Cosesaurus aviceps, Sharovipteryx mirabilis and Longisquama insignis Reinterpreted

Abstract

Originally published in 2000 (Rivista Italiana...) these taxa and 1300 others were subsequently added to a large phylogenetic analysis permitting a better understanding of the systematics of these taxa (they are lepidosaurs now, not prolacertiforms) and better understand their transitional anatomy. In essence, earlier errors are corrected here. Even so, these three taxa remain ancestral to pterosaurs after testing all other candidates.

Full text

1
Cosesaurus aviceps, Sharovipteryx mirabilis and Longisquama insignis
Reinterpreted
Or
Additional pterosaurian traits recovered from
Cosesaurus aviceps, Sharovipteryx mirabilis and Longisquama insignis
DAVID PETERS
Independent Researcher
12c San Miguel Drive, Saint Charles, Missouri, 63303, U.S.A.
davidpeters@att.net
314-323-7776

2
Key Words
Pterosauria, Fenestrasauria, Triassic, Madygen Formation, Upper Muschelkalk, Origin of
Active Flapping Flight
Abstract
Currently the majority of pterosaur and archosaur workers maintain the traditional
paradigms that pterosaurs appeared suddenly in the fossil record without obvious
antecedent and that pterosaurs were most closely related to archosaurs because they
shared an antorbital fenestra and a simple hinge ankle. Oddly, these hypotheses continue
despite the widely accepted acknowledgement that no archosauriformes document a
gradual accumulation of pterosaurian traits. The minority view provided four
phylogenetic analyses that documented a gradual accumulation of pterosaurian traits in
three fenestrasaurs, Cosesaurus aviceps, Sharovipteryx mirabilis, and Longisquama
insignis and their ancestors. These three also had an antorbital fenestra and a simple
hinge ankle by convergence. Unfortunately the minority view descriptions also included
several misinterpretations. Those are corrected here. The revised descriptions add further
support to the nesting of pterosaurs with fenestrasaurs, a clade that now nests within a
new clade of lepidosaurs between Sphenodontia and Squamata. The new data sheds light
on the genesis of active flapping fight in the nonvolant ancestors of pterosaurs.
Introduction
The search for pterosaur precursor taxa continues with no consensus in sight. Hone and
Benton (2007, 2008) summed up the majority view when they reported the Pterosauria

3
appeared suddenly in the fossil record without obvious antecedent. Representing the
minority view, Peters (2000a) redescribed four pterosaur precursor candidates,
Langobardisaurus (Renesto 1994, MCSNB 4860), Cosesaurus (Ellenberger & de Villalta
1974, MGB-V1), Sharovipteryx (Sharov 1971, PIN 2584/8) and Longisquama (Sharov
1970, PIN 2584/4, Figs. 1-12) and added these taxa to four prior phylogenetic analyses
(Evans 1988, Jalil 1993, Bennett 1996 and Benton 1999). In each analysis pterosaurs
nested with these four taxa rather than any included archosaur. Peters (2000a, 2002)
showed that these taxa document a gradual accumulation of pterosaurian traits and
erected the clade Fenestrasauria for Cosesaurus, Sharovipteryx, Longisquama and
pterosaurs based on the synapomorphy of an antorbital fenestra without a fossa. Later
workers (e.g. Brusatte et al. 2010, Nesbitt 2011) ignored these descriptions, analyses and
taxa. Nesbitt (2011) cited the arguments of Hone and Benton (2007) for this exclusion
(but see below). Brusatte et al. (2010: 10) preferred to, “follow the majority view,” and so
did not include the above taxa in their archosaur study, nor did they delete their
pterosaurs. Both studies nested pterosaurs within the Archosauriformes arising from the
Parasuchia and Proterochampsidae. In recent phylogenetic analyses of the Pterosauria
(Unwin 2003, Kellner 2003, Andres 2010 and all analyses derived from them) workers
likewise ignored Peters (2000a). Instead they included only archosauriforms as outgroup
taxa, treating them essentially as ‘all zero’ outgroups because none document a gradual
accumulation of pterosaurian traits. Notably, pterosaur workers did not employ taxa from
the Parasuchia or the Proterochampsidae as outgroups.
Prior Pterosaur Origin Phylogenetic Analyses and Hypotheses

4
Bennett (1996) nested pterosaurs with the small bipedal archosaur, Scleromochlus
(Woodward 1907, Benton 1999). The suprageneric taxa Ornithosuchidae, Parasuchia and
Suchia nested as successively more distant taxa. The nesting of bipedal Scleromochlus,
with its tiny hands and vestigial lateral manual and pedal digits is at odds with a later
work by Bennett (2008) in which he imagined a quadrupedal, arboreal, lizard-like proto-
pterosaur with elongate lateral manual and pedal digits similar to another hypothetical
model proposed by Wild (1978). Bennett (2008) imagined the first step toward the
pterosaurian-grade was the supination of the forelimb, followed by the loss of the digit 4
ungual and the hyperextension digit 4 to fold against the dorsal (now supinated posterior)
forelimb. Bennett (2008) did not explain how metacarpal 3 became detached from
metacarpal 4 so the three medial metacarpals could migrate together across the palmar
(now anterior) surface of the supinated manus where Bennett (2008) locates them in his
reconstruction. Bennett (2008) gave no explanation for the appearance or evolution of
other pterosaurian traits, such as the sternal complex, prepubis, attenuated tail and
metapodial proximal phalanx of pedal digit five.
In a follow-up to his 1996 paper, Bennett (2012) declined to include pterosaur
sister taxa recovered by Peters (2000a, see below). Instead Bennett reviewed his earlier
taxon inclusion set, restricted to members of the Archosauromorpha. Peters (2000a) was
mentioned only in regard to issues Bennett found with the Hone and Benton (2007)
review of Peters (2000a), discussed below.
Benton (1999) nested pterosaurs between Scleromochlus and Lagerpeton and all
three taxa were derived from Euparkeria, Proterochampsidae and Phytosauria (=
Parasuchia) in order of decreasing distance. In counterpoint, Peters (2002) listed traits

5
Scleromochlus shared with basal bipedal crocodylomorphs, none of which were included
in Benton (1999).
As mentioned earlier, Peters (2000a) added Langobardisaurus, Cosesaurus,
Sharovipteryx and Longisquama to prior analyses by Evans (1988), Jalil (1993), Bennett
(1996) and Benton (1999) and recovered these four as proximal sisters to pterosaurs.
Subsequent workers (Rieppel, Fraser & Nosotti 2003; Chatterjee & Templin 2004; Hone
& Benton 2007) labeled this study as “heterodox.” They did not check the specimens. In
this case, “heterodox” simply meant novel or different than orthodox.
Peters (2000a, 2002) showed that nearly every pterosaurian trait was already
present in basal fenestrasaurs, except the elongation of the wing elements and the axial
rotation of metacarpal 4. In those studies, basal pterosaurs shared with Langobardisaurus,
Cosesaurus, Sharovipteryx and Longisquama: (1) posterior teeth differentiated from
anterior teeth; (2) occiput at an obtuse angle to the jaw line; (3) pterygoids contact
vomers; (4) palatines reduced; (5) cervical vertebrae descend from back of skull in simple
curve (parallelogram-shaped cervicals); (6) procoelous presacral vertebrae; (7) posterior
dorsal ribs straight and fused to transverse processes; (8) chevrons short, grading to long
and parallel with each centrum distally, altogether producing an attenuated tail; (9)
scapula low, grading to parsagittally elongate; (10) radius and ulna straight and closely
appressed; (11) tibia and fibula straight and closely appressed; (12) fibula grading to
splint-like; (13) distal tarsal 1 absent; (14) metatarsal 5 short; (15) pedal 5.1 metapodial.
In Peters (2000a, 2002), basal pterosaurs shared with Cosesaurus, Sharovipteryx
and Longisquama (the Fenestrasauria): (1) antorbital fenestra; (2) anteriorly expanded
lacrimal overhangs antorbital fenestra; (3) premaxilla elongate; (4) quadratojugal spur

6
appears; (5) dorsal vertebrae transverse processes enlarged; (6) at least four sacral
vertebrae; (7) caudal ribs (transverse processes) reduced to seven; (8) clavicles overlap
medially and do not extend dorsally; (9) interclavicle with anterior process; (10)
anteriorly ilium elongated.
In Peters (2000a, 2002), basal pterosaurs shared with Sharovipteryx: (1) anterior
[maxillary] teeth enlarged; (2) reduced torso equal to or shorter than skull plus cervicals;
(3) mid and distal caudal vertebrae five times longer than tall; (4) humerus with
crescentic deltopectoral crest; (5) manual digit 4 longer than humerus plus ulna; (6) tibia
longer than femur; (7) dermal membranes (uropatagia) posterior to hind limbs.
In Peters (2000a, 2002), basal pterosaurs shared with Longisquama: (1) broad
sternal complex formed of fused clavicles, interclavicle and sternum (Wild 1993). From
the Sharovipteryx list, Longisquama also had 1) anterior teeth enlarged; 4) humerus with
crescentic deltopectoral crest.
Several other synapomorphies listed in Peters (2000a, 2002) are not listed here
because they were misinterpreted. Despite these problems, Peters (2000a, 2002)
presented the first, and so far only, series of taxa that document a gradual accumulation
of pterosaurian traits. That list has been expanded here (Fig. 12). See below.
In his PhD dissertation, Senter (2003) nested Cosesaurus with Langobardisaurus,
Sharovipteryx and Macrocnemus as a sister clade to Prolacerta and Protorosaurus.
Pterosaurs nested with Scleromochlus at the base of the Dinosauriformes derived from
Euparkeria and Proterosuchus. Longisquama nested with the rib-glider Coelurosauravus.
There are many scoring problems in Senter’s analysis. As an example, Senter (2003,
2004) scored Scleromochlus with an elongated pedal 5.1 and a large sternum, but both are

7
absent. Senter (2003, 2004) followed Sharov (1970) in reporting dorsal tubercles on the
skull of Longisquama and homologized them with the squamosal tubercles present in
Coelurosauravus. In Longisquama, those “tubercles” are actually fused parietals divided
medioposteriorly (Fig. 11). Senter (2003, 2004) considered the antorbital fenestra in
Longisquama the result of damage. He denied the presence of antorbital fenestra in
Cosesaurus, but traced it (Senter 2003, fig. 7).
In a two-part study Hone and Benton (2007, 2008) claimed they were going to
test the “prolacertiform” origin of pterosaurs (as recovered in Peters 2000a) against the
“archosaur” origin of pterosaurs (as recovered in Bennett 1996) by creating a supertree
with a supermatrix of prior published work. That method does not require examining
specimens either firsthand or with photographs. Despite being represented by complete
and articulated fossils, Cosesaurus and Langobardisaurus were scored for less than a
third of their character traits. Longisquama and Sharovipteryx were deleted from the
inclusion set. In contrast to their stated objectives, the matrices presented by Peters
(2000a) were deleted due to Hone and Benton’s (2007) inability to replicate Peters’
reanalysis of a modified Bennett (1996) dataset. Later Bennett (2012) reported their
inability was the result of seven coding errors in retyping Peters’ published data matrix,
along with other errors. In the second part of their two-part study, Hone and Benton
(2008) omitted all references to Peters (2000a) and credited the competing
“prolacertiform” hypothesis to Bennett (1996). Hone and Benton’s (2007, 2008)
conclusions were self-described as ‘inconclusive’ as they were unable to recover any taxa
that shared unambiguous pterosaurian traits. Their supermatrix also: 1) failed to nest four
generic choristoderes with their taxon Choristodera; 2) failed to nest their taxon

8
Squamata with or within the Lepidosauromorpha; 3) failed to nest Prolacerta within the
Prolacertiformes; 4) and nested the Pterosauria between the aquatic Proterosuchidae and
the lumbering Erythrosuchidae.
Peters (2009) identified a pteroid and preaxial carpal in Cosesaurus along with
fibers trailing the forelimbs, as in pterosaurs. Earlier Ellenberger (1993) illustrated these
bones and soft tissues, but did not associate them with pterosaur anatomy.
Earlier Descriptions of the Three Basal Fenestrasaurs
All earlier workers (see below) had difficulties describing the three basal
fenestrasaurs, Cosesaurus, Sharovipteryx and Longisquama. These difficulties can be
attributed to initial bias, the crushed nature of the fossils themselves and lack of
familiarity with sister taxa.
Cosesaurus aviceps
Ellenberger and de Villalta (1974, Figs. 1-4) described Cosesaurus aviceps as a
small (14 cm snout to tail tip), Middle Triassic (Ladinian) reptile represented by the
holotype and only known specimen, MGB-V1. Housed in the Museu de Geologia in
Barcelona, the hand-sized plate contains the natural mold of the complete articulated
skeleton and soft tissues in ventrolateral view together with an adhering medusa.
Cosesaurus was found in the Upper Muschelkalk of Alcover, Spain. That area is known
for marine fossils with soft tissue preservation (Ellenberger 1978, 1993).
Prior interpretations of elements within the pectoral and pelvic girdles of
Cosesaurus (Ellenberger and de Villalta 1974; Ellenberger 1978, 1993; Peters, 2000a)

9
were dissimilar to traits found in and shared by sister taxa recovered from prior analyses
(Peters 2000a) and the present analysis (Fig. 12). Cosesaurus was reported to have a
scapula acromion process (Ellenberger 1993; Fig. 2), but no candidate sister taxa have an
acromion process. Cosesaurus was reported to have a deep interclavicle keel and
enlarged, conjoined coracoids (Ellenberger 1993; Peters, 2000a, 2002), but no sister taxa
have these. Cosesaurus was reported to lack a sternum (Ellenberger 1993; Peters 2000a,
2002), but all candidate sister taxa have a sternum. Cosesaurus appeared to have a stem-
like process on its blade-like anterior ilium (Ellenberger 1993; Peters, 2000a, 2002), but
no related forms share this trait.
Ellenberger (1993) reconstructed Cosesaurus as a digitigrade biped with
lepidosaur and avian traits. He flipped the wrist and manus in order to make it appear that
digit 2 was the longest, as in birds, rather than digit 4. That brought the two preaxial
carpals to the postaxial side. A large, disarticulated sacral transverse process was
considered an ischium and a gastralium was considered a retroverted pubis with an
expanded articular ‘prepubis’ portion comprised largely of the anterior process of the
ilium. Feather impressions were reported emanating from the tail. These avian
interpretations were later dismissed (Milner 1985, Peters 2000a) as phylogenetic
evidence mounted that birds were derived from small theropod dinosaurs (e.g. Gauthier
1986).
Sanz and López-Martinez (1984) crudely reconstructed Cosesaurus as a juvenile
Macrocnemus, ignoring many traits (Fig. 1) not found in Macrocnemus (Peters 2000a).
Using a supplied photograph of Cosesaurus, Senter (2003) was able to trace only a few
crude outlines. He misidentified the coracoids as crescentic clavicles and denied the

