Systematic nomenclature

Pterosaur systematics are currently complicated by the existence of contrasting, sometimes incompatible interpretations of clade content, definition and nomenclature (see Witton 2013 for discussion). This paper follows pterosaur systematics and nomenclature as presented by Lü et al. (2012).

Gut content and stomach regurgitates

Gut content is known from a number of pterosaur specimens, but their state of preservation is highly variable. Among the poorest quality are masses of calcite crystals preserved in the posterior trunk regions of some articulated Solnhofen Formation pterosaurs. These are suspected to represent diagenetically altered stomach remains, but they present no characteristics allowing identification of ingested content (e.g. Bennett 2014; Hone et al. 2015). Other examples are of sufficient quality to permit identification of the consumed species and even the state of digestion before the death of the pterosaur individual.

The geochronologically oldest record pterosaur gut content is that of MCSNB 2888 (Fig. 1a), the holotype of the Triassic (Norian), Italian (Zorzino Limestone Formation) pterosaur Eudimorphodon ranzii (Wild 1978). MCSNB 2888 is a partial but articulated skeleton with partly digested bone preserved within the abdominal cavity. The preservation of the alleged food items within the dorsal rib cavity and their etched texture indicative of digestive acids confirms the identification of these remains as gut content. Mild disruption of the torso has distributed the remains throughout the trunk region, and the largest and most concentrated remains are clustered into a mass 8 mm long by 7 mm wide. Additional food remains are represented by smaller fragments of bone and indeterminable debris. Most of the remains cannot be identified but ganoid scales preserved within the gut cavity have been attributed to the fish genus Parapholidophorus (Wild 1978).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Evidence of pterosaur foraging and digestion. (a) Eudimorphodon ranzii torso (MSCSNB 2888) with fish scales and other gut content; (b–d) Rhamphorhynchus muensteri gut content and coprolites; (b) skull and neck of WDC CSG 255, showing partial fish remains in the throat region; (c) upper body of TMP 2008.41.001, showing poorly preserved gut content (GC) and a coprolite (CO); (d) SOS 4599 torso with substantial, part-digested fish remains; (e) Scaphognathus crassirostris skull and mandible with associated fish axial column (SMNS 59395), interpreted as regurgitate; (f) possible catastrophic feeding of Ludodactylus sibbicki (SMNK PAL 3828); (g) Pteranodon sp. mandible with fish regurgitate preserved between mandibular rami (AMNH 5098). Grey shading indicates ingested prey items. (a) After Wild (1978); (b) from Frey & Tischlinger (2012); (c) from Hone et al. (2015); (d) after Wellnhofer (1975); (e) after Bennett (2014).

As will become apparent throughout this review, fossils of the relatively abundant Jurassic (Tithonian), German rhamphorhynchid Rhamphorhynchus muensteri provide a wealth of information about the interactions of this species with other animals. Several specimens have been cited as showing gut content in the last century and, although not all are considered valid examples today, this species still offers the most complete and detailed picture of dietary preferences in any pterosaur. The best known Rhamphorhynchus gut content belongs to the Solnhofen Formation specimen SOS 4599 (Fig. 1d), a complete skeleton with large, partly articulated and somewhat digested fish remains in the trunk skeleton immediately anterior the pelvis (Wellnhofer 1975, 1991, p. 160). The relative extent of the gut content is impressive, it occupying much of the torso alongside five lunate structures that may be either additional food items (Wellnhofer 1975) or gut tissues (Unwin 2005). There is no doubt that this specimen records genuine gut content as the pterosaur torso is articulated, the fish remains are enclosed by ribs and they have etched textures consistent with other partly digested fossil remains. The caudal fin morphology can still be determined in the gut content and shows that the prey item was swallowed whole and head-first. The fish remains provide data on the size of the Rhamphorhynchus gut: at 30.3 mm long, they occupy 60% of the torso length (taken as the length from acetabulum to anteriormost dorsal vertebra), suggesting that much of the Rhamphorhynchus abdomen could be occupied by ingested food. The relatively enormous gut content of SOS 4599 suggests that Rhamphorhynchus must have possessed highly mobile and distensible throat tissues, much like those of modern fish-eating birds (Witton 2013).

Additional gut content has been found in the throat region of WDC CSG 255 (Fig. 1b), a complete and fully articulated Rhamphorhynchus notable for its close association with a predatory fish (see below; Frey & Tischlinger 2012). This gut content comprises a small and complete fish skeleton similar in morphology to the common Solnhofen teleost Leptolepides. It is proportionally much smaller than that seen in SOS 4599, but was also swallowed head-first. The fish skeleton is preserved in close association with the cervical vertebrae, suggesting it was in the throat region – perhaps being swallowed – when the pterosaur died. This interpretation is supported by its lack of etched textures (Frey & Tischlinger 2012). Hone et al. (2013) identified further evidence of fish prey for Rhamphorhynchus from piscine teeth and scales preserved around the ventral torso margin of the complete Rhamphorhynchus specimen CM 11427. Another Rhamphorhynchus with gut content, reported by Hone et al. (2015), is the articulated and virtually complete skeleton TMP 2008.41.001 (Fig. 1c). The arrangement of the ribcage, enveloping of the gut remains by the pterosaur skeleton and excellent articulation of the torso verify that these remains are genuine stomach content. However, what this ingested material represents is mysterious as it is not well preserved. It seems to be composed of bone, and thus represents a small vertebrate, but further identification is difficult.

