An unusual long-tailed pterosaur with elongated neck from western Liaoning of China

A new long-tailed pterosaur, Wukongopterus lii gen. et sp. nov, is described based on an almost complete skeleton (IVPP V15113) representing an individual with an estimated wing span of 730 mm. The specimen was discovered in strata that possibly represent the Daohugou Bed (or Daohugou Formation) at Linglongta, Jianchang, Liaoning Province, China. Wukongopterus lii is a non-pterodactyloid pterosaur diagnosed by the first two pairs of premaxillary teeth protruding beyond the dentary, elongated cervical vertebrae (convergent with Pterodactyloidea), and a strongly curved second pedal phalanx of the fifth toe. The specimen further has a broken tibia that indicates an injury occurred while the individual was still alive. Taphonomic aspects provide indirect evidence of an uropatagium, supporting the general hypothesis that at least all non-pterodactyloid pterosaurs show a membrane between the hind limbs. A phylogenetic analysis including most non-pterodactyloid pterosaurs shows that Wukongopterus lii gen. et sp. nov. lies outside the Novialoidea, being cladistically more primitive than the Rhamphorhynchidae and Campylognathoides. This analysis differs from previous studies and indicates that more work is needed before a stable picture of non-pterodactyloid pterosaur relationships is achieved.

Of the non-pterodactyloid pterosaurs associated to the Jehol Biota, two bear an elongated tail.The first is Pterorhynchus wellnhoferi, known from a complete skeleton collected from the Daohugou Bed (Czerkas and Ji 2002), whose age is controversial.Although this matter has not been settled yet, there is growing evidences that the layers of the Daohugou Bed (Wang et al. 2000) were formed between Late Jurassic to Early Cretaceous (Zhang 2002, He et al. 2004, Wang, Zhou et al. 2005) instead of Middle Jurassic as previously thought (Chen and Zhang 2004).The second long-tailed pterosaur is Changchengopterus pani that is based on a partial skeleton lacking the skull briefly described recently.According to Lü (2009) that specimen was collected in the Tiaojishan Formation that is regarded middle Jurassic in age.However after a thorough field investigation, Changchengopterus was probably collected from the same horizon as the specimen described in this paper (X.Wang, unpublished data).
Here we report another long-tailed flying reptile from a new locality (Linglongta, Jianchang County, Huludao City) of western Liaoning, China (Fig. 1).There is considerable controversy regarding those strata, with local geological map considering the fossil-bearing-bed the Lanqi Formation, which is equivalent to the Tiaojishan Formation.Based on sedimentological and field data, it is likely that the outcrop where the specimen was collected belongs to the Daohugou Bed (Formation) rather than Tiaojishan Formation (X.Wang, unpublished data).Therefore, along with Pterorhynchus wellnhoferi, the new find reported here (Wukongopterus lii gen.et sp.nov.) is potentially the youngest long-tailed nonpterodactyloid known to date, showing that those primitive pterosaurs were more diverse towards the end of the Jurassic perhaps entering in the Cretaceous.It further highlights the importance of the Jehol Biota that is starting to shape our understanding of the evolutionary history of this group of volant archosaurs.Scale bar: 50 mm.art, articular; cdv, caudal vertebrae; cri, cervical rib; cor, coracoid; cv, cervical vertebra; d, dentary; dv, dorsal vertebra; fe, femur; hu, humerus; il, ilium; is, ischium; l, left; m, maxilla; mcI-IV, metacarpal 1-4; ns, neural spine; ph1-4d4, first to fourth phalanx of manual digit IV; mtI-V, Metatarsal 1-5; ptd, pteroid; ra, radius; r, right; ri, rib; sca, scapula; sv, sacral vertebra; ti, tibia; ul, ulna; I-IV, 1-5, first through fifth pedal digits.
