Reassessment of body temperature and thermoregulation strategies in Mesozoic marine reptiles

Abstract

Ichthyosauria, Plesiosauria, and Metriorhynchidae were apex predators in Mesozoic oceanic trophic networks. Previous stable oxygen isotope studies suggested that several taxa belonging to these groups were endothermic and that some of them were homeothermic organisms. However, these conclusions remain contentious owing to the associated uncertainties regarding the δ¹⁸O value and oxygen isotope fractionation relative to environmental seawater. Here, we present new bioapatite phosphate δ¹⁸O values (δ¹⁸O_p) of Ichthyosauria, Plesiosauria, and Metriorhynchidae (Middle Jurassic to Early Cretaceous) recovered from mid- to high paleolatitudes to better constrain their thermophysiology and investigate the presence of regional heterothermies. The intraskeletal δ¹⁸O_p variability failed to reveal distinct heterothermic patterns within any of the specimens, indicating either intrabody temperature homogeneity or an overriding diagenetic overprint of the original biological δ¹⁸O_p bone record. Body temperature estimates have been reassessed from new and published δ¹⁸O_p values of well-preserved isolated teeth, recently revised Mesozoic latitudinal δ¹⁸O oceanic gradients, and ¹⁸O-enrichment factors of fully aquatic air-breathing vertebrates. Our results confirm that Ichthyosauria were homeothermic endotherms (31°C to 41°C), while Plesiosauria were likely poikilothermic endotherms (27°C to 34°C). The new body temperature estimates of the Metriorhynchidae (25°C to 32°C) closely follow ambient temperatures and point to poikilothermic strategy with no or little endothermic ability. These results improve our understanding of Mesozoic marine reptile thermoregulation and indicate that due to their limited body temperature variations, the δ¹⁸O_p values from Ichthyosauria fossil remains could be used as valuable archives of Mesozoic oceans δ¹⁸O_sw values that may help improve paleoenvironmental and paleoclimatic reconstructions.

Non-technical Summary

Some marine reptiles from the Mesozoic, such as ichthyosaurs, plesiosaurs, and metriorhynchids, were capable of reaching elevated body temperatures, and some could maintain body temperatures a few degrees above those of their marine environments, a characteristic similar to that observed in modern cetaceans. Nevertheless, the estimation of their body temperatures from the chemical oxygen signatures of their fossil remains (bones and teeth) is accompanied by uncertainties associated with the chemical oxygen signatures of the surrounding water and the mineralization processes of the bones and teeth. In this study, new data were collected from four ichthyosaurs, three plesiosaurs, and one metriorhynchid in order to gain a deeper understanding of the mechanisms by which these marine reptiles were able to maintain body temperatures higher than those of their environments. The chemical signatures of oxygen in the bones and teeth of the specimens did not exhibit any discernible patterns indicative of specific zones of heat production or loss, unlike what has been observed in modern marine vertebrates. Concurrently, we reassessed the estimated body temperatures of these marine reptiles, thereby corroborating the hypothesis that ichthyosaurs were homeothermic endotherms. Conversely, our new estimates suggest that plesiosaurs were likely poikilothermic endotherms, whereas metriorhynchids were probably also poikilothermic endotherms but with a limited capacity for heat production. Finally, the narrow range of body temperatures maintained by ichthyosaurs indicates that the oxygen chemical signatures of fossilized remains could serve as valuable markers for reconstructing variations in the oxygen isotope composition of the Mesozoic oceans, paving the way to enhance our understanding of the environment and climate of this period in Earth’s history.

Introduction

During the Mesozoic Era (251.9 to 66.0 Ma), marine reptiles such as ichthyosaurs, plesiosaurs, and metriorhynchids were distributed worldwide and played a key role in the trophic networks. Paleobiogeographic (Kear 2006b; Bardet et al. 2014; Vavrek et al. 2014; Delsett et al. 2016; Rogov et al. 2019; Zverkov et al. 2021), osteo-histological (Ichthyosauria: de Buffrénil and Mazin 1989, 1990; Anderson et al. 2019; Plesiosauria: Wiffen et al. 1995; Delsett and Hurum 2012; Fleischle et al. 2018), geochemical (Bernard et al. 2010; Séon et al. 2020; Leuzinger et al. 2023), and modeling studies (Brice and Grigg 2023) all indicate that Ichthyosauria Blainville, 1835 and Plesiosauria Blainville, 1835 were endothermic and probably homeothermic organisms, much like extant Cetacea Brisson, 1762. They would have been able to produce enough body heat to raise their body temperatures above those of the environments in which they lived (Bernard et al. 2010; Séon et al. 2020; Leuzinger et al. 2023). High and constant body temperature in Ichthyosauria would have been facilitated by their fusiform morphology favoring heat retention and the presence of a layer of fibroadipose tissue surrounding the trunk (Lindgren et al. 2018; Delsett et al. 2022). Evidence for adipose tissue is lacking for Plesiosauria, for which very few specimens preserving soft tissues have been found (Vincent et al. 2017). Thermophysiological studies of Metriorhynchidae Fitzinger, 1843, a group of fully aquatic marine Mesozoic crocodylomorphs, have not led to a consensus. The fossil occurrences confined to tropical paleolatitudes (Bardet et al. 2014) and the osteo-histological findings from studies by Hua and de Buffrénil (1996) and de Buffrénil et al. (2021) suggested that Metriorhynchidae had an ectothermic poikilothermic thermoregulatory strategy similar to that of modern crocodylomorphs. Conversely, the oxygen isotope analyses of Séon et al. (2020) suggested that Metriorhynchidae were able to raise their body temperatures a few degrees above that of the ambient environment by metabolic heat production but could not maintain a constant body temperature, indicating they were poikilothermic endotherms like extant tunas (Block and Finnerty 1994; Graham and Dickson 2004).

Previous estimates of body temperature reconstruction of Ichthyosauria, Plesiosauria, and Metriorhynchidae, calculated from the phosphate oxygen isotope composition of the bioapatite of their bones and teeth (δ¹⁸O_p), assumed a constant global oceanic δ¹⁸O value (δ¹⁸O_sw) of −1 ± 1‰ (Bernard et al. 2010; Séon et al. 2020; Leuzinger et al. 2023; Supplementary Information 1 for details about isotope-based body temperature estimations), an oversimplified assumption given latitudinal gradients in δ¹⁸O_sw values recorded in Jurassic and Cretaceous seas (Alberti et al. 2017, 2020; Letulle et al. 2022). Furthermore, recent work has shown that the oxygen isotope compositions of body water from extant air-breathing fully marine species—Orcinus orca Linnæus, 1758, Tursiops truncatus Montagu, 1821 (Séon et al. 2023), and Caretta caretta Linnæus, 1758 (Séon 2023)—are less ¹⁸O-enriched relative to their drinking water than that of semiaquatic vertebrates (crocodiles, turtles) used previously to reconstruct body temperatures of extinct marine reptiles (Bernard et al. 2010; Séon et al. 2020; Leuzinger et al. 2023).