Rhamphorhynchus is not the only Solnhofen Formation pterosaur known with dietary remains. SMNS 59395 (Fig. 1e), a complete skeleton of the Solnhofen species Scaphognathus crassirostris, preserves an articulated, but relatively poorly defined piece of gut regurgitate (Bennett 2014). This is represented by an articulated vertebral column preserved in a loose spiral adjacent to the mandibular tip of the pterosaur, as if ejected from the animal during the decay process. So far as can be established, this represents a small individual of the common Solnhofen fish Leptolepides. The exact size of the fish is not known as the vertebral column is not complete, but the proportions of the vertebrae suggest that it was much smaller, relatively, than that devoured by the SOS 4599 Rhamphorhynchus.

Compelling evidence of gut content is rare among pterodactyloid pterosaurs. Perhaps the only incontrovertible dietary remains from this group pertain to specimens of the Late Cretaceous (Santonian), North American ornithocheiroid Pteranodon. The best known example is AMNH 5098 (Fig. 1g), an isolated mandible from the Niobrara Formation with regurgitated gut content between the mandibular rami, probably held in place by throat tissues before burial. This specimen is the longest-known example of pterosaur dietary remains still considered valid (Brown 1943) and was initially reported as comprising ‘backbones of two species of fishes and the joint of a crustacean’ (Brown 1943). However, a reanalysis by Bennett (2001) suggests that only fish vertebrae are present. It is assumed that these jumbled, partly digested fish remains represent gut content regurgitated during death throes (Bennett 2001), as occurs in modern animals (Wings 2007). The bolus comprises disarticulated vertebrae, most of which are small (c. 7 mm in length), save for at least two that are much larger (c. 15 mm) and probably represent a second consumed species. Even these large vertebrae would have belonged to comparatively small fish, however, contrasting with the relatively enormous fish found in the gut of the SOS 4599 Rhamphorhynchus. Additional discoveries of possible Pteranodon gut content pertain to two specimens (SDSM 45719 and SDSM 69040) associated with Enchodus fish vertebrae from the Campanian Pierre Shale (Hargrave 2007). It is not uncommon for Pteranodon fossils to be associated with fish bones and these may represent additional dietary remains (Bennett 2001), although the widespread occurrence of fish fossils throughout the Niobrara and Pierre Shale deposits (Hargrave 2007) makes the palaeoecological significance of these associations uncertain.

Stomach content is known from two partial skeletons of the Cretaceous (Albian) ctenochasmatoid Pterodaustro guinazui, MIC-V263 and MIC-V243 (Codorniú et al. 2013). These Argentinian (Lagarcito Formation) remains are not food however, but sand and gravel interpreted as geogastroliths (Codorniú et al. 2013). The enclosure of these clasts within the body cavity of the pterosaur specimens, their distinction from local lithologies and evidence of a consistent diagenetic history all support their identification as inorganic gut content. It is thought that gastroliths were used by these pterosaurs to digest tough food items filtered from water through their unusual, sieve-like dentition. They remain the only pterosaur specimens known with gastroliths, despite the global distribution and relative abundance of pterosaurs with elongate, fine dentitions also potentially indicative of suspension-feeding.

Coprolites

The sole known pterosaur coprolite is associated with a complete Rhamphorhynchus specimen mentioned above, TMP 2008.41.001 (Fig. 1d; Hone et al. 2015). It is closely associated with the skeleton, being preserved immediately posterior to the pelvic region. Presumably, it was ejected from this individual shortly after it settled on the bed of the Solnhofen lagoon. The coprolite is small, broken into two roughly quadrangular pieces and variably composed of calcite masses (proximally) and hundreds of submillimetre, spike-shaped objects (distally). What the latter represent is not clear. A vertebrate identification is unlikely given that they do not conform to obvious vertebrate anatomy or mineralogy typical of Solnhofen vertebrate fossils. Cephalopod hooklets have been intimated as a possible, albeit problematic identification (Hone et al. 2015). The potential effects that a complete digestive process will have on the recognizability of these remains has been noted (Hone et al. 2015).