Diagnosis: A non-pterodactyloid pterosaur with the following combination of characters that distinguish it from other pterosaurs (autapomorphies are marked with an asterisk): first two pairs of premaxillary teeth protruding beyond the dentary*, at least 16 short peglike teeth on each side of the upper jaw (convergent with some archaeopterodactyloids), maxillary ramus of the jugal long, anteriorly projected and splint-like*,   ang, angular; art, articular; cbr, ceratobranchial; d, dentary; j, jugal; l, left; m, maxilla; nar, external naris; pm, premaxilla; pl, palatine; q, quadrate; r, right; san, surangular; spl, splenial; te, teeth; ?, unknown element.Cienc (2009) 81 (4) UNUSUAL LONG-TAILED PTEROSAUR FROM CHINA 797 quadrate inclined backwards for about 120 • *, cervical vertebrae more elongated than in any known nonpterodactyloid*, length of wing metacarpal about half the length of the first wing finger phalanx (convergent with the Pterodactyloidea), strongly curved second pedal phalanx of the fifth toe with the angle between the proximal and distal segments about 75 • *.

TAPHONOMY
The specimen (IVPP V15113) is preserved in a greydark shale and most elements are exposed in dorsal view.The skull is exposed on its right side and completely lacks the occipital and dorsal portions as well as part of the middle region, possible broken away during the collecting process.The anterior cervical vertebrae and parts of the left wing are also lacking.Bones tend to be flattened, a common condition of pterosaur material.The specimen is well articulated with almost all elements in their natural position, indicating that the carcass reached the bottom of the water column complete and remained practically undisturbed before final burial.The left wing is partially folded underneath the body and the left manus lies close to the right one.Some patches of soft tissue are preserved near the wing elements, particularly on the left side between the third and fourth wing phalanges.Although not as extremely well preserved as some other specimens from the Daohugou locality, Inner Mongolia, China (Wang et al. 2002, Kellner et al. 2009) and deposits from other countries (e.g., Wellnhofer 1991, Kellner andCampos 1999), the preserved material shows the structural fibers that are typical of the pterosaur wing (e.g., Unwin and Bakhurina 1994, Kellner 1996, Sayão and Kellner 2007).
A noteworthy feature of IVPP V15113 is the condition of the left hind limb.The left femur is articulated with the acetabulum and perpendicular to the body, almost as a mirror image of its right counterpart.The left tibia is broken below the proximal articulation, which is still articulated with the femur.The broken portion of the tibia and the foot, which also remained in anatomical position, are displaced towards the body, underneath the femur.The fracture is not a clean transverse break through the bone, which would have suggested a taphonomic origin, but rather longitudinal and therefore consistent with the rupture of fresh bone.No evidence of scavenging was observed that could have accounted for this breakage.The skeleton is rather undistorted and even small elements, such as the pedal phalanges of both feet, are preserved.Based on those observations it is likely that the breakage of this bone occurred while the animal was still alive.The fact that it is not healed as was observed in pterosaurs before (e.g., Kellner and Tomida 2000), there is a possibility that the broken tibia might be or resulted in the causa mortis of this individual.Similar conclusions on broken wing metacarpals were published by Wellnhofer (1970).
Lastly, the fact that the broken part of the left hind limb is still in close contact with the body argues for the presence of an uropatagium in Wukongopterus lii, as has been reported in a few primitive pterosaurs (Unwin andBakhurina 1994, Wang et al. 2002).