Previous osteo-histological and geochemical studies were exclusively based on isolated teeth or bone remains (de Buffrénil and Mazin 1990; Bernard et al. 2010; Fleischle et al. 2018; Séon et al. 2020) and thus provided only partial information on the thermoregulation of Ichthyosauria, Plesiosauria, and Metriorhynchidae. A key aspect of thermophysiology is to define how constant body temperature is at the level of vital organs (homeotherm vs. poikilotherm; Clarke and Pörtner 2010; Furukawa et al. 2015; Lovegrove 2017) and the distribution of body temperature within the body, known as regional heterothermies (Irving and Hart 1957; Folkow and Blix 1987; Favilla et al. 2022). An investigation into the mechanisms employed by Ichthyosauria, Plesiosauria, and Metriorhynchidae to regulate their body temperatures allows for an understanding of their adaptation to their environments and the explanation of their stratigraphic occurrences in the fossil record in view of fluctuating environmental temperatures during the Mesozoic (Takashima et al. 2006; Dera et al. 2011; Wierzbowski et al. 2013). Such climate changes do not seem to have affected their diversity (Bardet 1994; Martin et al. 2014; Stubbs and Benton 2016). Moreover, precisely defining their thermoregulatory strategies opens the way to investigating their behavior. For instance, homeothermic endotherms are typically active organisms, as they are fully independent of the temperatures of their environments. Nevertheless, this requires a significant energy input (Clarke and Pörtner 2010). Conversely, some species that are considered to be regional endotherms can produce heat locally, which can result in temperature heterogeneities or regional heterothermies (Carey 1982; Block 1986; Dickson and Graham 2004; Graham and Dickson 2004). In swordfish, for example, the heat production located close to the eyes improves visual acuity in foraging cold environments of great depth (Block 1987; Fritsches et al. 2005), whereas heat production in locomotory muscles of tunas and lamnid sharks enables them to swim faster or migrate over longer distances than poikilothermic ectothermic organisms (Blank et al. 2007; Bernal et al. 2012; Watanabe et al. 2015; Harding et al. 2021). The mapping of intraindividual temperature variations, and thus regional heterothermies, through the use of oxygen isotopes enables the identification of heat-producing zones within an organism, in addition to the delineation of thermal windows, the preferred sites for heat loss, which serve to regulate body temperature (Séon et al. 2022, 2024).

In this study, we performed an oxygen isotope analysis of 239 bones and 8 teeth from complete or subcomplete specimens, enabling the first evaluation of possible regional heterothermies within the bodies of ichthyosaurs, plesiosaurs, and metriorhynchids. We also reassessed body temperature estimates of Ichthyosauria, Plesiosauria, and Metriorhynchidae using new and published δ¹⁸O_p tooth data and an ¹⁸O-enrichment of fully aquatic animals and considering a global δ¹⁸O_sw gradient for the Jurassic and Cretaceous seas.

Institutional Abbreviations. MHNLM, Muséum d’Histoire Naturelle Le Mans; MPV, Musée Paléontologique de Villers-sur-Mer; PMO, Palaeontological collections of the Natural History Museum, University of Oslo, Norway.

Material and Methods

Sampled Specimens

Specimens of four Ichthyosauria, three Plesiosauria, and one Metriorhynchidae were sampled for their intraskeletal δ¹⁸O_p variability; taxonomic affiliation, collection number, size estimate, stratigraphic age, locality, and type of sampled material for each marine reptile specimen are reported in Table 1. In addition to these complete and subcomplete specimens, two teeth of Metriorhynchidae indet. specimens from the Marnes de Dives Formation (late Callovian, Les Vaches Noires Cliffs, France) were sampled for their δ¹⁸O_p values to reassess metriorhynchid body temperatures (Table 1), as well as a pachycormid fish tooth belonging to Hypsocormus Wagner, 1863 recovered from the same stratigraphic layer as the Ichthyosauria indet. specimen from Les Ardilles (Kimmeridgian, Auxerre, France) to determine the oceanic paleotemperature of this deposit. Overall preservation and appearance of the studied specimens vary between the localities. The Ichthyosauria indet. specimen from Les Ardilles (France) is slightly deformed and marked by numerous fractures secondarily filled with calcite (Mazin and Pavy 1995). The skeletal remains of the specimen of Metriorhynchus aff. superciliosus from Les Vaches Noires Cliffs (France) seem to have been little affected by compaction, except for the skull, which shows a few cracks with surface recrystallization (Le Mort et al. 2022). The ichthyosaurs (PMO 222.655, PMO 222.667, PMO 222.669) and plesiosaurs (PMO 212.662, PMO 222.663) from the Slottsmøya Member (Spitsbergen) are partly eroded, probably by the action of suspended particles and sediments transported by bottom currents (Martill 1985; Reisdorf et al. 2012) and show recrystallizations of calcite and barite in the pores of the skeletal elements (Kihle et al. 2012; Delsett et al. 2016). The skeletal remains are also for the most part highly fractured, partly due to local faults within the deposit and to frost weathering. For instance, the specimen of Keilhauia nui (PMO 222.655) was crossed by large fractures (Delsett et al. 2017). Its skeletal elements are particularly fractured and have a sandy appearance.

Table 1. Summary of information from Ichthyosauria, Plesiosauria, and Metriorhynchidae specimens sampled.

Sampling Method

Approximately 40 mg of each skeletal element and tooth were ground to fine powder using a Dremel micro-drill equipped with a diamond-studded drill bit; cortical bone was selected to maximize the chances of preserving a biological isotopic record. Skeletal elements of complete and subcomplete specimens were grouped in five categories, as follows: “Skull” groups all the bone forming the skull and the mandible; “Cervical region” groups the cervical vertebra only in Plesiosauria; “Dorsal region” groups the dorsal vertebrae, ribs, and skeletal elements forming the pectoral and pelvic girdles; “Caudal region” groups the caudal vertebrae; and “Appendicular region” groups the humerus, femur, radius, ulna, tibia, fibula, metacarpals, metatarsals, phalanges, and accessory bone elements located in the limbs. For the specimen of Colymbosaurus svalbardensis (PMO 222.663), the “Appendicular region” was split into anterior right limb (ARL), anterior left limb (ALL), posterior right limb (PRL), and posterior left limb (PLL), as most of the elements were found in articulation (Delsett et al. 2016). For Ichthyosauria and Metriorhynchus aff. superciliosus specimens, cervical vertebrae were included in the “Dorsal region.” A total of 247 skeletal elements and teeth were sampled, averaging about 20 and 40 samples per specimen. In addition to the bone and tooth samples, a sample of the recrystallization mineral observed in one dorsal vertebra from the Ichthyosauria indet. specimen from Les Ardilles was taken.

Bioapatite P₂O₅ Content and Phosphate Group Oxygen Isotope Analysis

Cortical bone and tooth powders (n = 247) were prepared using the wet chemical procedure detailed by Crowson et al. (1991) and modified by Lécuyer et al. (1993), which involves the isolation of phosphate ions (PO₄³⁻) from the bioapatite, which are then precipitated as silver phosphate crystals (Ag₃PO₄). For each sample, 20–30 mg of cortical bone or tooth enamel powder was dissolved in 2 ml of 2M HF. The CaF₂ residue was separated by centrifugation, and the solution was neutralized by adding 2.2 ml of 2M KOH. Amberlite IRN78 anion-exchange resin beads were added to the solution to isolate the PO₄³⁻ ions. After 24 h, the solution was removed, and the resin was rinsed with deionized water and eluted with 27.5 ml of 0.5M NH₄NO₃. After 4 h, the Amberlite IRN78 anion-exchange resin beads were removed from the solution. Then, 0.5 ml of NH₄OH and 15 ml of an ammoniacal solution of AgNO₃ were added, and the solutions were placed in a thermostatted bath at 70°C for 7 h, allowing the slow and quantitative precipitation of Ag₃PO₄ crystals. The Ag₃PO₄ crystals were filtered, dried, and cleaned. For each sample, the P₂O₅ content, a proxy for qualitatively estimating the dissolution of phosphate minerals of the mineralized skeletal elements bioapatite in fossil specimens (Nemliher et al. 2004), was estimated. It was calculated from the recovered mass of silver phosphate (Ag₃PO₄) after the chemical purification of the samples and phosphate chemical yields obtained for the NIST SRM 120c standards (P₂O₅ = 33.34 wt%), assuming stoichiometric conversion of bioapatite phosphate ions into silver phosphate (Lécuyer et al. 2003).