Jaw and prod marks

Several pterodactyloid trackways are thought to show evidence of pterosaurs feeding in a wader- or shorebird-like manner (Fig. 2). Shallow, paired, rounded impressions interpreted as beak traces are associated with Jurassic Pteraichnus footprints from the Del Monte Mines, Utah (Fig. 2b; Summerville Formation, Oxfordian; Lockley & Wright 2003), Lower Cretaceous Purbeckopus pentadactylus tracks from Dorset, England (Fig. 2a; Berriasian; Wright et al. 1997) and pterosaur tracks from the Late Cretaceous Blackhawk Formation, Utah, USA (Fig. 2c; Campanian; Wright et al. 1997; Mazin et al. 2003). Some controversy surrounds the track-maker identity of two of these tracks. Purbeckopus has been considered pterosaurian by most (e.g. Wright et al. 1997; Lockley et al. 2008), but not universally (Billon-Bruyat & Mazin 2003). The Blackhawk Formation tracks, which lack pes prints, have been suggested as pertaining to Hesperornithes (Parker & Balsley 1989) rather than pterosaurs, but the triradiate morphology of these traces shows strong similarity to pterosaur manus prints and suggested a pterosaurian origin to Unwin (1997) and Mazin et al. (2003). The lack of pes prints at this site has been attributed to the ‘manus-only’ ichnological phenomenon known from several pterosaur track sites (Mazin et al. 2003; see Lockley et al. 1995; Li et al. 2015).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Possible pterosaur feeding traces. (a) Early Cretaceous Purbeckopus pentadactylus tracks from the Purbeck Limestone Formation, England; (b) Late Jurassic foraging marks from the Del Monte Mines, Utah; (c) Late Cretaceous Blackhawk Formation tracks, Utah; (d) Pteraichnus sp. tracks and possible beak scrapes from the Late Cretaceous Cerro del Pueblo Formation, Mexico, argued here as more akin to the azhdarchid track Haenamichnus than the pterodactyloid track Pteraichnus (inset (d, i)). Shading represents trace identification: lighter grey shading shows pes prints; dark grey shading shows manus prints; and black shading shows possible foraging marks. (a) Modified from Wright et al. (1997); (b) modified from Lockley & Wright (2003); (c) modified from Parker & Balsley (1989); and (d) after Rodriguez-de la Rosa (2003).

That pterodactyloid tracks might be associated with foraging traces is not surprising given that modern animals often leave feeding traces when foraging on flat, muddy substrates. Traces left by birds pecking at soft substrate are very similar to those intimated for pterosaurs in being rounded, paired depressions which have profiles reflecting the track-maker's bill tip anatomy. It remains possible that non-pterosaurian animals could have produced these marks (perhaps common pterosaur prey, as suggested by Lockley & Wright 2003) but this would require the second animal to leave no additional evidence of its presence and coincidentally produce tracks closely resembling those of pecking beaked animals. Such a hypothesis is not as parsimonious as explaining these marks as being made by foraging pterosaurs.

As well as pecking, sweeping movements of pterodactyloid jaws may also be recorded in the pterosaur track record. These are long, shallow, sometimes curving channels in substrate closely associated with pterosaur manus or footprints. They are known from the Blackhawk Formation tracks mentioned above as well as alongside Pteraichnus prints from the Campanian Cerro del Pueblo Formation of North Mexico (Fig. 2d; Rodriguez-de la Rosa 2003). Similar, but broader marks attributed to sweeping pterosaur jaws are known from the Del Monte Mines tracksites (Lockley & Wright 2003). These marks were interpreted as ‘tail drags’ by Parker & Balsley (1989) rather than impressions of moving jaws, while those from the Cerro del Pueblo Formation were labelled as ‘tail drags of unknown origin’ by Rodriguez-de la Rosa (2003). If these traces were indeed made by pterodactyloid pterosaurs, tail drags are an unlikely explanation. The caudal series of pterodactyloids are characteristically short compared to their limb lengths, and unlikely to contact the ground without adoption of a pronounced crouch. In contrast, the long jaws of pterosaurs could easily reach ground level from a standing pose (Witton & Naish 2008, 2015).

The palaeoecological insight offered by pterosaur foraging traces is limited by the current inability to allocate pterosaur tracks to specific clades. Thus, while these tracks demonstrate that pterodactyloids sometimes fed terrestrially (e.g. Unwin 2005; Witton 2013), they provide little information about the habits of specific pterosaur groups. An exception to this problem might be the Cerro del Pueblo Formation tracks. Pterosaur pes prints from this formation are noted for being proportionally elongate compared to typical Pteraichnus prints (Rodriguez-de la Rosa 2003), a feature which characterizes pes prints of the Santonian/Campanian Korean pterosaur ichnotaxon Haenamichnus uhangriensis (Hwang et al. 2002). Further similarities between these prints are their distribution of padding around the heel and distal metatarsals, the lack of curving claw impressions and relatively short digits (Fig. 2di). These morphologies make the Cerro del Pueblo Formation tracks better approximations of Haenamichnus than Pteraichnus, and might suggest a similar track-maker. Haenamichnus is an exceptional pterosaur ichnotaxon in that a compelling case has been made for it belonging to a specific group, the Azhdarchidae, a group of often long-necked and gigantic edentulous pterodactyloids which dominated Late Cretaceous pterosaur evolution (Hwang et al. 2002). If elongate pes prints are characteristic of azhdarchids, the Cerro del Pueblo Formation tracks and possible jaw prints might also be referable to this group and present direct evidence of azhdarchid foraging behaviour.

