Friday, 24 October 2014

The pterodactyl tree

The relationships within Pterosauria have been the subject of phylogenetic debate since the 19th Century. In the very latest 20th century it was subject to rigorous cladistic analyses, which became more refined and comprehensive in the early two-thousands. There have been multiple approaches to pterosaur phylogeny, tackling both the Pterosauria as a whole and taking a ‘split approach’ -studying the stem-group non-pterodactyloids and the pterodactyloids separately.

In a recent study (Vidovic and Martill 2014) a contentious distribution of pterodacyloids was recovered during a cladistic analysis. The analysis was a pterodactyloid-specific study, and the first published to use non-pterodactyloid monofenestratans as the out-group. Being a pterodactyloid-specific analysis it was not a completely comprehensive analysis, thus it does not hold more “weight” than say the analysis of Andres et al. 2014. However, it is the largest pterodactyloid analysis to date, utilising more effective methods of extracting data from morphometrics (TNT’s continuous states function), and the characters were scrutinized to avoid compound characters (i.e. those characters that do not express all the information you tell them). For these reasons the new phylogeny presents an interesting hypothesis of pterosaur relationships that should not be dismissed without good reason.

A single most parsimonious tree of the Pterodactyloidea recovered using a TNT “new technology search”.Named nodes: 1 = Monofenestrata Lü et al. 2010 [30]; 2 = Pterodactyloidea Plieninger 1901 [31]; 3 = Ctenochasmatidae Nopcsa 1928 [32]; 4 = Lophocratia Unwin 2003 [33]; 5 = Aurorazhdarchidae fam. nov.; 6 = Ornithocheiroidea Seeley 1891 [34]; 7 = Istiodactylidae Howse et al. 2001 [35]; 8 = Pteranodontia Marsh 1876 [36]; 9 = Anhangueridae Campos and Kellner 1985 [37]; 10 = Tapejaroidea Kellner 1996 [38]; 11 = Azhdarchoidea Nesov 1984 [39]; 12 = Azhdarchidae Nesov 1984 [39]; 13 = Tapejaridae Kellner 1989 [40].

Now that I have explained why the cladogram is worth consideration, it is worth saying what is so different about it.

As I have previously mentioned the Vidovic and Martill (2014) analysis was the first to use non-pterodactyloid monofenestratans as the out-group. The significance of this out-group is that species of Darwinopterus (as well as Wukongopterus and Cuspicephalus) possess the confluent nares and antorbital fenestra (nasoantorbital fenestra) that pterodactyloids have, but in every other sense are more similar to non-pterodactyloids. Because this non-pterodactyloid group is so close to pterodactyloids we can be confident that Darwinopterus can polarise the characters effectively. Perhaps due to the use of Darwinopterus as the out-group, or the splitting down of compound characters, neither the Ctenochasmatoidea nor the Archaeopterodactyloidea have been recovered. The Ctenochasmatoidea and Archaeopterodactyloidea each contain the Gallodactylidae, the Ctenochasmatidae and Pterodactylus, the difference between the two is that the former is exclusive of Germanodactylus and the latter is inclusive of Germanodactylus. Maisch et al. (2004) conducted a specific analysis on dsugaripteroids (following Unwin 2003, with Ctenochasmatoidea and Germanodactylus included in Dsungaripteroidea) and concluded that Germanodactylus is paraphyletic –meaning it contains two or more closely related genera. In the Vidovic and Martill (2014) analysis the two Germanodactylus species were found to be incredibly distinct, perhaps even polyphyletic –meaning the inferred relationship is based on homoplasy (convergent characters) rather than symplesiomorphies (shared ancestral characters). In the analysis Germanodactylus cristatus was found to be the sister taxon to dsungaripterids and azhdarchoids, closer to Unwin’s (2003) hypothesis, whereas “Germanodactylus rhamphastinus” is the sister taxon to the new family group Aurorazhdarchidae, which is part of Archaeopterodactloidea sensu Kellner (2003). So, neither of the most favoured hypotheses seen in other analyses are entirely supported by this analysis. Moreover, the family groups that are normally grouped together -Pterodactylus, the ctenochasmatids (Ctenochasmatinae and Gnathosaurinae), Gallodactylidae and Ardeadactylus- are spread out stepwise between the node for Pterodactyloidea and the unnamed clade containing ornithocheiroids and Germanodactylus. The rest of the cladogram is mostly uncontroversial.

"Darwinopterus robustidens"

Is this distribution of “ctenochasmatoids” (From now on I will refer to both cteno. and archaeo. as “ctenochasmatoids”) illogical?
Perhaps not.

The new (since 2009) knowledge of non-pterodactyloid monofenestratans demonstrates that many of the features originally considered to define “ctenochasmatoids” are plesiomorphies. In the case of Vidovic and Martill (2014) these plesiomorphies are shared by several groups making the characters symplesiomorphies. More recently, after the paper and cladogram being discussed were submitted and peer reviewed for publication, a paper on a new ornithocheiroid, Hamipterus was published (Wang et al. 2014). Hamipterus is interesting, because it is clearly an ornithocheiroid, but possesses a long fibrous headcrest similar to that found in Darwinopterus and Cuspicephalus. This means that long fibrous headcrests are certainly not confined to “ctenochasmatoids”, nor are elongate necks, or elongate wing-metacarpals. In conclusion, there are no obvious characters uniting “ctenochasmatoids” and therefore the taxa are ‘strung out’ as a series of meta-taxa between the pterodactyloid node and the common ancestor of ornithocheiroids and Germanodactylus cristatus.

