Introduction
The name Thalattosuchia translates as ‘marine crocodiles’ and refers to their frequent occurrence in near-shore and pelagic deposits of Early Jurassic to Early Cretaceous age (e.g., Martin et al. Reference Martin, Amiot, Lécuyer and Benton2014). Thalattosuchia comprise Pelagosaurus typus representing a monospecific group (Pierce and Benton Reference Pierce and Benton2006), the monophyletic Metriorhynchoidea (Young and Andrade Reference Young and de Andrade2009) and Teleosauridae, a clade of uncertain monophyletic content according to recent hypotheses (Mueller-Töwe Reference Mueller-Töwe2005; Jouve Reference Jouve2009; Young and Andrade Reference Young and de Andrade2009). The Metriorhynchoidea exhibit a number of morphological and physiological specializations (paddle-like limbs, salt-excreting gland, a fluked tail) that are assumed to have been associated with a pelagic lifestyle (e.g., Hua and de Buffrénil Reference Hua and de Buffrénil1996; Fernández and Gasparini Reference Fernández and Gasparini2008). On the other hand, Pelagosaurus and the Teleosauridae, although not exhibiting such derived features are nonetheless adapted to a highly specialized aquatic lifestyle as exhibited by the size contrast between forelimbs and hind limbs. Their general appearance is reminiscent of that of modern gharials, their terrestrial locomotion was most probably restricted to crawling on sandy banks (Westphal Reference Westphal1962a). Teleosaurids occur in a variety of proximal deposits such as coastal or lagoonal sediments, therefore presumably in close proximity to emerged lands for laying eggs or basking. In addition, the teleosaurid Machimosaurus hugii has been reported from a brackish deposit in the lignite seam of Guimarota, Portugal (Krebs Reference Krebs1967, Reference Krebs1968) and this is probably also the case for some fragmentary remains of indeterminate teleosaurids from the Jurassic Khlong Min Formation of Peninsular Thailand (Buffetaut et al. Reference Buffetaut, Tong, Suteethorn and Raksaskulwong1994). More exceptional are the occurrences of teleosaurids in fully continental formations. This is the case for the Jurassic Peipehsuchus teleorhinus from the Ziliujing Formation of China (Young Reference Young1948).
Here, we document the unambiguous occurrence of a teleosaurid taxon in Late Jurassic freshwater deposits of northeastern Thailand (Fig. 1) providing a brief description and discussing its affinities. Then, we explore habitat preferences of these animals using strontium isotopes (Schmitz et al. Reference Schmitz, Ingram, Dockery III and Åberg1997; Balter et al. Reference Balter, Telouk, Reynard, Braga, Thackeray and Albarède2008, Reference Balter, Braga, Télouk and Thackeray2012; Tütken et al. Reference Tütken, Vennemann and Pfretzschner2011) and discuss the adaptation of thalattosuchians to the freshwater environment.
The geographic and geological context of the locality of Phu Noi. A, Geographic situation in northeastern Thailand with outcrops of the Phu Kradung Formation indicated in black. Number 1 refers to Phu Noi; B, Simplified stratigraphic framework; C, Traces fossils (rosettes) on the surface of a sauropod rib, indicating developmental activity of dermestid beetles; D, Gleyey structures part of a paleosoil horizon; E, General view of the locality of Phu Noi, Thailand.
Institutional abbreviations
KS, collection numbers for fossils from Kalasin Province at Sirindhorn Museum, Sahatsakhan, Thailand; PRC, Palaeontological Research and Education Centre, Maha Sarakham University, Thailand
Methods
Geological setting
The age of the Phu Kradung Formation is not precisely defined and has either been considered as Late Jurassic on the basis of vertebrate assemblages (Buffetaut et al. Reference Buffetaut, Suteethorn and Tong2001; Buffetaut and Suteethorn Reference Buffetaut and Suteethorn2007; Tong et al. Reference Tong, Claude, Suteethorn, Naksri and Buffetaut2009) or Early Cretaceous on the basis of detrital mineralogy and palynomorphs (Carter and Bristow Reference Carter and Bristow2003; Racey and Goodall Reference Racey and Goodall2009). Teleosaurids are unknown in Cretaceous deposits and if the presently described specimens are not an exception, their presence at Phu Noi may reject a Cretaceous age for this locality. On the basis of freshwater sharks, a Late Jurassic age has been suggested for most of the Phu Kradung Formation (Cuny et al. Reference Cuny, Liard, Deesri, Liard, Khamha and Suteethorn2014). The Phu Noi fossil site is located in the upper part of the Phu Kradung Formation but lies stratigraphically below most of the early Cretaceous fossiliferous localities of the Phu Phan Range (Liard and Martin Reference Liard and Martin2011), therefore a Late Jurassic age for the Phu Noi locality is likely.
