The dorsal appendages of the Triassic reptile Longisquama insignis: reconsideration of a controversial integument type
Elongated skin projections of the reptile Longisquama insignis from the Triassic of Kyrgyzstan are preserved as imprints on the only skeletal specimen and on seven additional pairs of fossil slabs and counter-slabs from the same locality and horizon. The integumentary structures became a matter of debate when they were assessed as “non-avian feathers” homologous to avian feathers. Conflicting interpretations of their morphology and relationship to other appendage types arose from the ambiguity of the fossil skin impressions. On the basis of comparative description of the individual morphology of all yet known Longisquama specimens we address aspects of taphonomy, development, and function and define to what extent Longisquama’s appendages share characteristics of avian vaned feathers. We explain the existing feather similarity by their development from a filamentous primordium and a complex sequence of individual processes, some of which are reminiscent of processes observed in feather development. Such an interpretation is in agreement with a set of homologous mechanisms of appendage morphogenesis in an archosauromorph clade including Longisquama and feather-bearing archosaurs but does not necessarily require that the appendages of Longisquama themselves are feathers or high-level feather homologues.
KeywordsDiapsida Fossil skin Feather development Deep homology Display structure Madygen Formation
Längliche Hautvorsprünge des Reptils Longisquama insignis aus der Trias Kirgisistans sind in Zusammenhang mit dem einzigen Fossilskelett sowie isoliert auf der Platte und Gegenplatte von sieben weiteren Exemplaren aus demselben Fundhorizont und von derselben Fundstätte erhalten. Die Hautstrukturen wurden Teil der Debatte um den Ursprung der Federn, als manche Bearbeiter sie als Federn eines Nichtvogels, die homolog zu Vogelfedern seien, auffassten. Im Widerspruch zueinander stehende Interpretationen zu ihrer Morphologie und Vergleichbarkeit mit anderen Typen von Hautanhängen waren besonders auch der Mehrdeutigkeit der fossilen Hauteindrücke geschuldet. Auf der Grundlage eines Vergleichs der Morphologien aller bekannten Longisquama-Anhänge nehmen wir zu Fragen der Taphonomie, Entwicklung und Funktion der Anhänge Stellung, ebenso zu der Frage, inwieweit sie den Vogelfedern mit Federfahne glichen. Wir erklären die bestehende Federähnlichkeit der Longisquama-Anhänge mit der Entwicklung aus einer filamentartigen Anlage und einer komplexen Abfolge von Einzelvorgängen, von denen manche Entwicklungprozessen heutiger Federn geglichen haben könnten. Solch eine Interpretation ist mit der Hypothese in Einklang, dass Mitglieder einer Archosauromorpha-Klade, die Longisquama und federtragende Archosaurier einschließt, eine Reihe homologer Enwicklungsmechanismen für Hautanhänge besitzen, erfordert jedoch nicht, dass die Longisquama-Anhänge selbst Federn oder auf einem hohen Level homolog mit Federn sind.
SchlüsselwörterDiapsida Fossile Haut Federentwicklung Tiefenhomologe Zurschaustellung Madygen-Formation
The fossil-bearing strata of the Madygen Formation have been investigated geologically for more than 70 years (short summaries of the historical background have been given by Dobruskina 1995 and Shcherbakov 2008a). A Ladinian to Carnian age has been derived by Dobruskina (1994, 1995) on the basis of macrofloral remains. Assignment to the Middle or earliest Late Triassic is concordant with the composition of the diverse entomofauna (Shcherbakov 2008a, b). Although the sedimentary succession of the Madygen Formation covers various depositional systems of an intermontane lake basin, including alluvial fans, rivers of a wet alluvial plain, lacustrine deltas, and a large freshwater lake, preservation of skin impressions in tetrapod fossils is restricted to marginal lake deposits which are represented by grey and brownish siltstones and claystones. Patches of interspersed sand, fossilized rhizomes with adherent substrate, and mass occurrences of bryophytes suggest that soft-tissue preservation took place in a lacustrine delta-front sub-environment.
