Thermophysiologies of Jurassic marine crocodylomorphs inferred from the oxygen isotope composition of their tooth apatite

1. Introduction

Extant crocodylians are archosaurs that rely on environmental sources of heat in order to raise and maintain their body temperature by behavioural thermoregulation in a restricted range bracketed by ‘critical minimum' and ‘critical maximum' temperatures [1,2]. Within this critical range, crocodylians tend to keep their body temperature within a narrower activity range from about 25°C to 40°C as determined empirically for a few extant species (see Markwick [3] for a review). Consequently, the spatial and temporal distribution of crocodylians is limited by the temperatures of their living environments and by their seasonal fluctuations. Owing to their rather conservative growth morphology and their restricted latitudinal distribution today, and in the geologic record, extant and fossil representatives of the crown group have been used as climate proxies for more than a century [3–6]. Based on observations of extant representatives of crown-group Crocodylia [7], Markwick [3] proposed that the occurrence of fossil representatives implies a living mean annual air temperature ≥ 14.2°C and a coldest month mean temperature ≥ 5.5°C. These minimum limits have been tentatively used to constrain the climatic environment of long-extinct crocodylomorphs [8], as well as the more distantly related clade Choristodera [9].

The crocodylomorph clade Thalattosuchia became secondarily adapted to marine life during the Mesozoic, and is subdivided into the families Teleosauridae [10] and Metriorhynchidae [11]. While teleosaurids retained a morphology reminiscent of typical semi-aquatic longirostrine crocodylomorphs, in having extensive osteoderm coverage and limbs adapted for terrestrial locomotion [12], metriorhynchids had a more hydrodynamic body plan, with a hypocercal tail, and hydrofoil-like forelimbs [13,14]. The bone histology of Callovian teleosaurids and metriorhynchids has been used to hypothesize that thalattosuchians were ectothermic (i.e. they relied on environmental sources of heat in order to raise their body temperature) and poikilothermic (i.e. their body temperature varied along with that of their environment), with teleosaurids being capable of mouth-gape basking on shore, whereas metriorhynchids thermoregulated differently by staying close to the water surface [15]. Hua & de Buffrénil [15] did note that young metriorhynchids may have had a faster growth rate than wild extant crocodylians.

Martin et al. [16] analysed the diversity of marine crocodylomorphs through the Mesozoic and Palaeogene, and observed a significant correlation between sea surface temperatures (SSTs) and generic diversity within four lineages, including the teleosaurids. Teleosaurids, at least European representatives, experienced a diversity crash at the end of the Jurassic during a global cooling of Tethyan waters [17], but according to Fanti et al. [18] they may have continued to survive until at least the Hauterivian along the southern coast of the Tethys Sea. Metriorhynchids, however, appear to have experienced an explosive radiation during the Callovian, surviving through the Late Jurassic until the Aptian [19]. When metriorhynchids became extinct is currently unclear. Young et al. [14] hypothesized a two-step extinction for Metriorhynchidae: first, a diversity crash at the end Jurassic, then a final extinction during the cold icehouse interval of the Valanginian [20]. However, recent re-evaluations of Cretaceous fossils and updated phylogenetic studies have disproved both steps of this hypothesis, with post-Valanginian metriorhynchid specimens known and no fewer than four metriorhynchid lineages that crossed the Jurassic–Cretaceous boundary [19,21,22]. The global thalattosuchian fossil record, however, is still poor; the known diversity of this clade is still heavily biased by the European rock record, as well as global marine record sampling biases for specific time spans (e.g. Aalenian, Oxfordian, Early Cretaceous). Regardless of the geological megabiases, there is still a conspicuous diversity mismatch within Thalattosuchia, which raised the question whether metriorhynchids evolved a distinct thermoregulatory strategy, such as endothermic capabilities, that would explain their diversity and survival under cool SSTs [16].