Doubtful or uncertain evidence of pterosaur foraging

A number of specimens intimated as recording pterosaur foraging behaviour can be considered uncertain, suspect or flawed. Issues associated with these specimens include a lack of evidence for flying reptiles being associated with certain trace fossils, misidentification of gut content or loss of specimens.

One such report concerns the earliest report of pterosaur gut content. This specimen of ‘Pterodactylus propinquus’ from the upper Jurassic of Germany (Broili 1938) is now lost (Wellnhofer 1970), but was said to preserve disarticulated fish debris in and around the region of the gular pouch, including a 10 mm-long series of articulated vertebrae and fin debris. The association of this material with the throat region of the pterosaur suggests that they might represent fish remains regurgitated from the gut during death. However, the unknown whereabouts of this specimen makes the validity of these claims difficult to ascertain.

Two specimens of Rhamphorhynchus once suggested to possess gut content have since been reinterpreted. Döderlein (1929) suggested that the Rhamphorhynchus specimen BSP AS 1 772 preserved regurgitated gut content in its throat region, indicating a broad diet of fish, crinoids and crustaceans. Reanalysis of this specimen by Wellnhofer (1975) revealed no trace of these elements and cast doubt on Döderlein's observations. Wellnhofer (1975) suggested that the maxilla, several ribs and vertebrae of a small Leptolepis-like fish were found among the disarticulated torso skeleton of PTH 1954.39b (now SOS 2819). These elements have now been reinterpreted as sternal ribs belonging to the pterosaur, and not the remains of a second animal (Claessens et al. 2009).

Sanz et al. (2001) described a gastric pellet comprising four partial skeletons of juvenile birds, representing at least three different species, from the Early Cretaceous (Barremian) Las Hoyas Lagerstätte of Spain (LH 11386). The articulation of the skeletons within the pellet are said to rule out a mammalian or fish origin, and the size of the pellet is too large to have stemmed from known avian or lizard taxa from this locality. This regurgitate is therefore suggested to have originated with a pterosaur or small theropod dinosaur (Sanz et al. 2001). Evidence from Rhamphorhynchus specimens verifies that at least some pterosaurs swallowed small animals whole and adds credence to this hypothesis, but the identity of this pellet-maker cannot be conclusively established at this stage.

Ambiguous gut content is preserved in IVPP V14377, the Aptian, Jiufotang Formation holotype specimen of ‘Nemicolopterus crypticus’ (Wang et al. 2008; a case for this specimen representing a juvenile of a contemporary tapejarid pterosaur species is made in Witton 2013). This comprises ‘yellow coloured matrix occasionally forming rounded structures’ thought to represent decayed soft tissues or amorphous gut content.

Aberrant evidence of possible pterosaur foraging is the association of a large leaf between the mandibular rami of the Early Cretaceous (Aptian) Brazilian pterosaur Ludodactylus sibbicki (SMNK PAL 3828, Fig. 1f; Frey et al. 2003). This Crato Formation specimen represents an articulated cranium, mandible and hyoid apparatus. Frey et al. (2003) argue that the likelihood of this leaf associating so intimately with the pterosaur fossil through taphonomic processes is low, suggesting that the animal must have apprehended the leaf when feeding. Given that the spear-like dentition of Ludodactylus appears better suited to piscivory than herbivory, this association has been interpreted as a fossilized calamity: the pterosaur mistook the leaf for a component of its regular diet or apprehended the leaf with other food items, after which its sharpened fronds became lodged in the gular pouch. The frayed leaf margin in the gular region of the pterosaur fossil is interpreted as evidence of attempts to dislodge the leaf from the gular pouch, and the pterosaur's inability to do this led to its death (Frey et al. 2003). This explanation is not untenable, but there are attributes of the specimen which seem peculiar within this scenario. For instance, the hyoid apparatus is preserved on top of the leaf rather than enveloping it. Evidently at least this part of the jaw apparatus – which supports much of the throat tissue supposedly stabbed and stuck to the plant material – was not significantly attached to the leaf itself. A discussion could be had about the likelihood of the hyoid breaking free after the leaf was embedded, how detachment of the skull from the body might have disrupted the throat region, and so on. However, this would be no more testable than existing interpretations of this specimen, and therefore of little value to those wanting to understand pterosaur behaviour, diets and foraging methods. It may be most sensible to consider any explanation of this unusual fossil association as conjectural: this pterosaur may have made a fatal mistake when attempting to swallow a leaf, or it may not.