DESCRIPTION AND COMPARATIVE ANATOMY
Although not complete, the skull of Wukongopterus lii is clearly elongated, a characteristic of non-anurognathid pterosaurs (length quadrate -anterior tip of the premaxillae: 101 mm, estimated length squamosal -tip of premaxillae: 120 mm).The rostral portion anterior to the external nares (32.6 mm) is proportionally shorter than in Angustinaripterus, Rhamphorhynchus and the Pterodactyloidea (He et al. 1983, Kellner 2003, 2004, Unwin 2003a).There is no evidence of a premaxillary crest that has been reported in some pterodactyloids (e.g., Campos and Kellner 1985), but also in the primitive Austriadactylus, Raeticodactylus, Angustinaripterus, Harpactognathus, and Pterorhynchus (He et al. 1983, Dalla Vecchia et al. 2002, Czerkas and Ji 2002, Carpenter et al. 2003, Stecher 2008).The ventral margin of the skull is straight as in most pterosaurs and differs from the undulating condition observed in Harpactognathus (Carpenter et al. 2003).The premaxillae are not laterally expanded, differing from Angustinaripterus.The alveolar margin of the maxilla is thickened, with a marked parallel sulcus running for most of the extension of this bone.The maxillary ramus of the jugal is an anteriorly directed long and very thin bone, clearly indicating that the antorbital fenestra must have been quite large in this pterosaur.The bony bar that forms the ventral margin of the middle portion of the skull (composed of the jugal and maxilla) is remarkably thin for a non-pterodactyloid pterosaur.The preserved part of the maxilla shows no evidence of a bony bar separating the external naris from the antorbital fenestra, and it is possible that Wukongopterus possessed a nasoantorbital fenestra.Although only the ventral portion of the left quadrate is preserved, it shows that this bone is inclined posteriorly for about 120 • (but not to the same degree as observed in the Archaeopterodactyloidea), differing from anurognathids, Austriadactylus, Cacibupteryx, and Dimorphodon (Wellnhofer 1991, Dalla Vecchia et al. 2002, Gasparini et al. 2004, Bennett 2007).There is no anterior bony projection (formed by the premaxillae) as present in Rhamphorhynchus (Wellnhofer 1975a, b).No detailed information of the palatal or occipital regions is available.
The lower jaw (total length: 103.2 mm) is articulated to the skull.The dentary is long and lacks a ventral sagittal crest as the one reported in anhanguerids (e.g., Kellner and Tomida 2000), some tapejarids (Wellnhofer andKellner 1991, Wang andZhou 2002), and in the primitive non-pterodactyloid Raeticodactylus (Stecher 2008).The anterior tip is straight and not downturned as in Campylognathoides and Eudimorphodon (Wild 1978, Padian 2008b) and does not show the laterally compressed anterior projection of Rhamphorhynchus.The mandibular symphysis is short (estimated length: 20 mm, less than 20% of mandibular length), differing from the longer symphysis of Rhamphorhynchus, Dorygnathus and pterodactyloids (Kellner 2003).The surangular is elongated and forms part of the dorsal margin of the adductor fossa.This bone also contacts the articular but does not take part in the articulation for the skull.The long angular composes the posteroventral border of the mandibular ramus.The articular is also elongated but not to the same degree as in Caviramus and Raeticodactylus, in which this bone is ventroposteriorly projected (Fröbisch andFröbisch 2006, Stecher 2008).The borders of this element are thickened and a prominent opening is present on the dorsal surface.It is not clear if this opening leads to a pneumatic foramen as observed in some pterodactyloids (e.g., Wellnhofer 1985).
The left ceratobranchial is preserved and forms an elongated rod-like bone running over the medial surface of the mandibular ramus.
Since the middle portion of the skull and mandible are not preserved, the total number of teeth cannot be established with certainty.Anterior to the broken area there are 10 upper teeth on the right side (anterior to the external naris), and one tooth is preserved on the posterior end of the maxilla right after the breakage.In the lower jaw the ninth tooth is positioned right under the anterior margin of the external naris.Based on the spacing between the alveoli, the total tooth counts estimated for the upper jaw is 16 and for the lower one might be 12, although this last number could be higher.In any case, the tooth count of the new Chinese specimen is higher than in the Anurognathidae, Sordes, Scaphognathus, and Rhamphorhynchus (Wellnhofer 1991).
The first two premaxillary pair of teeth are positioned anterior to the tip of the lower jaw and therefore have no matching mandibular teeth.Such a projection has not been observed in non-pterodactyloids with the exception of Scaphognathus, in which the dental margin of the upper jaw is slightly deflected dorsally (Wellnhofer 1991) contrasting to the straight condition of Wukongopterus.Within more derived pterosaurs, the archaeopterodactyloid Feilongus also has the anterior part of the upper jaw projecting relative to the lower jaw (Wang, Kellner et al. 2005), but is more pronounced than in Wukongopterus.The teeth are conical and peg-like, with an oval cross-section.All the teeth are quite short compared with rhamphorhynchids.Wukongopterus lacks anteroventrally-projecting fang-like teeth present in Dorygnathus and Angustinaripterus, and the particular heterodont dentition of Dimorphodon.The new species also lacks multicuspid teeth present in several Triassic pterosaurs (Wild 1978, Jenkins et al. 2001, Dalla Vecchia 2003, 2009, Fröbisch and Fröbisch 2006, Stecher 2008) and the finely serrated carinae observed in Austriadactylus (Dalla Vecchia et al. 2002).