For oxygen isotope composition measurements, five aliquots of 300 ± 20 μg of Ag₃PO₄ for each sample were mixed with 400 ± 50 μg of graphite in silver foil capsules. Oxygen isotope ratios were measured using a high-temperature vario PYRO cube elemental analyzer (EA), equipped with “purge and trap” technology (Fourel et al. 2011) and interfaced in continuous-flow mode to an IsoPrime isotope ratio mass spectrometer (Elementar UK, Cheadle, U.K.). To calibrate the measurements, the NBS 127 (barium sulfate precipitated from seawater from Monterey Bay, Calif., U.S.A., Sigma-Aldrich), and a silver phosphate precipitated from the international standards NIST SRM 120c (natural Miocene phosphorite from Florida, U.S.A., National Bureau of Standards) were used. The δ¹⁸O value for NBS 127 is fixed to the certified value of +9.3‰ Vienna Standard Mean Ocean Water (VSMOW; Hut 1987; Halas and Szaran 2001) and that for NIST SRM 120c to +21.7‰ VSMOW according to Lécuyer et al. (1993), Chenery et al. (2010), and Halas et al. (2011) for correction of instrumental mass fractionation during CO isotope analysis. Along with the silver phosphate samples derived from bone and tooth bioapatite, silver phosphates precipitated from standard NIST SRM 120c were repeatedly analyzed (δ¹⁸O_p = 21.7 ± 0.3‰, n = 59) to ensure that no isotope fractionation occurred during the wet chemistry. Values are reported as delta per mille values expressed with respect to VSMOW.

Oxygen and Carbon Isotope Analysis of Bioapatite Carbonate and Carbonate Content

Bone and tooth powders (n = 71) were pretreated to measure the oxygen and carbon isotope compositions of the carbonate group of the bone bioapatite (Plesiosauria: n = 40; Ichthyosauria: n = 29, Metriorhynchus aff. superciliosus: n = 1; and pachycormid fish tooth of Hypsocormus sp.: n = 1). The protocol used was that of Koch et al. (1997) and consisted of removing organic matter through a chemical reaction between bone powder and 0.4 ml of 2% sodium hypochlorite (NaOCl) for 24 h (0.4 ml NaClO per 10 mg of bone powder). The solution was centrifuged and rinsed three times with ultrapure water before 0.1 M acetic acid (CH₃COOH) was added for 24 h to remove any secondary carbonates that may have precipitated during the organic removal procedure. The solution was again centrifuged and rinsed three times. Then, the powders were dried at 50°C for 48 h before being collected. For each sample, three replicates of 2 mg were weighed into 3.7 ml round-bottomed glass vials and sealed (Exetainer, LABCO UK, Lampeter, Wales, U.K.). Oxygen (δ¹⁸O_c) and carbon (δ¹³C_c) isotope composition of the carbonate group of bone and tooth bioapatite was measured using an isoFLOW-type preparation system (Elementar GmbH, Germany) connected in continuous flow to a precisION isotope ratio mass spectrometer (Elementar UK). Each of the pretreated samples was reacted with saturated anhydrous phosphoric acid (H₃PO₄) prepared according to the protocol of McCrea (1950). The reaction took place at a constant temperature of 90°C. The CO₂ generated during acid digestion of the carbonate sample was then transferred to the mass spectrometer. Carbonate content (CO₃²⁻ %wt) of the bioapatite samples was measured based on the peak area of CO₂ detected by the mass spectrometer. Isotope measurements were corrected for instrumental drift and calibrated using two calcite isotopic standards: a Carrara marble (laboratory standard) with values of δ¹⁸O_VPDB = −1.84‰ and δ¹³C_VPDB = +2.03‰ (Fourel et al. 2015) and the NBS18 (international standard), whose values are δ¹⁸O_VPDB = −23.2‰ and δ¹³C_VPDB = −5.01‰ (Friedman et al. 1982; Hut 1987; Coplen et al. 2006); as well as NIST SRM 120c, whose values are δ¹⁸O_VPDB= −1.13‰ and δ¹³C_VPDB = −6.27‰ (Passey et al. 2007). For bone and tooth apatite, the acid fractionation factor α(CO₂–apatite carbonate) of 1.00773 determined for the NIST SRM 120c phosphate rock reference material was selected (Passey et al. 2007). Values are reported as delta per mille values expressed with respect to VSMOW for oxygen and Vienna Pee Dee Belemnite (VPDB) for carbon.

Raman Spectroscopy

Characterization of the mineralogical composition of the samples was carried out using an XploRA Raman microscope equipped with a diode-pumped Nd:YAG laser at 532 nm. For each sample analyzed (n = 84), 10 spectra of 10 s were acquired at 100× magnification. The scattered light was detected in the range of 150 and 2000 cm⁻¹. The first-order band of a pure silicon reference material at 520.7 cm⁻¹ was used to calibrate the spectrometer before each measurement. The full width at half maximum (FWHM) and the position of the ν1(PO₄³⁻) fully symmetric stretching band were measured to obtain information on the mineralogical composition and crystal structure of bioapatite (Pucéat et al. 2004; Thomas et al. 2007, 2011; Dal Sasso et al. 2018; Barthel et al. 2020; Kral et al. 2022).

Body Temperature Estimates

Body temperatures of Ichthyosauria, Plesiosauria, and Metriorhynchidae were re-estimated from published tooth δ¹⁸O_p values (Anderson et al. 1994; Bernard et al. 2010; Séon et al. 2020) as well as newly measured values from two teeth belonging to specimens of Metriorhynchidae indet. from Les Vaches Noires Cliffs (late Callovian, Villers-sur-Mer, France) and the teeth from the three specimens of ichthyosaur referred to as Palvennia hoybergeti (PMO 222.669), Keilhauia sp. (PMO 222.667), and Ichthyosauria indet. from Les Ardilles. The equation published by Lécuyer et al. (2013), adapted for air-breathing vertebrates, was used and an average ¹⁸O-enrichment of +0.8 ± 0.9‰, for air-breathing fully aquatic marine vertebrates was applied (Séon et al. 2023). Paleolatitudes were reconstructed using the software developed by van Hinsbergen et al. (2015) and then used for the estimation of the δ¹⁸O_sw values from the equation published by Alberti et al. (2020). For each locality, sea-surface temperature was calculated from the δ¹⁸O_p values of contemporary fish using the equation of Lécuyer et al. (2013), considering that body temperature (T _b) is equal to seawater temperature (T _sw), and that δ¹⁸O_bw ≈ δ¹⁸O_sw (Kolodny et al. 1983). The associated error in temperature calculation is equal to 3°C (based on the slope of 4.5 of the Lécuyer et al. [2013] equation used). As no contemporary fish tooth has been analyzed for the Slottsmøya Member fossil locality, the environmental temperature was estimated from the oxygen isotope composition of one brachiopod shell from Hammer et al. (2011) using the oxygen isotope fractionation equation provided by Letulle et al. (2023).

Statistical Analyses

We used descriptive statistics to explore body temperature re-estimates and intraskeletal δ¹⁸O_p variations in Mesozoic marine reptiles. The normality and homoscedasticity (the uniformity of the error associated with the variance for each of the values) of the δ¹⁸O_p values could not be verified. Considering the variable number of samples for each set (between 1 and 14 samples), the nonparametric Mann-Whitney-Wilcoxon test was used to compare median values between two sets of observations, each corresponding to a skeletal region. Statistical tests were performed using R software with a significance threshold set at p-value < 0.05. Pearson’s correlation coefficient r was employed to ascertain the degree of correlation between two quantitative variables.

Results

Bioapatite P₂O₅ Content and Phosphate Oxygen Isotope Composition (δ¹⁸O_p)

A total of 218 δ¹⁸O_p values were acquired from 4 ichthyosaurs, 3 plesiosaurs, and 1 Metriorhynchus aff. superciliosus specimen in order to map regional heterothermies and reassess body temperatures. Almost all samples from each specimen could be analyzed, except a few samples that failed to yield silver phosphate after wet chemistry preparation, especially from the specimen of Keilhauia nui (PMO 222.655). P₂O₅ content expressed in weight percentage (wt%) and measured oxygen isotope composition from the phosphate group of each bone and tooth of all the specimens are reported in Supplementary Table 1 and synthetized in Table 2. P₂O₅ content for all specimens ranged from 2.0% to 40.5% (Table 2), with more than 90% of the values ranging from 15.3% to 40.5%. These results are consistent with the %P₂O₅ measured for the skeletal and dental remains of modern marine vertebrates (26.0 ± 4.4%, range: [15.3, 33.92], n = 27; Nemliher et al. 2004). No significant differences were observed between the fossil deposits (Supplementary Figure 1A) or between specimens if we exclude the Keilhauia nui specimen (PMO 222.655), for which the percentage of P₂O₅ is significantly lower (Mann-Whitney-Wilcoxon test, p-value < 0.001; Supplementary Figure 1B). Globally, there are no significant differences in P₂O₅ content along the skeletal region, except for the Ichthyosauria indet. and Colymbosaurus svalbardensis specimens (Mann-Whitney-Wilcoxon test, p-value < 0.05; Supplementary Figure 2). Finally, no strong relationship between P₂O₅ content and δ¹⁸O_p values is observed in Ichthyosauria and in Plesiosauria specimens (Pearson correlation test, r = 0.32; Supplementary Figure 3).