The damaged wings of two fossil insects have been ascribed to pterosaur predation by Tischlinger (2001). These specimens, representing the odonatan Cymatophlebia longialata (accessioned in a private collection) and the neuropteran Archegetes neuropterum (MM P.V. 1, 1/16), are both from the Jurassic Solnhofen Formation and are perfectly preserved except for the removal of the posterior right wingtip in the odonatan and anterior left wingtip in the neuropteran. The odonatan also demonstrates minor damage to the posterior margin of the right anterior wing which is confluent with the damage seen elsewhere on the specimen, suggesting a common cause. The broken wing margins of the odonatan specimen cut across the venation pattern to suggest that the wings were sheared rather than torn, while the damage inflicted on MM P.V 1, 1/16 appears to have largely resulted from tearing or ripping.

There are a number of reasons to entertain the idea that pterosaurs might have antagonized these insects. The damage suffered by these wings is greater than that produced by typical insect wing wear. At most, this may reduce wing area by 10% (Hedenström et al. 2001). The excellent quality of the insect fossils suggests that they were not transported far before burial and thus are unlikely to have been damaged during taphonomic relocation. Aggressive acts from other organisms seem a likely cause of these pathologies, and the fact the insects are perfectly preserved other than their damaged wings might suggest that they were not attacked by an aquatic predator – it is unlikely that an aquatic animal would immediately ignore wounded, submerged prey after partly successful strikes. These points imply aggressive acts taking place above, rather than in, the Solnhofen lagoon. All three specimens share damage to their wing tips, matching areas considered likely to record failed feeding traces by modern birds predating lepidopterans (Wourms & Wasserman 1985), and the general extent of damage to each specimen might betray the involvement of a relatively large predator. Pterosaurs are perhaps the most suitable animals in the Solnhofen palaeobiota to predate volant insects and some taxa from these deposits – such as Anurognathus ammoni – seem specifically adapted for this form of predation (Wellnhofer 1991; Bennett 2007; Witton 2013). All this said, there is a limit to what damaged and torn insect wings can reveal about their origins, and direct evidence of pterosaur involvement with these insects remains wanting. Moreover, other scenarios (e.g. intraspecific disputes between insect individuals) provide equally compelling explanations of these specimens. They cannot, therefore, be considered wholly reliable insights into pterosaur foraging behaviour.

Bite marks and shed teeth

Many vertebrate specimens from Late Cretaceous Chalk deposits of North America are associated with the teeth and tooth marks of carnivorous vertebrates, particularly those of sharks and other predatory fish (e.g. Schwimmer et al. 1997; Shimada 1997; Everhart 2005; Everhart & Ewell 2006). Several fossils of the Santonian pterodactyloid Pteranodon bear characteristic tooth gouges made by the serrated teeth of Squalicorax and the teeth of this shark are often found alongside Pteranodon specimens (Bennett, pers. comm. 2007). Full details of many of these specimens remain unpublished, but Ehret et al. (2015) have recently outlined two Mooreville Chalk Formation Pteranodon wing bones (RMM 3274 and ALMNH 2014.1.200) with serrated bite marks probably attributable to the shark Squalicorax kaupi. The ALMNH Pteranodon also bears bite marks from a second fish, presumed to be a saurodontid such as Saurodon or Saurocephalus. The anterior cervical vertebra of the Pteranodon specimen LACM 50926 (Niobrara Formation) is preserved with a Cretoxyrhina mantelli tooth wedged between vertebral processes (Fig. 3f), providing evidence that this large lamiform shark also ate pterosaurs. Work on this specimen, including a full description and determination of the relationship between shark tooth and pterosaur bone, is currently underway (D. Hone, pers. comm. 2015).

Several accounts of theropod dinosaurs biting or ingesting pterosaur bone are known (Fig. 3). Two of these are represented by tooth material embedded in pterosaur specimens, allowing identification of these animals to species level in both cases. Currie & Jacobsen (1995) reported an embedded tooth tip and several tooth gouges from the dromaeosaur Saurornitholestes langstoni in the tibia of TMP 92.83 (Fig. 3b), a partial azhdarchid skeleton from the Late Cretaceous (Campanian) Dinosaur Park Formation, Canada. Buffetaut et al. (2004) demonstrated that larger theropods also ate pterosaurs with a spinosaurid tooth driven into an ornithocheirid cervical vertebra from the Early Cretaceous (Albian) Santana Formation of Brazil (WDC-SFB-001c, Fig. 3a). This tooth matches those described for the spinosaurid Irritator challengeri, also known from the Santana Formation (Sues et al. 2002). Owing to the lack of acid etching on the pterosaur material, it is suggested the vertebra was bitten, but not consumed, by this dinosaur.