Two skulls (KS33-209 and PRC-8) were surface collected during prospecting work on the Phu Noi hill. Subsequently, excavation of a single silty pedogenic horizon (surface area approximately 20 m×15 m) led to the discovery of several teleosaurid specimens (see description below). In this horizon, some dinosaur bones, ribs, and limbs bear on their external surface peculiar marks (Fig. 1) comparable in morphology to the rosettes described on sauropod bones from the Jurassic of Wyoming (Bader et al. Reference Bader, Hasiotis and Martin2009). Accordingly, the most likely trace makers are dermestid larvae in pupation chambers developing on decaying carcasses. This indicates that the fossiliferous horizon at Phu Noi was exposed under aerial conditions for extended periods of at least a few weeks, as implied by the development of necrophagous arthropods on the dinosaur carcasses. Gleyey structures (Fig. 1) indicative of temporarily waterlogged conditions are found at several horizons below and above the vertebrate-bearing horizon. This fossiliferous horizon is part of a thick sequence of alternating sandstone bars and silts, typical of floodplain deposits, which is thought to be indicative of a lake-dominated floodplain (Racey and Goodall Reference Racey and Goodall2009). Estimating a rough distance to the nearest coast is difficult, but according to paleogeographic reconstructions for the Late Jurassic, the Phu Noi locality was located far from the coast (>100 km?) when the Indochinese block was already connected to the South China block (Metcalfe Reference Metcalfe2009).
Laser ablation
All fossil teeth analyzed in this study originate from a single horizon of the Phu Noi excavation site (Fig. 1), part of the Phu Kradung Formation of northeastern Thailand. Fossil specimens include three teleosaurid teeth; four sauropod teeth; four theropod teeth; two Lepisosteiformes scales; two Hybodus teeth and one turtle bone. Teeth were broken and fixed in epoxy resin before polishing so as to obtain a flat surface of the sectioned enamel-dentine junction. Sintered SRM-1400 (Bone Ash) served as a bracketing reference, correcting instrumental biases during isotopic measurements (Balter et al. Reference Balter, Telouk, Reynard, Braga, Thackeray and Albarède2008). The 87Sr/86Sr ratio was measured at the LGLTPE, ENS Lyon on a Nu instrument-500 HR multicollector-ICP-MS (Inductively Coupled Plasma – Mass Spectrometer) with an Excite laser ablation system (193 nm excimer, Photon Machines/Cetac Analyte Teledyne). The standard SRM-1400 contains 249 μg/g of strontium, which yields about 3 volts on mass 88Sr for a laser power of 9 mJ with a spot diameter of 40 to 65 μm, a fluence of 15 J/cm2 and a repetition rate of 30 Hz. The strontium concentration was evaluated using the total strontium voltage, which was set at the certified value of 249 μg/g for the standard SRM-1400. Each sample was ablated at least three times across the dentine-enamel junction for teeth and across the ganoin-dentine junction for ganoid scales of lepisosteids (supplementary material). The turtle bone was ablated at eight different locations. In order to test for changes in 87Sr/86Sr during the mineralization of the teleosaurid teeth, two teeth were sectioned along a base-apex axis and were ablated across the dentine-enamel junction at eight different locations (Supplementary Appendix). The crown height of each tooth is about 20 mm and teeth were selected from specimen PRC-239. Finally, a polished section of a belemnite from the Lower Jurassic strata of Beaujolais, France (Toarcian, Suan et al. Reference Suan, Rulleau, Mattioli, Suchéras-Marx, Rousselle, Pittet, Vincent, Martin, Léna and Spangenberg E.2013) was ablated in order to have a marine reference in the sample list. An aliquot of this belemnite sample was dissolved in concentrated nitric acid, purified through a strontium-specific resin (Sr-Spec Eichrom®) and analyzed for 87Sr/86Sr ratios at the LGLTPE, ENS Lyon on a Neptune Plus multicollector-ICP-MS (Thermo Scientific).
Sediments were not rich enough to yield sufficient Sr voltage using laser ablation. Therefore, about 400 mg of sediment was leached over three days in 1M HCl. The leachate was centrifuged and purified through a strontium-specific resin (Sr-Spec Eichrom®). The purified sample was then analyzed for 87Sr/86Sr ratios at the LGLTPE, ENS Lyon on a Neptune Plus multicollector-ICP-MS (Thermo Scientific).
Results
Systematic paleontology
Superorder Crocodylomorpha Hay, Reference Hay1930; Suborder Thalattosuchia Fraas, Reference Fraas1901; Family Teleosauridae Geoffroy Saint-Hilaire, Reference Geoffroy Saint-Hilaire1825
(Fig. 2)
Two examples of teleosaurid skulls from Phu Noi. Dorsal (A) and ventral (B) views of PRC-239; dorsal (C) and ventral (D) views of PRC-238. Abbreviations: an, angular; bsph, basisphenoid; ch, choana; den, dentary; emf, external mandibular fenestra; ec, ectopterygoid; en, external nares; fr, frontal; j, jugal; l, lacrimal; ltf, lower temporal fenestra; mx, maxilla; n, nasal; or, orbit; p, parietal; pa, palatine; pfr, prefrontal; pmx, premaxilla; po, postorbital; pt, pterygoid; q, quadrate; ra, retroarticular process; san, surangular; sp, splenial; sof, suborbital fenestra; stf, supratemporal fenestra; sq, squamosal.
