Phosphatodraco

Illustration of the holotype cervical vertebra series as preserved, with interpretation of its order by Pereda-Suberbiola and colleagues, 2003 (above, C5–C9), and Kellner, 2010 (below, C3–C8)

Pereda-Suberbiola and colleagues originally interpreted the five preserved cervical vertebrae of Phosphatodraco as cervicals C5–C9. The frontmost preserved vertebra they interpreted as C5 consisted of two fragments; they found it unlikely that these belonged to two different vertebrae, since they lay in continuity with no sediment in between, and overlapped each other in some areas. They considered the sideways expansion at the front of this vertebra to be due to crushing, and pointed out that such preservation where fragile, yet well-preserved bones are associated with damaged material of the same individual is known from other vertebrate fossils in the same level. They identified the frontmost vertebra as a C5 because this is usually the longest cervical vertebra in pterosaurs, their length decreasing hindward.[1]

In 2007 paleontologist Alexander W. A. Kellner and colleagues noted that Phosphatodraco was one of the most interesting azhdarchids found in Africa, but used cautious language about the original interpretation of the vertebrae.[9] Kellner suggested in 2010 that interpretation of the holotype specimen had been affected by taphonomy (changes during decay and fossilisation); he instead proposed that the element originally described as a fifth cervical vertebra was actually the third and fourth vertebrae which had been compressed together, giving the impression that they were a single vertebra broken in the middle. This would shift the order of the vertebrae to C3–C8, and though this did not change the validity of the taxon (its distinctness being based on its stratigraphy, geography, and morphology), Kellner noted that the diagnosis (the suite of features that distinguish a taxon) and size estimate had to be reevaluated.[10] The frontmost cervical vertebrae which are missing from Phosphatodraco in either scheme are the atlas (C1, which connects with the back of the skull) and the axis (C2).[10][11]

Subsequent articles from 2011 and 2015 with Kellner among the co-authorship have concurred with Kellner's interpretation.[12][13] Paleontologist Alexander Averianov disagreed with Kellner's reinterpretation of the cervical vertebrae in 2014, and considered the original description accurate.[14] A 2015 article by paleontologist Mátyás Vremir and colleagues called the issue "controversial" and considered the specimen too crushed for proper comparison,[15] and Martill and Markus Moser concurred with this in 2018.[16] Paleontologists Darren Naish and Mark P. Witton (the co-authors of Vremir's article) followed Kellner's interpretation in 2017.[1][17] Paleontologist Rodrigo V. Pêgas and colleagues also followed Kellner's order in 2021.[18] Though the palaeontologist Alexandru A. Solomon and colleagues noted the suggested change in interpretation of the holotype order in 2019, they stated that even if the reinterpretation was correct, the specimen was too damaged for comparison with the single known cervical vertebra of their new genus Albadraco.[19]

Azhdarchid neck-lengths compared to their estimated wingspans. Phosphatodraco is fourth from the top, following Kellner's interpretation. (The last preserved vertebra is not shown)

The suite of features that distinguish a taxon from other related taxa is called a diagnosis, and in the case of Phosphatodraco, these features are all found in the cervical vertebrae. Since Pereda-Suberbiola and colleagues considered the preserved vertebrae to be C5–C9 of the series in their description, that is the diagnosis and description followed here.[1] Note that if Kellner's suggestion that the series actually represents vertebrae C3–C8 is correct, the diagnostic features listed by Pereda-Suberbiola and colleagues may possibly be inaccurate, and the following description would refer to different vertebrae in the series.[10] Pereda-Suberbiola and colleagues found Phosphatodraco to differ from other azhdarchids in that the hindmost vertebra, C8 in their order, is very elongated, 50% longer than C5, and has a prominent neural spine (the spine that projects upwards from the vertebra) that is almost as tall as the centrum, truncated in a square shape at the top, and located far back. Phosphatodraco is also distinct in that the ratio between the maximum length of the vertebrae and the front width between prezygapophyses (processes at the sides of the centrum which connected with the postzygapophyses of the previous vertebrae) of the middle cervical vertebrae is about 4.3 in C5 and 4.1 in C6.[1]

The five preserved cervical vertebrae have hollow centra and their cortical bone (the outer, thick layer) is about 1 mm (0.039 in) thick. The vertebrae vary in length, the longest being the frontmost of those preserved, a C5 broken in two according to Pereda-Suberbiola and colleagues (C3–C4 according to Kellner[10]), which they estimated to have been about 300 mm (0.98 ft) long when complete. The first of these fragments is 110 mm (4.3 in), the second is about 190 mm (7.5 in). When viewed from the side, the hind end of this vertebra shows a developed left postzygapophysis, and the convex articular condyle (the condyle that connected with the following vertebra) and left postexapophysial process (which connected with the preexapophys at the front of the preceding vertebra) is in front of it, all lying in the same plane due to crushing.[1]

C3–C5 vertebrae of Quetzalcoatlus (A–C) compared with those of Phosphatodraco (D)

