned by reconstructed δ¹⁸O_water (Huntington and Petersen, 2023

Huntington, K.W., Petersen, S.V. (2023) Frontiers of Carbonate Clumped Isotope Thermometry. Annual Review of Earth and Planetary Sciences 51, 611–641. https://doi.org/10.1146/annurev-earth-031621-085949

and refs therein).

Carbonates that originally precipitate out of thermodynamic equilibrium (e.g., corals, brachiopods, coccoliths; Davies et al., 2022

Davies, A.J., Guo, W., Bernecker, M., Tagliavento, M., Raddatz, J., Gischler, E., Floegel, S., Fiebig, J. (2022) Dual clumped isotope thermometry of coral carbonate. Geochimica et Cosmochimica Acta 338, 66–78. https://doi.org/10.1016/j.gca.2022.10.015

, 2023

Davies, A.J., Brand, U., Tagliavento, M., Bitner, M.A., Bajnai, D., Staudigel, P., Bernecker, M., Fiebig, J. (2023) Isotopic disequilibrium in brachiopods disentangled with dual clumped isotope thermometry. Geochimica et Cosmochimica Acta 359, 135–147. https://doi.org/10.1016/j.gca.2023.08.005

; Clark et al., 2025

Clark, A.J., Jaggi, M., Bernasconi, S.M., Fiebig, J., Stoll, H.M. (2025) Clumped Isotope Temperatures of Coccolithophores From Global Sediment Traps. Paleoceanography and Paleoclimatology 40, e2024PA005054. https://doi.org/10.1029/2024PA005054

) violate preconditions for Δ₄₇ palaeothermometry, but can be identified less ambiguously using dual clumped isotope analysis (Δ₄₇ plus Δ₄₈) (e.g., Bajnai et al., 2020

Bajnai, D., Guo, W., Spötl, C., Coplen, T.B., Methner, K., Löffler, N., Krsnik, E., Gischler, E., Hansen, M., Henkel, D., Price, G.D., Raddatz, J., Scholz, D., Fiebig, J. (2020) Dual clumped isotope thermometry resolves kinetic biases in carbonate formation temperatures. Nature Communications 11, 4005. https://doi.org/10.1038/s41467-020-17501-0

; Davies et al., 2022

Davies, A.J., Guo, W., Bernecker, M., Tagliavento, M., Raddatz, J., Gischler, E., Floegel, S., Fiebig, J. (2022) Dual clumped isotope thermometry of coral carbonate. Geochimica et Cosmochimica Acta 338, 66–78. https://doi.org/10.1016/j.gca.2022.10.015

, 2023

Davies, A.J., Brand, U., Tagliavento, M., Bitner, M.A., Bajnai, D., Staudigel, P., Bernecker, M., Fiebig, J. (2023) Isotopic disequilibrium in brachiopods disentangled with dual clumped isotope thermometry. Geochimica et Cosmochimica Acta 359, 135–147. https://doi.org/10.1016/j.gca.2023.08.005

). Because Δ₄₈ is similarly controlled by temperature but concerns different CO₂ isotopologues, a temperature dependent equilibrium between Δ₄₇ and Δ₄₈ exists, which has been constrained by Fiebig et al. (2021

Fiebig, J., Daëron, M., Bernecker, M., Guo, W., Schneider, G., Boch, R., Bernasconi, S.M., Jautzy, J., Dietzel, M. (2021) Calibration of the dual clumped isotope thermometer for carbonates. Geochimica et Cosmochimica Acta 312, 235–256. https://doi.org/10.1016/j.gca.2021.07.012

, 2024)

Fiebig, J., Bernecker, M., Meijer, N., Methner, K., Staudigel, P.T., Davies, A.J., Bayarjargal, L., Spahr, D., Winkler, B., Hofmann, S., Granzin, M., Petersen, S.V. (2024) Carbonate clumped isotope values compromised by nitrate-derived NO₂ interferent. Chemical Geology 670, 122382. https://doi.org/10.1016/j.chemgeo.2024.122382

considering ab initio thermodynamic predictions related to bond energies (Hill et al., 2014

Hill, P.S., Tripati, A.K., Schauble, E.A. (2014) Theoretical constraints on the effects of pH, salinity, and temperature on clumped isotope signatures of dissolved inorganic carbon species and precipitating carbonate minerals. Geochimica et Cosmochimica Acta 125, 610–652. https://doi.org/10.1016/j.gca.2013.06.018