10
presence of an antorbital fenestra, but his illustration Senter 2003, fig. 7) shows that he
traced it.
Initial bias has played a part in prior interpretations of Cosesaurus. Ellenberger’s
(1978, 1993) views were influenced by his preconception of bird homologies. Sanz and
López-Martinez (1984) clearly saw Cosesaurus as a juvenile Macrocnemus. Peters
(2000a, b) recognized that an elongate pedal 5.1 (fifth digit, first phalanx) in Cosesaurus
could be homologous with an elongate pedal 5.1 in pterosaurs, Sharovipteryx,
Langobardisaurus and Tanystropheus (Bassani 1886, Peyer 1931, Wild 1973). Peters
(2000a) agreed with Ellenberger (1978, 1993) on the presence in Cosesaurus of an
antorbital fenestra, a posteriorly displaced naris, a spike-like quadratojugal, procoelus
vertebrae, an elevated cervical series, four sacral vertebrae, an attenuated caudal series
with chevrons parallel to each centrum, a blade-like anteriorly expanded ilium, a slender
and appressed fibula, a simple hinge ankle joint, an elongate pedal 5.1, uropatagia trailing
the hind limbs and fibers trailing the forelimbs. These traits are shared with pterosaurs,
Longisquama and Sharovipteryx.
Sharovipteryx mirabilis
Originally Podopteryx [preoccupied] (Sharov 1971, Figs. 1, 5-9), Sharovipteryx is
a small (~10 cm snout-vent, ~21cm snout-tail tip length), long-legged reptile represented
by the only known specimen, PIN 2584/8 (plate and counterplate housed in the
Paleontological Institute, Russian Academy of Sciences, Moscow, Russia). Sharovipteryx
was found in the Madygen Formation (Ladinian–Carnian, Mid-Late Triassic),
Dzailauchou, on the southwest edge of the Fergana valley in Kyrgyzstan. This area is

11
known for its fossil insects (Shcherbakov 2008) and soft tissue preservation in reptiles.
While many elements are readily observable, magnified viewing of Sharovipteryx reveals
additional details. Scales, structures similar to ‘hairs’ found in pterosaurs (Wang et al.
2002), later called pycnofibers (Kellner et al. 2010), and extradermal membranes obscure
certain features and bone sutures. The middle area of the Sharovipteryx counterplate is
missing, replaced with filler.
Sharov (1971, figs. 3-5) traced Sharovipteryx and reconstructed it as a glider. He
also provided closer illustrations of the skull + neck and the torso + hind limb. Only the
hyoids were identified in the skull + neck figure. Sharov mistakenly considered the fossil
exposed in ventral view (Gans et al. 1987). He mistakenly identified the pubis as a
retroverted posterior pelvic element. Here (Fig. 5) that area is identified as the proximal
femur, displaced medially. Sharov (1971) tentatively, but correctly identified an
elongated finger extending to the pelvic area, a tiny antebrachium (slightly longer than
wide), carpals and an unidentified short curved line oriented medially that corresponds
here to an ungual (Fig. 5). Disregarding portions of his own tracings, Sharov (1971)
reconstructed Sharovipteryx along the lines of Scleromochlus with long, slender
forelimbs and small hands with short fingers extending just beyond the outstretched
knees. Sharov (1971) correctly traced small, convex, stiff, membranes anterior to the
middle of each femur. However, his reconstruction transformed those into elastic
membranes spanning the gap between the humerus and femur. Sharov (1971) identified
pedal digit 1 as a boneless curve tipped by an ungual, here (Fig. 5) identified as soft
tissue. Based on the presence of an elongate anterior ilium, Sharov considered

12
Sharovipteryx a ‘pseudosuchian’ (= advanced archosauriform) with possible affiliations
to pterosaurs and/or Scleromochlus.
Halstead (1975) proposed that Sharovipteryx was a direct ancestor to pterosaurs
basing his opinion on the presence in both of wing membranes. Confirming Sharov’s
(1971) figures, Gans et al. (1987) and Tatarinov (1989) reported the forelimbs were only
one-third to one-quarter the length of the hind limbs (Fig. 5). Gans et al. (1987) and Dyke
et al. (2006) both imagined the anterior membranes needed to stabilize Sharovipteryx in a
gliding configuration without identifying them on the fossil. The reconstruction by Dyke
et al. (2006) again attached the prefemoral membranes to the humerus, but Gans et al.
(1987) did not. Gans et al. considered Sharovipteryx either a lepidosaur or a protorosaur
with subthecodont teeth. Tatarinov (1989) considered Sharovipteryx a protorosaur (=
prolacertiform).
Contra the present and all previously published studies, Unwin et al. (2000)
reported the forelimbs ‘have yet to be clearly identified.’ They considered structures
previously identified as forelimb elements as ‘almost certainly remains of ribs’ and
supposed the forelimbs are still buried in the matrix. Unwin et al. (2000) agreed that
Sharovipteryx was likely a prolacertiform based on the elongate cervical vertebrae and
low neural spines, but also included a possible third trait, an incomplete lower temporal
bar described earlier by Tatarinov (1994). Unwin et al. (2000) were unable to identify the
tarsals or count the pedal phalanges, perhaps due to the displacement of the distal
phalanges (Fig. 8). They mistakenly reported the presence of elongate penultimate
phalanges on all digits. They reported the digits increased in length from one to five with
‘intermediate’ phalanges reduced, but those observations are not confirmed here. Unwin

13
et al. reported that digits 4 and 5 are subequal, as they appear in situ (Fig. 8), but a
reconstruction shows that digit 4 extends far beyond digit 5 in vivo. Digit 5 is much
longer than digit 4 when the disparate metatarsals are not included.
The skeleton of Sharovipteryx is preserved in a plate and a counterplate that
often separates bones not at their surfaces, but through their shattered midsections. Peters
(2000a) misinterpreted the posteriorly split cranial bones for pterygoids, assuming that
the top of the skull was on the counterplate, exposing the palatal bones in dorsal view. No
sister taxa had such robust palatal bones.
Longisquama insignis
Represented by the only known skeleton, PIN 2584/4 and its counterplate PIN
2584/5 (housed in the Paleontological Institute, Russian Academy of Sciences, Moscow,
Russia), Longisquama insignis (Sharov 1970, Figs. 1, 10-11) was a small (13 cm snout-
vent, 32 cm snout-tail tip length), reptile notable for its long dorsal plumes. A
contemporary of Sharovipteryx, Longisquama was also found in the Madygen Formation.
Sharov (1970) provided simple tracings of Longisquama (Fig. 10d) that included
the skull, cervicals, anterior dorsals, pectoral elements and forelimbs. Sharov traced soft
tissue in the crook of the elbow, trailing the humerus and ulna, filling the throat area and
a set of seven radiating plumes unlike those of any other reptile. The two “tubercle-like
structures” Sharov traced at the back of the skull (Figs. 10, 11) are actually the fused
parietals embayed posteromedially and rotated to a dorsal view. Sharov illustrated an
elongate scapula and a manual digit 4 equivalent to the length of the humerus. Here (Fig.
10) that ‘digit’ is composed of jumbled phalanges from both hands, some aligned as

14
Sharov drew them. Sharov (1970) suggested the elongate plumes represented an early
stage in the evolution of avian feathers, and that the clavicles represented an early avian
furcula. He considered Longisquama a “pseudosuchian” on the basis of an antorbital
fenestra (which is present) and a mandibular fenestra (which is not).
Agreeing with Sharov (1970), Haubold and Buffetaut (1987) reconstructed
Longisquama with paired plumes oriented laterally for gliding or parachuting. However,
when digital photographs of the plate and counterplate are aligned the left and right plates
do not reveal paired plumes, only a single row. So it is difficult to understand how the
concept of paired plumes arose other than to make them into parachutes. Jones et al.
(2000) agreed with the paired plumes interpretation, that the plumes were proto-feathers
and the clavicles were an avian furcula. Reisz & Sues (2000) and Prum et al. (2001)
disagreed with both.
Unwin et al. (2000) also agreed that the plumes were aligned as pairs dorsally.
Unwin et al. considered the skull holes of Longisquama the possible result of damage to
the fossil. Instead of a furcula, Unwin et al. identified the bones as paired clavicles along
with an interclavicle. They agreed with Sharov (1970) that Longisquama had acrodont
teeth. These traits, they noted, were more typical of lepidosaurs. Unwin et al. observed
only seven cervicals. All sister taxa have eight. One cervical, which may have been
missed, is visible below the skull (Fig. 10c). Unwin et al. (2000) observed that only the
anteriormost dorsal vertebrae were preserved. Sharov (1970) and all subsequent authors
(including Peters 2000a) agreed, but closer examination reveals the entire vertebral
column (Fig. 10). Unwin et al. (2000) reported a lack of cervical ribs, but they, too, are
visible. Unwin et al. reported a long, narrow scapula, a rod-like coracoid and crescent-

15
shaped clavicles articulating with a well-developed interclavicle, but did not associate
this suite of traits with pterosaurs. It is difficult to see how Unwin et al. were able to
report elongate manual penultimate phalanges without reconstructing the manus, because
the phalanges of one manus are on top of the other, scattered about (Fig. 10f, g). None of
the digits have elongated penultimate phalanges in the present reconstructions.
Martin (2004) agreed that Longisquama had feathers and was capable of gliding
with them. He provided an inaccurate skull drawing rather than a tracing (Fig. 11b).
Fraser (2006) considered the plumes to be plant material accidentally preserved in
coincidence with the vertebrae of Longisquama. Voight et al. (2009) and Buchwitz &
Voigt (2012) argued against the plant hypothesis, explaining how the plumes developed
in a feather-like process, convergent with avian dinosaurs. Due to inexperience working
with difficult materials, Peters (2000a) provided a sophomoric tracing of the skull (Fig.
11) and anterior portion of the skeleton without realizing that some of the elements were
palatal, occipital and from the hind limb. However, this was the first time a sternal
complex similar to those of pterosaurs (Wild 1993) was identified in a non-pterosaur. In
summary, virtually all work since Sharov (1970) has concentrated on the plumes with
only one fresh illustration of the skeleton from Peters (2000a).
Every worker who described the above three fenestrasaurs shed new light on each
specimen and made mistakes as well. Here those errors are corrected. The new
interpretations add support to the Peters (2000a) hypothesis that pterosaurs were indeed
derived from sister taxa to Cosesaurus, Sharovipteryx and Longisquama as all three had
additional pterosaurian traits not listed in Peters (2000a, 2002, see below).

16
Recently Kyrgyzsaurus (Alifanov & Kurochkin 2011) has been added to the list of
non-volant fenestrasaurs (Fig. 12) despite the fact that it was originally described as a
drepanosaur.
The basalmost pterosaur, MPUM 6009 (Wild 1978, Museo di Paleontologia,
Universitá di Milano, Italy) was originally considered a juvenile Eudimorphodon (Wild
1978) and later considered congeneric with Carniadactylus (Dalla Vecchia 2009) despite
the size difference. In Peters (2007) MPUM 6009 nested with neither, but formed a
transitional taxon between Longisquama and all other pterosaurs. Carniadactylus has a
relatively larger sternal complex without posterior indentations, a straighter coracoid with
a wider sternal articulation, and a longer scapula. The humerus of Carniadactylus has a
terminal deltopectoral crest, but it is not terminal in MPUM 6009. The pes is smaller,
relative to the tibia in MPUM 6009 and the metatarsus is also relatively smaller compared
to the pedal digits.
A new phylogenetic analysis (Fig. 12) documents a gradual accumulation of
pterosaurian traits arising within a new clade of lepidosaurs nesting between the
Sphenodontia (Rhynchocephalia) and Squamata.
Materials and Methods
The present taxon list of 24 is a subset of a larger taxon list of 389 taxa that includes
several archosauriformes including generic members of the Parasuchia,
Proterochampsidae and Scleromochlus. None of these nested with pterosaurs. That larger
matrix originally included 228 characters. In the present subset 77 of those characters

17
were constant and 140 of the remaining 151characters (Supp. Data) were parsimony
informative. That results in 5.8:1 character/taxon ratio.
Although some characters used here are similar to those from various prior
analyses, the present list of character traits (Supp. Data) was largely built from scratch.
Characters were chosen or invented for their ability to lump and split clades and for a
trait’s visibility in a majority of taxa. Small and hard-to-see foramina were not included.
Due to the wide range and size of the inclusion set, data were collected from
personal observation, photographs and the literature. I had direct access to the specimens
of Cosesaurus, Longisquama and the counterplate of Sharovipteryx from which most of
the present data were extracted. A first generation cast of Cosesaurus was a gift from Dr.
Ellenberger.
Taxa and characters were compiled in MacClade 4.08 (Maddison and Maddison,
1990) then imported into PAUP* 4.0b (Swofford, 2002) and analyzed using parsimony
analysis with the heuristic search algorithm. All characters were treated as unordered and
no character weighting was used. Bootstrap support figures were calculated (Fig. 12).
Results
The present phylogenetic analysis of 24 taxa and 151 characters recovered a single
optimal tree (Fig. 12) with a length of 403 steps, a Consistency Index (CI) of .529, a
Retention Index (RI) of .727, a Rescaled Consistency Index (RC) of .384 and a
Homoplasy Index (HI) of .471. All nodes in the lineage of pterosaurs had Bootstrap
scores over 86.
The present analysis is the first to include 24 taxa that document a gradual

18
increase in pterosaurian traits. This falsifies prior claims (e.g., Hone and Benton 2007,
2008) that pterosaurs appeared in the fossil record without obvious antecedents. Prior
workers had simply failed to include relevant taxa in their analyses and had looked
elsewhere in vain. The following list of lepidosaurs (Fig. 12) document a gradual
accumulation of pterosaur synapomorphies.
A. Homoeosaurus/Dalinghosaurus
1) Frontal/parietal suture straight and longer than frontal/nasal suture; 2) caudal series
longer than precaudal series; 3) pubic apron present and wide.
B. Carusia/Meyasaurus
4) Naris longer than 2x height; 5) last maxillary tooth at mid orbit.
C. Bavarisaurus
6) Naris displaced or posteriorly elongate; 7) caudal transverse processes absent beyond
the eighth caudal; 8) proximal metatarsals subequal in width.
D. Lacertulus/Daohugou lizard
9) Premaxilla ascending process invades nasals; 10) premaxilla extends beyond nares;
11) postorbital tentatively contacts parietal; 12) quadratojugal gracile; 13) pterygoid
narrow.