Referred material
KS33-209, a skull with part of the mandible in occlusion lacking the rostrum (maxillary and mandibular), PRC-238, PRC-239, PRC-240, three complete skulls with mandibles in occlusion; PRC-8, right skull table and basicranium and associated osteoderms; PRC-9, a fragmented skull table.
Geographic occurrence
Phu Noi, Phu Phan Range, Kham Muang district, Kalasin province, Thailand.
Geological provenance
Upper part of the Phu Kradung Formation, Upper Jurassic (?Tithonian).
Comparative description and affinities
The best preserved and most complete skulls including mandibles are PRC-238 and PRC-239 (Fig. 2) and four other skulls are currently under technical preparation and study. An in-depth description will be presented elsewhere.
The taxon from Phu Noi displays a combination of apomorphies for Thalattosuchia (Young and Andrade Reference Young and de Andrade2009) such as the absence of contact between nasal and premaxilla, a nearly tubular rostrum, a complex dorsal surface of the supratemporal roof, the postorbital located laterally to the jugal, a postorbital longer than the squamosal, a long mandibular symphysis about half of the skull length, a non-planar posterior margin of the skull table, the dorsal primary head of the quadrate contacting only the squamosal and not the laterosphenoid, the absence of a squamosal groove for the external earflap musculature and large pendulous basioccipital tubera. Furthermore, the taxon from Phu Noi is referred to Teleosauridae on the basis of the following apomorphies: the basisphenoid is long and projects further anteriorly than the quadrates and the cranial nerve XII opens in line with the foramen magnum.
The taxon from Phu Noi is longirostrine with a tubular rostrum that is ovoid in cross section. Its supratemporal fenestrae are longer than wide in dorsal view as in most teleosaurids except Teleosaurus cadomensis, which possesses large and as long as wide supratemporal fenestrae (Eudes-Deslongchamps Reference Eudes-Deslongchamps1870). Within members of the genus Steneosaurus, the supratemporal fenestrae are indeed longer than wide but appear also longer than in the taxon from Phu Noi. One exception concerns Steneosaurus bollensis, presenting supratemporal fenestrae of similar proportions to those of Peipehsuchus teleorhinus and the taxon from Phu Noi (Westphal Reference Westphal1962b: Figs. 1–2). Machimosaurus hugii also possesses longer supratemporal fenestrae than those of the taxon from Phu Noi. Also, the posterior margin of the fenestrae of Machimosaurus hugii are convex in dorsal view whereas in the taxon from Phu Noi and Peipehsuchus teleorhinus they widely expose the squamosal in this area, thus giving to the posterior margin of the fenestrae a concave outline in dorsal view, a condition moderately expressed among species of the genus Steneosaurus.
The orbits of the taxon from Phu Noi are circular as in all other teleosaurids. Their ventral margin is slightly protruding, a condition difficult to assess in other teleosaurids due to the dorsoventral compression of most specimens. However, a similar condition was reported for a nicely preserved skull of Machimosaurus hugii (Buffetaut Reference Buffetaut1982). The orbits of the Phu Noi teleosaurid are relatively large but their size does not depart much from the condition in Steneosaurus or Teleosaurus cadomensis. On the other hand, the orbits of Machimosaurus hugii are smaller relative to the total skull length.
The palate of the Phu Noi taxon is reminiscent of other teleosaurids in presenting choanae that are wide and opening both on the pterygoids and palatines. Among teleosaurids, the only exception is Teleosaurus cadomensis, which possesses choanae wider than the palatines (Jouve Reference Jouve2009). Another similarity of the Phu Noi taxon with teleosaurids concerns the presence of a relatively large external mandibular fenestra.
In teleosaurids, the antorbital fenestra is present but can easily be overlooked due to its small size. On the other hand, the antorbital fenestra consists of a large slit-like opening and is bordered dorsally by the lacrimal in the taxon from Phu Noi. This is also the case in Peipehsuchus teleorhinus from China. Further similarities between the two taxa are listed as follows: the nasal contributes to most of the rostrum length but is excluded from the narial border by a posteriorly V-shaped premaxilla; wide and dorsally facing posterolateral wall of the supratemporal fenestra, mostly built by the squamosal. The maxillary alveolar count in Peipehsuchus teleorhinus is 27 and the maxillary count in the taxon from Phu Noi is at least 25. As a comparison, this count can be approaching or surpassing 30 alveoli in Steneosaurus (Andrews Reference Andrews1913). Therefore, the taxon from Phu Noi exhibits characters reminiscent of the genus Peipehsuchus Young (Young Reference Young1948). As a side note, Peipehsuchus teleorhinus is possibly as old as Toarcian (Wang et al. Reference Wang, Liang, Kan, Li, Zhu and Ji2008). An in-depth description and phylogenetic analysis of the younger (?Tithonian) Phu Noi taxon is pending and for the moment, it is provisionally referred to Teleosauridae indet.