) plus empirical measurements of carbonates that crystallised closest to equilibrium. Moreover, the equilibrium Δ₄₇-Δ₄₈-T relationships for calcite remain valid for other common biomineral CaCO₃ polymorphs, including aragonite (Bernecker et al., 2025

Bernecker, M., Bonifacie, M., Staudigel, P., Meijer, N., Siebert, J., Wehr, N., Haussühl, E., Bernasconi, S.M., Petrash, D.A., Dietzel, M., Fiebig, J. (2025) Effects of mineralogy on Δ₄₇ and Δ₄₈ of carbonate-derived CO₂ below analytical resolution. Geochimica et Cosmochimica Acta 401, 89–103. https://doi.org/10.1016/j.gca.2025.06.004

). With Δ₄₇-Δ₄₈ measurements, disequilibrium signals can provide new insight into biomineralisation processes even if not primary growth temperature.

Bivalve shells are favoured palaeoclimate archives due to frequently good preservation in the fossil record and their ubiquity across aquatic environments. Schlidt et al. (2025)

Schlidt, V., Evans, D., de Winter, N.J., Bernecker, M., Arndt, I., Staudigel, P.T., Davies, A.J., Brand, U., Müller, W., Fiebig, J. (2025) Most bivalves and gastropods calcify indistinguishably from dual clumped isotope equilibrium. Geochimica et Cosmochimica Acta 410, 174–187. https://doi.org/10.1016/j.gca.2025.10.008

measured the Δ₄₇-Δ₄₈ composition of 14 modern marine bivalve species, finding only one analytically distinguishable from Δ₄₇-Δ₄₈ equilibrium. Curley et al. (2023)

Curley, A.N., Petersen, S.V., Edie, S.M., Guo, W. (2023) Biologically driven isotopic fractionations in bivalves: from palaeoenvironmental problem to palaeophysiological proxy. Biological Reviews 98, 1016–1032. https://doi.org/10.1111/brv.12940

, Gey et al. (2024)

Gey, C.J., Pfister, L., Türk, G., Thielen, F., Leonard, L., Schmitt, K.E., Schöne, B.R. (2024) Biologically driven isotope fractionation in ultrastructurally different shell portions of freshwater pearl mussels (Margaritifera margaritifera): Implications for stream water δ¹⁸O reconstructions. Limnology and Oceanography Letters 9, 827–836. https://doi.org/10.1002/lol2.10426

, and Lollos et al. (2025)

Lollos, D.M., Curley, A.N., Petersen, S.V. (2025) Regional variation in biologically driven isotopic fractionations in cosmopolitan bivalve genera Dosinia and Mactra. PALAIOS 40, 321–335. https://doi.org/10.2110/palo.2025.004

recently showed that some bivalves have a heterogeneous isotopic composition in different structural layers, frequently in δ¹⁸O and δ¹³C but only significantly in Δ₄₇ in two fossil species. Schlidt et al. (2025)

Schlidt, V., Evans, D., de Winter, N.J., Bernecker, M., Arndt, I., Staudigel, P.T., Davies, A.J., Brand, U., Müller, W., Fiebig, J. (2025) Most bivalves and gastropods calcify indistinguishably from dual clumped isotope equilibrium. Geochimica et Cosmochimica Acta 410, 174–187. https://doi.org/10.1016/j.gca.2025.10.008

were unable to address mixing effects from isotopically distinct layers as they crushed and homogenised valves wholesale. Lollos et al. (2025)

Lollos, D.M., Curley, A.N., Petersen, S.V. (2025) Regional variation in biologically driven isotopic fractionations in cosmopolitan bivalve genera Dosinia and Mactra. PALAIOS 40, 321–335. https://doi.org/10.2110/palo.2025.004

made the first attempt at measuring dual clumped isotopes in different layers of modern bivalve shells, but problems with standardisation and low replication led to relatively high uncertainty in Δ₄₈. So, to date there have still been no high precision Δ₄₇-Δ₄₈ measurements on the different layers of bivalve shells.