19
E. Tijubina
14) Premaxilla ventral orientation horizontal (not transverse); 15) orbit less than rostral
length; 16) jugal depth not gracile; 17) premaxillary teeth robust; 18) scapulocoracoid not
fused; 19) metatarsal one greater than half of four.
F. Huehuecuetzpalli
20) Skull table convex; 21) naris opening dorsolateral; 22) postfrontal does not contact
upper temporal fenestra; 23) quadratojugal articulation with quadrate insecure; 24)
olecranon process present; 25) fibula diameter not greater than half the tibia; 26)
metatarsal five axially twisted.
G. Macrocnemus
27) Preorbital skull length longer than postorbital length; 28) upper and lower temporal
fenestra with both arches present; 29) quadratojugal extends from jugal ramus; 30)
maxillary palatal processes present; 31) premaxillary teeth procumbent; 32) coronoid
process low or tiny; 33) eight cervicals; 34) cervical ribs with free anterior processes;
cervical rib orientation parallel to centra and elongate; 35) 25 or fewer presacral
vertebrae; chevrons parallel to centra; 36) gastralia rod-like; 37) pubis orientation ventral;
38) fibula appressed to tibia.
H. Jesairosaurus
39) Antorbital fenestra without maxillary fossa (tentative); 40) four premaxillary teeth;

20
41) dentary contributes to coronoid process; 42) three or more sacral vertebrae; 43)
scapula larger than coracoid; 44) ilium posterior process not longer than anterior process.
I. Langobardisaurus
45) Skull width less than 1.2x height at orbit; 46) prefrontal does not meet maxilla; 47)
paroccipital angle transverse; 48) supratemporal and squamosal fused; 49) internal nares
medial; 50) dorsal transverse processes present, some not shorter than centra; 51) second
caudal transverse process longer than the centrum; 52) advanced mesotarsal tarsus.
J. Cosesaurus
53) Antorbital fenestra without maxillary fossa; 54) naris not larger than antorbital
fenestra; 55) premaxilla excluded form choana; 56) vomers narrow; 57) vomer teeth
absent; 58) interclavicle fused to sternum; 59) coracoid reduced to strut; 60) ilium
anterior process longer than acetabulum width; 61) prepubis present; 62) ventral pelvis
without a thyroid fenestra (by convergence in Jesairosaurus); 63) tibia less than 2x ilium
length.
K. Kyrgyzsaurus
64) Major axis of naris greater than 30º; 65) nasals longer than frontals; 66) pineal
foramen absent; 67) quadrate not posteriorly excavated; 68) no medial fusion of frontals
and parietals; 69) postorbital posterior extends lateral to the medial posterior rim of
parietal; 70) vomer contacts maxilla; 71) posterior mandible depth even.

21
L. Sharovipteryx
72) Squamosal descending angle acute; 73) quadrate leans posteriorly; 74) internal nares
not close to premaxillary teeth; 75) some maxillary teeth with multiple cusps; 76) five or
more sacral vertebrae; 77) midcaudal centra 3x longer than tall; 78) radiale and ulnare
block-like; 79) manus subequal to pes; 80) manual unguals 1-3 trenchant with long
penultimate phalanges; 81) tibia/femur ratio not less than one; 82) metatarsus not
compact; 83) pes shorter than half the tibia; 84) pedal 3.1 not longer than p2.1.
M. Longisquama
85) Orbit does not enter anterior half of skull; 86) jugal quadratojugal process descends;
87) squamosal descends only to the dorsal cheek; 88) ectopterygoid fused to palatine; 89)
palatal teeth absent; 90) retroarticular angle descends; 91) cervical height equals length;
92) cervicals decrease in height cranially; 93) midcervicals shorter than mid dorsal
vertebrae; 94) reduced posterior dorsal ribs not present; 95) second caudal transverse
process greater than width of centrum; 96) clavicles medially fused.
The following tested traits separate the basal pterosaur MPUM 6009 from proximal
outgroup taxa.
N. MPUM 6009, basal pterosaur
1) Snout-occiput length not less than half the presacral length; 2) ventral aspect of
premaxilla vs. rostrum, a third or greater; 3) lacrimal not deeper than maxilla; 4) humerus
longer than femur; 5) forelimb longer than hind limb; 6) manus larger than pes; 7)

22
manual 4.4 longer than m4.3; 8) pedal digit five reduced to three phalanges (including
ungual).
Pterosaurian traits not employed in the present phylogenetic analysis: 1) anterior rotation
of the orbits: Cosesaurus; 2) anterior process of T-shaped interclavicle: tentatively in
Macrocnemus, fully in Cosesaurus; 3) metacarpal 4 axially rotated, palmar side posterior:
MPUM 6009 certainly, Longisquama less certainly; 4) ventral pelvis deeper than long:
Longisquama.
Soft tissue first appearance: 1) dorsal frill: Huehuecuetzpalli; 2. cervical pycnofibers
(pterosaur ‘hair’): Sharovipteryx; 3) caudal fibers (coalesce to form tail vane in certain
pterosaurs): Cosesaurus; 4) fibers trailing forelimb (aktinofibrils in pterosaurs):
Cosesaurus; 5) uropatagia: Cosesaurus; 6) volant wings and propatagium: preserved
rarely in pterosaurs, but likely present in basal pterosaurs and precursor taxa with a
pteroid.
Description
Despite being known from complete and articulated fossils, Cosesaurus, Sharovipteryx
and Longisquama are indeed difficult subjects. A cursory tracing or examination with
preconceptions will fail to provide the amount of data available to the more patient,
precise, experienced and unbiased observer familiar with their sister taxa. The following
are incomplete descriptions, focusing primarily on the misinterpretations of Peters
(2000a) and other workers.

23
Cosesaurus
As Ellenberger (1978, 1993) indicated (contra Sanz and López-Martinez 1984;
Peters 2000a), the scapula of Cosesaurus is indeed extremely narrow and posteriorly
elongate (Fig. 1), almost rib-like. This morphology follows the pattern of pterosaurs,
Longisquama, Sharovipteryx and Kyrgyzsaurus, and to a lesser extent Langobardisaurus,
Tanytrachelos (Olsen 1979) and Jesairosaurus (Jalil 1997). In situ the scapula
impressions appear undisturbed beneath the dorsal ribs and ventral gastralia (Fig. 2).
Neither scapula includes an acromion process (contra Ellenberger 1978, 1993).
The rim of the reported median “keel” in Cosesaurus (Ellenberger & de Villalta
1974; Ellenberger 1978, 1993; Peters 2000a, 2002; Fig. 2) is reinterpreted here as the
narrow, quadrant-shaped stem of a disarticulated coracoid (Fig. 3). Overall it is distinct
from the disc-like coracoids of Macrocnemus and Langobardisaurus, but similar to the
crescentic coracoids of Sharovipteryx, Longisquama and basal pterosaurs. There is a gap
between the purported plane of the see-through “keel” and the thicker “rim” (Fig. 2).
What lies beneath is reinterpreted as the anterior process of a cruciform interclavicle
beneath the crescentic coracoid that forms the thicker ‘rim.’ The in vivo dorsal portion of
the left coracoid lies on top of the sternum and interclavicle. The dorsal portion is
expanded for articulation with the scapula, as in pterosaurs. The similar right coracoid
extends in situ laterally with articular surfaces preserved closer to both the right scapula
and the middle of the interclavicle. The right coracoid mimics the shape and position of a
rim of the Macrocnemus scapula, hence my earlier misinterpretation (Peters 2000a; Fig.

24
2c). Ellenberger (1978, 1993; Fig. 2b) misinterpreted this curved bone as the acromion
process of the scapula.
The Cosesaurus interclavicle (Figs. 1-3) is cruciform, with an anterior process not
found in its phylogenetic precursor, Huehuecuetzpalli (Reynoso 1998), and only
tentatively developed in Macrocnemus (Peyer 1937). A large anterior process is also
present in Longisquama and basal pterosaurs where it produces a typically shallow keel.
In Cosesaurus the interclavicle is flat. The transverse processes are robust.
As in most basal tetrapods, the medial clavicles of Huehuecuetzpalli and
Macrocnemus articulate with the interclavicle, then curve dorsally to rim the anterior
coracoids and scapulae (Fig. 3a, b). This changes in Cosesaurus where the shorter
straighter clavicles do not extend dorsally, but rim only the anterior margin of the
trapezoidal sternum coincident with the transverse processes of the interclavicle. In
Longisquama and pterosaurs (Fig. 3d, e) the clavicles wrap around the sternum rim
lateroposteriorly (Wild 1993; Fig. 3), distinct from all other tetrapods.
Previously mistaken for conjoined coracoids (Ellenberger and de Villalta 1974;
Ellenberger 1978, 1993; Peters 2000a, 2002; Fig. 2), the thinly ossified sternum of
Cosesaurus is twice as wide as long. The sternal shape of Cosesaurus most closely
resembles that of Macrocnemus (Fig. 3b), but the placement, dorsal to the interclavicle
(Fig. 3c), is more like that of Longisquama and pterosaurs. Scattered ribs give the false
appearance of curved sternal borders in the impressed fossil.
Previously considered a stem-like process on the blade-like anterior of the
birdlike-like pubis (Ellenberger 1993) or pterosaur-like ilium (Peters 2000a), a
disarticulated fan-shaped prepubis (Fig. 4) rests on the expanded blade of the anterior

25
ilium beneath it. The other dislocated prepubis is beneath the right femur. Much of it
remains visible due to crushing and breakage of the hollow limb. These pterosaur-like
fenestrated prepubes are 3 mm in length, about as long as the anterior process of the
ilium, with a blade similar in size to the pubis. Prepubes are unknown in the more
primitive Macrocnemus and Langobardisaurus. Larger prepubes are present in
Sharovipteryx, Longisquama and pterosaurs. Based on available data, the prepubis
appears to be a new bone rather than a modified gastralium as no gastralia in more
primitive taxa are robust or articulate with the pubis.
Kyrgyzsaurus bukhanchenkoi
Published eleven years after Peters (2000a) and originally considered an odd sort
of drepanosaur, Kyrgyzsaurus bukhanchenkoi (Fig. 1) nests between Cosesaurus and
Sharovipteryx in phylogenetic analysis (Fig. 12). Larger than its sisters, Kyrgyzsaurus is
known from a skull, neck and pectoral girdle more robust than in Cosesaurus, but of
similar morphology. Note the parallelogram-shaped cervicals raising the skull over the
shoulders. As in other fenestrasaurs, extradermal soft tissue was also found dorsal to the
skeleton of Kyrgyzsaurus.
Sharovipteryx mirabilis
On the plate, the small, crushed skull of Sharovipteryx mirabilis (PIN 2584/8,
Figs. 1, 5-9) is exposed in dorsal view with the occiput and quadrates falling open in
medial view (Fig. 6). The mandible is largely hidden beneath the skull. Palatal elements
appear at the margins and through the posterior cranial split.

26
The reconstructed skull of Sharovipteryx greatly resembles those of Cosesaurus,
Kyrgyzsaurus, Longisquama and the basal pterosaur, MPUM 6009 (Fig. 1) sharing a
short rostrum, elongate posteriorly displaced naris, triangular antorbital fenestra, large
orbit, narrow temporal fenestrae and a straight jawline.
The premaxilla of Sharovipteryx is acutely pointed with a short jawline exposure.
The four tiny premaxillary teeth are about the size of the scales that layer its dorsal
surface. At mid-length the premaxilla expands laterally posterior to the naris, cutting off
most nasal contact. The ascending process continues as a narrow strip to the mid nasals.
The slit-like and posteriorly displaced naris extends for less than a quarter of the
premaxilla length. The nasals are twice the length of the frontals and are concave laterally
to accommodate the strip-like prefrontals that border laterally. The frontals are wider than
the nasals and include posterior lateral processes that substantially broaden the cranium
and accommodate small triangular postfrontals that do not contact the upper temporal
fenestrae. The frontal/parietal suture is obtuse, pointed posteriorly. The broad parietals
are slightly broader posteriorly and longer medially.
The Sharovipteryx maxilla ventral margin is straight and accommodates
approximately 17 teeth of various sizes. All are larger than those in Cosesaurus. The
largest anterior teeth are recurved. The ascending process of the maxilla arises at a
shallow angle (~25º) and contacts the nasal, prefrontal and lacrimal. These frame a
triangular antorbital fenestra without a fossa, as in pterosaurs. Gans et al. (1987)
considered the antorbital fenestra absent, but they did not attempt a reconstruction. Due
to crushing and slight scattering, the antorbital fenestra only becomes apparent with a
reconstruction, but phylogenetic bracketing also indicates its presence. Forming the

27
posterior margin of the antorbital fenestra, the lacrimal is a curved, anteriorly leaning T-
shape, concave along the orbit rim. The L-shaped jugal has an obtuse postorbital process.
While beetles dot the matrix, a previously overlooked wasp-like insect resides inside the
left orbit (Fig. 6). Unlike Cosesaurus, the jugal of Sharovipteryx lacks a quadratojugal
process. The quadratojugal is a short spur arising from the base of the jugal, but not in
contact with the quadrate. The postorbital is triradiate and half as tall as the orbit. The
squamosal is strongly curved in dorsal view and does not include a descending process,
unless it was broken off. The quadrate is a tall gently curved bone that leaned posteriorly
in life.
The paired vomers of Sharovipteryx are narrow rectangular strips. The strip-like
ectopterygoid and strip-like palatine probably combine to form an L-shaped bone, the
ectopalatine, but their intersection is hidden by overlying bone. Both sinusoidal
pterygoids are broken, but otherwise in place. They are more robust approaching the
quadrates. The paired basipterygoid processes of the basisphenoid are elongate and
narrow, as in Cosesaurus. The cultriform process is short and narrow. The supraoccipital
is tall and narrow. The small opisthotics are greatly expanded laterally. Together with the
supraoccipital they frame large posttemporal fenestrae.
The dentary is straight and narrows to a sharp Gothic tip in dorsal view. A few
large recurved teeth, twice as tall as other dentary teeth, appear below the
premaxilla/maxilla suture. The coronoid is low. The angular is less than half the depth of
the mandible. The retroarticular process (articular) turns up slightly.