Six cervical vertebrae from the middle and posterior part of the neck are preserved, exposed in dorsal view.The last one shows morphological similarities with the subsequent dorsal vertebrae but is much larger.All remaining cervical vertebrae bear ribs and are elongated, more than in any other non-pterodactyloid pterosaur and similar to more derived pterosaurs (e.g., Wang et al. 2007).Their length does not reach the condition of some archaeopterodactyloids like Pterodactylus kochi or the Azhdarchidae (Howse 1986, Kellner andLangston 1996), but are elongated similar to Germanodactylus cristatus.Although mainly exposed in dorsal view, the lateral surface of some can be observed and lacks a pneumatic foramen.The neural spine is blade-like and comparatively low, differing in this respect from other non-pterodactyloids.The complete set of dorsal vertebrae is preserved comprising 12 elements, none being fused into a notarium.The neural spine is high (proportionally higher than in the cervical series) and quadrangular.There are five sacral vertebrae, four original ones showing intercostal fenestrae and a fifth element incorporated from the caudal series, totally fused with the preceding vertebra.This last sacral has the transverse processes bent up and joins the postacetabular portion of the ilium on the medioventral surface.Despite being strongly connected to the ilium, the sacral vertebrae are not fused to this bone.The caudal vertebrae series is almost complete lacking the distal part.Individual vertebrae are difficult to distinguish.Rod-like structures formed likely by the elongation of the zygapophyses (as in most other non-pterodactyloid species except Austriadactylus and Changchengopterus [Dalla Vecchia 2002, Dalla Vecchia et al. 2002, Lü 2009]) are present indicating that Wukongopterus lii had a stiffened tail.Only one free caudal is preserved, but since the tail has drifted slightly from the pelvis, more might have been originally present.
The sternum is present and well ossified but lies under the body and cannot be described in detail.The scapula (length: 34.1 mm-32.9mm) is longer than the coracoid (∼ 25.3 mm) and both elements are not fused.The scapula is elongated and does not form a platelike structure as in some Triassic pterosaurs (Dalla Vecchia 2003, 2009).This bone is longer relatively to the coracoid compared to Changchengopterus pani.The coracoid shows a well developed biceps tubercle but lacks a deep coracoidal flange as reported for Changchengopterus (Lü 2009).Although not well preserved, the deltopectoral crest of the humerus (length: 38.7 mm) is positioned proximally and does not extend further down the shaft as in Raeticodactylus, Campylognathoides, and Eudimorphodon ranzii (Wild 1978, Padian 2008b, Stecher 2008).
The radius and ulna are elongated, with the diameter of the radius being sub equal to the one of the ulna (length: 62.1 mm).The carpus is best observed on the right side, the carpals are not fused, with both the proximal and distal series presenting two elements.This number could be higher and more carpal elements have been reported in pterodactyloid pterosaurs (Kellner and Tomida 2000).The pteroid (length: 7.3 mm) is very small and is attached to the proximal carpal series.The ratio of the wing metacarpal (length: 22.9 mm) and some bones (e.g., wing finger phalanges, humerus) suggests that this element is slightly larger compared to other non-pterodactyloid pterosaurs, but does not approach the extreme elongated pterodactyloid condition.Manual unguals are deeper and more curved than pedal unguals.The first wing finger phalanx is the smallest (length: 45.7 mm) followed by the fourth (51.4 mm), second (56.8 mm) and third (58.3-59.3mm) ones, respectively.The ratio of the first wing finger phalanx and the tibia is similar to "Eudimorphodon" cromptonellus (see Jenkins et al. 2001) and are the smallest values within non-pterodactyloid pterosaurs.