Here we present layer-specific dual clumped isotope measurements of four bivalve species. All specimens were collected live from natural environments with directly monitored water temperatures at two of four locations. Time synchronous sampling of the outer shell layer (OSL) and inner shell layer (ISL) allowed robust comparison of compositions. The Δ₄₇, Δ₄₈, δ¹⁸O, and δ¹³C values were assessed for their fidelity to environmental values and discussed for their implications on future palaeoclimate applications of dual clumped isotopes in bivalves.

top

Methods

Samples. Four bivalve species, aragonitic in both layers, were measured for dual clumped isotopes (Fig. 1): Arctica islandica (North Sea, Norway), Glycymeris bimaculata (Pag Bay, Croatia), Senilia senilis (Atlantic Ocean, Mauritania), and Geloina erosa (brackish mangrove, Indonesia). Specimens were collected live with articulated valves. Shells longer than 3 cm were selected to ensure sufficient sample sizes.

Full size image

We sampled one valve each from two different individuals of these species to assess reproducibility. Each valve was sectioned along the axis of maximum growth using a water cooled saw and cut faces were polished to a ∼15 μm finish to highlight growth banding and the OSL-ISL boundary. The organic-rich periostracum and ligaments were removed before sampling. Samples of ∼100 mg were milled with a hand drill at 3000 rpm, making consistent slow passes over the entire sampling area and taking frequent ∼10–30 s breaks to minimise frictional heating (Fig. 2).

Full size image

Clusters of at least two mid-life growth increments traceable across the OSL-ISL boundary under magnification were selected (Fig. 2a). These bands were milled out in the OSL and ISL separately, ensuring that the same part of ontogeny was represented in both samples from the same shell, but different years may have been sampled in the second shell for that species based on different sizes and band visibility. The thin shell of G. erosa required milling a strip the entire thickness of the ISL to yield 100 mg (Fig. 2b). This represents a lifetime average of material, so an equivalent transect from hinge to commissure of the OSL was drilled externally. It was possible to visually confirm that OSL and ISL material did not mix in cross sectional view, despite coming from different faces. Powders were manually homogenised and aliquoted in 7–9 replicates of 10.0 ± 0.2 mg. The powders remained in a vacuum oven at 30 °C for at least two weeks prior to analysis to prevent isotopic exchange with adsorbed water (Staudigel et al., 2025

Staudigel, P., Bernecker, M., Curley, A., Meijer, N., Bayarjargal, L., Dietzel, M., Fiebig, J. (2025) Isotopic resetting of CO₂ in the presence of carbonate samples: Implications for oxygen- and clumped-isotope analyses. Chemical Geology 694, 123012. https://doi.org/10.1016/j.chemgeo.2025.123012

).

Isotopic analysis. Samples were digested in 108 % phosphoric acid in a common acid bath at 90 °C as described in Fiebig et al. (2019)

Fiebig, J., Bajnai, D., Löffler, N., Methner, K., Krsnik, E., Mulch, A., Hofmann, S. (2019) Combined high-precision Δ₄₈ and Δ₄₇ analysis of carbonates. Chemical Geology 522, 186–191. https://doi.org/10.1016/j.chemgeo.2019.05.019

. Heated (1000 °C) and equilibrated (25 °C) gas standards were run as described in Bernecker et al. (2023)

Bernecker, M., Hofmann, S., Staudigel, P.T., Davies, A.J., Tagliavento, M., Meijer, N., Ballian, A., Fiebig, J. (2023) A robust methodology for triple (Δ₄₇, Δ₄₈, Δ₄₉) clumped isotope analysis of carbonates. Chemical Geology 642, 121803. https://doi.org/10.1016/j.chemgeo.2023.121803

. Mass spectrometric raw data was processed using the D4xgui app (Bernecker et al., 2026

Bernecker, M., Daëron, M., Staudigel, P.T., Hofmann, S., Fiebig, J. (2026) D4Xgui: A tool for baseline correction and standardization of carbonate clumped isotope raw data. SoftwareX 33, 102492. https://doi.org/10.1016/j.softx.2025.102492

), including correction of raw m/z 47–49 intensities for the pressure baseline, data anchoring relative to nominal Δ₄₇ and Δ₄₈ of equilibrated gases (Fiebig et al., 2021

Fiebig, J., Daëron, M., Bernecker, M., Guo, W., Schneider, G., Boch, R., Bernasconi, S.M., Jautzy, J., Dietzel, M. (2021) Calibration of the dual clumped isotope thermometer for carbonates. Geochimica et Cosmochimica Acta 312, 235–256. https://doi.org/10.1016/j.gca.2021.07.012

), and full error propagation considering mass spectrometric measurements and data standardisation. Δ₄₇ and Δ₄₈ are reported in CDES90 and 95 % confidence intervals (CI) are considered for data interpretation. When aggregating data from multiple samples, Monte Carlo simulations (n = 10,000) considering sample means and standard error resulting from the former data processing were used.