The two small diapsids became known for their peculiar morphology: Sharovipteryx mirabilis Sharov 1971 (renamed by Cowen 1981, because the genus name “Podopteryx” introduced by Sharov was found to be occupied) was an early limb-supported glider with an exceptionally large uropatagium and probably an archosauromorph (Gans et al. 1987; Unwin et al. 2000; Dyke et al. 2006); Longisquamainsignis Sharov 1970 had unusual integumentary structures, which are the object of this study. The only skeletal specimen features a series of seven elongated skin projections attached along the animal’s back in a fan-like arrangement. Further fossil specimens show skin structures unrelated to skeletal remains. The nature of these appendages and their attribution to one of the recent integument types has been a matter of much discussion (Sharov 1970; Reisz and Sues 2000; Jones et al. 2000; Prum 2001; Unwin and Benton 2001; Martin 2004). In an earlier study (Voigt et al. 2009) we introduced recently recovered material of the Longisquama appendages and focussed on feather-like anatomical features, their interpretation in a specific developmental model, and the question of whether an aerodynamic function of the appendages is plausible.
Here we follow as approach which goes further: to form a broader basis for discussion of appendage characteristics we document the morphological range of all yet known Longisquama appendage specimens. This is reasonable, because all exemplars are fragmentary and differ substantially in appearance. They cover a range of sizes (Fig. 2) and possibly varying morphology and developmental stages. The mode and delicateness of preservation differ, contributing to their ambivalent perception. On the basis of all documented facts we discuss evidence for:
structural differentiation; and
We also address the question of whether and how Longisquama’s integumentary structures might have been linked to the evolution of elongated skin appendages in archosaurs.
Indications for function and functional analogues in extant reptiles are also briefly discussed.
FG, Geological Institute of TU Bergakademie Freiberg, Germany; PIN, Palaeontological Institute of the Russian Academy of Sciences, Moscow, Russia.
A, anterior; ab, anterior bar; af, anterior flange; al, anterior lobe; co, rib; drl, distal rear lobe; ml, middle lobe; mx, middle axis; ors, oblique rear lobe seam; P, posterior; pb, posterior bar; pl, posterior lobe; ru, ruga; sp, spine impression; tr, appendage transition zone.
The type material of Longisquama insignis comprises the only skeletal specimen PIN 2584/4 as the holotype and, as paratypes, four specimens with isolated appendage fragments—PIN 2584/5 and PIN 2584/7 are single exemplars, PIN 2584/6 has two appendages, and PIN 2584/9 a series of five appendages. Recent material was recovered from the type locality in 2007: FG 596/V/1 to 596/V/3 are single exemplars with lengths of 28.9, 3.7, and 3.6 cm, respectively. FG 596/V/1 is more than twice as large as the longest fragments of the type series (Fig. 2).
General appearance and homogeneity of the appendage structure
The general appearance of the appendages has been characterized by Sharov (1970) and further defined in subsequent approaches (Haubold and Buffetaut 1987; Jones et al. 2000; Reisz and Sues 2000; Voigt et al. 2009): According to the most complete appendage fossils in PIN 2584/4 (appendages nos. 4–6, numbering from anterior to posterior), PIN 2584/7, PIN 2584/9 (nos. 1, 3, 4), and FG 596/V/1, the outline is hockey-stick-like with a relatively long and narrow proximal section and a short, wide and curved distal section. In the holotype specimen the convex margin of all appendages is on the cranial side and the concave margin is on the caudal side. Similarly, all other appendages are curved and we refer to the right slab of each specimen as the one with the appendage(s) dextrally curved and the left slab as the one with the sinistrally curved appendage(s).
Sharov’s (1970) observation that the narrow proximal and broad distal sections of the relatively complete appendage fossils are morphologically disparate—the proximal section has smooth cranial and caudal lobes enclosing a corrugated middle lobe whereas the distal section has two asymmetric corrugated lobes separated by a distinct middle axis—has been confirmed by later studies, even though interpretation of these structures has varied substantially (Jones et al. 2000; Voigt et al. 2009). Sharov found that this conspicuous proximal–distal disparity is a result of the distal tapering of the (proximal) posterior lobe. Accordingly, the two distal lobes correspond to the anterior and middle lobes of the proximal section. Some of the appendage fragments cover either the proximal section (PIN 2584/4, nos. 1–3, 7; PIN 2584/6; PIN 2584/9, no. 2) or the distal section (PIN 2584/5; PIN 2584/9, no. 5; FG 596/V/2; FG 596/V/3) and display the respective surface morphology.