Phylogenetic analysis
The Phu Noi teleosaurid was coded and included in a matrix for a total of 73 taxa and 240 characters (Young et al. Reference Young, Brusatte, de Andrade, Desojo, Beatty, Steel, Fernández, Sakamoto, Ruiz-Omeñaca and Schoch2012) using the software TNT (Goloboff et al. Reference Goloboff, Farris and Nixon2003). The analysis resulted in a strict consensus of 10 trees with a best tree length of 663 steps. Bremer decay indices were computed and are superimposed on Figure 3. The recovered topology follows what has been reported previously (Young et al. Reference Young, Brusatte, de Andrade, Desojo, Beatty, Steel, Fernández, Sakamoto, Ruiz-Omeñaca and Schoch2012; Martin and Vincent Reference Martin and Vincent2013); only the Thalattosuchia are represented in figure 3 because this work focuses on the intrarelationships of teleosaurids, not on the interrelationships of Thalattosuchia with other crocodylomorphs (see also recent work, Wilberg Reference Wilberg2015). The topology recovered within Thalattosuchia differs from previous works (Goloboff et al. Reference Goloboff, Farris and Nixon2003; Young et al. Reference Young, Brusatte, de Andrade, Desojo, Beatty, Steel, Fernández, Sakamoto, Ruiz-Omeñaca and Schoch2012; Martin and Vincent Reference Martin and Vincent2013) in the placement of Metriorhynchoidea relative to Teleosauridae. Here, the Teleosauridae are non-monophyletic (for similar results see: Mueller-Töwe Reference Mueller-Töwe2005; Jouve Reference Jouve2009; Young and Andrade Reference Young and de Andrade2009) and Metriorhynchoidea are not the sister group to Teleosauridae but are sister taxon to a group comprising the most derived teleosaurids (Phu Noi teleosaurid, Steneosaurus obtusidens and Machimosaurus hugii). Pelagosaurus typus is recovered as the basalmost thalattosuchian.
Strict consensus tree recovered in this work showing the relationships within Thalattosuchia. The habitat in which the different taxa were recovered (marine, freshwater, brackish) is indicated by a symbol and are superimposed on the tree. Numbers refer to Bremer Decay indices.
Isotopic analysis
Calculated strontium concentrations in samples from Phu Noi range from ~100 to ~2000 μg/g (electronic supplementary material, table S1) with an average value of 631±872 μg/g. Theropods exhibit Sr concentrations (208±127 ppm) that are significantly lower than sauropods (764±294 ppm; Student’s t-test: p**=0.0017) or total fauna (766±413 ppm; Student’s t test: p***<10−4; Fig. 4).
Isotope ratios of 87Sr/86Sr against 1/strontium concentrations (in ppm−1) for all the vertebrate samples analyzed in this study. We consider the inverse of the concentration to linearize potential relationship with isotopic ratio in case of diagenesis. For the sake of clarity, we took the average value of the Sr isotopic ratio and concentration. The mixing with a diagenetic end-member is recognized for turtle bone aliquots (grey symbols) arising from a single sample.
During the Neptune session, the international standard SRM-987 gave an average value of 0.71025±0.00001 (2 SD, n=3). During the laser ablation session, the standard SRM-1400 yielded an average 87Sr/86Sr value of 0.71371±0.00526 (2 SD), which was compared to a TIMS value of 0.71310±0.00002 (2 SD) (Schweissing and Grupe Reference Schweissing and Grupe2003) for sample bracketing during laser ablation sessions. The belemnite yielded a value of 0.70697±0.00097 (2 SD) using laser ablation and a value of 0.70719±0.000007 (2 SD) using wet chemistry (Neptune session). The 87Sr/86Sr values of the belemnite obtained by means of wet chemistry and laser ablation compared well with the accepted values of the strontium isotope curve for Toarcian seawater based on belemnites (e.g., McArthur et al. Reference McArthur, Howarth and Bailey2001).
The range of the 87Sr/86Sr values is narrow, from 0.71088 to 0.71179, with the exception of three values that deviate from this range, namely a theropod tooth at 0.70754 and two turtle bone values at 0.71879 and 0.72144 (Fig. 5; Supplementary Appendix, Table S1), which are indicative of diagenesis as discussed below. The leachate of the sediment at Phu Noi yielded a 87Sr/86Sr value of 0.71334±0.000007 (2 SD). Results of the intra tooth analyses show a narrow range of values from 0.710282 to 0.711840.
Strontium isotopic variability (87Sr/86Sr) in the Upper Jurassic vertebrate continental assemblage of the locality of Phu Noi. Sample names of individual taxa are indicated on the left hand side. The vertical stipple line on the left is the approximate 87Sr/86Sr value for the marine environment during the Late Jurassic (after Prokoph et al. Reference Prokoph, Shields and Veizer2008). Error bars represent 2SD.