Environmental data sets. Environmental temperature data for the sampling locations were compiled from various sources. In all but one case, the duration of these records includes the actual lifetime of the shells based on collection dates. For G. bimaculata, the in situ record begins the year after sample collection, and the low inter-year variability in that record indicates that extrapolation to the recent lifetime of the animal is reasonable.

For A. islandica, monthly mean water temperature data (direct measurements, 5 m depth) were extracted from publicly accessible data from the Ingøya monitoring station of the Norwegian Institute for Marine Research (https://www.imr.no/forskning/forskningsdata/stasjoner/view?station=Ingoy). We compiled data from 2013–2020 (samples collected in 2020) to obtain at least five observations for each month. For G. bimaculata, daily water temperature data (direct measurements, 3–4 m depth) from 2015–2020 (samples collected in 2014) for Pag Bay were contributed by the Institute of Oceanography and Fisheries (Split, Croatia) (e.g., Vilibić et al., 2022

Vilibić, I., Dunić, N., Peharda, M. (2022) Near-surface ocean temperature variations across temporal scales in the coastal eastern Adriatic. Continental Shelf Research 245, 104786. https://doi.org/10.1016/j.csr.2022.104786

). For S. senilis and G. erosa, we used the interpolated data product from the World Ocean Atlas (WOA) 2023 at 0.25° resolution (Locarnini et al., 2024

Locarnini, R.A., Mishonov, A.V., Baranova, O.K., Reagan, J.R., Boyer, T.P., Seidov, D., Wang, Z., Garcia, H.E., Bouchard, C., Cross, S.L., Paver, C.R., Dukhovskoy, D. (2024) World Ocean Atlas 2023, Volume 1: Temperature. NOAA Atlas NESDIS 89. https://doi.org/10.25923/54bh-1613

). Monthly mean temperature data for 0–10 m were extracted from the nearest appropriate grid cell to the sampling location. The 2015–2022 and 2005–2014 data products were used for S. senilis (samples collected in 2021) and G. erosa (samples collected in 2010), respectively.

top

Results

Dual clumped isotopes. All samples but one G. bimaculata OSL were in equilibrium (i.e. the 95 % CI overlapped the equilibrium line of Fiebig et al., 2024

Fiebig, J., Bernecker, M., Meijer, N., Methner, K., Staudigel, P.T., Davies, A.J., Bayarjargal, L., Spahr, D., Winkler, B., Hofmann, S., Granzin, M., Petersen, S.V. (2024) Carbonate clumped isotope values compromised by nitrate-derived NO₂ interferent. Chemical Geology 670, 122382. https://doi.org/10.1016/j.chemgeo.2024.122382

) (Fig. 3a). Within each valve, the difference between OSL and ISL temperatures was smaller than the 95 % CIs, which overlapped significantly, rendering them all analytically indistinguishable. For all samples, the mean annual water temperature (MAT) was contained within the 95 % confidence window (Fig. 3c–f), including the disequilibrium G. bimaculata sample, indicating that only Δ₄₈ was significantly positively biased.