In order to investigate the thermophysiology of metriorhynchids, we analysed the oxygen isotope composition of the enamel phosphate of their teeth (δ¹⁸O_p). Indeed, the δ¹⁸O_p value of vertebrate apatite (the mineral constituting bone, teeth and some fish scales) depends on the animal's body water δ¹⁸O_bw value, as well as its body temperature [23–25]. For air-breathing vertebrates, the body water has a δ¹⁸O_bw value controlled by oxygen input coming from drinking water, food and inhaled oxygen (through metabolic water production), as well as oxygen loss as water vapour through transcutaneous evaporation, sweat, exhaled vapour, and liquid water in urine and faeces, some of these losses being associated with oxygen isotope fractionation [23,26]. A fractionation equation that relates the apatite phosphate δ¹⁸O_p value to that of body water and body temperature (T_b) can be adapted from the phosphate-water temperature scale previously established by Longinelly & Nuti [24] and recently updated by Lécuyer et al. [27]. Such an equation has proven to be valid for a large range of invertebrate and vertebrate bioapatites [24,28–30]:

This relationship is commonly used to reconstitute past SSTs based on fish apatite, as most fish have a body temperature similar to that of their surrounding water, and a body water δ¹⁸O_bw value equal to the ambient one [17,20,31,32]. Air-breathing vertebrates have a body water ¹⁸O-enriched relative to environmental water, the magnitude of which depends on the amount of body water loss through exhaled water vapour and transcutaneous evaporation. Direct measurements have shown that terrestrial mammals and birds have body water from about 4 to 7‰ ¹⁸O-enriched relative to their drinking water [33–35], whereas the body waters of semi-aquatic to aquatic crocodylians, turtles and birds are about 2 to 3‰ ¹⁸O-enriched relative to their drinking water [33,36,37].

In this study, we estimate the body temperatures of metriorhynchid and teleosaurid thalattosuchians recovered from five Jurassic localities in England and France. By comparing T_b of thalattosuchians with those of associated ichthyosaurs or plesiosaurs, as well as with their ambient SST, we show that metriorhynchids may have evolved some degree of endothermic-like thermophysiology, whereas teleosaurids retained a more typical ecto-poikilothermy.

2. Material and methods

(a) Sample collection

We analysed 88 fossil teeth of Jurassic fish and marine reptiles for their oxygen isotope composition of phosphate (δ¹⁸O_p), as well as for their oxygen (δ¹⁸O_c) and carbon (δ¹³C_c) isotope composition of apatite carbonate. The first English locality is Smallmouth Sands, Dorset (figure 1), from the lower part of the Kimmeridge Clay Formation, dated as lower Kimmeridgian [39]. The second English locality is the Oxford Clay Formation Peterborough clay pits, a world-famous fossiliferous collection of sites, dated to the middle Callovian, which has yielded a rich marine fauna along with terrestrial elements [40]. The first French locality is Les Vaches Noires of Normandy, dated as upper Callovian, where the Marnes de Dives Formation crops out, consisting of marls where a rich fauna composed of marine and terrestrial elements has been recovered [41]. The second French locality is the excavation of Les Lourdines near the eponymous quarry, dated as middle Callovian, which is composed of white limestone (calcaire des Lourdines), where a marine fauna and some terrestrial plants have been recovered [42]. Finally, the locality of Cintheaux, dated as Bathonian, consists of the Pierre de Caen limestone, which has yielded a marine fauna with some terrestrial elements [43]. Studied tooth specimens include those of metriorhynchid and teleosaurid thalattosuchians, ichthyosaurs and plesiosaurs, as well as of chondrichthyans and osteichthyans (electronic supplementary material, table S1). When possible, tooth enamel was sampled using a spherical diamond-tipped drill bit. For smaller teeth, the bulk enamel and dentine were ground using an agate mortar and pestle.