Discussion
Strontium isotopes and teleosaurid provenance
The ingested strontium is incorporated into the crystal lattice of bioapatite during mineralization of teeth and bones (Reynard and Balter Reference Reynard and Balter2014) and therefore its isotopic composition reflects that of the environment in which the animal lived, as derived from food and drinking water (e.g., Graustein Reference Graustein1989; Blum et al. Reference Blum, Taliaferro, Weisse and Holmes2000; Maurer et al. Reference Maurer, Galer, Knipper, Beierlein, Nunn, Peters, Tütken, Alt and Schöne2012). That fossil teeth do not retain an in vivo strontium signal is always an issue because biogenic strontium may be rapidly replaced after burial by diagenetic strontium (e.g., Nelson et al. Reference Nelson, DeNiro, Schoeninger, De Paolo and Hare1986; Sillen Reference Sillen1986; Hoppe et al. Reference Hoppe, Koch and Furutani2003). Diagenesis can be detected by several means, but here we take advantage of the lower Sr concentrations in theropod teeth to assume, as a general rule, that the Phu Noi assemblage has not been overprinted significantly by diagenesis. Strontium, and barium concentrations decrease up the trophic chain because these are two non-essential trace elements that tend to mimic Ca, which is bio-essential (Elias et al. Reference Elias, Hirao and Patterson1982; Burton et al. Reference Burton, Price and Middleton1999; Balter Reference Balter2004). Although interpretations can be complicated by the ingestion of trace elements from dust or soil (Kohn et al. Reference Kohn, Morris and Olin2013), this process has been widely used as a paleodietary tool, but has been restricted to the reconstruction of the diet of Late Paleolithic Neanderthals (Balter et al. Reference Balter, Person, Labourdette, Drucker, Renard and Vandermeersch2001, Reference Balter, Bocherens, Person, Labourdette, Renard and Vandermeersch2002) and of Early Pleistocene hominins (Sillen Reference Sillen1986; Sponheimer et al. Reference Sponheimer, Deruiter, Leethorp and Spath2005; Balter et al. Reference Balter, Braga, Télouk and Thackeray2012), i.e., of Quaternary age. Thus, the low Sr concentrations in the teeth of theropods compared to other taxa reflect that the most parsimonious interpretation is that these animals were feeding at a higher trophic level. It is noteworthy that the preservation of a Ca biopurification pattern in the Phu Noi assemblage dated at about 150 Ma is exceptional and to our knowledge, has never been described before for Mesozoic fossils. Also, the large range of strontium isotope values in the various fossils analyzed here argue that all the samples have not converged on a single diagenetic value. However, possible diagenetic effects have to be scrutinized sample by sample in order to discard samples suspected of diagenetic overprinting. Diagenetic process can be modeled as a conventional mixing balance between biogenic and diagenetic end-members with distinct strontium concentrations [Sr] and isotopic ratios 87Sr/86Sr: samples affected by diagenesis would fall on a mixing hyperbola in a [Sr] vs 87Sr/86Sr space (Albarède Reference Albarède1995; Balter et al. Reference Balter, Braga, Télouk and Thackeray2012). Here, no such correlation is observed, except for the two turtle bone outliers PN-TU1-6 and PN-TU1-7 (Fig. 4; Supplementary Appendix, Table S1), for which both the strontium concentration and the isotopic ratio are different from the rest of the samples. Both samples have more radiogenic 87Sr/86Sr values than the soil strontium soluble fraction (Fig. 5). The 87Sr/86Sr values of the two turtle bone outliers must have been driven by another diagenetic pool with a higher 87Sr/86Sr value, at least ~0.722, possibly a secondary mineral phase that precipitated in the porous structure of bone (e.g., calcite, gypsum). In this condition, the soil strontium soluble fraction (i.e., “soil water” on Fig. 5) cannot be considered as the prevalent diagenetic end-member for the PN-TU1-6 and PN-TU1-7 samples. However, three other turtle bone samples exhibit suspicious 87Sr/86Sr value, i.e., higher than ~0.711 (which is the average value of the dataset) and lower than the soil soluble fraction value at 0.71334. These samples are PN-TU1-2, PN-TU1-3 and PN-TU1-8 (Supplementary Appendix, Table S1). For these three samples, slight diagenetic strontium overprinting originating from the “soil water” strontium can be suggested, but is not demonstrated, as no increase of strontium concentration is concomitantly observed. In the present study, diagenesis has affected bone samples, leaving dentine and enamel seemingly unaffected by diagenetic overprints. This is in contrast, for instance, with the results obtained at the Middle Eocene conservation-Lagerstätte Messel site, Germany (Tütken Reference Tütken2014), where dentine samples have systematically lower 87Sr/86Sr values (~0.706) than related enamel (~0.711). Dentine are thus lower by ~0.06 units of 87Sr/86Sr as compared with enamel, which is an important offset that is not observed in the present study, indicating very different diagenetic processes observed for the fossils of the Messel deposits and those of the Phu Kradung Formation. Moreover, that enamel and dentine strontium isotope values are not significantly different from each other is in agreement with an in vivo signature, as reported in modern faunas (Budd et al. Reference Budd, Montgomery, Barreiro and Thomas2000).
For the reasons presented above, we consider at least our enamel data to be most likely representative of an in-vivo signal. On average, the 87Sr/86Sr values measured on the vertebrate assemblage (~0.711, Fig. 5, Supplementary Data, tables S1 and S2) reflect a continental environment, with values significantly more radiogenic than the values reported for marine environment of Late Jurassic age (~0.707 as compiled in McArthur et al. Reference McArthur, Howarth and Bailey2001). This implies that vertebrates recovered from Phu Noi mineralized their teeth locally in a freshwater setting incorporating the isotopic ratio of 0.71334. This is in line with sedimentological and petrographic observations of polycrystalline quartz grains and lithic fragments of quartz-mica schists that indicate a crystalline, metamorphic bedrock substrate as a source area for the clastic sediments of the Phu Kradung Formation (Racey et al. Reference Racey, Love, Canham, Goodall, Polachan and Jones1996: p. 20).