Full size image

Temperatures derived from sample mean Δ₄₇ values were within 3.8 °C of either MAT or growing season temperature (GST) in all cases (Fig. 3c–f, Table 1). The mean difference between the Δ₄₇ temperature and relevant environmental mean was 1.6 ± 2.6 °C (2 s.d.). Binning four A. islandica samples, the mean Δ₄₇ temperature was 7.1 ± 4.7 °C (2 s.e.), where MAT and GST were 6.9 °C and 7.2 °C, respectively. The total temperature range from individual samples including the 95 % CIs (3.3–10.3 °C) is almost the same size as the total annual water temperature range (3.9–10.7 °C). Binning four G. bimaculata samples, the mean Δ₄₇ temperature was 18.9 ± 6.0 °C (2 s.e.), where MAT and GST temperature were 16.5 °C and 19.2 °C, respectively. For all individual samples, the entire 95 % CI (13.8–23.9 °C) was within the total seasonal temperature range (6.7–28.6 °C) but skewed towards the GST. Binning four S. senilis samples, the mean Δ₄₇ temperature was 23.0 ± 5.0 °C (2 s.e.), where MAT was 21.1 °C. The total temperature range from individual samples including the 95 % CIs (19.9–27.3 °C), aligned better with warmer months within the full seasonal water temperature range (16.7–27.2 °C). Binning four G. erosa samples, the mean Δ₄₇ temperature was 28.7 ± 7.0 °C (2 s.e.), where MAT was 28.3 °C. Although all 95 % CIs (23.7–34.8 °C) overlapped the total seasonal temperature range (25.7–29.9 °C), together their range is several degrees wider than the recorded environmental temperatures.

ID	Lat	Long		δ13C (VPDB) Mean ± 2 s.d.	δ18O (VPDB) Mean ± 2 s.d.	Δ47 (CDES90) Mean ± 95 % CI	Δ48 (CDES90) Mean ± 95 % CI	Δ47-T (°C) min–max (mean)	Water T (°C)
Arctica islandica (collected 11/1/2020)
AI-A	71.06	24.10	OSL ISL	2.05 ± 0.12 1.44 ± 0.06	2.58 ± 0.07 3.14 ± 0.08	0.65 ± 0.01 0.65 ± 0.01	0.27 ± 0.02 0.27 ± 0.02	5.6–9.4 (7.4) 3.3–7.5 (5.4)	Range MAT GST	3.9–10.7 6.9 7.2
AI-B	71.06	24.10	OSL ISL	1.89 ± 0.05 1.72 ± 0.05	2.55 ± 0.17 3.20 ± 0.11	0.65 ± 0.01 0.65 ± 0.01	0.27 ± 0.02 0.27 ± 0.02	5.4–9.8 (7.6) 5.9–10.3 (8.1)
Glycymeris bimaculata (11/1/2014)
GB-A	44.46	15.03	OSL ISL	1.51 ± 0.08 0.29 ± 0.06	1.27 ± 0.12 1.27 ± 0.17	0.61 ± 0.01 0.62 ± 0.01	0.27 ± 0.02 0.25 ± 0.02	17.1–21.4 (19.2) 13.8–18.0 (15.9)	Range MAT GST	6.7–28.6 16.5 19.2
GB-B	44.46	15.03	OSL ISL	1.46 ± 0.05 1.22 ± 0.03	1.21 ± 0.08 1.14 ± 0.15	0.60 ± 0.01 0.61 ± 0.01	0.24 ± 0.02 0.26 ± 0.02	19.2–23.9 (21.5) 16.7–21.3 (18.9)
Geloina erosa (collected 2010)
GE-A	−7.71	108.88	OSL ISL	−12.13 ± 0.05 −10.74 ± 0.04	−4.56 ± 0.07 −4.77 ± 0.10	0.59 ± 0.01 0.59 ± 0.01	0.25 ± 0.02 0.22 ± 0.02	23.7–27.8 (25.7) 24.5–29.2 (26.8)	Range MAT	25.7–29.9 28.3
GE-B	−7.71	108.88	OSL ISL	−12.26 ± 0.15 −11.26 ± 0.08	−4.51 ± 0.11 −6.17 ± 0.10	0.58 ± 0.01 0.58 ± 0.01	0.25 ± 0.02 0.23 ± 0.02	29.5–34.8 (32.1) 27.7–32.6 (30.1)
Senilia senilis (collected 9/1/2021)
SS-A	19.90	−16.25	OSL ISL	−0.98 ± 0.08 −0.81 ± 0.11	0.95 ± 0.17 0.90 ± 0.12	0.60 ± 0.01 0.60 ± 0.01	0.25 ± 0.02 0.25 ± 0.02	21.0–25.1 (23.0) 19.9–24.0 (21.9)	Range MAT	16.7–27.2 21.1
SS-B	19.90	−16.25	OSL ISL	−1.41 ± 0.03 −0.16 ± 0.06	1.09 ± 0.20 0.81 ± 0.11	0.60 ± 0.01 0.60 ± 0.01	0.24 ± 0.02 0.25 ± 0.02	19.9–24.6 (22.2) 22.5–27.3 (24.9)