• Download figure
• Open in new tab
• Download PowerPoint

(b) Analytical techniques

(i) Oxygen isotope analysis of biogenic apatite phosphate

Apatite powders were treated following the wet chemistry protocol described by Crowson et al. [44] and slightly modified by Lécuyer et al. [45]. This protocol consists in the isolation of phosphate $(\mathrm{P}{\mathrm{O}}_4^{3-})$ from apatite as silver phosphate (Ag₃PO₄) crystals using acid dissolution and anion-exchange resin. For each sample, 20–30 mg of enamel powder was dissolved in 2 ml of 2 M HF. The CaF₂ residue was separated by centrifugation and the solution was neutralized by adding 2.2 ml of 2 M KOH. Amberlite™ IRN78 anion-exchange resin beads were added to the solution to isolate the $\mathrm{P}{\mathrm{O}}_4^{3-}$ ions. After 24 h, the solution was removed, and the resin was rinsed with deionized water and eluted with 27.5 ml of 0.5 M NH₄NO₃. After 4 h, 0.5 ml of NH₄OH and 15 ml of an ammoniacal solution of AgNO₃ were added and the solutions were placed in a thermostated bath at 70°C for 7 h, allowing the precipitation of Ag₃PO₄ crystals. Oxygen isotope compositions were measured using a high-temperature elemental analyser interfaced in continuous flow mode to an isotopic ratio mass spectrometer [46] at the Plateforme d'Ecologie Isotopique du Laboratoire d'Ecologie des Hydrosystèmes Naturels et Anthropisés (LEHNA, UMR5023, Université Claude Bernard Lyon 1). For each sample, five aliquots of 300 μg of Ag₃PO₄ were mixed with 300 μg of pure graphite powder loaded in silver foil capsules. Pyrolysis was performed at 1450°C using a vario PYRO cube™ elemental analyser (Elementar, Germany) interfaced in continuous flow mode with an Isoprime™ isotopic ratio mass spectrometer (Elementar UK). Measurements have been calibrated against silver phosphate precipitated from the NBS120c (natural Miocene phosphorite from Florida), as well as against the NBS127 (barium sulfate precipitated using seawater from Monterey Bay, California, USA). The value of NBS120c was fixed at 21.7‰ (Vienna standard mean ocean water, V-SMOW) according to Lécuyer et al. [45], and that of NBS127 set at the certified value of 9.3‰ (V-SMOW; see [47,48]) for correction of instrumental mass fractionation during CO isotopic analysis. Silver phosphate precipitated from standard NBS120c along with the silver phosphate samples derived from fossil bioapatites was repeatedly analysed (δ¹⁸O_p = 21.72 ± 0.22‰, n = 20) to ensure that no fractionation occurred during the wet chemistry. Data are reported as δ¹⁸O values with respect to V-SMOW (in ‰ δ units).

(ii) Oxygen and carbon isotope analysis of biogenic apatite carbonate

In order to remove potential organic contaminants as well as secondarily precipitated calcite, about 10 mg of apatite powder was pre-treated following the protocol of Koch et al. [49]. Powders were washed with a 2% NaOCl solution to remove organic matter, then rinsed five times with double deionized water and air-dried at 40°C for 24 h. Acetic acid (0.1 M) was then added and left for 24 h, after which the powder was again rinsed five times with double deionized water and then air-dried at 40°C for 24 h. The powder/solution ratio was kept constant at 0.04 g ml⁻¹ for both treatments. Stable isotope compositions of carbonate oxygen and carbon were carried out at the Plateforme d'Ecologie Isotopique du Laboratoire d'Ecologie des Hydrosystèmes Naturels et Anthropisés (LEHNA, UMR5023). The measurements were performed using an isoFLOW system connected on-line in continuous flow mode to a precisION mass spectrometer (Elementar UK). For each sample, two aliquots of 2 mg of pretreated apatite powder were loaded in LABCO Exetainer^® 3.7 ml soda glass vials, round bottomed with Exetainer caps (LABCO UK), and were reacted with anhydrous phosphoric acid. The reaction took place at 90°C in a temperature regulated sample tray. The CO₂ gas generated during the acid digestion of the carbonate sample was then transferred to the mass spectrometer via the centrION interface. Calibrated CO₂ gas was used as the monitoring gas. Typical reproducibilites are 0.05 and 0.07‰, respectively, for δ¹³C and δ¹⁸O measurements. For tooth apatite, the acid fractionation factor α (CO₂–apatite carbonate) of 1.00773 determined for the NBS120c phosphate rock reference material was selected [50]. The calibrated materials used were Carrara marble (δ¹⁸O_V-PDB = −1.84‰, δ¹³C_V-PDB = +2.03‰ [51]), NBS18 (δ¹⁸O_V-PDB = −23.2‰, δ¹³C_V-PDB = −5.01‰) and NBS120c (δ¹⁸O_V-PDB = −1.13‰, δ¹³C_V-PDB = −6.27‰ [50]), Isotopic compositions are quoted in the standard δ notation relative to V-SMOW for oxygen and V-PDB (Vienna Pee Dee belemnite) for carbon.