Hamipterus skulls with fibrous bony headcrests

Reconstruction of an ornithocheiroid pterosaur

Given that the new hypothesis of pterodactyloid phylogeny is not just plausible, but in the light of new evidence quite likely, the major group of pterodactyloids with dentitions confined to the anterior rostrum and long pteroid bones (~65% the length of the ulna) required a name. Part of the study was funded by the Linnean Society and Systematics Association, in order to test the phylogenetic position of Aurorazhdarcho and correct the family name from “Protazhdarchidae” to Aurorazhdarchidae if it required one. Indeed the cladistic analysis found Aurorazhdarcho shared characters like a long pteroid bone with Aerodactylus, thus Aurorazhdarchidae was erected for them and their closest relatives. Cycnorhamphus and Gladocephaloideus were also included in the Aurorazhdarchidae, so should the name Gallodactylidae have been used instead?
Reconstruction of aurorazhdarchid, Aerodactylus - adult and juvenile
Gallodactylidae was erected without definition in a coffee table book (albeit an excellent one [Wellnhofer 1991]). Much later Gladocephaloideus (Lü et al. 2012) was included in the family, accompanied by a formal definition which excludes Ardeadactylus, Aerodactylus and possibly Aurorazhdarcho. Later still, Bennett (2013) performed a taxonomic review, in which he made Cycnorhaphus the only genus to occupy Gallodactylidae. Indeed there is quite a morphological distinction between Gladocephaloideus and Cycnorhamphus, although undoubtedly Gladocephaloideus more similar to Cynorhamphus than any of the other aurorazhdarchids with its reduced dentition. It is likely that the grouping of Aerodactylus with Gladocephaloideus in the cladogram is the result of juvenile specimens being used in the analysis, thus sharing juvenile features. This is the same problem that led to Aerodactylus being considered Pterodactylus for so long! So for now I will follow Bennett (2013),  Gallodactylidae can be considered monotypic -at least until an analysis proves Lü et al.’s (2012) hypothesis that all pterosaurs with less than 50 teeth group together (note that amending the diagnosis to pterosaurs possessing 64 teeth would include “G. rhamphastinus” which shares little in common with Cycnorhamphus). In my opinion Gallodactylidae shouldn't have its diagnosis amended excessively, instead it should be reclassified to Gallodactylinae and nest within the more inclusive Aurorazhdarchidae.

In conclusion, the new phylogenetic hypothesis is a challenge to the established phylogenies of the past 10 years. It will be interesting to see what happens when comprehensive analyses begin to become more refined in their character construction.


Bennett SC (2013) The morphology and taxonomy of the pterosaur Cycnorhamphus. Neues Jahrbuch Fur Geologie Und Palaontologie-Abhandlungen 267: 23–41. doi: 10.1127/0077-7749/2012/0295
Kellner AWA (2003) Pterosaur phylogeny and comments on the evolutionary history of the group In: Buffetaut E, Mazin MJ, editors. Evolution and Palaeobiology of Pterosaurs: Geological Society of London, Special Publications 105–137. 
Maisch MW, Matzke AT, Sun G (2004) A new dsungaripteroid pterosaur from the Lower Cretaceous of the southern Junggar Basin, north-west China. Cretaceous Research 25: 625–634. doi: 10.1016/j.cretres.2004.06.002
Lü J, Ji Q, Wei X, Liu Y (2012) A new ctenochasmatoid pterosaur from the Early Cretaceous Yixian Formation of western Liaoning, China. Cretaceous Research 34: 26–30. doi: 10.1016/j.cretres.2011.09.010 
Unwin DM (2003) On the phylogeny and evolutionary history of pterosaurs. In: Buffetaut E, Mazin MJ, editors. Evolution and Palaeobiology of Pterosaurs: Geological Society of London, Special Publications. 139–190. 
Wellnhofer, P. (1991): The illustrated encyclopedia of pterosaurs. – 192 pp.; London (Salamander Books).
Wang X, Kellner AWA, Jiang S, Wang Q, Ma Y, Paidoula Y, Cheng X, Rodrigues T, Meng X, Zhang J, Li N, and Zhou Z (2014). Sexually Dimorphic Tridimensionally Preserved Pterosaurs and Their Eggs from China. Current Biology. Forthcoming. doi:10.1016/j.cub.2014.04.054

Wednesday, 22 October 2014

Gotta catch ‘em all – Pterosaur taxonomy

Despite the name being brand spanking new, i.e. Aerodactylus Vidovic and Martill 2014, the specimens have been known to science for over a century and a half!
Type specimen of Aerodactylus scolopaciceps
Aerodactylus is a new genus name for the species Pterodactylus scolopaciceps Meyer 1860. If you don’t know your Velociraptor from your Deinonychus and you’re worried this has gone over your head, don’t worry… P. scolopaciceps was synonymised (subsumed under the name) with Pterodactylus kochi in the 1880’s by Zittel – who famously (in pterosaur research) has a pterosaur wing with soft tissues named after him. Zittel (1883) even suggested that P. kochi might be synonymous with P. antiquus. The name P. scolopaciceps was briefly used by Broili (1938), who considered the species to be valid due to the preservation of many of the features listed by Meyer (1860) in a second specimen. However, the name fell out of use again, and by the time Wellnhofer (1970) performed his review of German pterodactyloids the name had become a mere footnote.