The outlier theropod with the lowest strontium value (~0.708) mineralized its tooth on a different substrate, possibly reflecting migration from another area. From the general distribution of 87Sr/86Sr, there is no indication that the teleosaurid material is allochtonous. A detailed sampling of two teleosaurid teeth belonging to PRC-239 along incremental growth lines allows tracking the evolution of the strontium isotope signal (and therefore of the provenance) during the time necessary for mineralization. Tooth replacement takes about two months to complete as reported from a study on a meter long extant Alligator mississippiensis (Erickson Reference Erickson1996). Erickson (Reference Erickson1996) also concluded that tooth replacement slows down during ontogeny. Although not assessed for large animals, it is reasonable to admit that in several meter long specimens, several months are necessary to renew teeth. In the case of PRC-239 representing an adult, it is reasonable to assume that the teeth analyzed here did record more than two months of environmental signal. Electronic supplementary material, table S2 reports strontium isotope values, which are all around 0.711 (n=48) for these two teeth with a minimum value of 0.710 and a maximum value of 0.712. These 87Sr/86Sr values fall in the range of the seasonal variability of the weathering flux of a homogeneous catchment, which is about 2.10-3 in the study of the Xijiang River (Wei et al. Reference Wei, Ma, Liu, Xie, Lu, Deng, Ren, Zeng and Yang2013). Indeed, the timeline of the mineralizing tooth is perhaps too short to capture significant shifts in 87Sr/86Sr ratios that would provide evidence for migration. Therefore, it cannot be excluded that migration between the freshwater and the marine habitat could take place in the Phu Noi teleosaurid. Nevertheless, it can be said that in the time necessary to mineralize the teeth (over 2 months), there is no evidence of migration of teleosaurids between the freshwater environment of Phu Noi and a marine environment. Thus, the strontium isotopic values of the Phu Noi teleosaurid fall in the range of values retrieved for the other vertebrates from Phu Noi, and the most probable conclusion is that this teleosaurid is a long-term (over 2 months) resident of this continental freshwater paleoenvironment.
Colonization of freshwater drainages
The earliest known thalattosuchians are Early Jurassic in age (early Toarcian) and are exclusively recorded from marine deposits. Several million years later, the teleosaurid from northeastern Thailand illustrates an unambiguous example of a thalattosuchian recovered from a freshwater setting. From an evolutionary perspective, a reconstruction of the evolution of osmoregulation in thalattosuchians is hampered by the unclear interrelationships of its main lineages (Pelagosaurus, Teleosauridae, Metriorhynchoidea). Therefore, osmoregulation within Thalattosuchia and more specifically within Teleosauridae can only be tentatively superimposed on an evolutionary hypothesis, as has been implied for the capacity of members of this clade to osmoregulate (Fernández and Gasparini Reference Fernández and Gasparini2008). Based on a transitional stage model (Mazzotti and Dunson Reference Mazzotti and Dunson1989), Fernández and Gasparini (Reference Fernández and Gasparini2008) consider teleosaurids to be estuarine/marine and metriorhynchoids to be pelagic swimmers. The presence of salt-excreting glands in Geosaurus and Metriorhynchus represents a solid basis for the latter assumption (Fernández and Gasparini Reference Fernández and Gasparini2000, Reference Fernández and Gasparini2008; Gandola et al. Reference Gandola, Buffetaut, Monaghan and Dyke2006; Herrera et al. Reference Herrera, Fernández and Gasparini2013). However, characterization of a marine versus estuarine lifestyle for teleosaurids and Pelagosaurus is impossible based solely on morphology, hence there is a lack of resolution regarding their adaptation to transitional paleoenvironments. As a comparison, salt-excreting glands are active in extant Crocodylus (Mazzotti and Dunson Reference Mazzotti and Dunson1989), but do not leave any osteological information because they occur on the tongue.
According to our results, three hypotheses could be advanced for the occurrence of some teleosaurids in riverine/lacustrine habitats. First, an adaptation to the freshwater environment could be an autapomorphy of some teleosaurids and evolved once in the derived teleosaurid from Thailand (and possibly also in Machimosaurus) (Fig. 3) as a secondary adaptation from a primitively marine teleosaurid ancestor. A second possibility is the retention of an adaptation inherited from a basal member of Thalattosuchia or from the ancestors of Thalattosuchia, if we assume they lived in freshwater environments during the Late Triassic-Early Jurassic time interval. In the latter hypothesis, metriorhynchoids were exclusively living in the marine environment (supported by the presence of salt-excreting glands (Fernández and Gasparini Reference Fernández and Gasparini2008) and we could predict that all other thalattosuchians (Pelagosaurus and Teleosauridae) would likely be recovered in riverine environments during the Toarcian-Tithonian interval. The latter two hypotheses seem likely, considering Peipehsuchus teleorhinus from the Ziliujing Formation of China as the oldest representative of a freshwater thalattosuchian (Lower Jurassic, Peng et al. Reference Peng, Ye, Gao, Shu and Jiang2005). Nevertheless, it should be stressed that the capacity to tolerate salt may not have been lost at all as is seen today with the following example: the extant species of Crocodylus do have a capacity for salt excretion although many of them spend their entire life in freshwater environments. This does not prevent many individuals of Crocodylus porosus or Crocodylus acutus to spend a considerable amount of time in brackish or marine habitats. Therefore, a third possibility may be that teleosaurids were able to migrate between marine and freshwater habitats without losing the capacity for salt excretion.