3. Results

Oxygen isotope compositions of apatite phosphate (δ¹⁸O_p) and apatite carbonate (δ¹⁸O_c), and carbon isotope compositions of apatite carbonate (δ¹³C_c) are reported in electronic supplementary material, table S1 along with published δ¹⁸O_p values of teeth and bones of marine vertebrates [52–54]. Analysed teeth have δ¹⁸O_p values ranging from 18.3 to 21.8‰ V-SMOW, δ¹⁸O_c values ranging from 24.1 to 28.7‰ V-SMOW and δ¹³C_C values ranging from −11.2 to 7.7‰ V-PDB. At the three sites of Smallmouth Sands, Peterborough and Les Vaches Noires, ichthyosaurs and plesiosaurs have slightly lower mean δ¹⁸O_P values than those of co-occurring thalattosuchians (electronic supplementary material, table S2).

For each locality, SST was calculated from the δ¹⁸O_p values of fish using equation (1.1) (electronic supplementary material, table S2), considering that T_b ≈ T_sw, and that δ¹⁸O_bw ≈ δ¹⁸O_sw [28]. Because average seawater δ¹⁸O_sw value may have varied between −1 and 0‰ V-SMOW depending on the amount of seawater stored as polar ice [55], an average value of −0.5‰ was arbitrarily selected for temperature calculation, keeping in mind that the associated error in temperature calculation is about 2.3°C (based on the slope of 4.5 of equation (1.1)).

The body temperatures of studied marine reptiles have also been estimated using equation (1.1) and assuming a seawater–body water ¹⁸O-enrichment of 2‰, a general enrichment previously observed among semi-aquatic and aquatic air-breathing vertebrates, including extant crocodylomorphs [33,36,37]. While ichthyosaurs and plesiosaurs have a body temperature within the 32–40°C range compatible with their known endo-homeothermy [53], teleosaurids have lower T_b ranging from 27 to 31°C, and metriorhynchids show intermediate body temperatures ranging from 29 to 37°C (electronic supplementary material, table S2).

4. Discussion

(a) Original preservation of the stable isotope compositions

Before discussing the thermophysiological significance of the oxygen and carbon isotope compositions of vertebrate apatites, pristine preservation of the isotopic record needs to be assessed. Indeed, biotic and abiotic processes leading to the decomposition, burial and fossilization of living organisms may alter the original isotopic composition of bioapatite through processes of secondary precipitation, ion adsorption or dissolution–recrystallization of bioapatite [56–61]. Although no method can definitely demonstrate whether the original isotope compositions have been kept, several ways to assess the preservation state of the isotopic record have been proposed [57,58,61–65]. In modern skeletal tissues of vertebrates, carbonate and phosphate precipitate close to equilibrium with body water, so the δ¹⁸O_p and δ¹⁸O_c values are positively correlated. Because isotopic exchange rates between carbonate-water and phosphate-water are significantly different, re-equilibration of both compounds during diagenesis is not expected and altered enamel should show isotopic shifts from the empirical δ¹⁸O_p–δ¹⁸O_c line. Therefore, it is expected that the distribution of pristine δ¹⁸O values of fish and reptile tooth enamel should follow a line with a slope close to unity, mimicking those established between the δ¹⁸O_c and δ¹⁸O_p values of modern mammals [61,63,66–68]. Despite the narrow range in the distribution of oxygen isotope compositions (figure 2), both δ¹⁸O_c and δ¹⁸O_p values fall within the range observed in extant and fossil marine vertebrates [17,69,70]. This correlation shows that the oxygen isotope compositions of structural carbonate in both tooth and bone apatites have preserved to a certain degree their original record.