Table 2. Synthesis of isotopic values (δ¹⁸O_p, δ¹⁸O_c, δ¹³C_c) and CO₃²⁻ and P₂O₅ content (wt%) of the bone and tooth bioapatite of the specimens of Ichthyosauria, Plesiosauria, and Metriorhynchidae. Abbreviations: ARL, anterior right limb; ALL, anterior left limb; PRL, posterior right limb; and PLL posterior left limb.

The intraskeletal variability of δ¹⁸O_p is illustrated in Figures 1, 2, and 3, respectively for Ichthyosauria, Plesiosauria, and Metriorhynchidae. The Hypsocormus tooth recovered from the same stratigraphic layer as the specimen of Ichthyosauria indet. from Les Ardilles yielded a δ¹⁸O_p value of 18.1 ± 0.2‰, and the δ¹⁸O_p values of the Metriorhynchidae teeth VN6 and VN7 equaled 21.4 ± 0.2‰ and 19.8 ± 0.2‰, respectively (Supplementary Table 2). Intraskeletal variability in δ¹⁸O_p values was observed for all the specimens of Ichthyosauria, the specimen of Kimmerosaurus sp., and the specimen of Metriorhynchus aff. superciliosus, but these differences are not significant between body regions (Mann-Whitney-Wilcoxon test, p-value > 0.05; Fig. 4). However, a trend seems to emerge in specimen PMO 222.667 of Keilhauia sp. and Palvennia hoybergeti specimen PMO 222.669, in which the teeth have overall higher δ¹⁸O_p values than those of the bones, but the statistical significance of this observation could not be tested because of the small number of samples for each set (n < 5 for teeth). The only significant differences in δ¹⁸O_p values along body regions observed are in the Colymbosaurus svalbardensis (PMO 222.663) specimen. Mann-Whitney-Wilcoxon statistical test results (p-value < 0.05; Fig. 4) clearly indicate that skeletal elements in the anterior part of the skeleton (dorsal vertebrae, ribs, left and right forelimbs) have significantly higher δ¹⁸O_p values than skeletal elements in the posterior part (left and right hindlimbs and caudal vertebrae).

Figure 1. Regional heterothermies in ichthyosaurs. Phosphate oxygen isotope and temperature variability within the skeleton from A, the specimen of Ichthyosauria indet., B, the specimen of Palvennia hoybergeti (PMO 222.669), and C, the specimen of Keilhauia sp. (PMO 222.667). Color in bones corresponds to the δ¹⁸O_p difference between bone and the midrange value, (δ¹⁸O_p-max + δ¹⁸O_p-min)/2, of the skeleton. For paired skeletal elements, the mean value is illustrated. Available skeletal elements are shown in light gray, while unavailable elements and skeletal elements with potentially altered δ¹⁸O_p values are shown in dark gray. Skeletal elements (e.g., “limb bones”; Supplementary Table 1) with precise location unknown are not illustrated, while the locations of vertebrae, ribs, and phalanges that have not been found in articulation have been established arbitrarily. The representation of the organisms is not to scale.

Figure 2. Regional heterothermies in plesiosaurs. Phosphate oxygen isotope and temperature variability within the skeleton from A, the specimen of Elasmosauridae indet. (MHNLM.2005.16.1), B, the specimen of Colymbosaurus svalbardensis (PMO 222.663), and C, the specimen of Cryptoclididae indet. (PMO 212.662). Color in bones corresponds to the δ¹⁸O_p difference between bone and the midrange value, (δ¹⁸O_p-max + δ¹⁸O_p-min)/2, of the skeleton. For paired skeletal elements, the mean value is illustrated. Available skeletal elements are shown in light gray, while unavailable elements and skeletal elements with potentially altered δ¹⁸O_p values are shown in dark gray. Skeletal elements (e.g., “limb bones”; Supplementary Table 1) with precise location unknown are not illustrated, while the location of vertebrae, ribs, and phalanges that have not been found in articulation have been established arbitrarily. The representation of the specimens is not to scale.

Figure 3. Regional heterothermies in Metriorhynchus aff. superciliosus. Phosphate oxygen isotope and temperature variability within the skeleton from the specimen of Metriorhynchus aff. superciliosus (MPV 2010.3.610). Color in bones corresponds to the δ¹⁸O_p difference between bone and the midrange value, (δ¹⁸O_p-max + δ¹⁸O_p-min)/2, of the skeleton. When both vertebrae centra and neural spine were measured, the mean value is illustrated. Available skeletal elements are shown in light gray, while unavailable elements are shown in dark gray. The representation of the specimen is not to scale.

Figure 4. Box plots showing the δ¹⁸O_p value distribution of bones sets for the specimens of Ichthyosauria (A, Ichthyosauria indet.; B, Keilhauia sp.; C, Palvennia hoybergeti), Metriorhynchus aff. superciliosus (D) and Plesiosauria (E, Elasmosauridae indet.; F, Colymbosaurus svalbardensis; and G, Cryptoclididae indet.). Asterisks indicate the significance of the observed differences between pair of groups: *p-value < 0.05; **p-value < 0.01; ***p-value < 0.001. Outliers are plotted as white circles. The horizontal bars in the boxes correspond to the medians and the whiskers to the minimum and maximum values. The average analytical error for each bone analyzed is on the order of 0.3‰. Abbreviations: App., Appendicular region; ALL, anterior left limb; ARL, anterior right limb; PLL, posterior left limb; PRL, posterior right limb.

Plesiosaur specimens (Colymbosaurus svalbardensis, Cryptoclididae indet.) have been recovered in articulation (Vincent et al. 2007; Delsett et al. 2016); consequently, the relation between δ¹⁸O_p values and distance from the trunk was investigated (Fig. 5A). No relationship between δ¹⁸O_p values and position in the limbs (Colymbosaurus svalbardensis, Cryptoclididae indet.) or for the vertebral position in the neck (Cryptoclididae indet., Elasmosauridae indet.) was observed (Fig. 5B). To summarize, phosphate oxygen isotope compositions reveal variations for all studied Mesozoic marine reptiles (from 1.0‰ to 4.1‰), comparable to those observed in present-day marine vertebrates (from 1.5‰ to 2.8‰; Séon et al. 2022), and for most there are no significant differences in δ¹⁸O_p from the individual skeletal regions.

Figure 5. A, δ¹⁸O_p values (‰, VSMOW) according to bone position within the limb for the specimen of Cryptoclididae indet. and the specimen of Colymbosaurus svalbardensis. Stylopod set corresponds to the femur or the humerus, Zeugopod set to radius and ulna or tibia and fibula, Metapod set to metacarpals or metatarsals, and Acropod set to phalanges. Abbreviations: AL, Anterior limb indet.; PRL, posterior right limb; PLL, posterior left limb. B, δ¹⁸O_p values (‰, VSMOW) according to position of the vertebra within the cervical region for the specimen of Elasmosauridae indet. and the specimen of Cryptoclididae indet.