28
A set of elongate and tapering hyoids emerges from below the base of the skull.
The longest extends to the fifth cervical. Their asymmetrical preservation suggests they
were mobile.
In Sharovipteryx there are eight procoelus cervicals, probably seventeen dorsals,
seven sacrals and about forty caudals. The presacral count matches that of Cosesaurus.
Cervicals 3-6 are elongate, like those of long-necked pterosaurs, with very low to absent
neural spines. Langobardisaurus also has a long neck, but proximal sister taxa (Figs. 5,
12) do not. As in pterosaurs, the morphology of the cervicals permits substantial
dorsoventral mobility, but little movement to the left and right. Extremely gracile cervical
ribs are oriented parallel to and are the length of each centrum.
The shorter dorsal vertebrae have short neural spines and short transverse
processes. The first dorsal rib is gracile. The next seven are wide and flat distally.
Together they form a disc in dorsal view. The next seven ribs are not preserved, but were
likely extremely gracile (if present at all) as there is little room for any other shape. The
two posterior dorsals do not have ribs. Gracile gastralia are barely perceptible. Seven
sacrals are aligned between the elongated ilia. Their ribs (sacral transverse processes) are
robust and expand distally. There is no sacral fusion.
Overall the tail of Sharovipteryx is extremely attenuate and stiff. The anterior four
or five caudals are hidden beneath a rectangular patch of tiny scales, the product of a
dorsal ‘kink’ (Gans et al. 1987) that raised the tail base. The first visible caudal is a
robust bone with elongate articular surfaces permitting great dorsoventral flexion. The
next one is narrower and elongate, like a cervical, with similar articular surfaces. The
remaining caudals vary from three to seven times longer than tall. The posterior caudals

29
gradually decrease in diameter and length, likely to the tail tip, which is missing from the
broken edge of the plate. Most caudal chevrons are subequal to and parallel each
centrum.
The strap-like scapula is covered with a layer of scaly epidermis that looks like
netting. It is longer than the stem-like coracoid, which lies beneath the ribs. The scapula
and coracoid are unfused and widest at their suture. The ventral coracoid includes a
posteriorly directed articular surface, probably connected to the unexposed sternal
complex. The shape of the Sharovipteryx sternal complex can be estimated by
phylogenetic bracketing, taking the rectangular sternal complex of Cosesaurus or the
triangular one of MPUM 6009 and modifying either to fit the first few dorsal ribs of
Sharovipteryx (Fig. 7). This hypothetical sternal complex would have been wider than
long, narrowing anteriorly. In like fashion, the sternal complex also narrows anteriorly in
Longisquama (Peters 2000a, 2002) and the basal pterosaur MPUM 6009 (Fig. 1).
In contrast to prior workers, Unwin et al. (2000) considered the forelimbs of
Sharovipteryx buried or missing. The radius and ulna do indeed have a similar shape to
the paddle-like rib tips, but all of the actual rib tips are accounted for. The right ulna and
radius are oriented parallel to the ribs, but none of the other forelimb elements resemble
ribs. The present interpretation (Fig. 1) matches and enhances what Sharov (1971) and
Peters (2002) traced and Tatarinov (1994) described. Here the left and right elements
match and all elements share traits and proportions with Cosesaurus, Longisquama and
pterosaurs (Fig. 1). The tiny forelimbs of Sharovipteryx are best considered vestiges.
The short, robust humerus of Sharovipteryx has a large deltopectoral crest and a
prominent capitulum/trochlea with flattened articular surfaces. So the elbow was likely

30
inflexible. The antebrachium is slightly longer than wide with an appressed radius and
ulna incapable of pronation or supination, as in pterosaurs. The radiale and ulnare are
block-like as in pterosaurs. The distal carpals are proportional to the proximal
metacarpals. The pteroid is a short pawn-shaped bone that would have anchored a tiny
propatagium if present. The square preaxial carpal is also preserved. Metacarpals 1-3
increase in length laterally with more asymmetry than in Cosesaurus. As in pterosaurs
and Longisquama, metacarpal 4 is more robust that the others with an expanded distal
articular surface. In pterosaurs metacarpal 4 rotates axially 90º (Peters 2002) so the finger
flexes in the plane of the manus (contra Bennett 2008). In Sharovipteryx, if any rotation
were present, it would have been minor because the PILs (parallel interphalangeal lines,
Peters 2000b, 2010, Fig. 3) were continuous across all four medial digits. These lines
indicate that digit 4 phalanges flexed in sets with their medial counterparts. In pterosaurs
PILs are not continuous with digit 4 because it flexes independently in the plane of the
manus and is so much larger. In Sharovipteryx, digits 1-4 are slender and elongate,
increasing in length from 1 to 4 and all had trenchant unguals.
In Sharovipteryx the ilial processes are much longer. The pubis and ischium of the
right pelvis are visible through the sacrals (Fig. 5). The left ventral pelvis is largely
buried beneath uropatagia, so its shape can only be determined by the contours of the soft
tissue that blankets them. As in Cosesaurus, the ventral pelvic elements of Sharovipteryx
appear to be sutured, not fused, and without a thyroid fenestra.
Based on subtle surface contours, one prepubis appears to be buried beneath the
fingers of the right hand (Fig. 5). If so, it is similar in shape to those found in Cosesaurus
and MPUM 6009 and proportional to the Sharovipteryx femur, which makes the prepubis

31
appear large compared to the pelvis (Figs. 1, 7). If I have not correctly identified and
outlined the prepubis, phylogenetic bracketing between Cosesaurus and MPUM 6009
provides Sharovipteryx with this shape and size of a prepubis.
The Sharovipteryx femur is as long as the glenoid-acetabulum length. That is
more than twice the relative length present in Cosesaurus (Fig. 1). The basal pterosaur,
MPUM 6009, has a femur nearly as long as the glenoid-acetabulum length. Most derived
pterosaurs do not have such a long femur (exceptions include stork-like azhdarchids).
The distal end of the Sharovipteryx femur expands anteriorly to form a pulley-lie process
(Unwin et al. 2000), which also acts as a stop to prevent overextension. Distinct from
Cosesaurus, the tibia/fibula in Sharovipteryx is slightly longer than the femur, as in
Longisquama and pterosaurs. Substantial soft tissue at the knee fills the gap between the
femur and tibia, permitting great flexibility. A simple hinge ankle joint is present with
two proximal tarsals (broader astragalus and spool-shaped calcaneum) along with three
distal tarsals (centrale, distal tarsal 3 and distal tarsal 4; Peters 2000a).
Distinct from Cosesaurus the metatarsals in Sharovipteryx, Longisquama and
MPUM 6009 (Fig. 1), spread widely, framing interdigital membranes. In situ pedal digit
1 is displaced lateral to metatarsal 2 and also displaced proximally, similar to the
displacement of the other distal phalanges. In the new pedal reconstruction metatarsal
four is longer than the others, as in Cosesaurus and basal pterosaurs. Here the interdigital
membranes and uropatagium extend to the penultimate phalanges, further than in either
Cosesaurus or pterosaurs. Pedal digit 4 is subequal to metatarsal 4 in Cosesaurus. Digit 4
is slightly longer in Sharovipteryx and much longer in MPUM 6009. As in Cosesaurus

32
and pterosaurs, pedal 5.1 is metapodial and digit 5 includes an ungual. Pedal 5.2 and p5.3
remain distinct in Cosesaurus, Sharovipteryx and Longisquama, but fuse in pterosaurs.
The premaxilla, shoulders and the base of the tail in Sharovipteryx are covered
with scales. Long ‘hairs’ (pycnofibers in pterosaurs, Kellner et al. 2010) are visible on the
scaleless skin surrounding the cervicals.
Cosesaurus and Longisquama both have dorsal plumes. Phylogenetic bracketing
indicates that plumes should also be present in Sharovipteryx. A closer look reveals
several short and gracile plumes in Sharovipteryx (Figs. 1, 6, 7) all slightly deeper than
the torso. If we were looking at the same structures, Gans et al. (1987) described these as,
“A series of grooves or folds suggests that a flap of skin may have attached to the
dorsolateral surface of the trunk.” No propatagium or forelimb trailing membranes are
visible, but phylogenetic bracketing and the presence of a short pteroid indicates that
Sharovipteryx probably also had extradermal forelimb membranes. The fiber-enforced
uropatagia are readily visible. Small, stiff pre-femoral membranes are also present, as
traced by Sharov (1971). In a gliding configuration (Fig. 7) these would have filled the
gap between the disc-like torso and the extended femora without connecting to the torso
or forelimbs (Gans et al. 1987). In the pes interdigital membranes are visible between
digits 4 and 5. Similar membranes were likely present between the medial pedal digits.
Considering the long list of traits shared by Sharovipteryx and pterosaurs, it is a
wonder that they were never tested together in any published phylogenetic analysis other
than Peters (2000a). The long forelimbs of pterosaurs and the short forelimbs of
Sharovipteryx seem to have created an impasse. However, a variety of theropods,
including several flightless birds, also vary widely in forelimb length.

33
Longisquama insignis
The skeleton of Longisquama insignis (PIN 2584/4) is severely crushed. Soft
tissue is well preserved here, but it often obscures bones and sutures. Contra earlier
reports (Sharov 1970; Unwin et al. 2000; Peters 2000a, 2002), the skeleton is complete
(Fig. 10). The forequarters and the separated tail are largely articulated, but the left
scapula + coracoid and the sternal complex are displaced anteriorly. The pelves are
flipped posteriorly and the prepubes are displaced posteriorly. The posterior dorsal
vertebrae rise sharply and the anterior sacrals descend sharply, as if they were
taphonomically pulled between vertebrae #32 and #33. A set of posterior dorsal ribs is
displaced near the mid caudals. The posterior 15 caudal vertebrae are preserved in a
twisted spiral, as if they were the last to fall to the substrate, held aloft by the spinning
vane at the tail tip. What Sharov (1970) illustrated as a displaced plume roughly parallel
to the forelimbs is actually the left hind limb (Fig. 10). What Jones et al. (2000) and
Buchwitz & Voigt (2012) considered plume stems dorsal to the dorsal vertebrae are
actually metatarsals 1-5 of the left pes. The left pedal digits extend dorsally. The right
hind limb is buried beneath posterior plumes. The left hand is essentially on top of the
right one, which creates a jumble of digits. Even so, all the phalanges can be identified
and both reconstructions match one another (Fig. 10). In essence, others have already
seen many of the so-called “missing” post-cranial elements. They were hiding in plain
sight, just misidentified. Here the elements are identified, segregated and reconstructed
(Fig. 1).

34
With one Longisquama plume now identified as a hind limb, only six large
plumes remain along with several smaller ones arising from the skull and cervicals. By
homology with Cosesaurus and Sharovipteryx, the plumes are dorsal frills. No more
dorsal frills are identified from the disturbed posterior torso and pelvic areas and no other
dorsal frills are scattered about. So it is possible that only six large frills were present in
Longisquama. They were sometimes shed (Sharov 1970, Buchwitz & Voigt 2012).
A V-shaped tail vane tips the tail. A scaly and hairy gular sac fills the throat area.
A trachea is preserved (Fig. 11). As Sharov (1970) noted, membranes fill the crook of the
arm (= propatagium) and trail the forelimb elements (= brachiopatagium). The uropatagia
are detached from the hind limbs, covering some bones. Gastralia are present, but were
thoroughly scattered when the torso and pelvis were disturbed. Nearly every aspect of the
skeleton of Longisquama greatly resembles that of Cosesaurus and/or Sharovipteryx—
only more exaggerated (Fig. 1).
A survey of past attempts at tracing the skull of Longisquama (Fig. 11) shows
that Sharov (1970), Peters (2000a), Senter (2003) and Martin (2004) all struggled with
tracing it or they opted not to go into detail. Those problems are rectified here (Fig. 11)
with a more precise and labeled set of tracings and reconstructions.
The small, crushed skull of Longisquama is exposed largely in left lateral view on
the plate (Fig. 11), missing only the pre-narial portion of the premaxilla and dentary. The
L-shaped ectopalatines (fused ectopterygoid + palatine), vomers and sinusoidal
pterygoids were rotated to the parasagittal plane during crushing, exposing their ventral
surfaces while crossing the antorbital fenestra and orbit. The hyoids and occipital bones

35
were likewise rotated during crushing. Crushing exposed the dorsal surfaces of the fused
parietals.
Of the premaxilla only the ascending process is preserved. It is a stem-like
structure that rises as far as the anterior orbit. An impression of a posterior premaxilla
tooth is also visible. The nasal anteriorly borders the posterior naris and extends slightly
beyond the premaxilla ascending process. The maxilla includes a curved, wave-shaped
ascending process that meets the lacrimal on the dorsal rim of the triangular antorbital
fenestra. The maxilla includes 16 triangular to recurved teeth, deepest below the
antorbital fenestra. Posterior maxillary teeth are either very closely appressed or
multicusped. The jugal is slightly deeper than the maxilla. The lacrimal process of the
jugal is small. The postorbital process of the jugal is twice as tall and leans posteriorly.
The quadratojugal process of the jugal descends and produces a tiny, straight
quadratojugal aligned parallel with the posteriorly tilted quadrate. The lacrimal is a
slender splintered stem with a robust sloping T-shape top. The postfrontal is a narrow
strip, narrower anteriorly, bordering the nasal and frontal. The frontal includes a narrow
anterior process interdigitating between the nasal and prefrontal. The posterior frontal
extends laterally in a tongue-like process to back a small triangular postfrontal not in
contact with the upper temporal fenestra. The postorbital is gracile and triradiate with a
long, straight jugal process/stem. The parietals are robust and fused together with large
and deep upper temporal fenestrae. A deep medial embayment divides the posterior
parietals. The squamosal borders the upper temporal fenestra and may contribute a
descending process lateral to the quadrate, but this is difficult to determine in dorsal view.