• Download figure
• Open in new tab
• Download PowerPoint

A clue to the primary preservation of stable carbon isotope composition of apatite carbonate is the systematic and significant difference in δ¹³C_c values between fish and marine reptiles (figure 3). Air-breathing reptiles have δ¹³C_c values mainly reflecting those of their diets, with an isotope fractionation that depends on their digestive physiology [71], as expected. In aquatic environments, the relationship between fish carbonate and diet δ¹³C values is complicated as a substantial amount of the carbon may be derived from dissolved inorganic carbon in their ambient water [72,73] with a higher δ¹³C value [74]. Finally, the weight percentage of carbonate in analysed fossil apatites (electronic supplementary material, table S1) lies within the expected biological range of modern vertebrate apatites of 2–13% [70,75,76]. From these lines of evidence, we can assume that oxygen isotope compositions of apatite phosphate and carbonate have kept at least a significant part of their original information, and might be interpreted in terms of seawater temperature and thermophysiology.

• Download figure
• Open in new tab
• Download PowerPoint

(b) Body temperature reconstruction

In a previous study of marine reptile thermophysiology, Bernard et al. [53] interpreted the oxygen isotope differences between marine reptiles and coexisting fish from a large range of water temperatures (estimated from 14 to 34°C) in terms of T_b differences. They concluded that the three lineages Ichthyosauria, Plesiosauria and Mosasauridae were most likely endothermic and homeothermic marine reptiles, although for mosasaurids this was later debated [77,78]. The method used in Bernard et al. [53] cannot clearly identify the thermophysiology of thalattosuchians (figure 4). Indeed, the available sample-set of thalattosuchian specimens is restricted to five localities with a narrow range of low palaeolatitudes from about 29° to 36° N (electronic supplementary material, table S2; palaeolatitudes calculated using the online application of van Hinsbergen et al. [79]), as well as a narrow range of estimated palaeotemperatures from 22 ± 2 to 27 ± 2°C. Consequently, the sampled sites do not allow the two strategies of thermophysiology to be clearly distinguished. Most metriorhynchids and teleosaurids show intermediate values between the expected range for endotherms and ectotherms, with the exception of the metriorhynchids from the middle Callovian of Les Lourdines, which fall within the expected range of endotherms, and one teleosaurid from the Bathonian of Cintheaux, which has a typical ectotherm signature (figure 4).

• Download figure
• Open in new tab
• Download PowerPoint

Using equation (1.1) and assuming that thalattosuchians have a δ¹⁸O_bw value about 2‰ more positive than their ambient seawater, both metriorhynchids and teleosaurids have a body temperature above their environmental one, metriorhynchids being slightly warmer than teleosaurids on average (electronic supplementary material, table S2; figure 5). However, thalattosuchian T_b values are systematically below the calculated ones of co-occurring ichthyosaurs and plesiosaurs, for which the body temperature was calculated from the same equation (1.1) and using a body water δ¹⁸O_bw value similar to that of thalattosuchians.

• Download figure
• Open in new tab
• Download PowerPoint

An endothermic-like thermophysiology interpreted from metriorhynchid δ¹⁸O_p values seems likely according to their known morphology and ecology as active pelagic predators [15,80–84]. Active predation would require elevated metabolic rates compatible with an endothermic thermophysiology. This seems plausible considering the large suite of evidence for an endothermic ancestral condition for Archosauria and a reversal within the crocodylomorph lineage to an ectothermic state coming from different fields of biology, including developmental biology [85], physiology [86], anatomy [87], palaeohistology [88] and phylogenetic signal extraction [89]. Endothermy within metriorhynchids might have been inherited from their archosaur ancestors, and ‘reactivated' along with the acquisition of morphological adaptations to active pelagic predation. However, body temperature regulation seems to have been limited as observed T_b values vary with varying seawater temperature (figure 5). Limited thermoregulatory capacities would be compatible with the restricted range of palaeolatitudinal occurrences compared with fully endo-homeothermic ichthyosaurs and plesiosaurs having been found from equatorial to polar seas [90].