Carbonate Bioapatite Oxygen and Carbon Isotope Compositions (δ¹⁸O_c and δ¹³C_c) and Carbonate Content (wt%)

Oxygen and carbon isotope compositions of the carbonate group of bioapatite as well as carbonate content were measured to assess the preservation of the oxygen isotope composition of the phosphate group. All the values are reported in Supplementary Table 1, a synthesis is provided in Table 2 for each specimen, with a graphical representation in Figure 6. The tooth from the pachycormid fish Hypsocormus sp. recovered from the same stratigraphic layer as the specimen of Ichthyosauria indet. yielded a δ¹⁸O_c value equal to 27.30 ± 0.20‰ (VSMOW), a δ¹³C_c value equal to −4.27 ± 0.15‰ (VPDB), and a carbonate content of 4%. The δ¹⁸O_c of calcite recrystallization sampled in the dorsal vertebra of the specimen of Ichthyosauria indet. from Les Ardilles is equal to 24.20 ± 0.10‰ (VSMOW) and to −0.10 ± 0.10‰ (VPDB) for δ¹³C_c.

Figure 6. The δ¹⁸O_c − δ¹⁸O_p differences between teeth and bone types plotted against carbonate content (wt%) of bioapatite. The dashed rectangle zone corresponds to the combination of parameters that would indicate a potential biological preservation of the oxygen isotope composition of the bioapatite of teeth and bones. Silhouettes were made by Gareth Monger for Ichthyosauria and Metriorhynchidae and T. Michael Keesey for Plesiosauria. Images downloaded from PhyloPic and used under a CCBY 3.0 license: https://creativecommons.org/licenses/by/3.0.

Mineralogical Characteristics of Bones and Teeth from Mesozoic Marine Reptiles

Raman spectroscopy was used to assess possible mineralogical changes associated with fossilization and diagenetic processes. The Raman spectral parameters from ichthyosaurian and plesiosaurian bone elements and teeth are reported in Supplementary Table 3 and illustrated in Figure 7. The values of position of the ν1(PO₄³⁻) stretching band range from 939.6 cm⁻¹ to 1023.4 cm⁻¹, whereas FWHM values range from 8.2 cm⁻¹ to 34.7 cm⁻¹ (Supplementary Table 3). No significant differences in ν1(PO₄³⁻) stretching band values between individuals is observed (Mann-Whitney-Wilcoxon statistical test, p-value > 0.05), although some skeletal elements of the specimen of Keilhauia nui appear to have higher values. The Raman results indicate that there is little mineralogical change in the skeletal elements, except for some bone elements belonging to the Keilhauia nui, the Keilhauia sp., and the Colymbosaurus svalbardensis specimens.

Figure 7. The ν1(PO₄³⁻) stretching band position and full width at half maximum (FWHM) measured for Ichthyosauria and Plesiosauria bones and teeth (Supplementary Table 3). Empty symbols represent enamel samples, while solid symbols represent bone samples. The gray zone corresponds to the combination of parameters that would indicate a mineralogical alteration of the bioapatite of teeth and bones. The black arrows indicate the evolution of the position of the ν1(PO₄³⁻) stretching band position and the FWHM depending on the type of ionic substitution within the bioapatite (Thomas et al. 2011).

Ichthyosauria, Plesiosauria, and Metriorhynchidae Body Temperature

Ichthyosauria, Plesiosauria, and Metriorhynchidae body temperature estimates were compared to environmental paleotemperatures, which range from 14 ± 2°C (Slottsmøya Member, Spitsberg, 65.5°N) to 34 ± 1°C (Les Ardilles, France, 34.5°N; Supplementary Table 2). Ichthyosauria, Plesiosauria, and Metriorhynchidae body temperature estimates are available in Supplementary Table 2 and reported against corresponding sea-water temperature estimates in Figure 8.

Figure 8. Body temperature estimates of Ichthyosauria (A), Plesiosauria (B), and Metriorhynchidae (C) from tooth enamel δ¹⁸O_p values (new data in Supplementary Table 1 and published data in Supplementary Table 2) reported against environmental seawater temperature of their living environments. The body temperature range of extant homeothermic endotherm Cetacea is illustrated in red (Morrison 1962; Hampton et al. 1971; Yeates and Houser 2008), while the body temperature range shown in blue corresponds to the temperature range of a strictly ectothermic poikilothermic model organism whose body temperature is equal to the environmental temperature +2°C. Body and environmental temperatures were calculated using the equation of Lécuyer et al. (2013) and considering the latitudinal gradient of δ¹⁸O_sw values of Alberti et al. (2020) and an ¹⁸O-enrichment of body water relative to environmental water of +0.8 ± 0.9‰ for Mesozoic marine reptiles (Séon et al. 2023). Silhouettes were made by Gareth Monger for Ichthyosauria and Metriorhynchidae and T. Michael Keesey for Plesiosauria. Images downloaded from PhyloPic and used under a CCBY 3.0 license: https://creativecommons.org/licenses/by/3.0.

Our estimates indicate that ichthyosaurs had body temperatures ranging from 31°C to 45°C, close to extant Cetacea (Morrison 1962; Hampton et al. 1971; Yeates and Houser 2008; Fig. 8A) and systematically higher than those of the seawater in which they lived (between 3°C and 31°C higher; Supplementary Table 2). No relationship is observed between Ichthyosauria body temperatures and environmental oceanic temperatures (Pearson correlation test, r = 0.3; Fig. 8A). Plesiosauria body temperature range from 27°C to 34°C; these estimates are, as for Ichthyosauria, higher than those of their environment (between 5°C and 11°C; Supplementary Table 2) and correlate moderately with environmental temperatures (Pearson correlation test, r = 0.42; Fig. 8B). The estimated body temperatures of Metriorhynchidae, ranging from 25°C to 32°C, are 3°C to 5°C higher than environmental temperatures (Supplementary Table 2), while being highly correlated (Pearson correlation test, r = 0.96; Fig. 8C).

Discussion

Preservation of the Biological δ¹⁸O_p Values

Before discussing the thermophysiological significance of the δ¹⁸O_p values of bone and tooth bioapatite, pristine preservation of the isotope record needs to be assessed. Biotic and abiotic processes leading to the decomposition, burial, and fossilization of living organisms may alter the pristine isotope composition of bioapatite through processes of secondary precipitation, ion adsorption, or dissolution–recrystallization of bioapatite (Kolodny et al. 1996; Blake et al. 1997; Trueman et al. 2003; Zazzo et al. 2004a,b). Although no method can definitely demonstrate whether the original isotope compositions have been preserved, several ways to assess the preservation state of the oxygen isotope record have been proposed (Iacumin et al. 1996; Kolodny et al. 1996; Fricke et al. 1998; Lécuyer et al. 2003; Pucéat et al. 2004; Zazzo et al. 2004a,b; Tütken et al. 2008; Keenan 2016; Lécuyer and Flandrois 2023; Turner-Walker et al. 2023).

In general, enamel is widely considered to be more resistant to diagenetic processes compared with bones (e.g., Lowenstam and Weiner 1989; Sillen and LeGeros 1991; Kohn et al. 1999; Lee-Thorp and Sponheimer 2003; Gehler et al. 2011), as it possesses a high crystallinity index, being composed of 95% bioapatite crystals. In addition, bioapatite crystals are larger and more compact than those in bone (Pasteris et al. 2008). As a result, the space between the crystals is reduced, limiting fluid circulation, interaction with bacterial and microbial organisms, and precipitation of secondary minerals (LeGeros 1981; Driessens and Verbeeck 1990). The precipitation of secondary minerals leads to mineralogical changes observable through mineralogical characterization methods such as Raman spectroscopy. According to Thomas et al. (2007, 2011), bones and teeth altered by diagenetic processes have peak widths (FWHM) of ν1(PO₄³⁻) at half height of less than 9 cm⁻¹ and band positions of ν1(PO₄³⁻) greater than 964.7 cm⁻¹. Among the analyzed samples, few skeletal elements have parameters that might suggest that they are mineralogically altered (Fig. 6), including different type of bones (skull bones, vertebrae, and phalanges) from specimens of Palvennia hoybergeti, Keilhauia nui, and Colymbosaurus svalbardensis. The sandy appearance of the remains from the Keilhauia nui specimen could explain why so few silver phosphate crystals were recovered at the end of the chemical preparation. Their low P₂O₅ content compared with the other specimens from the same stratigraphic layer and the mineralogical change observed by Raman spectroscopy would indicate the presence of bone alteration (Table 2, Supplementary Fig. 1B).