Not all fossils are simple imprints of either a left or a right appendage surface—Reisz and Sues (2000) clarify that a continuous sedimentary core separating the two sides of a former voluminous appendage is preserved in some specimens (PIN 2584/5; PIN 2584/7; see also FG 596/V/1 and FG 596/V/3). Moreover substantial thinning of the sediment filling from proximal to distal occurs in the single appendage specimens PIN 2584/7 and FG 596/V/1.
Anchoring and base of the appendage
To what degree the former left and right surfaces of the proximal appendage corresponded can be inferred from FG 596/V/1, in which a relatively thick sedimentary core is preserved. In this specimen the tripartite structure, including the complex relief of the middle lobe, occurs mirror-inverted on the left and right sides—rugae and interstices are synchronously arranged in a convex–convex/concave–concave fashion (Fig. 4a–d; Voigt et al. 2009). A small difference in the anterior and posterior lobe widths between the left and right imprints results in an anterior–posterior mismatch of the left and right middle lobes.
The relief of the proximal section varies in distinctness: In PIN 2584/7 the proximal surface is almost untextured and on the right plate of the holotype in some of the proximal appendage parts the typical corrugated morphology seems to be concealed as if by a cover layer. Depicted and discussed by Jones et al. (2000), the third appendage of the holotype displays most clearly the demarcation between the uncovered imprint and a more proximal part where a thin proximal cover layer only weakly reproduces the underlying tripartite structure. These features have been regarded as indicative of the former existence of an envelope (Jones et al. 2000; Voigt et al. 2009), but given the preservation of thin sediment cores in other specimens, the cover layer might instead be the product of a similar taphonomic process and represent an infill rather than an additional envelope.
The length of the proximal section has a documented range between 7.5 cm in appendage no. 6 of the holotype and over 20 cm in FG 596/V/1. The latter exemplar has also a somewhat wider proximal section (0.55 cm) than the narrowest holotype appendages whose anterior–posterior width is 0.35 cm.
The transition from a narrow proximal to a broad distal section is a principal feature of the Longisquama appendages (Sharov 1970; Reisz and Sues 2000; Voigt et al. 2009). By using length and angle measurements we have demonstrated for the recently discovered specimen FG 596/V/1 that the proximal to distal rise in anterior–posterior width is correlated with a increase in curvature (in the sagittal plane) and in the width of the anterior lobe, with an increase in the density of middle to rear lobe rugae, a change in the orientation of rugae, and with thinning and tapering of the posterior lobe (Voigt et al. 2009; see also Fig. 4e).
Summary of appendage characteristics observed in eight Longisquama insignis specimens
Tapering base contacting thoracal spine
Middle lobe: ruga long axes oblique
Sheath-like outer membrane
Change in the orientation of rugae
Seam crossing the rear lobe
Corrugated longitudinal lobes separated by middle axis
1, 4, 5
Left/right asymmetry, spine imprints
Apical end preserved
Not parameterized by Voigt et al. (2009) further morphological change occurs in the anterior lobe, which has a more differentiated appearance in the distal section: Besides the emergence of rugae, resembling the rear lobe corrugation pattern, a broadening edge (anterior flange, Figs. 4a, b, 5b, 6) splits off in the distal third of the proximal section and runs up to the appendage’s apex.
Related to the proximal–distal transition in the appendages PIN 2584/4, no. 4, 2584/5, 2584/7, and FG 596/V/1, a narrow, slightly curved seam runs from the posterior margin of the proximal section to the middle axis of the distal section, thereby crossing the rugae of the rear lobe. Furthermore in PIN 2584/7, the transitional zone is marked by two transverse cuts and a mismatch is observed in the left and right appendage surfaces (Fig. 5b).
The distal appendage section is asymmetrically bipartite. The anterior lobe resembles the slightly broader rear lobe in that they both display a pattern of transverse corrugation. The asymmetrical partition is reversed close to the apex of PIN 2584/5, 2584/7, 2584/9, nos. 3, and 5, in which the rear lobe thins out continuously while the anterior lobe tapers off only at the very end. A smooth anterior flange, separated from the broader corrugated part of the anterior lobe by a longitudinal incision, is observed in most exemplars covering the distal section, namely PIN 2584/5, PIN 2584/7, PIN 2584/9, nos. 1, 3, 4, and 5, FG 596/V/1, and FG 596/V/3. Left and right slab reliefs in PIN 2584/5 show that the flange is biconvex. Only on the left imprint of this specimen a narrow edge delimits the anterior flange cranially.