The C6 (Kellner's C5[10]) is the best-preserved of the vertebrae, and is shorter than the preceding C5, about 225 mm (8.9 in) long. It is distorted so that the front part is visible in lower side view, and the hind part is visible in left side view. Its centrum is procoelous (concave at the front surface), its prezygapophyses are horn-like, and nearly concave and parallel when seen from below. The right prezygapophys has a small tubercle (a rounded projection) at the midline, and there is no indication of additional processes at the front end of the vertebra or pneumatic foramina (holes) on the side surface of the centrum. The front cotyle (the concave front end of the centrum) is distorted, but it appears to be twice as wide as high, ovoid (or egg-shaped), and with a slightly concave upper margin. The lower margin has a prominent hypapophysis (a downwards projection), and this keel's height diminishes towards the centrum's mid-length. There is a longitudinal oval sulcus (a groove) on the right side of the lower surface, near the base of the prezygapophysis. The lower surface of the centrum is nearly flat, and the postexapophys is well-developed at the lower side of the condyle, like in the preceding vertebrae.[1]

The following C7 vertebra (Kellner's C6[10]) is visible in bottom view, is missing the hind part, the preserved part being 190 mm (7.5 in) long, and the complete length perhaps being the same or shorter as the preceding C6. Its prezygapophyses are similar to those of the preceding vertebra, and the cotyle is oval and compressed from top to bottom. There is a sulcus below the left prezygapophys, apparently without a ridge extending hindward. The centrum bulges slightly hindward, becoming narrower at its mid-length. The postzygapophyses are well-developed, and diverge widely from the longitudinal midline of the centrum. A small protuberance between the postzygapophyses perhaps indicates where the upper margin of the neural canal (through which the spinal cord passed) was located, though its features cannot be accurately determined.[1]

Characteristics of azhdarchid cervical vertebrae; J is the C7 of Phosphatodraco following Kellner's order (it is C8 in Pereda-Suberbiola and colleagues' order)

The next to last vertebra is C8 (Kellner's C7[10][17]) which is visible in side view, and despite missing some of the front, the centrum is very elongated, measuring 150 mm (5.9 in). The most notable feature of this vertebra is its tall neural spine, which is placed at its hindmost part. The neural spine is 40 mm (1.6 in) high measured from the upper surface of the postzygapophys to the top, almost the same height as the centrum, which is 45 mm (1.8 in). The front and hind margins of the neural spine are vertically parallel to each other, and its top is truncated in a square shape, and perpendicular to the side edges. The left postzygapophys is located at the base of the hind end of the neural arch (which forms the arch of bone through which the spinal cord passed). It is similar to the same vertebra of Quetzalcoatlus in that the neural spine is square on top, but differs in being placed so far back. The left postexaphophyseal process is well-developed at the back and below, but does not extend past the condyle as it does in the preceding vertebrae.[1]

The last vertebra is the C9 according to Pereda-Suberbiola and colleagues (Kellner's C8,[10]) which is visible in hind view, and its preserved part is 75 mm (3.0 in) high. Its neural arch has a large neural spine and well-developed transverse processes (which projected from the sides of the centrum and acted as attachment points for muscles and ligaments). The neural spine terminates in a blunt process above, and the hind side of the neural spine has an oval depression, which has thick vertical edges at its sides. The transverse processes are long and slender, and project to the sides and slightly downwards. The neural canal is small and nearly circular, is about 22 mm (0.87 in) in diameter, and there are no pneumatic foramina near it. Its condyle is broad, around five times wider than high, and is crescent-shaped in cross-section. The left postexapophysis is placed at the side of the condyle and almost vertical. Though none of the vertebrae preserve cervical ribs, the development of the transverse processes of the last vertebra indicates that it probably had ribs.[1] The indeterminate bone fragment associated with the two last vertebrae has a similar texture to them, is flat and crescent-shaped, and is about 9 mm (0.35 in) wide and 44 mm (1.7 in) long.[1]

Life restoration showing Phosphatodraco in terrestrial pose

Pereda-Suberbiola and colleagues found that the frontmost preserved cervical vertebrae of Phosphatodraco (their C5–C7) were similar in form to those of the mid-series cervical vertebrae of other long-necked pterodactyloid pterosaurs (the group consisting of short-tailed pterosaurs). The last hindmost preserved vertebrae (their C8–C9) have some features in common with the rest of the vertebrae, including broad, ovoid cotyles and condyles, as well as postexapophyses, but they differ in their neural canal being demarcated from the centrum and in having a prominent neural spine. Pereda-Suberbiola and colleagues suggested the C8–C9 could be cervicalized dorsal vertebrae, back vertebrae which have been incorporated into the neck. The total number of cervical vertebrae in pterosaurs varies between seven and nine, and the first dorsal vertebra is considered to be the first one that connects with the sternum (breast bone). Early pterosaurs like Rhamphorhynchus had eight cervical vertebrae with cervical ribs on at least C3–C8; later pterodactyloids had seven vertebrae and no ribs. In later pterdactyloid groups, nine cervical vertebrae are present, two of them being cervicalized dorsals, and adults have a notarium (a structure consisting of fused vertebrae in the shoulder region, also seen in birds).[1]