The femur (length: 33.3-35.6 mm) is shorter than the tibia (length: 52.8 mm) and has a large head.Both pedes are well preserved with metatarsal III (16.8-17.2mm) being larger than metatarsal IV (14.5-15.0).The phalangeal formula (2.3.4.5.2) is typical for non-pterodactyloid pterosaurs that have two elongated phalanges on pedal digit V.The first phalanx of pedal digit V is longer than in some non-pterodactyloids (e.g., Rhamphorhynchus).The last phalanx of pedal digit V differs from all pterosaurs in being more curved with the distal and proximal portion forming an angle of about 75 • .This "boomerang-shaped" last phalanx of pedal digit V is also observed in Sordes, Dorygnathus and Scaphognathus, but all show a larger angle (135 • ) between the distal and proximal parts (Wellnhofer 1991, Unwin andBakhurina 1994).The pedal unguals are long, curved with a deep lateral sulcus, and have a broad, flattened ventral surface.

DISCUSSION AND CONCLUSIONS
The long tail, short wing metacarpals and long pedal digit V clearly indicate that Wukongopterus lii is not a member of the Pterodactyloidea.In order to assess the phylogenetic position of the new species, we performed a cladistic analysis using the original data set An Acad Bras Cienc (2009) 81 (4) of Kellner (2003), which is being continuously updated by the inclusion of new characters and taxa, as well as by the review of character states (Kellner 2004, Wang, Kellner, 2005, 2008, see Appendix).The present analysis shows that Wukongopterus lii is a primitive non-pterodactyloid placed outside the Novialoidea, but more derived relative to the Anurognathidae, Austriadactylus, Sordes, Preondactylus, and Scaphognathus (Fig. 4).Wukongopterus lii lacks the synapomorphic features of the Rhamphorhynchinae and also the ones which unite the two species of Campylognathoides.These taxa are nested within the Novialoidea.
Although the phylogeny of non-pterodactyloid taxa is beyond the scope of this paper, it is interesting to note that the results that were achieved in the present study differ from previous analyses (e.g., Kellner 2003, 2004, Unwin 2003a, b, Wang, Kellner et al. 2008, Dalla Vecchia 2009).Among the most striking differences is the paraphyly of the Campylognathoididae, having Eudimorphodon ranzii forming a monophyletic group with other Triassic pterosaurs instead of Campylognathoides as has been previously advocated (e.g., Wild 1978, Kellner 2003, Unwin 2003a, b).The particular position of Peteinosaurus as sister-taxon of "Eudimorphodon" cromptonellus further supports the suspicion that Eudimorphodon as presently defined is paraphyletic and there are doubts if all specimens attributed to this taxon do represent the same species or even the same genus (e.g., Kellner 2003, Dalla Vecchia 2003, 2009).Caution is therefore needed before compiling all Eudimorphodon specimens in one single terminal taxon as has been done before (Unwin 2003b).
In addition, the search conducted by PAUP (Swofford 2000) with the inclusion of Harpactognathus, Angustinaripterus, and Cacibupteryx, known from cranial material only, and Changchengopterus, known from a partial skeleton (lacking cranial elements), resulted in thousands of most parsimonious cladograms.Part of the explanation for this can be attributed to the incompleteness of several species.Due to the fragility of their bones and based on the fact that they are flying animals, pterosaur remains are generally difficult to preserve and are mostly fragmentary (e.g., Dalla Vecchia et al. 2001, Fröbisch and Fröbisch 2006, Calvo et al. 2007, Martill et al. 2008, Costa and Kellner 2009).Even deposits where exceptionally specimens are occa-sionally found like the Romualdo Formation (e.g., Fara et al. 2005) have yielded incomplete and fragmentary material (e.g., Wellnhofer 1985).Regarding nonpterodactyloid pterosaurs, with the exception of Campylognathoides, Dorygnathus and Rhamphorhynchus (Wellnhofer 1975a, b, Bennett 1995, Padian 2008a, b), all other are based on incomplete or badly preserved specimens (e.g., Preondactylus).Even more complete taxa such as Austriadactylus and Raeticodactylus lack important post-cranial bones that have yielded diagnostic characters of several non-pterodactyloid clades (e.g., Kellner 2003, Unwin 2003b).When only Changhengopterus pani is included in the phylogenetic analysis, the consensus cladogram differs even more from all published pterosaur phylogenies so far by having Austriadactylus as the most basal pterosaur, the Anurognathidae as the sister group of the Pterodactyloidea, and all remaining non-pterodactyloid united in a monophyletic group (see Appendix).Since we were not able to examine the holotype and only known specimen of Changhengopterus pani first hand, this result must be seen with caution.