To assess the effect of diagenetic processes on the oxygen isotope composition of the bones and the teeth, we used the difference between oxygen isotope compositions of biogenic bioapatite phosphate and corresponding structural carbonate (δ¹⁸O_c − δ¹⁸O_p), the carbonate content (CO₃²⁻ %wt) of bioapatite, and the carbon isotope composition of bioapatite structural carbonate (δ¹³C_c) (Fig. 6). Samples that have either δ¹⁸O_c − δ¹⁸O_p differences lower than +7‰ and higher than +11‰ (Iacumin et al. 1996; Pellegrini et al. 2011; Sisma-Ventura et al. 2019) or a carbonate content higher than +13.4% must be regarded with caution, as they may reflect a potential diagenetic alteration (Brudevold and Soremark 1967; Vennemann et al. 2001; McElderry et al. 2013; Wingender et al. 2021). All the samples from Ichthyosauria, Plesiosauria, and Metriorhynchidae have δ¹⁸O_c − δ¹⁸O_p differences lower than the critical value of +11‰, but most of them are below +7‰. Nonetheless, some of them possess carbonate content greater than +13.4%, especially those from bones belonging to the specimen of Ichthyosauria indet. from Les Ardilles (Fig. 6), in line with the secondary calcite mineralization observed during sampling. Moreover, a trend is observed between carbonate contents and δ¹⁸O_c values considering the tooth δ¹⁸O_c value as the pristine biological signal and that of the secondary calcite mineralization, which would form the diagenetic end-member (Supplementary Fig. 4). This relationship also highlights the link between the occurrence of high carbonate content and the ultrastructure (e.g., connected porosity) of the bones, here the vertebrae, as among all the skeletal elements analyzed in the specimen of Ichthyosauria indet., six (n = 6) are vertebrae and two (n = 2) are limb bones. These skeletal elements are among the bones with the highest porosities within Ichthyosauria (Anderson et al. 2019).

Finally, most of the δ¹³C_c values, when plotted against CO₃²⁻ (wt%), conform to two different mixing lines. In the specimen of Cryptoclididae indet. from Spitsbergen, the data point to variable amounts of secondary calcite with δ¹³C_c values of −15‰ (Fig. 9), which indicates a ¹³C-depleted organic carbon source typical of the sulfate reduction zone (Meister and Reyes 2019). In the specimen of Ichthyosauria indet., the data point to variable amounts of more ¹³C-enriched secondary calcite with δ¹³C_c values of ~0‰, in agreement with the value of −0.1‰ measured on intra-bone calcite sampled from the same specimen. These higher isotopic values could either reflect an early precipitation of calcium carbonate close to the sediment–seawater interface with a significant seawater influence or formation below the sulfate reduction zone close to a ¹³C-enriched methanogenic CO₂ source (Meister and Reyes 2019). In both specimens, δ¹³C_c values point to an almost identical, well-preserved end-member with low CO₃²⁻ contents (<10%) and δ¹³C_c values of −8‰ to −10‰ VPDB dominated by structural bioapatite carbonates. The range of δ¹³C_c values recorded for these CO₃²⁻-poor samples of marine reptiles are identical to that previously reported for ancient air-breathing vertebrates (Séon et al. 2020), but substantially lower than the δ¹³C_c values recorded in the pachycormid Hypsocormus sp. teeth measured from Les Ardilles and from coeval fossil fishes (−5.47‰ to +3.15‰ VPDB; Séon et al. 2020). As expected, air-breathing reptiles have lower δ¹³C_c values than fish. In fact, the δ¹³C values of air-breathing reptiles mainly reflect those of their diet, with isotopic fractionation depending on their digestive physiology (Passey et al. 2005). In fish, on the other hand, the δ¹³C values of carbonate come from the diet, but also from the contribution of a large amount of dissolved inorganic carbon in the surrounding water (McConnaughey et al. 1997), which has a high δ¹³C value (~0‰ to 2‰; Santos et al. 2011). These features demonstrate that, despite the different taphonomic history suggested by the diagenetic δ¹³C_c values at the two sites, the geochemical signature of the CO₃²⁻-poor samples is preserved and hence paleobiologically informative.

Figure 9. The δ¹³C_c values (‰, VPDB) of bone elements and teeth in ichthyosaurs and plesiosaurs as a function of carbonate content (wt%). The values corresponding to the Hypsocormus sp. tooth found in the same stratigraphic layer as the Ichthyosauria indet. specimen are also plotted. The dotted line marks the boundary between mineralized elements with a carbonate content greater than or less than +13.4%.

From here on, based on the evidence provided by mineralogical and isotope proxies, we assume that δ¹⁸O_p values from most of the Ichthyosauria, Plesiosauria, and Metriorhynchidae samples, except the Keilhauia nui specimen (PMO 222.655), have kept at least a significant part of their original oxygen isotope compositions and might be interpreted in terms of thermophysiology. As a precaution, we removed all the skeletal elements for which doubt was expressed regarding the preservation of the original isotopic signal (mineralogical, carbonate content, and isotope clues; Supplementary Table 1).

Intraskeletal Body Temperature Distribution in Ichthyosauria, Plesiosauria, and Metriorhynchidae

The oxygen isotope composition of air-breathing aquatic tetrapods is controlled by their body temperatures, the proportion and isotopic composition of metabolic and drinking waters, and the breathing-induced isotope enrichment factor relative to these sources of waters. The possible occurrence of salt glands in at least some taxa of Metriorhynchidae (Fernández and Gasparini 2000, 2008; Gandola et al. 2006; Cowgill et al. 2023), Ichthyosauria (Wahl 2012; Campos et al. 2020; Massare et al. 2021), and Plesiosauria (Buchy et al. 2006; O’Gorman and Gasparini 2013; Páramo-Fonseca et al. 2019) indicates that they would have been able to drink seawater to maintain their water balance. We thus consider that seawater was the main source of water and thus oxygen, as is the case in extant reptiles with salt glands (Rash and Lillywhite 2019). Because the δ¹⁸O_sw value influencing δ¹⁸O_p values can vary over time and specific regions, the timing of oxygen incorporation into the bone and tooth bioapatite needs to be identified to accurately compare the δ¹⁸O_p values of different skeletal elements and identify potential regional heterothermies.

Similar to most reptiles, Ichthyosauria, Plesiosauria, and Metriorhynchidae continuously replaced their teeth. According to the studies conducted by Kear et al. (2017) and Maxwell et al. (2011a,b, 2012), the mineralization time for the teeth was estimated to be at most 2 to 3 years for plesiosaurs and approximately 1 year for ichthyosaurs. Although there is currently no available information regarding the tooth renewal time of Metriorhynchidae, it can be assumed that it was comparable to that of extant crocodylomorphs, which is approximately 3 to 15 months (Erickson 1996; Finger et al. 2019). In Ichthyosauria and Plesiosauria, bones underwent regular remodeling throughout life, as evidenced by studies conducted by Kolb et al. (2011), Delsett and Hurum (2012), Houssaye et al. (2014), and O’Keefe et al. (2019). For the Metriorhynchidae, cyclical growth marks in some bones indicate little bone remodeling and lack of or few secondary osteons (Hua and de Buffrénil 1996; de Buffrénil et al. 2021); thus, the time recorded in the bones is likely to be slightly longer. The δ¹⁸O_p values of the teeth and bones of Ichthyosauria, Plesiosauria, and Metriorhynchidae can be compared, as these tissues record a period ranging from one to several years. Furthermore, the annual and seasonal fluctuations in modern δ¹⁸O_sw values in the open ocean and the depositional environments in which the marine reptiles studied were discovered (Hammer et al. 2011; Lebrun and Courville 2013; Delsett et al. 2016) are typically only a few tenths of a per mille (~ ±0.5‰; LeGrande and Schmidt 2006), so this alone cannot fully explain the observed intraindividual δ¹⁸O_p variations in ichthyosaurs and plesiosaurs of more than 1‰ to 3‰. Thus, we postulate that the δ¹⁸O_p variability recorded in bones and teeth from the Ichthyosauria, Plesiosauria, and Metriorhynchidae studied mostly reflects temperature variations. Nevertheless, this supposition is contingent upon the absence of migration for these animals. It is possible that migrations between high and low latitudes, as well as between open-marine environments and restricted or brackish environments, may exert a considerable influence in intraskeletal δ¹⁸O_p variability. A further limitation of this study is the lack of information available regarding the migratory patterns of these taxa, which precludes an investigation into this possibility.