Bulk isotopic composition. Absolute OSL δ¹⁸O values from the two individuals of each species were analytically indistinguishable in all cases (Table 1). Absolute OSL δ¹³C values were all analytically indistinguishable except S. senilis, where the two differed by 0.43 ‰ (2 s.d. < 0.2 ‰) (Table 1). In contrast, ISL δ¹⁸O values were all indistinguishable except for G. erosa, where the two differed by 1.40 ‰ (2 s.d. < 0.2 ‰), and ISL δ¹³C values were distinguishable in every case with differences ranging from 0.28 to 0.93 ‰ (2 s.d. 0.03–0.11 ‰) (Table 1).

All but one shell (G. bimaculata) had measurable differences between ISL and OSL δ¹⁸O and/or δ¹³C values (Δδ¹⁸O_ISL-OSL, Δδ¹³C_ISL-OSL) (Fig. 3b). Altogether, Δδ¹⁸O_ISL-OSL values spanned −1.66 to +0.65 ‰ and Δδ¹³C_ISL-OSL spanned −1.22 to +1.39 ‰. Among individuals of the same species, the magnitudes of these offsets differed but their directionality was largely the same. A. islandica fell in quadrant 1 (+Δδ¹⁸O_ISL-OSL, −Δδ¹³C_ISL-OSL). G. bimaculata had Δδ¹⁸O_ISL-OSL ≈ 0 ‰ and −Δδ¹³C_ISL-OSL values, the larger of which come from the specimen biased in Δ₄₈. S. senilis and G.erosa fell in quadrant 4 (−Δδ¹⁸O_ISL-OSL, +Δδ¹³C_ISL-OSL values).

top

Discussion

Dual clumped isotopes record MAT/GST highly accurately. It is uncertain whether the S. senilis Δ₄₇ temperatures were truly warm season biased as it appears (Fig. 3e) because WOA data down to 10 m were averaged, potentially including waters cooler than the true shallow growth environment. A similar consideration is true for G. erosa, plus the real seasonality in a shallow, tidal mangrove where summer water temperatures can exceed 30 °C (Soeprobowati et al., 2023

Soeprobowati, T.R., Jumari, J., Wasiq Hidayat, J., Muhammad, F., Hanif Al Falah, M., Kadek Dita Cahyani, N., Gell, P. (2023) Water Quality Status of Mangrove Ecosystem in Bedono, Sayung, Demak, Central Java. Pollution 9, 1374–1385. https://doi.org/10.22059/poll.2023.355899.1803

) is very likely greater than the offshore data point. In both cases employing WOA data, density of instrumental measurements affects the fidelity of temperatures, and the multi-year monthly averaging masks extremes in given months/years that the instrumental data resolves.

Both the OSL and ISL record comparable temperatures. The one non-equilibrium sample could be strictly fortuitous (a 5 % probability) or might indicate slightly elevated levels of organics (Fiebig et al., 2024

Fiebig, J., Bernecker, M., Meijer, N., Methner, K., Staudigel, P.T., Davies, A.J., Bayarjargal, L., Spahr, D., Winkler, B., Hofmann, S., Granzin, M., Petersen, S.V. (2024) Carbonate clumped isotope values compromised by nitrate-derived NO₂ interferent. Chemical Geology 670, 122382. https://doi.org/10.1016/j.chemgeo.2024.122382

). Even with accurate equilibrium clumped isotope temperatures, different bulk δ¹⁸O in the two layers would still bias δ¹⁸O_water reconstructions. Inter-layer homogeneity should be assessed prior to using Δ₄₇ and δ¹⁸O to calculate δ¹⁸O_water. Good agreement between OSL values from the same location across different specimens suggests that the OSL is better for reconstructing δ¹⁸O_water than the ISL (Table 1).