The present study shows the lack of stability regarding the relationships of non-pterodactyloid taxa and several changes are expected in the future when additional and more complete non-pterodactyloid material comes to light.Until a more stable picture of the relationships of non-pterodactyloids is achieved, we refrain from naming nodes.
The only non-pterodactyloid pterosaurs from China are the anurognathids Jeholopterus (from the Daohugou Bed) and Dendrorhynchoides (from the Jianshangou Bed of the Yixian Formation), and three putative rhamphorhynchids: Angustinaripterus from the Middle Jurassic Xiashaximiao Formation, Changchengopterus reported from the Tiaojishan Formation (although we suspect that it comes also from the Daohugou Bed) and Pterorhynchus (from the Daohugou Bed).The anurognathid Dendrorhynchoides was first thought to possess an elongated tail (Ji andJi 1998, Ji et al. 1999), which was convincingly dismissed by Unwin et al. (2000).Angustinaripterus longicephalus is only known by a skull and lower jaw but, despite its uncertain phylogenetic position, is not a pterodactyloid (e.g., separated external nares and antorbital fenestra).The dentition and the presence of a premaxillary sagittal crest are some fea- tures that distinguish Angustinaripterus from Wukongopterus.
The non-pterodactyloid Changchengopterus pani is known from a partial skeleton to which we had no access that was only briefly described by Lü (2009).Based on the original publication, Wukongopterus differs from Changchengopterus by several features, including the presence of elongated pre-and postzygaphophyses, a comparatively larger ulna, and the strongly curved second phalanx of pedal digit V.In the data matrix, Lü (2009) also points out that Changchengopterus bears short cervical vertebrae, which contrasts to the elongated condition of those bones in Wukongopterus.Changchengopterus also appears to have a proportionally smaller tibia, which can be regarded as a potential diagnostic feature of this taxon.Another difference is found in the proportion of the wing finger elements, with Wukongopterus having the first wing finger phalanx the shortest of all (including the fourth), while in Changchengopterus this bone has the same size of the third wing finger phalanx (Lü 2009).
The only long-tailed pterosaur that comes from the same deposit of Wukongopterus is Pterorhynchus wellnhoferi, which is based on a nearly complete specimen from the Daohugou Bed that was still unprepared when described (Czerkas and Ji 2002).Unfortunately, we also did not have any access to the specimen (as is apparently the case for other researchers, C. Bennett, pers. comm. 2009).Based on the published illustrations, Wukongopterus differs from Pterorhynchus by features such as the lack of a premaxillary sagittal crest and the larger number of teeth.Pterorhynchus was classified in the Rhamphorhynchidae (Czerkas and Ji 2002) but is unlikely a member of this clade since it lacks cranial rhamphorhynchid synapomorphic features (e.g., Kellner 2004).It would be interesting to make a detailed comparison between Wukongopterus and Pterorhynchus, particularly regarding postcranial elements, in order to establish if there is a close relationship among those taxa.
Among the interesting features of Wukongopterus are the several traits of the skull that are similar to pterodactyloids.Those include the thin ventral margin of the skull, suggesting the presence of a large antorbital fenestra, and the inclination of the quadrate.Also some postcranial elements, mainly the elongated cervical ver-tebrae that have not been reported in any non-pterodactyloid before, are a derived trait of this taxon.The different sizes of the cervical vertebrae in pterosaurs, now also in the non-pterodactyloids, is a quite interesting subject for further research that might try to understand mechanical consequences of such arrangement.Nonetheless the remaining skeleton clearly shows primitive non-pterodactyloid characteristics such as the elongated tail and the developed fifth pedal digit.