Assuming that the temporal variation in the oxygen isotope composition of the body water of Ichthyosauria, Plesiosauria, and Metriorhynchidae is approximately constant over time, as was shown in modern vertebrates (cetaceans: Séon et al. 2023; sheep: Green et al. 2018), we can infer the range of mineralization temperatures across their bodies using the δ¹⁸O_p values of the geochemically best-preserved skeletal elements. Thus, using the equation of Lécuyer et al. (2013), we obtain a mineralization temperature range of approximately 8°C to 12°C for Ichthyosauria (Fig. 1), 4°C to 12°C for Plesiosauria (Fig. 2), and on the order of 4°C to 5°C for Metriorhynchus aff. superciliosus (Fig. 3). In comparison, the mineralization temperature range of skeletal elements estimated from bone δ¹⁸O_p is around 8°C to 12°C for extant homeothermic endotherms Cetacea and Pinnipedia, 10°C for poikilothermic endotherm Atlantic bluefin tuna, and 15°C for poikilothermic endotherm swordfish (Séon et al. 2022, 2024), while direct measurement of body temperature range for poikilothermic ectotherms such as crocodiles and turtles has been estimated to be ~10°C (Dunham et al. 1989; Markwick 1998). Note that for specimens of Elasmosauridae indet. and Metriorhynchus aff. superciliosus, we were unable to analyze distal skeletal elements belonging to the appendicular skeleton (not preserved), which may explain why the mineralization temperature ranges for these specimens are narrower than for the other analyzed taxa (Figs. 2, 3). For all specimens, the comparison between the different skeletal regions does not allow us to identify heterothermies in Mesozoic marine reptiles. Indeed, we did not observe the expected colder limbs and tails and differences in neck temperatures in Plesiosauria. This could, in part, be the result of a homogenization of the δ¹⁸O_p values of bones, or at least a reduction in the original differences, by the various postmortem diagenetic processes and not detected by the methods used here (Raman spectroscopy, δ¹⁸O_c − δ¹⁸O_p offset), as has been hypothesized previously by other authors who have studied the intraskeletal variations in the δ¹⁸O of extinct organisms (Barrick and Showers 1994). Significant differences in δ¹⁸O_p values of the dorsal region (vertebrae, ribs, girdle bones) and the limbs are only observed for the specimen of Colymbosaurus svalbardensis (Fig. 4). However, the differences observed are contrary to what we would have expected. This would indicate a higher average temperature of mineralization of the hindlimbs compared with the skeletal elements of the trunk (vertebrae, ribs, pectoral, and pelvic bones). In the specimen of Colymbosaurus svalbardensis (PMO 222.663), skeletal elements of the anterior region (dorsal vertebrae, ribs, left and right forelimbs) have significantly higher values than those of the posterior region (left and right hindlimbs and caudal vertebrae). These significant differences in δ¹⁸O_p appear unusual in terms of thermoregulatory strategies and could be the result of a preservation difference between the more weathered anterior and posterior skeletal elements of the specimen (Delsett et al. 2016). Regarding the limbs of Plesiosauria, the osteo-histological study carried out by Delsett and Hurum (2012) on two specimens of Colymbosaurus svalbardensis (PMO 216.838, PMO 216.377) indicates that the most distal limb bone elements display growth striations, perhaps resulting from cyclic growth. In our study, our data reveal no clear relationship between δ¹⁸O_p values and the position of the skeletal element within the limb (Fig. 5A). Thus, no body temperature gradient can be identified in the limb of Plesiosauria based on the geochemical approach. An identical conclusion can be drawn regarding the possible presence of a temperature gradient in the neck of Plesiosauria, as our detailed mapping of the cervical region shows no relationship between δ¹⁸O_p and vertebral position (Fig. 5B). Finally, the only systematic trend recorded in our dataset concerns teeth, which, when analyzed alongside bones from the same organism, exhibit higher δ¹⁸O_p values than the bones, including skull elements located close to the analyzed teeth in Keilhauia sp. specimen PMO 222.667 and Palvennia hoybergeti specimen PMO 222.669 (Fig. 4). Interestingly, similar observations have been made in odontocetes (Barrick et al. 1992; Séon et al. 2022), which would indicate that teeth mineralize at lower temperatures than skull bones (lack of insulating tissue on the rostrum). However, the number of teeth analyzed makes it impossible to determine whether these differences are significant or not. Furthermore, in the monophyodont odontocetes, there is the influence of lactation, which probably includes uncertainties in the interpretation of odontocete tooth δ¹⁸O_p and makes them incomparable with Ichthyosauria. On the other hand, Séon et al. (2022) observed no δ¹⁸O_p differences between teeth and skull bones in the polyphyodont Atlantic bluefin tuna. Differences in δ¹⁸O_p values between teeth and bones in Ichthyosauria could be explained by a difference in mineralization temperature at the rostrum of about 3°C to 4°C, but the lower number of values available for each of these observation series means that we must be cautious about such considerations. Indeed, the observed higher values from the teeth could also result from differences in isotopic preservation between tooth enamel and bones.

Reassessment of Ichthyosauria, Plesiosauria, and Metriorhynchidae Body Temperature and Thermoregulatory Strategies

Previous isotope-based reconstructions of extinct marine reptile body temperatures opted for an ¹⁸O-enrichment of body water relative to drinking water of +2‰ (Bernard et al. 2010; Séon et al. 2020; Leuzinger et al. 2023), originally estimated by Amiot et al. (2007) from body fluids of young Nile crocodiles (Crocodylus niloticus), and Barrick et al. (1999) for turtles (Chrysemys sp.). These species have a semiaquatic ecology, low body mass (1.2 kg to 5 kg), and relatively low metabolic rates. We therefore consider that a +2‰ enrichment of body water relative to environmental water may not be the most appropriate value for Mesozoic fully marine reptiles. We considered instead the +0.8 ± 0.9‰ enrichment value measured by Séon et al. (2023) for fully aquatic air-breathing vertebrates (Orcinus orca and Tursiops truncatus; 150 to 3600 kg), which appears more appropriate considering the ecology, the size, and the putative thermo-metabolism of the Ichthyosauria, Plesiosauria, and Metriorhynchidae studied here.

To reassess Ichthyosauria, Plesiosauria, and Metriorhynchidae body temperatures, we have favored the use of tooth enamel, as its robustness against diagenetic alteration has been demonstrated in previous studies (Lowenstam and Weiner 1989; Sillen and LeGeros 1991; Kohn et al. 1999; Lee-Thorp and Sponheimer 2003; Gehler et al. 2011). A last important change in our work relative to previous studies is the use of a spatially variable δ¹⁸O_sw values calculated using the hemispheric gradient proposed by Alberti et al. (2020). This gradient places δ¹⁸O_sw highest values in subtropical areas subject to high evaporation and lowest values in high-latitude seas receiving high fluvial input from surrounding large continental areas, consistent with both modeling (Zhou et al. 2008) and empirical evidence (Letulle et al. 2022). We thus consider this scenario as far more realistic than the δ¹⁸O_sw hypothesis used previously (Bernard et al. 2010; Séon et al. 2020). Together, the use of revised values and spatial changes in δ¹⁸O_sw values produce new body temperature estimates of Ichthyosauria, Plesiosauria, and Metriorhynchidae that are 3°C to 6°C lower than estimated previously (Bernard et al. 2010; Séon et al. 2020; Supplementary Table 2, Supplementary Fig. 5).