Our accurate temperature reconstructions evidence the validity of drilling as a preparation method, despite previous experiments where frictional heating induced with high drill speeds in small areas partially reset Δ₄₇ temperatures (e.g. Staudigel and Swart, 2016

Staudigel, P.T., Swart, P.K. (2016) Isotopic behavior during the aragonite-calcite transition: Implications for sample preparation and proxy interpretation. Chemical Geology 442, 130–138. https://doi.org/10.1016/j.chemgeo.2016.09.013

; Staudigel et al., 2023

Staudigel, P., Davies, A.J., Bernecker, M., Tagliavento, M., van der Lubbe, H.J.L., Nooitgedacht, C., Looser, N., Bernasconi, S.M., Vonhof, H., Fiebig, J. (2023) Fingerprinting Kinetic Isotope Effects and Diagenetic Exchange Reactions Using Fluid Inclusion and Dual-Clumped Isotope Analysis. Geochemistry, Geophysics, Geosystems 24, e2022GC010766. https://doi.org/10.1029/2022GC010766

). Although there still may be some reordering or aragonite-to-calcite conversion (Staudigel and Swart, 2016

Staudigel, P.T., Swart, P.K. (2016) Isotopic behavior during the aragonite-calcite transition: Implications for sample preparation and proxy interpretation. Chemical Geology 442, 130–138. https://doi.org/10.1016/j.chemgeo.2016.09.013

noted 95–100 % aragonite at drill speeds of 4000 rpm), comparisons between our hand drilled samples and environmental temperatures indicate that any effect from this preparation bias is dwarfed by analytical uncertainty and natural seasonality. Frictional heating from drilling should be a minimal concern when following cautious procedures as done here (low speed and pressure, constant motion, and short intermittent contact periods), especially when the sample material is not very resistant to milling as are most recent and well preserved fossil bivalves.

Considering the near total lack of intra-specific differences in δ¹⁸O and δ¹³C values deriving from the OSL, it appears that external dissolved inorganic carbonate (DIC) is likely compositionally identical for the two specimens collected at each location. Different OSL δ¹³C among the two S. senilis shells could be due to the samples reflecting different growth years wherein bottom water δ¹³C varied, but indistinguishable δ¹⁸O and temperatures limit confidence in upwelling or freshening as potential causes. Alternatively, variable DIC of porewaters surrounding the burrowing shells could be important and tend to vary in δ¹³C more than δ¹⁸O due to microbial activity.

The intra-specific variation in Δδ¹³C_ISL-OSL and Δδ¹⁸O_ISL-OSL could be driven by either kinetic or source effects. Considering all samples, there is no strong correlation between Δδ¹⁸O_ISL-OSL and Δδ¹³C_ISL-OSL, not supporting prevalent kinetic control on the bulk isotopic composition. Some evidence suggests that δ¹⁸O may be more susceptible to kinetic effects than Δ₄₇ or Δ₄₈ due a combination of higher measurement precision and slower equilibration at the CaCO₃-H₂O interface depending on mineral surface characteristics (Schlidt et al., 2025

Schlidt, V., Evans, D., de Winter, N.J., Bernecker, M., Arndt, I., Staudigel, P.T., Davies, A.J., Brand, U., Müller, W., Fiebig, J. (2025) Most bivalves and gastropods calcify indistinguishably from dual clumped isotope equilibrium. Geochimica et Cosmochimica Acta 410, 174–187. https://doi.org/10.1016/j.gca.2025.10.008

), but even if minor kinetic effects exist here, source effects likely overshadow them.

The immediate DIC source for the ISL might become measurably distinct from ambient water firstly by isotopic modification of internal DIC by numerous metabolic processes (e.g., respiration, tissue maintenance, digestion, reproduction), and secondly when that modified internal DIC is transformed into extrapallial (calcifying) fluid. If the same or different fractionating processes occur when secreting the outer extrapallial fluid (which may or may not be the case), mixing with ambient water may overprint this physiologically derived signal, leading to very consistent OSL δ¹⁸O and δ¹³C values, whereas the ISL values preserve it.