Non-pterodactyloid pterosaurs also show considerable variation in the shape of the second (and last) pedal phalanx that can be short (Campylognathoides), straight (Jeholopterus, Dimorphodon), slightly curved (Rhamphorhynchus), and strongly curved ("boomerang" shaped) (Sordes, Scaphognathus, Dorygnathus).Although falling into the last category, the angle between the proximal and distal portion of the last phalanx of pedal digit V in Wukongopterus lii is strongly curved, with the proximal and distal segment at an angle of lesser than 90 • .In Sordes the last phalanx of pedal digit V has been demonstrated to be connected with the uropatagium (Unwin and Bakhurina 1994) and it is generally accepted that this was also the case of other pterosaurs (e.g., Wellnhofer 1991).Therefore, the variation of morphology and size of the phalanx of pedal digit V might indicate a variation in the shape of the uropatagium in non-pterodactyloid pterosaurs, a hypothesis that might be explored in the future with more findings.
To conclude, most researchers agree that the primitive long-tailed pterosaurs went extinct by the end of the Jurassic (Wellnhofer 1991), some even using them for dating deposits (e.g., Lü 2009).Despite the disputed age of the Daohugou Bed, that might extend into the lower Cretaceous, Wukongopterus indicates that longtailed pterosaurs were more diverse towards the end of the Jurassic than previously thought.This discovery further enhances the importance of the ancient ecosystems of the Jehol Group for the understanding of pterosaur evolutionary history.genes de Almeida Campos, Chris Bennett and Mark Witton are acknowledged for comments on the earlier draft of this ms.This study was supported by the National Science Fund for Distinguished Young Scholars (40825005), National Natural Science Foundation of China ( 40121202 Palavras-chave: Pterosauria, Wukongopterus, ?Jurássico Superior-Cretáceo Inferior, Liaoning, China.

APPENDIX
In order to access the phylogenetic position Wukongopterus lii gen.et sp.nov., we performed a phylogenetic analysis using PAUP 4.0b10 for Microsoft Windows (Swofford 2000).The large dataset (3 outgroups + 57 pterosaur taxa, 89 characters, several multistate) dictated the heuristic search option.Characters were given equal weight and treated unordered (ACCTRAN setting).The search conducted by PAUP including all three outgroups (Ornithosuchus longidens, Herrerasaurus ischigualastensis and Scleromochlus taylori) and all 57 pterosaur taxa produced 140.698 of equally parsimonious trees with a length of 233 steps (consistency index = 0.6781, retention index = 0.8806, rescaled consistency index = 0.5971).A second search excluding the outgroup Scleromochlus and the pterosaur taxa Changchengopterus pani, Harpactognathus gentryii, Angustinaripterus longicephalus and Cacibupteryx caribensis, PAUP found 15893 equally parsimonious trees of 225 steps (consistency index = 0.6933, retention index = 0.8831, rescaled consistency index = 0.6122).A strict consensus cladogram of this last search is shown in Figure 4. See the main text for more details.15.Premaxillary sagittal crest, position: 0 -absent; 1 -confined to the anterior portion of the skull; 2 -starting anterior to the anterior margin of the nasoantorbital fenestra, not reaching the skull roof above the orbit; 3 -starting anterior to the anterior margin of the nasoantorbital fenestra, extending beyond occipital region; 4 -starting at about the anterior margin of the nasoantorbital fenestra, reaching the skull roof above the orbit but not over the occipital region; 5 -starting close or at the anterior portion of the skull and extended over the occipital region; 6 -starting close or at the anterior portion of the skull, reaching orbit but not extended over the occipital region; 7 -starting at the posterior half of the nasoantorbital fenestra Germanodactylus rhamphastinus (Wagner, 1851) 0010100-1-0000400021 00000121?0??????0?00 0000000120 00???11??0 ?001???211 ??7?021??? 31???2???