According to our new estimates, Ichthyosauria maintained their body temperatures at a constant value well above environmental values (Fig. 8A) and can be considered as homeothermic endotherms, consistent with previous isotope-based studies (Bernard et al. 2010; Séon et al. 2022). Nevertheless, the body temperature values estimated (45 ± 2°C) for the ichthyosaurs of the Slottsmøya Member (Palvennia hoybergeti and Keilhauia sp.) are higher than those of present-day homeothermic endothermic marine vertebrates, based on a δ¹⁸O_sw value of −1.5‰ from the equation of Alberti et al. (2020). However, this equation is calibrated between 0° and 40°N, and it is probable that this δ¹⁸O_sw value of −1.5‰ is overestimated, given that ocean basins such as the Arctic Ocean at the time were landlocked and were therefore likely to be subject to freshwater input from precipitation runoff. Estimates from numerical modeling for the middle Cretaceous indicate δ¹⁸O_sw values of −5 ± 2‰ for this region (Zhou et al. 2008). Based on this modeled δ¹⁸O_sw value, the estimated body temperature of ichthyosaurs from the Slottsmøya Member would have been 31 ± 4°C, confirming the homeothermic and endothermic nature of Ichthyosauria. In contrast to previous studies, however, our results indicate that Plesiosauria, as well as Metriorhynchidae, had body temperatures higher than but covarying with environmental temperatures, and hence should be considered poikilothermic endotherms. We note that the estimated Metriorhynchidae body temperatures are closer to those of their environmental values than in Plesiosauria (Fig. 8B,C). These thermoregulatory strategies are largely consistent with other indices linked to paleobiogeography, morphology, locomotion, and estimated metabolic rates.

Indeed, Ichthyosauria would have been able to live in high-latitude environments (Delsett et al. 2016; Rogov et al. 2019) where ocean paleotemperatures were too low for ectothermic turtles and crocodylomorphs to survive, as evidenced by their absence in high-latitude deposits from the same stratigraphic levels (Rich et al. 2002; Bardet et al. 2014). This would have been greatly facilitated by their high metabolic rates (de Buffrénil and Mazin 1990; Nakajima et al. 2014; Anderson et al. 2019) and their fusiform morphology, which is very effective in limiting heat loss to the environment (Innes et al. 1990; Gearty et al. 2018), coupled with a subcutaneous layer of adipose tissue reducing thermal conductivity (Lindgren et al. 2018; Delsett et al. 2022). Nevertheless, this could only be the case for the most recent ichthyosaurs in the evolution of the lineage, such as parvipelvian and thunniform ichthyosaurs of the Ophthalmosauridae family. Indeed, primitive Ichthyosauria had more anguiliform morphologies and, as far as we know, no adipose tissue has been found in these specimens (Motani 2005; Moon and Stubbs 2020).

For Plesiosauria, the characterization of their thermoregulatory strategy is more ambiguous, and our results differ interestingly from those for ichthyosaurs and to some extent contrast with previous studies (Bernard et al. 2010; Séon et al. 2020; Leuzinger et al. 2023). High metabolic rates close to those of modern endotherms are supported by the plesiosaurs’ paleobiogeographic distribution (Bardet et al. 2014; Delsett et al. 2016; Zverkov et al. 2021), osteo-histological evidence (Wiffen et al. 1995; Kear 2006a; Delsett and Hurum 2012; Fleischle et al. 2018; O’Keefe et al. 2019; Sander and Wintrich 2021), and the identification and quantification of compounds resulting from the degradation of lipids and carbohydrates by infrared spectroscopy (Wiemann et al. 2022). So far, no evidence for a subcutaneous layer of adipose tissue has been reported in Plesiosauria, but this might be a sampling artifact, as specimens exhibiting soft-tissue preservation are exceptionally rare (Vincent et al. 2017). On the other hand, our new data from teeth show that their body temperatures covaried to some extend with environmental temperatures. One possible explanation that may reconcile these apparently contradictory observations is that Plesiosauria produced heat through their locomotor muscles. This heat production strategy is indeed used by sea turtles with the same type of locomotion and enables them to have core body temperatures higher than the rest of the body (Standora et al. 1982; Sato et al. 1994). The heat generated by muscle contraction may then partly be conserved by the inertia of their body mass (gigantothermy), as in the case of leatherback turtles (Paladino et al. 1990; Sato 2014), whose bone histology studies show evidence of elevated growth rate and high metabolic rate (Nakajima et al. 2014; Wilson 2023). The existence of a comparable heat production system in Plesiosauria may thus partly explain the recorded difference between their body temperatures and their environments as well as their inability to maintain their body temperature at a constant level in the thermally least favorable environments (high latitudes).

Metriorhynchidae differed from the previous two in terms of their geographic distribution, which was restricted to tropical zones (Bardet et al. 2014), and show osteo-histological features (Hua and de Buffrénil 1996; de Buffrénil et al. 2021) indicative of a poikilothermic ectothermic thermoregulatory strategy. While Metriorhynchidae were initially considered to be active hunters (Massare 1987; Young and de Andrade 2009; de Andrade et al. 2010; Young et al. 2012), it was later postulated that some species could also have been scavengers and opportunists (Hua et al. 2024), a strategy that does not require the production of long and intense efforts, in agreement with the estimates of their movement speed (Massare 1988; Massare et al. 1994; Gutarra et al. 2023). Although their body temperatures have been considered as slightly higher than those of their environments (Séon et al. 2020), our revised estimates point to much lower differences than previously estimated. Interestingly, Gienger et al. (2017) showed that the standard metabolic rate values measured in saltwater crocodiles (Crocodylus porosus) are 36% higher than those of other taxa such as American alligator (Alligator mississippiensis) or freshwater crocodile (Crocodylus johnstoni). However, this difference in metabolic rate is not visible in the osteo-histological sections taken from these species (de Buffrénil et al. 2021). Metriorhynchidae may therefore have had a higher metabolic rate, but this difference would not be recorded in bone tissue. We therefore speculate that the production of metabolic heat could have been linked to vital organs such as intestines or liver as well as the locomotor muscles, which, as in tuna, would have been internalized and close to the spinal column and would have limited heat loss to the surrounding aquatic environment (Graham and Dickson 2004).

Conclusion

The reassessment of oxygen isotope–based body temperature estimates of Ichthyosauria, Plesiosauria, and Metriorhynchidae have led to a reconsideration of their thermoregulatory strategies. The intraskeletal δ¹⁸O_p variability in four specimens of Ichthyosauria, three specimens of Plesiosauria, and one specimen of Metriorhynchus aff. superciliosus did not allow us to characterize and locate regional heterothermies. This may be due to the impact of diagenetic processes on the original oxygen isotope composition of bone bioapatite. The newly measured δ¹⁸O_p from the teeth of Ichthyosauria specimens from high paleolatitudes confirms that these organisms had body temperatures higher than those of their living environments and were probably homeothermic endotherms with body temperature ranging between 31°C and 37°C. In contrast to Ichthyosauria, it appears that the body temperatures of Plesiosauria and Metriorhynchidae were influenced by changes in environmental temperatures. Therefore, they were most likely poikilothermic endotherms, such as some extant sea turtles and tunas. Nevertheless, the thermoregulatory strategy of Metriorhynchidae remains difficult to characterize, because their body temperature estimates are very close to those of their environments, and they could be defined as either poikilothermic endotherms or poikilothermic ectotherms. Further investigations are needed to clearly define their thermoregulatory strategy. Finally, the independency of Ichthyosauria body temperatures relative to environmental temperatures, confirmed by our study, places these organisms as promising tracers of spatiotemporal changes in δ¹⁸O_sw values that could allow for improved Mesozoic paleoclimate reconstructions.