36
The robust, tall right quadrate is undisturbed. The left quadrate is rotated ~80º in situ with
its articular surface in the middle of the orbit.
The gracile occipital bones were apparently fused together because they were
preserved as a unit during taphonomic crushing. The supraoccipital portion is very tall
and the opisthotics are elongate. These frame large, posttemporal fenestrae. The long
basipterygoid processes of the basisphenoid are more than twice the length of the short,
gracile cultriform process.
The vomers are short narrow strips, extending from the premaxilla to contact the
anterior pterygoids. The short rectangular ectopterygoid is fused to the narrow palatine
probably creating an L-shaped ectopalatine, contacting the maxilla below the ascending
process and the jugal suture. The pterygoids are gracile and sinusoidal. In palatal view the
palatal elements produce a triangular rostrum. Several sets of hyoids are present. The Y-
shaped hyoids are as long as the skull. More robust shorter hyoids are below the
mandible. Two slender tapering bones cross the antorbital fenestra in situ, but probably
were placed medially, between the other hyoids in vivo.
The mandible includes 13 conical teeth, several deeper than the gracile dentary.
Posterior teeth are packed together so closely that some may represent multicusped teeth
with multiple roots. The coronoid is not visible because it is obscured by the jugal, but
the coronoid is not large in clade sisters. The surangular is nearly twice the depth of the
angular. The ventral mandible margin is straight, but the posterior angular descends
slightly.
In Longisquama there are eight small procoelus cervicals, 34 dorsals (two long
ones between the anterior ilia), five sacrals and about 40 caudals. The presacral count is

37
nine more than in Cosesaurus (Fig. 1) contributing to a torso that is twice as long. The
cervicals are much smaller than those in Cosesaurus, despite the greater size of
Longisquama overall. Tiny cervical ribs are present. The anterior dorsals are similar in
size to the cervicals. However the newly identified posterior dorsals, especially those
between the elongate ilia, are increasingly longer and larger posteriorly. Based on the
number of displaced dorsal ribs and the number of dorsal vertebrae without them, ribs
attend every dorsal vertebra except those between the ilia. The longest ribs are at mid
torso. There is no sacral fusion. Nor are the short sacral ribs expanded laterally, as in
Sharovipteryx. The anterior three caudals have transverse ribs of increasing lengths, but
the next three caudals reduce their transverse ribs caudally. Distinct from Cosesaurus, the
transverse ribs in Longisquama are fused to each vertebra. The remaining caudals are
elongate and apparently without ossified chevrons. Caudals 21 to 31 are extremely
elongate. Thereafter the caudals gradually decrease in diameter and length to the tail tip,
which is marked by an oval of eight tiny, sometimes paired ossifications supporting the
tail vane. Overall the tail is extremely attenuate and apparently stiff for most of its length,
but the in situ spiraling of the posterior 15 vertebrae suggests they were more flexible
than in Sharovipteryx or pterosaurs.
The strap-like scapula of Longisquama is longer than the stem-like coracoid. Both
are more robust than in Cosesaurus (Figs. 1, 3) and the coracoid has a straighter stem.
The Longisquama sternal complex is crescentic with perhaps a very shallow anterior
interclavicle keel. The clavicles overlap medially and rim the sternum posterolaterally, as
in pterosaurs. The transverse processes of the interclavicle are likewise crescentic. By

38
way of expansion or embayment or both, the sternum develops a posteromedial process
that creates a trident shape posteriorly, as in MPUM 6009, a basal pterosaur (Fig. 1).
The humerus of Longisquama is no larger than that of Cosesaurus, but it has a
large deltopectoral crest and expanded distal condyles, as in basal pterosaurs (Fig. 1). The
radius and ulna of Longisquama are likewise subequal to those in Cosesaurus. The
radiale and ulnare are block-like, as in Sharovipteryx and pterosaurs. The pteroid and
preaxial carpal are larger than in Cosesaurus. As in Cosesaurus, metacarpals 1-3 only
slightly increase in length laterally. Distinct from Cosesaurus, and similar to
Sharovipteryx and pterosaurs, metacarpal 4 in Longisquama is more robust that the others
with an expanded distal articular surface. Here, as in Sharovipteryx, the manual PILs
remained continuous with digit 4 joints, indicating that digit 4 phalanges still flexed in
sets with their counterparts. Similar to Sharovipteryx, manual digit 4 in Longisquama is
longer than digit 3 by two phalanges. Like MPUM 6009, digit 4 is more robust in
Longisquama. The largest, most trenchant ungual in Longisquama is at the tip of manual
digit 4. Like Cosesaurus, manual digit 5 in Longisquama is similar in length to the
metacarpus and not a vestige, as in Sharovipteryx and pterosaurs (Fig. 1).
In Longisquama the short posterior process of the ilium rises 60º, much higher
than in more primitive fenestrasaurs. This angle approaches that of the basal pterosaur
MPUM 6009 (Fig. 1). Most derived pterosaurs revert to a lower posterior ilium angle of
elevation. The pelvic elements (ilium, pubis, ischium) are all fused in Longisquama and
the ventral elements are much deeper than in Cosesaurus and Sharovipteryx. One and a
half prepubes are preserved. The other half is lost past the matrix break. The prepubis

39
articulates with the ventral pubis at an immobile butt joint (contra Claessens et al. 2009)
and approaches the size and shape of the MPUM 6009 prepubis.
Relative to the torso, the femur of Longisquama is shorter than in Cosesaurus. No
distinct femoral neck is present, as in sister taxa. The distal femur expands to form a
concave cap for the convex proximal tibia. The tibia/fibula in Longisquama is longer than
the femur. Distinct from all sister taxa, the tibia and fibula of Longisquama are robust.
The fibula is slender, but not a splint. Tarsals have not been identified in the chaos of
plume bases and vertebrae.
Distinct from sister taxa, the short, robust metatarsals of Longisquama are nearly
subequal, only slightly increasing in length laterally. Metatarsal 5 is less than a third as
long as metatarsal 4 (Fig. 1), as in sister taxa. The proximal phalanges on digits 1–3 are
subequal to their metatarsals. The penultimate pedal phalanges are shorter. Digits 2–4
align distally as in MPUM 6009. The unguals are sharp, but not strongly curved. Pedal
5.1 is bent at mid-length with a distal trochlea articulation enabling retroverted distal
phalanges, as in pterosaurs (Fig. 1). In Longisquama pedal 5.2 contacted the substrate on
its dorsal surface, as in basal pterosaurs and the trackmaker of Rotodactylus (Peabody
1948, Peters 2000b, 2011). Pedal 5.2 is short. Pedal 5.3 is shorter and provided with an
ungual (p5.4). Basal pterosaurs have one less phalanx than in Longisquama resulting
from the fusion of pedal 5.2 and p5.3, maintaining the sum of their lengths.
Physiology and Behavior
Pterosaurs were the first vertebrates to achieve powered flight. Here, based on additional
data, are the surrounding environnments and gradual changes that produced a wide

40
variety of anatomy, physiology and behavior in the Fenestrasauria, only one of which
successfully culminated in powered flight.
Environment
With only four known specimens, nonvolant fenestrasaurs are exceedingly rare in
the fossil record. However, Rotodactylus ichnites, matched to fenestrasaur trackmakers
(Peters 2000b), are more common. They are known from the lower Triassic of the
Southwest United States, Europe and Algeria (Brusatte et al. 2011).
Rotodactylus tracks were formed on floodplains of large meandering rivers, far
inland from the coast (Brusatte et al. 2011). Cosesaurus was washed out to marine
sediments, which provide few clues to nearby terrains. The other three basal fenestrasaurs
are all from the Madygen formation.
The Madygen megaflora was rich in pteridosperms (seed ferns). It also contained
conifers, ferns, horsetails, and lycopsids (Shcherbakov 2008). In addition to the multitude
of insects, freshwater bivalves, tadpole/shield shrimps and fish were also found there. A
primitive cynodont, Madysaurus (Tatarinov 2005), was the only possible predator of the
fenestrasaurs thus far discovered.
Diet and Predation
Found in insect-laden sediments, fenestrasaurs were likely insectivores catching
them on the substrate. In Sharovipteryx certain maxillary teeth were twice as large as
others and recurved. These could have pierced the exoskeletons of insects. Posterior teeth
were in contact with each other or multicusped for oral processing, a trait that continued

41
only to the end of the Triassic in pterosaurs. In palatal view (Figs. 7, 11) the rostrum of
fenestrasaurs is broader cranially, providing a measure of binocular vision by rotating the
orbits anteriorly. That would have been useful for judging distances to landing locations,
prey, mates and rivals.
Bipedal Locomotion and Flapping
In the basal lepidosaur, Huehuecuetzpalli, as in many living lizards, the ossified
scapula is short, anteriorly fenestrated and dorsally truncated (Reynoso 1998).
Impressions indicate that a cartilaginous extension, the suprascapula, extended the
ossified portion, as in living lizards. Derived taxa took two paths. In Macrocnemus and
Tanystropheus the scapula became short, not fenestrated, and without a trace of a
suprascapula. This loss of fenestration has obscured their lepidosaurian affinities. Taking
the other path, in Tanytrachelos and Langobardisaurus the continued erosion of the
anterior rim gives the posterior remainder a strap-like appearance. In Jesairosaurus the
scapula is also strap-like, but oriented vertically to the height of the neural spines, as in
related drepanosaurs. In Cosesaurus, Sharovipteryx, Kyrgyzsaurus, Longisquama and
pterosaurs the scapula is narrow and extends posterodorsally over several ribs. This
change in morphology from ‘wide, short, mobile and vertical’ in Huehuecuetzpalli to
‘narrow, elongate, less mobile and diagonal’ in Cosesaurus signals a change in
locomotion.
The evolution of the coracoid matches that of the scapula in this clade. In living
lizards (and Huehuecuetzpalli) the coracoid is fenestrated. In Macrocnemus and
Tanystropheus the coracoid fenestrae fill with bone, reverting back to a disc shape.

42
Jesairosaurus and the proto-drepanosaur, Hypuronector (Olsen 1979), retain a coracoid
embayment but more derived drepanosaurs do not. In Cosesaurus and later fenestrasaurs
the anterior coracoid fenestrae expanded greatly, leaving only the quadrant shaped
posterior rim. MPUM 6009 retains that quadrant-shaped coracoid. More derived
pterosaurs more or less straighten out the coracoid stem (Fig. 3e).
In living lizards (and Huehuecuetzpalli) each scapulocoracoid is free to move
anteroposteriorly (Jenkins & Goslow 1983), riding along the sliding joint bordered by the
interclavicle stem and the anterior sternal rim. By contrast the narrow and posterodorsally
elongate scapula of birds is relatively immobile, especially so owing to the great firmness
of the ventral coracoid articulation to the sternum. Huntington (1918) reported the bird
scapula combined great strength and rigidity, enabling a powerful ventro-appendicular
musculature to move the anterior limb within a limited range in a few directions with
great force, but not adapted to a wider extent of more diversified motion. In birds such a
pectoral girdle serves as an anchor to the bones and muscles of the flapping wing.
Analogous due to similarities in morphology, the strap-like scapula and stem-like
coracoid in Cosesaurus, Sharovipteryx, Longisquama and basal pterosaurs would have
likewise delivered great force over a limited range, especially so considering the
socketing of the coracoid ventral stem to the sternal complex. Overall Cosesaurus had the
proportions of a typical small quadrupedal lepidosaur (Fig.1), but the immobilization of
the pectoral girdle must have had distinct behavioral and locomotory consequences.
Pterosaurs are universally considered flapping reptiles, capable of powered flight.
As in birds, those similar pectoral girdles provided anchors for the muscles and bones
that create thrust and lift in the forelimbs. That a basically identical pectoral girdle is

43
found in the tiny nonvolant pterosaur precursor, Cosesaurus attests to its nascent flapping
ability. Obviously with such a short manus flight would have been impossible. Instead,
perhaps Cosesaurus practiced some form of wing-assisted running (Dial 2003). The
flapping winglets could have acted as secondary sexual signals or to intimidate rivals.
Short trailing fibers on the forelimb of Cosesaurus (Ellenberger 1993, Peters 2009) were
the precursors to the longer aktinofibrils that support pterosaur wing membranes. Even in
their nascent stages they would have extended the surface area of the forelimbs, creating
proto-wings. The presence of a tiny pteroid and preaxial carpal on Cosesaurus (Peters
2009) provides a starting point for the development of the pterosaur propatagium that
Sharov (1970) illustrated (Fig. 1, 10) in Longisquama.
Among pterosaur workers there is universal agreement that a broad sternal
complex with a robust anterior (interclavicle) process anchored the large pectoralis
muscles of adduction that pterosaurs used for flapping (e.g. Wellnhofer 1991). A
precursor sternal complex in Cosesaurus (Figs. 2, 3) and Longisquama (Fig. 1) would
have been similarly muscled and used, even if less developed. The evolution of the
sternal complex from the plesiomorphic state is well documented in fenestrasaurs (Peters
2002) and their lepidosaur ancestors.
In the basal lepidosaur, Huehuecuetzpalli, as in most tetrapods, the clavicles rise
dorsolaterally to contact the anterior borders of the coracoid and scapula. In pterosaurs
the clavicles extend posterolaterally to rim the sternum. The short straight clavicles in
Cosesaurus represent a transitional phase between these two configurations (Peters
2000a, 2002). Since the dorsal orientation of the clavicles was lost in Cosesaurus, it is
likely that any muscles anchored there were also lost or transferred in the process. As the

44
clavicles later expanded along the lateral rims of the sternum in Longisquama and
pterosaurs, sternal muscles would have found new anchorage there.
The sternum of the basal lepidosaur, Huehuecuetzpalli, is rhomboidal and poorly
ossified (Reynoso 1998, Fig. 3). It is located at the posterior tip of the interclavicle. In the
more derived Macrocnemus the sternum is relatively smaller and without a pointed
anterior or posterior, and it remains at the posterior tip of the interclavicle. In Cosesaurus
the sternum is dorsal to the interclavicle without fusing to it. In Longisquama and
pterosaurs the sternum is fused to the interclavicle and clavicles, creating anchors for
strong flapping muscles.
Prepubes provide immobile ventral extensions to the pubis and medial anchors to
muscles of femoral adduction (contra Claessens et al. 2009). Their appearance in
fenestrasaurs coincides with the adoption of bipedal locomotion (Peters 2002). Pterosaur
ichnites (Peters 2011) indicate that certain derived pterosaurs reverted to a secondary
form of quadrupedal locomotion with their fingers oriented laterally to posteriorly. This
could only occur when their forelimbs became long enough to reach the substrate (Peters
2000b, 2002). Note, that is not the case with the basal pterosaur, MPUM 6009 (Fig.1)
with those long hind limbs. Rarely mentioned by other pterosaur workers, bipedal
pterosaur tracks are known (Conrad et al. 1987, Lee et al. 2009, Peters 2011, Kim et al.
2012). A bipedal pose produced a long lever arm of the elevated torso, neck, head and
forelimbs with its fulcrum at the first sacral vertebra. Without support the presacral
region would tend to drop were it not for the ventral support provided by the gastralia and
the prepubis, which guided compressive forces back to the pelvis.