Other animals known to have consumed pterosaurs include Crocodyliformes. Bite marks from these reptiles are recorded on the Romanian holotype specimen of the azhdarchid Eurazhdarcho langendorfensis (EME VP 312, Sebeş Formation, Maastrichtian), specifically the cervical vertebrae and distal metacarpal IV (Fig. 3e; Vremir et al. 2013). These rounded punctures, breaks and regions of crushing are reminiscent of marks left by biting Crocodyliformes, and such traces are common to bones from this formation (Vremir et al. 2013).

A series of tooth marks are seen on an isolated, Australian ornithocheiroid wing metacarpal (NMV P197962, Fig. 3c) from the Early Cretaceous (Albian) Toolebuc Formation (Kellner et al. 2010). These shallow punctures, associated with slight crushing, are aligned in a row along the anterodorsal margin of the bone. The identity of the animal which created these bite marks is not known. A further possible record of pterosaur tissue consumption was reported by Kellner & Langston (1996), who suggested that three unusual openings in the quadrate of the azhdarchid, Quetzalcoatlus sp. specimen TMM 4422-30 (Fig. 3d; Javelina Formation, Maastrichtian), may record puncture wounds. Unfortunately, these openings also reveal little about the identity of the animal which produced them.

Pterosaurs as gut content

Pterosaur bone has also been found within the abdominal cavity of MPC-D100/54 (Fig. 3g), a specimen of the Campanian Mongolian dromaeosaur Velociraptor mongoliensis (Hone et al. 2012). The pterosaur material comprises only scant remains – fragments of limb bone shafts – but demonstrates the very thin bone compacta characteristic of Cretaceous pterodactyloids and is enclosed by ribs within the fully articulated dinosaur skeleton. Circumstantial evidence suggests that an azhdarchid identity is likely for the pterosaur material because, although it is too fragmentary to be linked with any specific pterodactyloid clade, azhdarchid pterosaurs seem to have dominated latest Cretaceous faunas and are present in contemporary east-Asian deposits (Hone et al. 2012).

Pterosaur-consumer gut regurgitates

Fossil gut regurgitate containing pterosaur remains is rare, although a number of examples, now considered doubtful, have been reported (see below). Presently, only a gastric pellet from the Kimmeridgian Nusplingen limestones of Germany (SMNS Inv.-No. 63990, Fig. 3h) seems to be a valid example of a regurgitated pterosaur. This is thought to represent the partly digested remains of a Rhamphorhynchus (Schweigert et al. 2001), the specimen including at least a wing metacarpal, caudal vertebrae and wing finger fragments. It has been suggested that the regurgitate originated with a marine crocodylomorph or large predatory fish, both of which are known from the same deposit. Regurgitates of modern crocodiles do not contain any bony material but are instead composed of keratinous tissues such as hair and claw sheaths (Fisher 1981). If these characteristics were true for Mesozoic crocodylomorphs, a piscine origin for SMNS Inv.-No. 63990 seems more likely.

Pterosaurs preserved with their predators

A set of fossils from the Tithonian Solnhofen Formation show associated, complete skeletons of Rhamphorhynchus muensteri and the much larger predatory ganoid fish Aspidorhynchus acutirostris (Frey & Tischlinger 2012; Weber 2013). These associations, of which five are known, are believed to be the result of predatory acts from the Aspidorhynchus. Accessioned examples of these specimens belong to the Wyoming Dinosaur Centre (WDC CSG 255; Fig. 3i, described in detail by Frey & Tischlinger (2012)) and Bergér Museum (no specimen number), with the three remaining specimens being housed in private collections. All five examples are illustrated by Weber (2013). That these specimens are not just chance associations but have palaeoecological significance is suggested by several points. Associations between large vertebrates are very rare in the Solnhofen plattenkalk, making the recovery of five associations with identical taxonomic compositions unlikely to be a chance occurrence. All skeletons in these associations are in the same state of excellent preservation and articulation (soft-tissues occur in at least two of the pterosaurs – Weber 2013), suggesting the associated animals have similar taphonomic history. The arrangement of the associations is also consistent in that the pterosaurs are always located around the skull and jaws of their respective Aspidorhynchus. In some examples (e.g. WDC CSG 255) the wing bones of the Rhamphorhynchus are preserved under the jaws of the neighbouring fish. These details point to the pterosaurs being seized, carried by and ultimately preserved with Aspidorhynchus individuals. That five near-identical examples of the same phenomenon are known suggests that interactions between Rhamphorhynchus and Aspidorhynchus may have once been common occurrences.