Whether those teleosaurids recovered in marine deposits were temporarily visiting freshwater ecosystems, for example to seek food, reproduce or avoid competitive interactions with other marine reptiles is plausible (see below for examples of extant taxa that migrate in freshwater ecosystems). The fact is that two teleosaurid genera have been recorded in non-marine habitats. The oldest record appears to be Peipehsuchus teleorhinus from the Lower Jurassic Ziliujing Formation of China, which originates from a lacustrine deposit (Wang et al. Reference Wang, Liang, Kan, Li, Zhu and Ji2008) (a freshwater plesiosaur was also reported from this same formation by Sato and Wu (Reference Sato and Wu2003)). Secondly, the present study reports a teleosaurid from a freshwater habitat of the Late Jurassic, therefore younger than the Chinese occurrence of Peipehsuchus. In addition, fragmentary remains of teleosaurids have been reported from a brackish environment of the Middle to Upper Jurassic Khlong Min Formation in Peninsular Thailand (Buffetaut et al. Reference Buffetaut, Tong, Suteethorn and Raksaskulwong1994). These few records indicate a capacity for some teleosaurids to live in non-marine habitats, at least for a few months as highlighted with the present study. Conversely, do the abundant teleosaurids recovered in coastal deposits of Europe preserve a trace of their passage in freshwater habitats? The Kimmeridgian Machimosaurus hugii occurs in the lignite deposits of Guimarota, Portugal (Krebs Reference Krebs1967, Reference Krebs1968), which has been interpreted as a brackish environment (Buffetaut Reference Buffetaut1982). This taxon is often recovered in marine deposits (Martin and Vincent Reference Martin and Vincent2013) and the occurrence of the Portuguese specimen in a transitional environment might be viewed as an evidence for migration between marine and brackish habitats and questions its ability to visit freshwater environments.
For historical reasons, more teleosaurids have been discovered in Europe than in other parts of the world. The vast majority of Jurassic and Lower Cretaceous sediments of Europe were deposited in marine environments. Thus, with the exception of terrestrial faunas washed into marine environments (e.g., Buffetaut Reference Buffetaut1994), fossil vertebrates including teleosaurids found in such deposits most likely have a marine origin. Whether teleosaurids were present in contemporaneous drainage basins has been questioned (Westphal Reference Westphal1962b). That teleosaurids spent some time of their life in the freshwater habitat could be tested by analyzing the strontium isotopic composition of their teeth.
Other predominantly marine predators are also mentioned in freshwater settings of Jurassic and Cretaceous age. Plesiosaurs have been reported from non-marine Lower Cretaceous deposits of Australia (Bartholomai Reference Bartholomai1966; Kear Reference Kear2006), Britain (Andrews Reference Andrews1922; Kear and Barrett Reference Kear and Barrett2011), Germany (Wegner Reference Wegner1914), Canada (Sato et al. Reference Sato, Eberth, Nicholls and Manabe2005; Vavrek et al. Reference Vavrek, Wilhelm, Maxwell and Larsson2014) and China (Sato and Wu Reference Sato and Wu2003). Mosasaurs have also been reported from the Upper Cretaceous (Santonian) freshwater deposits of Hungary (Makádi et al. Reference Makádi, Caldwell and Ősi2012) and strontium and oxygen isotope studies confirmed they remained in this freshwater setting for some time (Kocsis et al. Reference Kocsis, Ősi, Vennemann, Trueman and Palmer2009). There are also undescribed large isolated varanoid vertebrae from Campano-Maastrichtian freshwater deposits of southern France (Buffetaut et al. Reference Buffetaut, Leloeuff, Tong, Duffaud, Cavin, Garcia and Ward1999). Hybodont sharks derive from a marine ancestor (Cuny et al. Reference Cuny, Liard, Deesri, Liard, Khamha and Suteethorn2014) and are frequently recovered in Mesozoic brackish and fresh water deposits (Klug et al. Reference Klug, Tütken, Wings, Sun and Martin2010; Fischer et al. Reference Fischer, Voigt, Franz, Schneider, Joachimski, Tichomirowa, Götze and Furrer2012; Reference Fischer, Schneider, Voigt, Joachimski, Tichomirowa, Tütken, Götze and Berner2013; Cuny et al. Reference Cuny, Liard, Deesri, Liard, Khamha and Suteethorn2014). Our results, including strontium isotope measurements on two hybodont teeth confirm the long-term occurrence of hybodonts in the freshwater habitats of Phu Noi (Fig. 5).