45
Peters (2000b) found a match between the pes of Cosesaurus and certain
Rotodactylus ichnites, which are typically digitigrade and narrow gauge with an extended
pedal digit 5 planted behind the other digits (Peabody 1948). The tracks also demonstrate
occasional bipedalism. The tiny forelimbs and elongate hind limbs of Sharovipteryx and
Longisquama further attest to the practice of obligate bipedal locomotion in this clade
(Peters 2000a, 2002). Torso elevation during bipedal standing and running would have
freed the forelimbs to do something else, like flapping. A functionally bipedal
configuration has no need for the sliding coracoid joint found in quadrupedal lepidosaur
and tanystropheid pectoral girdles.
Early Triassic archosaurs or dinosaurs could not have made Rotodactylus and
Prorotodactylus imprints (contra Brusatte et al. 2011). Based on relative toe lengths the
better match is with macrocnemids and fenestrasaurs, taxa not considered by Brusatte et
al. (2011).
Sharovipteryx was an obligate biped due to the great discrepancy in the length of
its fore and hind limbs (Peters 2000a, 2002). The elongate ilium and seven sacral
vertebrae provided support for the cantilevered forequarters leveraged over the fulcrum at
the pelvis. Sharovipteryx would have been an able sprinter in the sprawling manner of
living lizards capable of bipedal locomotion (Snyder 1954). Hollow leg bones kept their
mass low. Major muscle anchors were concentrated at the pelvis and large prepubis. The
distal femur and proximal tibia do not bear large muscle anchor scars or processes.
The unfused ankle of fenestrasaurs was a simple hinge joint, convergent with
dinosaurs. This ankle morphology was inherited from a Late Permian sister to the basal
lepidosaur, Huehuecuetzpalli (Fig. 12), which also had unfused proximal tarsals

46
(Reynoso 1998), distinct from squamates and sphenodontids. Related tanystropheids
retained an unfused tarsus, but drepanosaurs more derived than Hypuronector fused the
proximal tarsal elements.
The metatarsals of Sharovipteryx could radiate 60º (Fig. 4), unlike the parallel
metatarsals and toes of Cosesaurus and the trackmakers of Rotodactylus (Peabody 1948,
Peters 2000b). Distinct from other fenestrasaurs, Sharovipteryx did not retrovert distal
pedal digit 5, probably due to the more distal extent of its uropatagia, nearly to the tip of
digit 5 (Fig. 8). Interdigital spaces were filled with membranes. Like web-footed ducks,
Sharovipteryx would have walked on these fiber-reinforced membranes, creating unique
ichnites. One wonders if Sharovipteryx could also run on water, like the living lizard,
Basiliscus basiliscus.
Whether Sharovipteryx ever hopped is a mystery. Libby et al. (2012) described
how leaping lizards use their tail for pitch control to redirect angular momentum from
their bodies to their tails, stabilizing body attitude in the sagittal plane. Extant hopping
mammals, like the kangaroo rat (Bartholomew & Caswell 1951, Biewener & Blickhan
1988) and the kangaroo (Alexander & Vernon 1975) approximate the gross morphology
of Sharovipteryx with their short torso, long tail, short forelimbs and long hind limbs.
Odd little Longisquama has no analog in the present day, other than to some
extent, the lemur, (e.g. Propithecus). Both share a small skull with large eyes, a short
neck, a long, flexible torso, strong hind limbs and an attenuate tail typically elevated at
the pelvis.
The tibia/fibula and feet of Longisquama are much more robust than in
Sharovipteryx or basal pterosaurs (Fig. 1). A robust hind limb suggests that Longisquama

47
was a leaper or runner, not a gracile glider. The nearly subequal metatarsals of
Longisquama radiated less than 30º, a product of its bipedal, narrow gauge configuration.
That made the pes better suited for terrestrial locomotion. Relative to the metatarsals, the
digits were much longer in Longisquama. Like most terrestrial amniotes, the penultimate
pedal phalanges were relatively short. By contrast, in the pterosaur MPUM 6009, the
pedal digits were much longer relative to the metatarsals and the unguals were more
trenchant, ideal for tree clinging.
Like the trackmaker of Rotodactylus and pterosaurs, Longisquama had a
retroverted, dorsal side down, distal pedal digit 5. Peters (2000a, 2002, 2011) showed
how such a toe could be used as a universal wrench for perching on horizontal branches,
along with keeping an elevated metatarsus steady on a flat substrate. The center of
balance remained over the anterior toes, with relatively little pressure on the retroverted
digit.
Respiration and Metabolism
In the basal pterosaur MPUM 6009 the naris is much larger than in Sharovipteryx
and Cosesaurus. Perhaps that reflects an increase in respiration to oxidize a higher energy
lifestyle based on feeding the large muscles necessary to continuously flap the much
larger wings. The larger naris could also represent weight savings on a larger skull.
As lepidosaurs, fenestrasaurs (including pterosaurs) may have lacked a
diaphragm, so respiration was driven by expansion and contraction of the ribs (contra
Claessens et al. 2009). In fenestrasaurs, the locking of the sternal complex to the anterior
ribs prevented expansion of the anterior torso. So only the posterior ribs were free to

48
drive respiration in fenestrasaurs. The deeper pelvis and prepubis on Longisquama and
MPUM 6009 made room for longer gastralia and longer posterior dorsal ribs. Thus, both
deeper torso pterosaurs and longer torso longisquamids had a greater capacity for rib-
driven respiration than shallow and short torso cosesaurs and sharovipterygids.
In basal tetrapods, including most lizards, an undulating torso constricts the
ability of both lungs to fill and empty at the same time. As bipeds, basal fenestrasaurs
were no longer tied to Carrier’s constraint (Carrier 1987). They could breathe while
running, an evolutionary novelty that was retained by pterosaurs, convergent with
dinosaurs.
The presence of ‘hair’ on the neck skin of Sharovipteryx (Fig. 6) is the first
indication of the pycnofibers that would eventually insulate the bodies of pterosaurs, as
preserved in Jeholopterus (Wang et al. 2004, Kellner et al. 2010) and Sordes (Sharov
1971). Altogether the above traits point to a rising metabolic rate in this clade.
Gliding
If Sharovipteryx prepared for a glide the way, the living rib-glider, Draco volans,
does, then it would have reoriented itself head down on a vertical tree trunk (Fig. 9). In
any orientation, glides would have been initiated with a coordinated leap to create initial
airspeed (Gans et al. 1987).
While the gliding abilities of Sharovipteryx have been almost universally accepted
(except by Tatarinov 1989), the anterior control surfaces that would have stabilized that
glide have only been imagined (Sharov 1971, Gans et al. 1987, Dyke et al. 2006). A

49
closer look reveals there was never any need to imagine anterior membranes. Everything
Sharovipteryx needed to glide is readily visible on the fossil (Figs. 5-7).
The vestigial forelimb canards of Sharovipteryx would have been only a minor
component in the gliding and steering abilities of Sharovipteryx. Far more important was
the lateral expansion of the extensible neck skin by the elongate hyoids (Figs. 6, 7). These
provided Sharovipteryx with anterior strakes. When deployed, simply raising or lowering
the long neck set pitch. In this way the strakes acted like horizontal elevators on an
airplane, only in front of the main wing. Heretofore overlooked, the extensible neck skin
of Sharovipteryx represents the long sought anterior control surface. Widespread ribs
flattened the torso of Sharovipteryx, as in Draco volans. Stiff pre-femoral membranes
filled gaps between the torso and hind limb creating a more or less continuous lifting
surface.
The small uropatagia in Cosesaurus (Ellenberger 1978, 1993, Fig. 1) became
greatly enlarged in long-legged Sharovipteryx. These fiber reinforced membranes created
hind limb wings (Fig. 7). If each femur extended anteriorly 60º and each tibia extended
from the femur 120º (similar to a reconstruction in Gans et al. 1987), the pedes would
have aligned with the center of gravity, as in the bipedal configuration. The center of
gravity of the hind limb wing would have passed just aft of the main spar, at the deepest
camber of the wing, as in birds and pterosaurs.
While gliding, the webbed feet of Sharovipteryx became aerodynamic control
surfaces. In flight, the pedes were configured near the center of gravity, outboard on the
wing tips where they could act as ailerons and/or spoilers. By contrast, in pterosaurs the
webbed toes were located far aft of the center of balance were they could act like twin

50
vertical stabilizers, flexing and extending at the tarsus to control yaw. In pterosaurs
slightly flexing the webbed toes gave the pedal membranes camber that provided laterally
directed lift (with sufficient airspeed) and so relieved the stress of extending the hind
limbs laterally by muscle power alone. In the presence of large forelimb wings and the
absence of neck strakes, pterosaurs retained small uropatagia to act as posterior
horizontal stabilizers. Like Sharovipteryx, and distinct from illustrations in Peters (2002),
pterosaurs would have extended their entire hind limbs laterally while flying.
The elevated posterior sacrals and posterior ilium of Longisquama kept the tail
elevated, both to keep it from dragging and to provide an elevated staff enhanced with a
decorative tail vane. The tail vane was not an aerodynamic control surface. Rather like
feathers on an arrow or a weather vane, it kept the rest of the tail aligned with the rest of
the body in the airstream (Peters 2002).
The small uropatagia in short-legged Cosesaurus (Ellenberger 1978, 1993)
became greatly enlarged in long-legged Sharovipteryx, Longisquama and basal
pterosaurs. Other than their decorative function, uropatagia in Longisquama might have
helped dissipate metabolic heat from the large muscles in the hind limbs, acting like large
surface area radiators. In long leaps trailing membranes might have added lift to extend
those leaps and some measure of aerial braking to soften landings.
Reproduction and Ontogeny
The tiny pelvic opening of Cosesaurus indicates that only tiny eggs could pass
through it (Figs. 1). So Cosesaurus probably laid many tiny eggs at a time, whether
buried or in a nest. A hypothetical hatchling from a 5 mm cosesaur egg would be about 1

51
cm in length or 1/14 the size of the adult. It would have been easy prey for most animals,
whether insect or tetrapod. Only a moist leaf-litter environment would ensure that such
tiny hatchlings would not die of desiccation (Hedges and Thomas 2001) due to their high
surface area-to-volume ratio. Sharovipteryx had a deeper pelvic opening that could have
produced a larger egg and a hatchling about 2 cm in length or 1/8 the size of the adult.
In Longisquama and pterosaurs reproductive strategies continue to evolve. Both
Longisquama and MPUM 6009 had a much deeper pelvic opening capable of passing
relatively much larger eggs. From such a large egg a hatchling could be 4cm in total
length or 1/5 the size of the adult. In short-torso flying pterosaurs perhaps only one egg of
this size at a time could be carried within a mother. In long-torso, nonvolant
Longisquama, several more could be carried.
When only one young is produced at a time great care is usually given to that
embryo/hatchling. The most protective place to develop is within the mother, especially
when she has a high metabolism and the eggshell is extremely thin (Chiappe et al. 2004).
These data support the hypothesis of egg retention in longisquamids and pterosaurs until
just before hatching, a trait that often arises in lepidosaurs (Pyron & Burbrink 2013), but
never in archosaurs. In support of this, pterosaur eggs preserved without an adult nearby
include full term ossified embryos (Wang & Zhou 2004, Ji et al. 2004, Chiappe et al.
2004). These eggs have been conventionally laid and the adult has flown away. Pterosaur
eggs preserved with an adult nearby do not include ossified embryos. In Lü et al. (2011)
an immature egg was expelled upon the death of the mother. In Wang et al. (2014)
immature eggs containing no ossified bones were fossilized along with the bones of adult
pterosaurs. Likely each egg was inside a female at the time of her death, but became

52
scattered along with the rest of her bones during the weather event (perhaps a desert
storm) that swept the colony to the bottom of the lake. Buried eggs would have been
protected from such an event.
The only known adult/embryo pairing among pterosaurs is in the genus
Pterodaustro (Chiappe et al. 2004; Chinsamy et al. 2008). The Pterodaustro embryo is
1/8 the size of the adult, reached sexual maturity after two years and full size in five
years. If longisquamids and pterosaurs followed the pattern of other tetrapods, smaller
genera, like Longisquama and MPUM 6009, probably reached sexual maturity more
quickly and had shorter lifespans.
In fenestrasaurs, including Longisquama, mating probably occurred as it does in
birds, by pressing the male cloaca against that of the female for sperm transfer. The
raising of the tail base in Longisquama and basal pterosaurs, like MPUM 6009, may have
facilitated mating in these bipeds. Hemipenes, a trait restricted to squamates, may or may
not have been present because Sphenodon lacks these and the clade including
Huehuecuetzpalli and fenestrasaurs nests between them (Peters 2007, Fig. 12).
Mating and territorial display rituals
Flapping probably originated as an addendum behavior to a suite of secondary
sexual traits already in place. Cosesaurus had a dorsal frill, fibers trailing the forelimbs,
uropatagia trailing the hind limbs and fibers (not feathers) emanating from the tail (Fig.
1). However, these are only small precursor decorations to the exotica found on
Longisquama, in which the dorsal frill had greatly expanded to form elongate plumes. In
the clade that produced pterosaurs, bigger wings, rather than longer plumes, enhanced

53
their display. At first the fore limb membranes may have signaled with a flash of bright
membrane color (Peters 2002). Later, larger, flapping wings may have increased the
height of leaps. Natural selection would have favored the best performers. A wagging tail
tipped with a colorful vane may have been another secondary sexual trait in
Longisquama. It is not impossible to imagine that fenestrasaurs also became vocal.
Variation and Autapomorphies
While Kyrgyzsaurus, Sharovipteryx and Longisquama include many traits transitional
between the primitive morphology of Cosesaurus and the derived morphology of
pterosaurs (listed earlier, Peters 2002), it is clear that all three represent clades that split
off from the direct line that led to pterosaurs. Sharovipteryx autapomorphies include the
reduction of the forelimbs, elongation of the cervicals, development of neck strakes,
flattening of the torso, elongation of the posterior ilium, development of pre-femoral
membranes and the expansion of much larger uropatagia. Longisquama autapomorphies
include the reduction of the cervicals, elongation of the torso, the increase in tibia
diameter and the development of hyper-elongate plumes. The robust morphology of
Kyrgyzsaurus represents a fourth evolutionary direction for cosesaurs, that of increased
size, and probably with no propensity to glide or fly while retaining vigorous flapping as
a secondary sexual or territorial behavior, judging by the size of its pectoral girdle. Given
the proportions of Cosesaurus and MPUM 6009 (Fig. 1), especially in the forelimb, it
seems likely that the development of the pterosaur wing did not involve a period of
reduction, as seen in Sharovipteryx and Longisquama.