As much as the appearance of the middle axis impressions, the corrugation pattern varies among specimens. In PIN 2584/5 the two sets of anterior and rear lobe rugae have similar frequencies and two individual rugae appear to form pairs (Fig. 7). However, for the other specimens this cannot be confirmed—mostly the anterior rugae are less distinct (as in PIN 2584/9) or not well matching the posterior rugae pattern (as in PIN 2584/7). The orientation of the anterior lobe rugae is slightly anterodistal or perpendicular to the middle axis trend in PIN 2854/5, 2854/7, 2854/9, and FG 596/V/1. By contrast the rear lobe rugae mostly display a rather pronounced posterodistal trend and only in PIN 2584/5 do they have a sigmoidal shape and, close to the middle axis, a perpendicular trend. Given that the ends are identified correctly, an unusual anteroproximal/posteroproximal orientation of anterior/rear lobe rugae occurs in the small fragment FG 596/V/3 (Fig. 8b).
A series of characteristic features is exclusively seen in PIN 2584/5, arguably because of the exceptional preservation of the filigree surface morphology: The density of the particularly long and narrow rugae is high, rising up to 5 per mm close to the apical end, whereas it does not exceed 1.5 per mm in other exemplars. In the rear lobe of this specimen the relief of the rugae is shallow close to the middle axis and at the caudal margin and more pronounced in between. From the middle axis to the caudal margin, which is formed by a continuous apically thinning edge, branching and joining of the rugae occur (Fig. 7). In the distalmost anterior lobe corrugation seems to be irregular with bent rugae obliquely crossing and cutting off other rugae (Jones et al. 2001). Comparing the slab and counter-slab of PIN 2584/5 rugae on the left and right surfaces correspond in a convex–convex/concave–concave fashion, similar to the pattern seen in the proximal FG 596/V/1. Accordingly more voluminous segments and very thin intervening sections would have alternated in the former appendage (Voigt et al. 2009).
The most striking feature in PIN 2584/5 is a series of approximately 20, up to 1.5 mm long, distinct spine-like impressions on the right anterior lobe. The impressions’ broad end is adjacent to the middle axis, from which they diverge anterodistally (Fig. 7a, b). The periodic onset of the imprints is synchronous with the anterior lobe corrugation pattern. On a small fragment of the sedimentary core and, less distinct, on the left slab of PIN 2584/5, structures corresponding to the spine-like impressions occur: On the core surface they constitute short off-branching grates attached to the middle axis; on the left slab short rounded impressions and shallow grates occur close to the groove of the middle axis. Although the rugae of other appendages can have a thin and strut-like appearance, as in the distal PIN 2584/7, none has spine-like imprints in addition to a corrugation pattern. It has to be noted that the somewhat tendential terms “spine-like impression” and “middle axis (impression)” refer to peculiarities of the skin imprint topography that can be clearly distinguished from the transverse corrugation pattern (“rugae”) but these structures are not necessarily the marks of actual spines or an actual middle rod.
As the holotype is the only specimen with skin imprints associated with a skeleton, the possible arrangement of appendages in the living animal may be inferred from the configuration in the holotype: The first six appendages are attached to the thorax with nos. 1–4 contacting the vertebral column. Arguably they are adjacent to the spines of subsequent thoracal vertebrae (but vertebral features are difficult to discern (Fig. 3). No. 7 seems not to be in a natural position but tilted off caudoventrally. All holotype appendages lack the apical end, so the question whether they shorten caudally cannot be answered definitely. The lengths of fragments 1–7 are 7.6, 9.7, 12.2, 14.2, 7.6, 10.2, and 2.8 cm, respectively, with the proximal–distal transition occurring after approximately 12.7 cm in no. 4 and after 7.5 cm in no. 6. The straight or only slightly curved proximal sections of appendages 1–6 are fanned out caudodorsally, diverging at angles of 20, 10, 8, 13.5, and 23° (from the anteriormost appendages).