Colonization of the freshwater environment by marine vertebrates could be tied to a variety of ecological factors including new available niches, lack of competition on newly available resources, or refuges from predators. Beside thalattosuchians, members of another clade of marine crocodylomorphs, Dyrosauridae, have been recorded from transitional environments (an account is presented elsewhere (Hastings et al. Reference Hastings, Bloch and Jaramillo2011)) and from freshwater deposits in the Maastrichtian of India (Khosla et al. Reference Khosla, Sertich, Prasad and Verma2009) and in the Eocene of Punjab, Pakistan (Buffetaut Reference Buffetaut1977, Reference Buffetaut1978). The rarity of large specimens and the abundance of small ones led Buffetaut (Reference Buffetaut1977) to hypothesize that those freshwater environments were used as refuges for juveniles against large marine predators represented at that time by eusuchians. Sato et al. (Reference Sato, Eberth, Nicholls and Manabe2005) observed that freshwater plesiosaurs from the Cretaceous of Canada are of small size, and possibly represent juveniles. More recently Vavrek et al. (Reference Vavrek, Wilhelm, Maxwell and Larsson2014) interpreted a new freshwater locality from the Canadian High Arctic as a possible plesiosaur nursery. The situation for the Phu Noi teleosaurid is however different with specimens representing different size classes. The abundance of large gynglymodian fish skeletons in this locality (Deesri Reference Deesri2012) hints at the availability of food resources for these teleosaurids.
Modern-day species of Crocodylus and cetaceans offer interesting ecological parallels with extinct marine reptiles. Out of the 23 species of living crocodilians, only two species, C. porosus and C. acutus are frequently found in estuarine or marine habitats and two others, C. niloticus and C. johnstoni have known estuarine populations (e.g., Mazzotti and Dunson Reference Mazzotti and Dunson1989; Wheatley et al. Reference Wheatley, Peckham, Newsome and Koch2012). Occasional accounts also exist for other members of Crocodylinae in brackish environments. There are four extant species of freshwater cetaceans (including the recently extinct Chinese species Lipotes vexillifer) living permanently in large fluvial systems of South America and Asia (Cassens et al. Reference Cassens, Vicario, Waddell, Balchowsky, Van Belle, Ding, Fan, Mohan, Simoes-Lopes, Bastida, Meyer, Stanhope and Milinkovitch2000; Hamilton et al. Reference Hamilton, Caballero, Collins and Brownell2001). Because it spent a considerable amount of time in freshwater, the Phu Noi teleosaurid could be considered an ecological equivalent of these freshwater dolphins. A closely related species to the freshwater dolphins (see debate about monophyly (Cassens et al. Reference Cassens, Vicario, Waddell, Balchowsky, Van Belle, Ding, Fan, Mohan, Simoes-Lopes, Bastida, Meyer, Stanhope and Milinkovitch2000; Hamilton et al. Reference Hamilton, Caballero, Collins and Brownell2001), Pontoporia blainvillei inhabits coastal and estuarine habitats of Argentina and Brazil but does not venture far inland. This species represents the marine/estuarine equivalent of teleosaurids following the interpretation of (Fernández and Gasparini (Reference Fernández and Gasparini2008). A last case is represented by the delphinid Orcaella brevirostris, which is euryhaline (Baird and Beasley Reference Baird and Beasley2005): this species spreads along the shorelines of SE Asia and penetrates deep within the Mahakam, Mekong, and Irrawaddy rivers. The species of cetaceans mentioned above are found in a variety of habitats, some living permanently in freshwater habitats, other migrating back and forth between marine and freshwater habitats and others not venturing inland but having a distribution in estuaries and coastal environments. It is therefore probable that thalattosuchians and particularly teleosaurids were exploiting such a variety of environments too. Further isotopic studies on various fossil assemblages may reveal specific migratory patterns. According to the present study, the Phu Noi teleosaurid could either be viewed as a pioneering taxon into a novel habitat or as an evidence for the widespread behavior of teleosaurids to visit and durably inhabit riverine ecosystems of the Jurassic. Strontium isotope studies provide a means with which to test these hypotheses through the analysis of tooth enamel from the numerous specimens recovered throughout European Jurassic localities.
Acknowledgments
The authors thank all of the members of Palaeontological Research and Education Centre (PRC) of Mahasarakham University, students of Biology Department, Faculty of Science of Maharasakham University and especially the members of the staff of Sirindhorm Museum and of Department of Mineral Resources (DMR) of Thailand who participated in the fieldworks. The fieldwork was funded by the Department of Mineral Resources and the Palaeontological Research and Education Centre. U. Deesri, S. Suteethorn, and V. Suteethorn were supported by the research in Higher Education Project year 2014/Higher Education Research Promotion (HERP) and National Research University Development in the office of the Higher Education Commission, the National Research council of Thailand (NRCT), and the Budget Revenue Fiscal Year 2557, Mahasarakham University. Some of the animal outlines in Figures 4 and 5 were retrieved from www.phylopic.org. This manuscript benefited from the constructive remarks of S. Salisbury, T. Tütken and one anonymous reviewer.
Supplementary Material
Supplemental materials deposited at Dryad: doi: 10.5061/dryad.pj69f