54
Conclusions
Much has been learned about pterosaurs and their outgroup taxa since the publication of
the minority view on pterosaur origins, which most workers have ignored. Earlier
interpretations of Cosesaurus, Sharovipteryx and Longisquama are enhanced and
corrected here. Odd autapomorphies are now replaced with synapomorphies linking these
fenestrasaurs to their lepidosaurian precursors, their pterosaurian descendants and to each
other. The pterosaur-like pectoral girdle of Cosesaurus gives new insights into the timing
and development of characters associated with flapping flight in pterosaurs. These
include support for the hypothesis of wing-assisted bipedal running along with energetic
flapping during mating rituals and threat displays. The prepubis acted as an extension of
the pubis providing new anchors for femoral adductor muscles used in bipedal
locomotion. Despite its lack of large forelimbs, Sharovipteryx shared a suite of traits with
pterosaurs. Other workers may have overlooked the hind limbs, posterior torso and tail of
Longisquama, but it is clear from their data and figures, they did not provide the attention
to detail provided here (Figs. 10-11).
Despite the long list of pterosaurian synapomorphies documented by these taxa
(Fig. 12), paleontologists have been reticent to employ these taxa in phylogenetic
analyses that include pterosaurs. Some of that reticence may be due to tradition and the
widespread use of previously published matrices (even though included sister taxa do not
demonstrate gradual accumulations of derived characters). Some of that reticence may be
due to distrusting interpretation of difficult materials (even though phylogenetic
bracketing recovers the same traits). It is time for the old paradigm to shift.

55
Acknowledgements— I am indebted to J. Gallemí, J. Gomez-Alba and the staff of the
Museu de Geologia, Barcelona, for access to the holotype of Cosesaurus. D. Pruitt of The
City Museum, St. Louis, along with V. Alifanov and A. Karhu of the Paleontological
Institute, Moscow, permitted access to the holotypes of Longisquama and Sharovipteryx.
I thank A. Tintori for providing valuable comments on an earlier version of the
manuscript.
References
Alexander R.M. & Vernon A. (1975) - The mechanics of hopping by kangaroos
(macropodidae). J. Zoo. 177:265-303.
Alifanov V.R. & Kurochkin E.N. (2011) - Kyrgyzsaurus bukhanchenkoi gen. et sp. nov. a
new reptile from the Triassic of southwestern Kyrgyzstan. Paleo. J. 45: 639-647.
Andres B. (2010) - A new rhamphorhynchoid pterosaur from the Upper Jurassic of
Xinjiang, China, and the phylogenetic relationships of basal pterosaurs. J. Vert. Paleo.
30: 163-187.
Bartholomew G.A. & Caswell H.H. (1951) - Locomotion in kangaroo rats and its
adaptive significance. J. Mam. 32:155-169.
Bassani F. (1886) - Sui Fossili e sull’ età degli schisti bituminosi triasici di Besano in
Lombardia. Atti della Società Italiana di Scienze Naturali 19:15-72.
Bennett S.C. (1996) - The phylogenetic position of the Pterosauria within the
Archosauromorpha. Zoo. J. Linn. Soc. 118: 261-309.
Bennett S.C. (2008) - Morphological evolution of the forelimb of pterosaurs: myology
and function. In: Buffetaut E. and Hone D.W.E. (Eds) - Flugsaurier: pterosaur papers in
honour of Peter Wellnhofer. Zit., B28: 127-141.

56
Bennett. S.C. (2012) - The phylogenetic position of the Pterosauria within the
Archosauromorpha re-examined. Hist. Bio. iFirst article, 2012, 1-19.
Benton M.J. (1999) - Scleromochlus taylori and the origin of dinosaurs and pterosaurs.
Phil. Trans. R. Soc. London, B 354: 1423-1446.
Biewener A.A. & Blickhan R. (1988). Kangaroo rat locomotion: design for elastic energy
storage or acceleration? J. Exp. Biol. 140:243-255.
Brusatte S.L., Benton M.J., Desojo J.B. & Langer M.C. (2010) - The higher-level
phylogeny of Archosauria (Tetrapoda: Diapsida), J. Syst. Palaeo. 8: 3-47.
Brusatte S.L., Niedźwiedzki G. & Butler R. (2011) - Footprints pull origin and
diversification of dinosaur stem-lineage deep into Early Triassic. Proc. R. Soc., London,
B, 278, 1107-1113.
Buchwitz M. & Voigt S. (2012) - The dorsal appendages of the Triassic reptile
Longisquama insignis: reconsideration of a controversial integument type. Paläo. Zeit. 86
(3): 313-331.
Carrier D.R. (1987) - The evolution of locomotor stamina in tetrapods: circumventing a
mechanical constraint. Paleobio. 13: 326-341.
Chatterjee S. & Templin. R.J. (2004) - Posture, Locomotion, and Paleoecology of
Pterosaurs. Geo. Soc. America, Sp. P. 376: 1-64.
Chiappe L.M., Codorniú L., Grellet-Tinner G. & Rivarola D. (2004) - Argentinian
unhatched pterosaur fossil. Nature, 432: 571.
Chinsamy A., Codorniú L. & Chiappe L.M. (2008) - Developmental growth patterns of
the filter-feeder pterosaur, Pterodaustro guinazui. Bio. Let. 2008: 282-285.

57
Claessens L.P.A.M., O’Connor P.M. & Unwin D.M. (2009) - Respiratory Evolution
Facilitated the Origin of Pterosaur Flight and Aerial Gigantism. PLoS ONE 4(2): e4497.
online PLOS paper.
Conrad K., Lockley M.G. & Prince N.K. (1987) - Triassic and Jurassic vertebrate-
dominated trace fossil assemblages of the Cimarron Valley Region: Implications for
paleoecology and biostratigraphy. New Mex. Geo. Soc. Guidebook. 38th Field Conf., NE
New Mex. 1987. 127-138.
Dial K. (2003) - Wing-Assisted Incline Running and the Evolution of Flight. Science
299: 402-404.
Dyke G.J., Nudds R.L. & Rayner, J.M.V. (2006). Flight of Sharovipteryx mirabilis: the
world’s first delta-winged glider. J. Evo. Bio. 19: 1040-1043.
Ellenberger P. (1978) - L’Origine des Oiseaux. Historique et méthodes nouvelles. Les
problémes des Archaeornithes. La venue au jour de Cosesaurus aviceps (Muschelkalk
supérieur) in Aspects Modernes des Recherches sur l’Evolution. - In: Bons J. (Ed) -
Comptes Ren. Coll. Montpellier 12-16 Sept. 1977. Vol. 1. - Montpellier, Mém. Trav.
Ecole Prat. Hautes Etudes, Inst. Montpellier, 4: 89-117
Ellenberger P. (1993) - Cosesaurus aviceps. Vertébré aviforme du Trias Moyen de
Catalogne. Étude descriptive et comparative. Mémoire. Avec le concours de l’École
Pratique des Hautes Etudes. Lab. de Paléo. Vert.. U. Sci. Tech. Languedoc, Montpellier
(France). 1-664.
Ellenberger P. & de Villalta J.F. (1974) - Sur la presence d’un ancêtre probable des
oiseaux dans le Muschelkalk supérieure de Catalogne (Espagne). Note preliminaire. Acta
Geo. Hisp. 9: 162-8.

58
Evans S.E. (1988) - The early history and relationships of the Diapsida. In: Benton M.J.
(Ed) - The Phylogeny and Classification of the Tetrapods, Volume 1: Amphibians,
Reptiles, Birds. The Syst. Assoc. Sp. Vol. 35A. Clarendon Press, Oxford: 221-260.
Fraser N. (2006). Dawn of The Dinosaurs: Life in the Triassic. Bloomington: Indiana U.
Press. 310 pp.
Gans C., Darevski I. & Tatarinov L.P. (1987) - Sharovipteryx. A reptilian glider? –
Paleobio. 13: 415-426.
Gauthier J. (1986) - Saurischian Monophyly and the origin of birds. In Padian K. (Ed) -
The Origin of Birds and the Evolution of Flight. Mem. Calif. Acad. Sci. 8: 1-55.
Halsted L.B. (1975) – The Evolution and Ecology of the Dinosaurs. London: Peter Lowe,
116 pp.
Haubold H. & Buffetaut E. (1987). Une novelle interprétation de Longisquama insignis,
reptile énigmatique du Trias supérieur d'Asie centrale [A new interpretation of
Longisquama insignis, an enigmatic reptile from the Upper Triassic of Central Asia].
Compt. Ren. Acad. Sci. Paris 305 (serie II): 65-70.
Hedges S.B. & Thomas R. 2001. At the Lower Size Limit in Amniote Vertebrates: A
New Diminutive Lizard from the West Indies. Carib. J. Sci. 37:168-173.
Hone D.W.E. & Benton M.J. (2007) - An evaluation of the phylogenetic relationships of
the pterosaurs to the archosauromorph reptiles. J. Syst. Palaeo. 5: 465-469.
Hone D.W.E. & M. J. Benton. (2008) - Contrasting supertree and total evidence methods:
the origin of the pterosaurs. Zitt. B28: 35-60.

59
Huntington G.S. (1918) - Modern problems of evolution, variation, and inheritance in the
anatomcal part of the medical curriculum. In: Bardeen C.R., Hardesty I., Bremer J.L. &
Boyden E.A. (Eds) Anat. Rec. A.R. Liss, 14: 359-387
Jalil N.-E. (1997) - A new prolacertiform diapsid from the Triassic of North Africa and
the interrelationships of the Prolacertiformes. J. Vert. Paleo. 17(3): 506-525.
Jenkins Jr., F.A. & Goslow Jr. G.E. (1983) - The functional anatomy of the shoulder of
the Savannah Monitor lizard (Varanus exanthematicus). J. Morph. 175:195-216.
Ji Q., Ji S.-A., Cheng Y.-N., You H.L., Lü J.-C., Liu Y.-Q. & Yuan C.X. (2004) -
Pterosaur egg with leathery shell. Nature 432: 572.
Jones T.D., Ruben J.A., Martin L.D., Kurochkin E., Feduccia A., Maderson P.F.A.,
Hillenius W.J., Geist N.R. & Alifanov V. (2000). Nonavian Feathers in a Late Triassic
Archosaur. Science 288 (5474): 2202-2205.
Kellner A.W.A. (2003) - Pterosaur phylogeny and comments on the evolutionary history
of the group. In: Buffetaut E. & Mazin J.M. (Eds) - Evolution and palaeobiology of
pterosaurs: Geo. Soc. Sp. Pub. 217: 105-137.
Kellner A.W.A., Wang X., Tischlinger H., Campos D.A., Hone D.W.E, & Meng X.
(2010) - The soft tissue of Jeholopterus (Pterosauria, Anurognathidae,
Batrachognathinae) and the structure of the pterosaur wing membrane. Proc. R. Soc. B
277: 321-329.
Kim J.Y., Lockley M.G., Kim K.S., Seo S.J. & Lim J.D. (2012) - Enigmatic Giant
Pterosaur Tracks and Associated Ichnofauna from the Cretaceous of Korea: Implication
for the Bipedal Locomotion of Pterosaurs. Ichnos 19 (1-2): 50-65.

60
Lee Y.N., Azuma Y., Lee H.-J., Shibata M. & Lu J. (2009) - The first pterosaur
trackways from Japan. Cret. Res. 31, 263–267.
Libby T., Moore T.Y., Chang-Siu E., Li D., Cohen D.J., Jusufi A. & Full R.J. (2012) -
Tail-assisted pitch control in lizards, robots and dinosaurs. Nature 481(7380): 181-184.
Lü J., Unwin D.M., Deeming D.C., Jin X., Liu Y. & Ji Q. 2011. An egg-adult association,
gender, and reproduction in pterosaurs. Science 331(6015): 321-324.
Martin L.D. (2004) - A basal archosaurian origin for birds. Acta Zoo. Sin. 50 (6): 978-
990.
Milner A.R. (1985) - Cosesaurus–The last proavian? Nature 315: 544.
Nesbitt S.J. (2011) - The early evolution of archosaurs: relationships and the origin of
major clades. Bull. Am. Mus. Nat. Hist. 352: 292 pp.
Olsen P.E. (1979) - A new aquatic eosuchian from the Newark Supergroup Late Triassic-
Early Jurassic) of North Carolina and Virginia. Postilla 176: 1-14.
Peabody F.E. (1948) - Reptile and amphibian trackways from the Lower Triassic
Moenkopi formation of Arizona and Utah. U. Calif. Pub. Bull. Dept. Geo. Sci. 27: 295-
468.
Peters D. (2000a) - A Redescription of Four Prolacertiform Genera and Implications for
Pterosaur Phylogenesis. Riv. It. Paleo. Strat. 106(3): 293-336.
Peters D. (2000b) - Description and Interpretation of Interphalangeal Lines in Tetrapods.
Ichnos 7: 11-41.
Peters D. (2002) - A New Model for the Evolution of the Pterosaur Wing—with a twist.
Hist. Bio. 15: 277-301.

61
Peters D. (2007). The origin and radiation of the Pterosauria. Flugsaurier. The
Wellnhofer Pterosaur Meeting, Munich 27.
Peters D. (2009) - A Reinterpretation of Pteroid Articulation in Pterosaurs. J. Vert. Paleo.
29: 1327-1330.
Peters D. (2010) In defence of parallel interphalangeal lines, Hist. Bio. 22: 437-442.
Peters D. (2011) - A catalog of pterosaur pedes for trackmaker identification. Ichnos
18(2): 114-141.
Peyer B. (1931). Tanystropheus longobardicus Bass sp. Die Triasfauna der Tessiner
Kalkalpen. Abhandlungen Schweizerische Paläontologie Gesellschaft 50:5-110.
Peyer B. (1937) - Die Triasfauna der Tessiner Kalkalpen XII. Macrocnemus bassanii
Nopcsa. Abhand. Schweiz. Palaeo. Geo. Gesell. 1-140.
Prum R.O., Unwin D.M. & Benton M.J. (2001) - Longisquama Fossil and Feather
Morphology. Science 291 (5510): 1899-1902.
Pyron R.A. & Burbrink F.T. (2014) - Early origin of viviparity and multiple reversions to
oviparity in squamate reptiles. Eco. Let. 17: 13-21.
Reisz R.R. & Sues H.-D. (2000) - The "Feathers" of Longisquama. Nature 408 (6811):
428.
Renesto S. (1994) - A new prolacertiform reptile from the Late Triassic of Northern Italy.
Riv. It. Paleo. Strat. 100(2): 285-306.
Reynoso V.-H. (1998) - Huehuecuetzpalli mixtecus gen. et sp. nov: a basal squamate
(Reptilia) from the Early Cretaceous of Tepexi de Rodríguez, Central México. Phil.
Trans. R. Soc., London B 353: 477-500.