Apart from the holotype, specimen PIN 2584/9 comprises several appendages whose regular association is probably not because of a taphonomic effect. It features a series of five sequentially overlapping appendage fragments with the apical ends of nos. 3, 4, and 5 and probably also of the first appendage preserved. Apparently the anterior appendages are longer with their bipartite distal sections ending up higher than in their posterior neighbours (Fig. 6). The overlapping proximal sections are parallel or very acute-angled and only the different onsets of the strongly curved distal sections result in divergence of the distal ends.
Aspects of preservation
A flattened sediment core reproducing the surface morphology of a former three-dimensional skin appendage, but lacking evidence of fossil substance is unusual. Within the sample of feather fossils from 39 localities analysed by Davis and Briggs (1995), Archaeopteryx feathers from Solnhofen were the only ones preserved as sediment casts. The preservation of overlapping appendage surfaces in PIN 2584/9 is reminiscent of the Berlin specimen of Archaeopteryx. Possibly a model of early cementation upon the organic structures, as discussed by Rietschel (1985) for Archaeopteryx feathers, is an appropriate explanation for Longisquama appendage fossils also (see also “death mask” preservation discussed by Briggs 2003). Colour preservation of many insect wings (Shcherbakov 2002, 2011) suggests that in the Longisquama specimens organic remnants of skin tissue might be present also but not visible with an optical microscope.
In the studied specimens several examples of deviant morphology occur and can be explained taphonomically: The thin seam connecting the caudal margin and middle axis in some specimens (Figs. 4, 5b) may be because of compression of a tubular appendage where it becomes increasingly curved—in the transition zone from the proximal to the wide and flat distal appendage (Voigt et al. 2009). Furthermore the transverse cuts modifying the outer appendage shape in PIN 2584/7 may be a consequence of destruction during transport or diagenetic compression. Orientation and sigmoidal shape of the rear lobe rugae and the muddled appearance close to the apex in PIN 2584/5 may have been caused by shearing and relative movement between different parts of the appendage before its complete decay.
The depositional environment of the fossiliferous Sharov locality shales can be characterized as marginal lacustrine with an accumulation of non-aquatic fossils because of a nearby inflow of a feeder river. In this context the regular association of appendages in PIN 2584/6 and PIN 2584/9 suggests that their bases were still connected by tissue before their deposition and embedment.
Why the appendages of Longisquama are not plant remains
The arrangement is regular; there is no sign that the association was coincidental. Except for the caudalmost impression, all elongate imprints set in along the dorsal thorax, some are more strongly curved and tapering close to the vertebral column. They do not appear to continue below or above the skeleton.
Type of preservation: There is no indication of carbonaceous preservation, otherwise occurring for many plant fossils from the Longisquama type locality. Moreover, plant organs preserved three-dimensionally, with the impressions of the outer surfaces separated by a core of fine-grained sediment, are not known from macrofloral remains of the Madygen Formation.
We know of no plant organ which matches all the principal characteristics of the appendages. Mesenteriophyllum kotschnevii (Sixtel 1962), an endemic plant fossil of unknown affinity, bears some similarity, because it has a folded surface; it lacks the characteristic curvature, hockey-stick-like outline, and asymmetric partition, however. If their belonging to the skeleton is questioned, a plausible alternative, explaining what the elongate projections could be, would add weight to these claims.
For fossil integuments which cannot be synonymised with recent integument types the question whether they grew from a cylindrical epidermal invagination comparable with the follicles of avian feathers and mammalian hair is not easily answered. In the holotype the proximal appendage sections are generally flat as in the other specimens, but their bases display a slight convexity above the tapering proximal end (Fig. 3). This led to the hypothesis that they were tubular before diagenetic compression (Jones et al. 2000; Reisz and Sues 2000; Prum 2001; Voigt et al. 2009). Furthermore, the thickening of the sediment core towards the proximal end of the single appendage specimens PIN 2584/7 and FG 596/V/1 is in agreement with a tubular rather than flat appendage base.
However, a tubular base is also known from recent non-follicular appendages, for example the beard bristles of the wild turkey and the cylindrical dorsal scales of some squamates (Lucas and Stettenheim 1972; Sawyer et al. 2003b; Wu et al. 2004; Chang et al. 2009) but neither has the degree of differentiation observed in feathers and in the appendages of Longisquama (see below). The proximity of the appendage bases to the thoracic vertebrae could be regarded as indicative of the appendages’ deep anchoring within the dermis, justifying their interpretation as having descended from a follicle (Jones et al. 2000; Martin 2004; Voigt et al. 2009; this approach). If the function of the appendages required support by muscles or ligaments a follicular invagination as an attachment site would have been advantageous.