Skeletal diagrams showing Zhejiangopterus (left, as preserved without skull, right, as reconstructed), one of the two other azhdarchids known preserved with a relatively complete neck (the other is Quetzalcoatlus)

In 2008 Witton and Naish pointed out that although azhdarchids have historically been considered to have been scavengers, probers of sediment, swimmers, waders, aerial predators, or stork-like generalists, most researchers until that point had considered them to have been skim-feeders living in coastal settings, which fed by trawling their lower jaws through water while flying and catching prey from the surface (like skimmers and some terns). In general, pterosaurs have historically been considered marine piscivores (fish-eaters), and despite their unusual anatomy, azhdarchids have been assumed to have occupied the same ecological niche. Witton and Naish noted that evidence for this mode of feeding lacked support from azhdarchid anatomy and functional morphology; they lacked cranial features such as sideways compressed lower jaws and the shock-absorbing adaptations required, and their jaws instead appear to have been almost triangular in cross-section, unlike those of skim-feeders and probers.[6]

Witton and Naish instead stated that azhdarchids probably inhabited inland environments, based on the taphonomic contexts their fossils have been found in (more than half the fossils surveyed were from for example fluvial or alluvial deposits, and most of the marine occurrences also had fossils of terrestrial lifeforms), and their morphology made them ill-suited for lifestyles other than wading and foraging terrestrially, though their feet were relatively small, slender, and had pads, not suited for wading either. These researchers instead argued that azhdarchids were similar to storks or ground hornbills, generalists they termed "terrestrial stalkers" that foraged in different kinds of environments for small animals and carrion, supported by their apparent proficiency on the ground and relatively inflexible necks, for example the well-preserved neck of Phosphatodraco providing information about their morphology. Witton and Naish suggested that their more generalist lifestyle could explain the group's resilience compared to other pterosaur lineages, which were not thought to have survived until the late Maastrichtian like the azhdarchids did (pterosaurs went extinct along with the non-bird dinosaurs during the Cretaceous-Paleogene extinction event 66 million years ago).[6][2]

Witton elaborated in a 2013 book that the proportions of azhdarchids would have been consistent with them striding through vegetated areas with their long limbs, and their downturned skull and jaws reaching the ground. Their long, stiffened necks would be an advantage as it would help lowering and raising the head and give it a vantage point when searching for prey, and enable them to grab small animals and fruit.[24] In their 2021 study that reinterpreted Tethydraco as an azhdarchid, and possibly the same as Phosphatodraco, Labita and Martill noted that azhdarchids might have been less terrestrial than suggested by Witton and Naish, since the Moroccan fossils were from marine strata, as was Arambourgiania from the phosphates of Jordan. They noted that no azhdarchids had been found in truly terrestrial strata, and proposed they could instead have been associated with aquatic environments, such as rivers, lakes, marine and off-shore settings.[7]

Pterosaurs are generally thought to have gone gradually extinct by decreasing in diversity towards the end of the Cretaceous, but Longrich and colleagues suggested this impression could be a result of the poor fossil records for pterosaurs (the Signor-Lipps Effect). Since they found multiple lineages (Pteranodontidae, Nyctosauridae, and Azhdarchidae, the latter including Phosphatodraco and others) to have co-existed during the late Maastrichtian of Morocco, this is the most diverse pterosaur assemblage known from the Late Cretaceous. Pterosaurs during this time thereby had increased niche-partitioning compared to earlier faunas from the Santonian and Campanian ages, and they were able to outcompete birds in large size based niches, and birds therefore remained small, not exceeding 2 m (6.6 ft) wingspans during the Late Cretaceous (most pterosaurs during this time had larger wingspans, and thereby avoided the small-size niche). To these researchers, this indicated that the extinction of pterosaurs was abrupt instead of gradual, caused by the catastrophic Chicxulub impact. Their extinction freed up more niches that were then filled by birds, which led to their evolutionary radiation in the Early Cenozoic.[2]

Possible azhdarchid trackway Haenamichnus from Korea (left) and feeding posture inferred from the tracks

Witton summarized ideas about azhdarchid flight abilities in 2013, and noted they had generally been considered adapted for soaring, although some have found it possible their musculature allowed flapping flight like in swans and geese. Their short and potentially broad wings may have been suited for flying in terrestrial environments, as this is similar to some large, terrestrially soaring birds. Albatross-like soaring has also been suggested, but Witton thought this unlikely due to the supposed terrestrial bias of their fossils and adaptations for foraging on the ground. Studies of azhdarchid flight abilities indicate they would have been able to fly for long and probably fast (especially if they had an adequate amount of fat and muscle as nourishment), so that geographical barriers would not present obstacles.[25]

Azhdarchids are also the only group of pterosaurs to which trackways have been assigned, such as Haenamichnus from Korea, which matches this group in shape, age, and size. One long trackway of this kind shows that azhdarchids walked with their limbs held directly underneath their body, and along with the morphology of their feet indicates they were more proficient on the ground than other pterosaurs. According to Witton, their proportions indicate they were not good swimmers on the other hand, and though they could probably launch from water, they were not as good at this as some other pterosaur groups.[25]