Members of the same species occupying the same quadrant in Figure 3 demonstrate an inherent consistency in the pathways fractionating internal DIC on the species level due to shared physiology, but the differences in absolute magnitudes of Δδ¹⁸O_ISL-OSL and Δδ¹³C_ISL-OSL values could be due to individual level factors like metabolic rate, diet, reproduction, and valve closure variably influencing internal DIC (e.g., Curley et al., 2023

Curley, A.N., Petersen, S.V., Edie, S.M., Guo, W. (2023) Biologically driven isotopic fractionations in bivalves: from palaeoenvironmental problem to palaeophysiological proxy. Biological Reviews 98, 1016–1032. https://doi.org/10.1111/brv.12940

and refs. therein). Additionally, the spread of values further affirms that these fundamental fractionating pathways can vary among different species of bivalves, as previously reported (e.g., Lollos et al., 2025

Lollos, D.M., Curley, A.N., Petersen, S.V. (2025) Regional variation in biologically driven isotopic fractionations in cosmopolitan bivalve genera Dosinia and Mactra. PALAIOS 40, 321–335. https://doi.org/10.2110/palo.2025.004

; Curley et al., 2023

Curley, A.N., Petersen, S.V., Edie, S.M., Guo, W. (2023) Biologically driven isotopic fractionations in bivalves: from palaeoenvironmental problem to palaeophysiological proxy. Biological Reviews 98, 1016–1032. https://doi.org/10.1111/brv.12940

).

Some of G. erosa’s large Δδ¹⁸O_ISL-OSL and Δδ¹³C_ISL-OSL values could be due to their unique ecology as a tidal species (Purba et al., 2025

Purba, M.M., Fajri, N.E., Yuliati, Y. (2025) Bioecology of lokan calm (Geloina erosa) in the mangrove ecosystem of Buruk Bakul village, Bukit Batu district, Bengkalis regency, Riau province. Aurelia Journal 7, 55–66. https://doi.org/10.15578/aj.v7i1.14329

), where in different years they may have experienced more meaningfully different behaviours, freshwater inputs, and DIC sources. They also have δ¹³C values in either layer consistently at least 10 ‰ lower than any of the other species (Table 1), possibly consistent with relatively high incorporation of metabolic DIC in the calcifying fluid carbon budget due to more time spent with valves closed during low tide (e.g., Gillikin et al., 2007

Gillikin, D.P., Lorrain, A., Meng, L., Dehairs, F. (2007) A large metabolic carbon contribution to the δ¹³C record in marine aragonitic bivalve shells. Geochimica et Cosmochimica Acta 71, 2936–2946. https://doi.org/10.1016/j.gca.2007.04.003

).

top

Conclusions

Δ₄₇ temperatures from hand drilled A. islandica, G. bimaculata, S. senilis, and G. erosa shells were consistently within 3.8 °C (on average 1.6 °C) of environmental mean temperatures. Δ₄₇-Δ₄₈ values fell on the equilibrium line in all except one case, wherein the Δ₄₇ temperature remained accurate. Temperatures reconstructed from either OSL or ISL carbonate in the same shell were analytically indistinguishable, but δ¹⁸O and δ¹³C varied systematically. Inter-layer δ¹⁸O and δ¹³C offsets indicate some amount of physiological influence, likely driven by source effects. The directionality of these is consistent among individuals of the same species, suggesting a common mechanism. In some cases, Δδ¹⁸O_ISL-OSL values are great enough to meaningfully bias reconstructed δ¹⁸O_water if the ISL is sampled, despite accurate Δ₄₇ temperatures, highlighting the importance of considering intra-shell variability when using bivalve shells as hydroclimate archives.

top

Acknowledgements

We would like to acknowledge assistance from Jan van Gils (S. senilis), Rob Witbaard (A. islandica) and Simon Troelstra (G. erosa) in acquiring some of the specimens used in this work. We thank Miguel Bernecker, Sven Hofman, Niels Meijer, and Philip Staudigel for assistance in the lab. JF was supported through DFG, Reinhart Koselleck project FI-948/13–1, NJW and BO were financially supported by the Netherlands Research Council (NWO) through a personal VENI fellowship awarded to NJW (MACRO, VI.Veni.222.354). NJW was supported by funds from the “Sectorplan Beta and Techniek” by the Dutch Ministry for Education, Culture and Science. MP was supported by CSF project BivalveSPEECH (IP-2024-05-2190). This manuscript was improved by the contribution of editorial and anonymous peer review.

Editor: Eric Oelkers

top

References

top

Supplementary Information

The Supplementary Information includes:

• Tables S-1 to S-7
• Supplementary Information References

Download the Supplementary Information (PDF)

Download Tables S-1 to S-5 (xlsx)

Download Tables S-6 and S-7 (xlsx)