Interpretation of the oxygen isotope composition of teleosaurid apatite in terms of body temperature may be strongly biased by their possible semi-aquatic and eurhyaline ecology, some species having been found in estuarine or freshwater environments [91–94]. As most teleosaurids retained a typical semi-aquatic crocodylomorph morphology (external mandibular fenestrae, extensive osteoderm cover and limbs adapted to terrestrial locomotion), it is hypothesized that they were ectothermic and poikilothermic ambush predators spending most of their time motionless, and mouth-gape basking like modern crocodylians. However, during the late Kimmeridgian–early Tithonian there is evidence for a subclade of teleosaurids that became more pelagic [95]. The oxygen isotope composition of teleosaurid apatite indicates that they kept a body temperature lower than those of ichthyosaurs and plesiosaurs, but close to those of metriorhynchids. However, more estuarine living environments can be characterized by more negative δ¹⁸O values as a result of the mixing between seawater and river waters having negative δ¹⁸O values such as in the case of San Francisco Bay [96]. In this example, waters from the Sacramento River having δ¹⁸O values ranging from about −12 to −10‰ mix with seawater of 0 and result in estuarine waters that can have δ¹⁸O values of −3 to −10‰. Keeping this in mind, then the calculated body temperature of teleosaurids living in estuarine environments would be lower, and their apatite δ¹⁸O values would fit within the expected range of ectotherms. Moreover, the possible semi-aquatic lifestyle of teleosaurids could at least partly account for their elevated δ¹⁸O_p values as a result of body water loss through transcutaneous evaporation. In the studied localities where teleosaurid specimens were found, fossil remains of continental vertebrates and plants have been found, indicating a proximity to landmasses and a possible estuarine origin of teleosaurids. Gigantothermy and behavioural thermoregulation may also account for the calculated body temperatures of teleosaurids close to those of metriorhynchids. Today, large marine crocodylians, such as approximately 1000 kg adult individuals of the saltwater crocodile (Crocodylus porosus), are able to raise their body temperature well above ambient temperatures through mouth-gape basking behaviours and can retain this elevation by the thermal inertia of their large body size [97,98]. Large teleosaurids may have used a similar behavioural thermoregulation as extant marine crocodiles (C. porosus) and would have raised their body temperatures close to that of Jurassic metriorhynchids and maintained it within a narrow range. This would explain the apparent tendency of teleosaurid-fish isotope difference to parallel those of endo-homeothermic ichthyosaurs and plesiosaurs (figure 4). Most metriorhynchid specimens were of smaller body-size than teleosaurids [99,100], and metriorhynchids would have been unable to mouth-gape bask onshore [14].

Based on the available isotopic dataset, it seems likely that teleosaurids retained a typical ecto-poikilothermic thermophysiology in agreement with their morphology and ecology, whereas metriorhynchids may have been endothermic, being able to raise their body temperature above the ambient one, and close to that of other warm-blooded marine reptiles. However, metriorhynchids could not have achieved efficient thermoregulation, as suggested from their varying body temperature along with varying SST.

5. Concluding remarks

The possible difference in thermophysiologies between metriorhynchids and teleosaurids inferred from their stable oxygen isotope composition of apatite can at least partly explain the peculiar thalattosuchian biodiversity pattern of the Jurassic and Early Cretaceous [16,22]. Teleosaurid diversity crashed at the end of the Jurassic, a time of global marine temperature decline, probably as a result of the temperature change or the global regression affecting their ecological niches [90]. Endothermy may have helped metriorhynchids to cope with global cooling, and owing to their pelagic ecology, they might not have been as affected by the marine regressions as teleosaurids. This may explain their success across the Jurassic–Cretaceous boundary [19,22,101]. At this point, we cannot speculate when metriorhynchids became extinct, or why. All we can say is that Jurassic metriorhynchids had an imperfect endothermic and poikilothermic thermophysiology. We hope future studies will investigate whether the Late Jurassic pelagic subclade of teleosaurids began to develop endothermic capabilities, and test whether Cretaceous metriorhynchids, the most marine-adapted of all thalattosuchians [14,102], evolved towards enhanced thermoregulatory abilities.

Data accessibility

Stable oxygen and carbon isotope compositions are provided in Excel tables as electronic supplementary material.

Authors' contributions

All authors contributed to the study and the writing of the manuscript.

Competing interests

We declare we have no competing interests.

Funding

N.S. and R.A. were financially supported by the ANR OXYMORE. M.T.Y. is financially supported by Leverhulme Trust Research Project grant no. RPG-2017-167.

Footnotes

One contribution of 15 to a theme issue ‘Vertebrate palaeophysiology’.