62
Rieppel O., Fraser N.C. & Nosotti S. (2003) - The monophyly of Protorosauria (Reptilia,
Archosauromorpha): a preliminary analysis. Atti Soc. It. Sci. Nat. Mus. Civ. Storia Nat.,
Milano 144(2): 359-382.
Sanz J.L. & López-Martinez N. (1984) - The prolacertid lepidosaurian Cosesaurus
aviceps Ellenberger & Villalta, a claimed ‘protoavian’ from the Middle Triassic of Spain.
Géobios 17: 747-753.
Senter P. (2003) - Taxon Sampling Artifacts and the Phylogenetic Position of Aves.
[dissertation]. DeKalb, Illinois, No. Ill. U.
Senter P. (2004) - Phylogeny of Drepanosauridae (Reptilia, Diapsida). J. Syst. Paleo.
2:257-268.
Sharov A.G. (1970) - A peculiar reptile from the lower Triassic of Fergana. Paleo. Z. (1):
127-130.
Sharov A.G. (1971) - New flying reptiles from the Mesozoic of Kazakhstan and
Kirghizia. Trans. Paleo. Inst., Akad. Nauk, USSR, Moscow, 130: 104-113 [in Russian].
Shcherbakov D.E. (2008) - Madygen, Triassic Lagerstätte number one, before and after
Sharov. Alavesia 2:113-124.
Snyder R. (1954) - The anatomy and function of the pelvic girdle and hind limb in lizard
locomotion. Am. Journ. Anat. 95: 1-45.
Tatarinov L.P. (1989) - [The systematic position and way of life of the problematic Upper
Triassic reptile Sharovipteryx mirabilis.] Paleo. Zh. 1989(2): 110-112. [in Russian].
Tatarinov L.P. (1994) - Terestrial vertebrates from the Triassic of the USSR with
comments on the morphology of some reptiles. In: Mazin J.-M. & Pinna G. (Eds.)
Evolution, ecology and biogeography of the Triassic reptiles. Paleo. Lomb. New Ser. 2.

63
Tatarinov L.P. (2005) - A new cynodont (Reptilia, Theriodontia) from the Madygen
Formation (Triassic) of Fergana, Kyrgyzstan. Paleo. J. 39:192-198
Unwin D.M. (2003) - On the phylogeny and evolutionary history of pterosaurs. In:
Buffetaut E. & Mazin J.M. (Eds) - Evolution and Palaeobiology of Pterosaurs. Geo. Soc.
Sp. Pub. 217: 139-190.
Unwin D.M., Alifanov V.R. & Benton M.J. (2000) - Enigmatic small reptiles from the
Middle-Late Triassic of Kyrgyzstan. In: Benton M.J., Shishkin M.A. & Unwin D.M.
(Eds) The Age of Dinosaurs in Russia and Mongolia. Cambridge: Cambridge U. Press:
177-186.
Voight S., Buchwitz M., Fischer J., Krause D. & Georgi R. (2009). Feather-like
development of Triassic skin appendages. Naturwiss. 96: 81-86.
Wang X., Zhou Z., Zhang F. and Xu X. (2002) - A nearly completely articulated
rhamphorhynchoid pterosaur with exceptionally well-preserved wing membranes and
“hairs” from Inner Mongolia, northeast China. Chin. Sci. Bull. 47(3): 226-230.
Wang X.-L. & Zhou Z. (2004) - Palaeontology: pterosaur embryo from the Early
Cretaceous. Nature 429: 623.
Wang X., Kellner A.W.A., Jiang S.-X., Wang Q., Ma Y., Paidoula Y., Cheng X.,
Rodrigues T., Meng X., Zhang J.-L., Li N. & Zhou Z. (2014) - Sexually Dimorphic
Tridimensionally Preserved Pterosaurs and Their Eggs from China, Cur. Bio.
http://dx.doi.org/10.1016/j.cub.2014.04.054
Wellnhofer P. (1991) - The Illustrated Encyclopedia of Pterosaurs, Salamander, 192 pp.

64
Wild R. (1973). Die Triasfauna der Tessiner Kalkalpen XXIII. Tanystropheus
longobardicus (Bassani) (Neue Ergebnisse). Schweizerische Paläontologische
Abhandlungen 95: 1-16.
Wild R. (1978) - Die Flugsaurier (Reptilia, Pterosauria) aus der Oberen Trias von Cene
bei Bergamo. Boll. Soc. Paleo. It. 17: 176-256.
Wild R. (1993) - A juvenile specimen of Eudimorphodon ranzii Zambelli (Reptilia,
Pterosauria) from the upper Triassic (Norian) of Bergamo. Riv. Mus. Civ. Sci. Nat. “E.
Caffi” Bergamo 16: 95-120.
Woodward A.S. (1907) - On a new dinosaurian reptile (Scleromochlus taylori, gen. et sp.
nov.) from the Trias of Lossiemouth, Elgin. Qtr. J. Geo. Soc. 1907 63:140-144.
Figure Captions
Figure 1. Fenestrasaurs to scale. A. Cosesaurus in lateral view plus pelvis, hypothetical
egg and hatchling, ventral view of pectoral girdle, dorsal view of right pes. B.
Sharovipteryx in lateral view along with pes in dorsal view, hypothetical sternal
complexes in ventral view, hypothetical egg and hatchling. C. Longisquama in lateral
view including pectoral girdle in ventral view, manual unguals in anterior view, sacral
area in dorsal view, right pes in dorsal view, hypothetical egg and hatchling. D. MPUM
6009 in lateral view along with pes in dorsal view, manus in dorsal and anterior views,
sternal complex in ventral view, hypothetical egg and hatchling. E. Kyrgyzsaurus in
lateral view, plus parietals in dorsal view, sternal complex as preserved and palate in
palatal view. Soft tissue is gray. Scale bar equals 10 cm.

65
Figure 2. The pectoral girdle of Cosesaurus aviceps. A. The in situ specimen. B.
Elements according to Ellenberger (1993) in which the clavicles are aligned transversely,
the scapulae are posteriorly elongated and provided with a disc-like acromion process
anteriorly, the coracoids are fused medially to form a ventral plate and a thick-rimmed
keel is present medially. C. Interpretation of elements according to Peters (2000a) in
which the clavicles are aligned transversely, the scapulae are short and crescentic, the
coracoids are fused medially and a thick-rimmed interclavicle keel is present. D. Present
reinterpretation in which the clavicles are aligned transversely, the scapulae are
posteriorly elongated without an acromion process, the coracoids are quadrant-shaped,
the interclavicle is flat and cruciform, the sternum is flat, broader than the interclavicle
and dorsal to it. Abbreviations: Ac, acromion process of scapula; Cl, clavicle; Co,
coracoid; Hu, humerus; Keel, interclavicle keel; Ic, interclavicle, Pp, prepubis; Pt,
pteroid; Sc, scapula; and St, sternum. Black dots are preserved bubbles. Scale bar = 1 cm
divided into 10 mm.
Figure 3. Comparison of pectoral girdles. A. The basal lizard, Huehuecuetzpalli mixtecus
(IGM 7389 and IGM 4185, Instituto de Geología, Universidad Nacional Autónoma de
México, Ciudad Universitaria, Mexico; modified from Reynoso 1998). Here the clavicles
and humerus of the larger specimen, IGM 7389, have been scaled down to fit the pectoral
girdle of the smaller specimen, IGM 4185. B. The tanystropheid, Macrocnemus bassanii,
(modified from Peyer 1937). C. The fenestrasaur, Cosesaurus aviceps. D. Longisquama
insignis, traced in situ by Sharov (1970, above) and traced here (below). Note the
taphonomic displacement of the sternal complex to a position that returns the clavicles to

66
the base of the neck, their plesiomorphic position in tetrapods. This has been a source of
confusion regarding the affinity of these clavicles to the furcula in birds (Jones et al.
2000). E. Sternal complex of MCSNB 8950 (Museo Civico di Scienze Naturali,
Bergamo; modified from Wild 1993). Abbreviations as in figure 1. Scale bars equal 3
mm.
Figure 4. Cosesaurus pelvic region. An earlier misinterpretation of the stem shape arising
from the anterior ilium (Peters 2000a) is corrected here with the identification of two
disarticulated and fenestrated prepubes, similar to those in basal pterosaurs. Scale bar
equals 2 mm.
Figure 5. Sharovipteryx mirabilis in situ. A. Counterplate. B. Traced image of bones and
soft tissues (center). C. Plate. Counterplate flipped to match plate. Note: The darkened
middle of the plate is replaced with filler. Abbreviations: 1, 2, 3, 4, manual digits; Ca,
caudal vertebrae; Ce, cervical vertebrae; Co, coracoid; DP, dorsal plumes; EDM,
extradermal membrane; Fe, femur; Hu, humerus; Pt, pteroid; R/U, radius/ulna; Sc,
scapula; Sk, skull; T/F, tibia/fibula; Ur, uropatagium. Scale bar equals 1 cm. See figure 6
for skull details.
Figure 6. Sharovipteryx mirabilis plate, close-up of skull and cervicals in situ. A. Tracing
of the plate. B. The plate. These interpretations correct earlier mistakes by Peters (2000a).
Note the impression of a wasp-like insect in the left side of the skull. Abbreviations: Bs,
basisphenoid; De, dentary; Ect, ectopterygoid; Fr, frontal; Hy, hyoid; Ju, jugal; La,

67
lacrimal; Mx, maxilla; Na, nasal; Oc, occipital bones; Pal, palatine; Par, parietal; Pmx,
premaxilla; Po, postorbital; Pof, postfrontal; Prf, prefrontal; Pt, pterygoid; Qj,
quadratojugal; Qu, quadrate; Sq, squamosal; St, stapes; Vo, vomer; (Scale bar equals 1
cm.
Figure 7. Sharovipteryx reconstructed. A. Dorsal view, left hind limb extended to gliding
configuration, hyoids laterally extended creating neck skin strakes, vestigial fore limbs
extended creating canard wings. Vertical arrow indicates hypothetical center of gravity.
A1. Hypothetical sternal complex in dorsal view, based on Cosesaurus (left) and MPUM
6009 (right) both distorted to match rib articulations in Sharovipteryx. B. Dorsal view of
skull and neck, hyoids not laterally extended. C. Skull in dorsal, occipital, lateral and
palatal views. D. Lateral view, bipedal configuration, note dorsal frill/plumes. Vertical
arrow indicates center of gravity over the toes. E. Pelvis doubled in size in lateral view.
F. Forelimb doubled in size. Unguals shown in lateral and anterior views. Trailing
membrane is hypothetical based on phylogenetic bracketing between Cosesaurus and
pterosaurs. G. Pes doubled in size. Interdigital membranes between digits 1 and 3 are
hypothetical. Continuous PILs (parallel interphalangeal lines, Peters 2000b) indicate
phalanges that operate in sets during flexion and extension. Scale bar equals 1 cm.
Figure 8. Sharovipteryx left pes in ventral view. Left: In situ tracing. Note distal elements
have drifted proximally. Right: The same reconstructed, PILs added. Interdigital
membrane between digits 1-4 based on 4-5 membrane. Note three distal tarsals (centrale,

68
dt3 and dt4), plus an unfused astragalus and calcaneum comprise the tarsus, producing a
simple hinge ankle joint. Scale bar equals 1 cm.
Figure 9. Three views of Sharovipteryx on a vertical surface. Here is how Sharovipteryx
might have clung to and climbed on a seed fern trunk.
Figure 10. Longisquama insignis (PIN 2584/4, Sharov 1970). A. Plate. B. Traced image
of right forelimb, soft tissue not included. C. Traced image of left forelimb and
associated soft tissue. D. Tracings by Sharov (1970). Note the soft tissue trailing the
forelimbs, anterior to the elbow, along the dorsal spine and filling the throat area. All are
confirmed here. E. Current tracing of skeletal and soft tissue elements. Right
appendicular elements are lightened for clarity. Sharov’s lowest plume is actually the
hind limb. Plume bases are beneath the left metatarsus. F. Reconstruction of left manus.
G. Reconstruction of matching right manus. H. Reconstruction of pedal elements. Note
pedal 3.2 (in gray) is hidden beneath pedal 2.2 in situ. Abbreviations: 1.1, first manual
phalanx of the first digit and other phalanges follow this pattern; circled numbers, pedal
digits; reverse numbers, sacral vertebrae, other numbers, presacral and caudal vertebrae;
Co, coracoid; DP, dorsal plumes; DR, dorsal ribs; Fe, femur; Hu, humerus; Mc,
metacarpal; Pl, pelvis; Pp, prepubes; Pt, pteroid; Pxc, preaxial carpal; Ra, radius; Re,
radiale; S, sternal complex; Sc, scapula; Sk, skull; T/F, tibia/fibula; Ue, ulnare; Ul, ulna;
Ur, uropatagium. Scale bar equals 1 cm. See figure 3 for skull details and figure 1 for
reconstruction.

69
Figure 11. Longisquama insignis skull. A. According to Sharov (1970). B. According to
Martin (2004). C. According to Senter (2003). D. According to Peters (2000a). E.
Tracing of the plate, skull elements ghosted to clarify palatal and occipital elements. F.
Tracing of the plate, palatal and occipital elements omitted for clarity. G. Skull
reconstructed, dorsal view; H. Skull reconstructed, lateral view; I. Skull reconstructed,
palatal view. J. Fused occipital bones. These interpretations correct earlier mistakes by
Peters (2000a). Abbreviations: Bs, basipterygoid process of the basisphenoid; De,
dentary; Ep, ectopalatine (ectopterygoid + palatine); Fr, frontal; Hy, hyoid; Ju, jugal;
La, lacrimal; Mx, maxilla; Na, nasal; Oc, occiput; Par, parietal; Pmx, premaxilla; Po,
postorbital; Pof, postfrontal; Prf, prefrontal; Pt, pterygoid; Qj, quadratojugal; Qu,
quadrate; ScR, sclerotic ring; Sq, squamosal; Vo, vomer.
Figure 12. Accumulation of pterosaurian traits in precursor taxa. The following taxa
document a gradual accumulation of pterosaurian traits unmatched by any series of
archosauriforms. Letters reference nodes. Bootstrap scores indicated. See text for list of
gradually accumulating pterosaurian traits at each node.