Doubtful or uncertain instances of pterosaurs as food sources

A specimen of Triassic gastric ejecta (MFSN no. 1891), once thought to contain remains of cf. Preondactylus, has traditionally been regarded as the stratigraphically oldest evidence of a predator–pterosaur interaction (Dalla Vecchia et al. 1989). Successive re-examination of the remains has seen their pterosaurian nature look increasingly doubtful, however. Dalla Vecchia (2003) thought them no more diagnosable than Pterosauria incertae sedis, and more recently Holgado et al. (2015) provided detailed, compelling evidence for the pellet comprising the bones of a protorosaurian archosauromorph. A further regurgitated pterosaur skeleton from Barremian/Aptian deposits of southern Mongolia was described by Kurochkin (1991), but subsequent analysis has revealed that bones within the specimen are far more bird-like than pterosaurian (Bakhurina & Unwin 1995). The association of two preserved feathers with the pellet corroborates this observation, and the specimen is now regarded as a small, possibly enantiornithine bird (Bakhurina & Unwin 1995).

Evidence of Santonian plesiosaurs ingesting pterosaurs in the Western Interior Seaway was recorded by Brown (1904). The plesiosaur specimen AMNH 5803, from the Campanian Pierre Shale Formation of South Dakota, was reported as having gut content containing broken pterodactyloid limb elements. However, the specimen is badly weathered and the alleged pterosaur material is poorly preserved, casting some doubt over its identity (C. Mehling, pers. comm. 2007). An additional case of a marine reptile consuming a pterosaur was reported by Martill (1986) based on the discovery of delicate, hollow bones in the gut region of the Oxford Clay (Callovian) marine crocodile Metriorhynchus (LEIUG 90985). Subsequent analysis has suggested these bones may have belonged to a fish rather than a pterosaur (Forrest 2003).

Kellner et al. (2010) suggested that insect borings had been found in pterosaurs, citing the skull description of Quetzalcoatlus sp. authored by Kellner & Langston (1996) as evidence of this fact. This suggestion is confusing as no insect borings were mentioned in this description or other documentation of Quetzalcoatlus remains, and they may be erroneous. Confirmed utilization of pterosaur tissues by invertebrates is known in the form of invertebrate rasping traces, possibly representing activity by bone-consuming osteopeltid gastropods, on pterosaur bones from the Albian Cambridge Greensand Formation (Unwin 2001). Encrusting organisms in the form of serpulid worms and remnants of oyster shells also occur on pterosaur fossils from this unit. All cases reflect the use of settled, probably long-dead pterosaur carcasses and therefore provide relatively little information concerning the lifestyles of flying reptiles, although they are notable for being rare examples of invertebrates utilizing pterosaur bones.

Pterosaurs in Mesozoic foodwebs

An obvious question concerning the fossils reviewed here is whether they represent predatory acts on living pterosaurs, or scavenging acts on dead ones. It seems likely some of these examples represent genuine predation. Specifically, the multiple Aspidorhynchus and Rhamphorhynchus associations, which are perfectly preserved, consistently arranged and of identical taxonomic composition probably record predatory behaviour. Frey & Tischlinger (2012) provide a plausible explanation for these fossils as failed predatory attempts from Aspidorhynchus. They suggest that, once apprehended, at least some Rhamphorhynchus were too large, complex or entangled for the fish to ingest, and that the animals died when mistakenly entering the toxic bottom waters ultimately responsible for the excellent preservation of Solnhofen fossils. Frey & Tischlinger (2012) invoke skim-feeding as a possible reason for Rhamphorhynchus to be close enough to water to be seized by a fish. Numerous authors have been critical of the skim-feeding pterosaur hypothesis (e.g. Chatterjee & Templin 2004; Humphries et al. 2007; Witton & Naish 2008, 2015) and it seems unreasonable – and unnecessary – to invoke the involvement of this behaviour here. A less speculative interpretation is simply that Rhamphorhynchus routinely entered aquatic settings and their surroundings, perhaps entering water directly (see discussion, below), and were potential prey for aquatic predators.

Circumstances surrounding other examples of pterosaurs being ingested by carnivorous animals are less clear. It is apparent that the pterosaurs recorded in these specimens were dead or dying when being eaten because none show evidence of recovery from their wounds. All records of theropods feeding on pterosaurs are suggested to be acts of scavenging rather than predation. Currie & Jacobsen (1995) cite the size disparity between the 2 m-long dromaeosaur Saurornitholestes and a 6 m wingspan azhdarchid as reason for assuming that the dinosaur was eating carrion. The Brazilian pterosaur vertebra associated with an Irritator tooth reverses this scenario with a large theropod and comparatively small pterosaur (Buffetaut et al. 2004), but scavenging is still cited as favourable over active predation. Greater equality of size is represented by the Velociraptor and azhdarchid specimen described by Hone et al. (2012), the pterosaur being thought to span 2–3 m across the wings and the dromaeosaur being of similar length. Hone et al. (2012) suggest that the pterosaur would have been a challenging prey item for Velociraptor, despite the estimated 44% weight advantage given to the dinosaur, and conclude that this specimen also represents scavenging behaviour. Shark feeding traces on Pteranodon bones are also interpreted as scavenging behaviour owing to the lack of healing exhibited on the pterosaur fossils (Ehret et al. 2015). The relatively small size of Squalicorax (typically less than 3.0 m long) is often used as reasoning for its bite marks representing scavenging behaviour on contemporary marine animals (Schwimmer et al. 1997).