Much later, in 2013 Chris Bennett published a paper that followed Zittel’s suggestion and synonymised P. kochi with P. antiquus, including the two specimens referred to P. scolopaciceps. This synonymisation was based on the growing support from Mateer (1976), Jouve (2004) Bennett (1996) himself, and new evidence from principal component analyses (PCA) and Nopcsa curves. It seemed a closed case. However, PCA is not statistics or repeatable with an altered dataset, unless a population shows a normal distribution. Likewise, Nopcsa curves are open to interpretation. Although PCA can be used to support, or refute hypotheses (whilst being treated with appropriate caution) a cluster does not necessarily provide meaningful data to a taxonomist. So, a -small, unassuming- question mark hung over the synonymy. 
Referred specimens of Aerodactylus scolopaciceps

Whilst studying many Pterodactylus specimens for a much larger project, I noticed that there were a few slight differences between some specimens of P. kochi (note: this was before Bennett’s paper). So when Bennett (2013) synonymised the species with P. antiquus I set to work, constructing hypotheses and testing them. Ideally I would have used a discriminant function analysis (DFA), which is a bit like a PCA but far more appropriate for answering taxonomic problems. Unfortunately data limitations meant that DFA was not an option. Instead, I used an exhaustive bivariate technique, which used statistics to tell me if removing or adding specimens significantly improved or diminished the relationship support for a group. The results were clear, the hypothesis of P. kochi and P. antiquus being different from specimens now called Aerodactylus scolopaciceps was supported, while confusion over P. kochi’s taxonomic validity remains.
Cladogram of Pterodactyloidea from Vidovic & Martill 2014

A cladistic analysis (which deserves its own article) demonstrated that Aerodactylus scolopaciceps is more closely related to Ardeadactylus longicollum, Aurorazhdarcho micronyx, Gladocephaloideus and Cycnorhamphus than Pterodactylus. The group containing Aerodactylus was named Aurorazhdarchidae, which increases our understanding of the family level diversity in the Bavarian plattenkalks. Aurorazhdarchidae is made up of pterosaurs traditionally considered ctenochasmatoids or archaeopterodactyloids with long necks, depressed rostra (concave beaks) and long pteroid bones.

The species name for Aerodactylus scolopaciceps comes from its superficial resemblance to a Woodcock or Snipe. In the paper we suggested that it could have occupied a similar ecological niche, probing sediments, or browsing on ground level for invertebrates. Admittedly in our paper we provide little evidence for these interpretations, but we would welcome a palaeoecological study on these specimens. Considering it is one of the best represented pterosaur taxa (with soft tissue) in the Solnhofen Plattenkalk the possibilities for future research are vast.

Finally, why name the genus after Aerodactyl?
Well, as with most things in science, it was a conversation in a pub that was responsible.
Bennett SC (1996) Year-classes of pterosaurs from the Solnhofen limestone of Germany: Taxonomic and systematic implications. Journal of Vertebrate Paleontology 16: 432-444.
Bennett SC (2013) New information on body size and cranial display structures of Pterodactylus antiquus, with a revision of the genus. Palaeontologische Zeitschrift 87: 269-289.
Broili F (1938) Beobachtungen an Pterodactylus. Sitzungsberichte der Bayerischen Akademie der Wissenschaften, Mathematisch-naturwissenschaftliche Abteilung 1938: 139-154.
Jouve S (2004) Description of the skull of a Ctenochasma (Pterosauria) from the Latest Jurassic of eastern France, with a taxonomic revision of European Tithonian Pterodactyloidea. Journal of Vertebrate Paleontology 24: 542-554.
Mateer NJ (1976) A statistical study of the genus Pterodactylus. Bulletin of the Geological Institutions of the University of Uppsala 6: 97-105.
Meyer H (1860) Zur Fauna der Vorwelt: Reptilien aus dem lithographischen Schiefer des Jura in Deutschland und Frankreich. Frankfurt. 1-84.
Wellnhofer P (1970) Die Pterodactyloidea (Pterosauria) der Oberjura-Plattenkalke Suddeutschlands. Bayerische Akademie der Wissenschaften, Mathematisch-Wissenschaftlichen Klasse, Abhandlungen 141: 1-133.
Zittel KA (1883) Über Flugsaurier aus dem lithographischen Schiefer Bayerns. Palaeontographica 29: 47-80.