How pertinent these conclusions are remains uncertain, as equally compelling arguments could be made for these specimens representing predatory scenarios. Volant animals are routinely killed by terrestrial or even semi-aquatic animals of all sizes in modern times (e.g. Emmons 1986; Hutton 1987; Pérez-Higareda et al. 1989; Thorbjarnarson 1990; Wang 2002), and it is not unreasonable to suppose that dromaeosaurid and spinosaurid theropods might have been capable of subduing pterosaurs. Likewise, the mass disparity between even large Pteranodon and sharks such as Squalicorax and Cretoxyrhina was probably tremendous: using modern shark masses as a guide, Witton (2013) suggested that even a modestly sized, 3 m-long Mesozoic shark would weigh between 100 and 250 kg, considerably more than the highest reasonable mass estimates for large Pteranodon (Witton & Habib 2010). Such a discrepancy in mass suggests that, so long as they could reach them, these large, sharp-toothed predators would not struggle to overpower swimming or floating Pteranodon. A modern equivalent of this behaviour is seen when living seabirds are attacked and eaten by sharks (Wetherbee & Cortés 2004), suggesting that this scenario is not implausible for ancient communities. Ultimately, ascertaining the predatory or scavenging nature of these specimens is difficult, perhaps impossible, from scraps of ingested bone, rare bite marks and solitary embedded teeth.

A similar conclusion might be drawn about the preserved gut content of pterosaurs themselves. Being mostly disarticulated fish bones, they provide little insight into whether these pterosaurs predated or scavenged their meals. Exceptions are the SOS 4599 and WDC CSG 255 Rhamphorhynchus and SMNS 59395 Scaphognathus. The articulated dietary remains of these specimens suggest that they were eaten whole, or nearly whole, and imply that the pterosaurs in question were either involved in killing the ingested animals or able to scavenge them relatively quickly after their death from other causes.

Less speculative use of the pterosaur palaeoecological record is using its basic data – such as the nature of the predatory animals eating pterosaurs, or the dietary preferences of pterosaur species – to test functional and biomechanical hypotheses of flying reptile habits. The predicted lifestyles of Rhamphorhynchus as a soaring-adapted, marginal marine predator of pelagic prey and of Pteranodon as a sea-going specialist (e.g. Wellnhofer 1975; Bennett 2001; Unwin 2005; Tütken & Hone 2010; Witton 2013) are entirely consistent with their palaeoecological records. Both demonstrate multiple interactions with pelagic species, showing them both as consumers and the consumed in their respective ecosystems in the manner predicted by functional studies. Recent studies have shown that these animals have numerous adaptations for efficient take-off from aquatic settings (including reduced hindlimbs, small torsos, ‘hatchet-shaped’ or ‘warped’ deltopectoral crests and others – see Habib & Cunningham 2010), suggesting they may have routinely entered water (Fig. 4b). This may account for their relatively good palaeoecological records, elevated quality of fossil preservation and common encounters with piscine carnivores (also see Lockley & Wright (2003) and Hone & Henderson (2014) for discussions of pterosaur swimming).

The azhdarchid palaeoecological record also matches lifestyle predictions for this group made through functional studies. The habits of these pterosaurs have long been controversial, but recent, seemingly largely accepted work on azhdarchid functional anatomy and taphonomy suggests that they foraged for small animals in terrestrial environments (Witton & Naish 2008, 2015). The Mexican pterosaur trace fossils and associated possible beak scrapes, suggested here to belong to azhdarchids, might record a member of this group engaging in ground-based foraging. This possible feeding trace has further significance in demonstrating that azhdarchids could reach the ground with their beak tips, despite their relatively inflexible necks. Neck arthrology has been a point of contention for many azhdarchid lifestyle hypotheses (see discussions in Witton & Naish 2008, 2015), making possible trace fossils of azhdarchid jaw tips useful evidence for the terrestrial foraging hypothesis. Additional support for this idea comes from examples of azhdarchid bodies being eaten by typical ‘terrestrial’ Mesozoic faunas, such as dinosaurs and Crocodyliformes (Fig. 4c). This meets taphonomic assessments showing that the azhdarchid record is biased towards inland, continental habits and not waterways or seas (Buffetaut et al. 1996; Witton & Naish 2008).

Studies of pterosaur lifestyles remain in their infancy, but the fossils reviewed here provide useful tests of palaeoecological hypotheses proposed for these often controversial animals. The expansion of this dataset through new discoveries and reappraisals of historic remains should be pursued by those wanting to understand the role of pterosaurs in Mesozoic ecosystems. To that end, curators, collectors and researchers are encouraged to carefully check for and record pterosaur specimens with evidence of their places in Mesozoic food webs.