Patterns of differentiation
Differentiation along the proximal–distal axis has been demonstrated to be a marked feature of the Longisquama appendages, materializing in the conspicuous change from the proximal to the distal section (Sharov 1970; Reisz and Sues 2000; Voigt et al. 2009; this approach). Only the transition from feather vane to calamus seems to represent a likewise extensive changeover (Lucas and Stettenheim 1972; Yu et al. 2004; Alibardi 2007) and in fact the sections have been synonymised with vane and calamus (Jones et al. 2000). A proximal–distal polarity also occurs in the development of less complex integumentary structures, for example elongated reptilian scales and mammalian hair (Chuong et al. 2000; Wu et al. 2004).
Anterior–posterior differentiation is a common feature in amniote integuments (Wu et al. 2004). In the Longisquama appendages such differentiation is indicated by the curvature in the sagittal plane, by the anterior to posterior longitudinal partition which can be asymmetric as in the two distal lobes, but also by the oblique proximal corrugation pattern. Unlike the shaft of a variety of avian feathers, the middle axis does not form the anterior end of the distal Longisquama appendage, which is instead occupied by the flange of the anterior lobe.
Reviewing developmental aspects of recent and fossil skin appendages, Wu et al. (2004) list the appendages of Longisquama as having determinate anterior–posterior and proximal–distal axes along which differentiation occurs, but no right–left asymmetry. The latter is a characteristic feature of remiges linked to the formation of wings (Prum 1999; Chuong et al. 2000; Prum and Brush 2002), and usually not occurring in other appendage types. However we find that PIN 2584/5 has left–right asymmetry dissimilar to the state in wing feathers, represented by the distinct middle axis reliefs of the left and right imprints (Fig. 7). Moreover the spine-like impressions on the right and left plates differ in shape and distinctness.
We interpret the “middle axis imprints” and related “spine impressions” in PIN 2584/5 as indicative for an actual branched internal frame, consisting of an axial rod and rigidly attached spines (Voigt et al. 2009). Surrounded by an outward membrane, a branched structure of this kind would not be in conflict with the observed appendage outline which is continuous and rather smooth (leading to the controversial interpretation that appendage branches were outwardly fused as in some feathers; Jones et al. 2000, 2001). Apart from the particular morphology in PIN 2584/5, we find likely that the complex surface reliefs found in other specimens are markings of a differentiated internal structure also. Accordingly, the conspicuous corrugations may reflect separate internal chambers or internal layers of changing thickness, which are enclosed by the more flexible outer membrane.
As an alternative to our hypothesis of a differentiated internal structure including a branched frame, the structuring of the appendage surface could also reflect material or thickness differences in a single, extensively folded outer membrane and not the markings of internal tissues. Whereas our interpretation implies a clear differentiation between internal and external structures, the idea that the complete appendage topography only reflects morphological variation in an outward membrane still leaves the possibility of internal–external differentiation within such a membrane.
If the smooth basal cover in some holotype appendages, introduced as a “feather sheath” by Jones et al. (2000), truly represents no part of the usual core-like sediment filling but an additional outer cover layer, this would be another indication of differentiation of the appendage into successive layers.
Is there compelling evidence for the appendages’ mode of development?
The obvious elongate shape which is more pronounced than in elongated lizard scales indicates the predominance of unidirectional growth. In extant appendage types, for example feathers and hair, this kind of growth is sustained by a zone of cell proliferation at the base of the appendage (Lucas and Stettenheim 1972; Wu et al. 2004; Dhouailly 2009). Assuming the same mechanism in Longisquama, the distalmost appendage parts would be the oldest.
The marked proximal–distal differentiation, including a transition in appendage width, curvature, and surface morphology (Voigt et al. 2009), is a strong indication of a differentiation zone at the base of the appendage (above the zone of cell proliferation). Arguably the rachigenic and ramogenic zones within the multilayered epidermal collar of the feather follicle are the only feasible extant analogues of such a differentiation zone (Lucas and Stettenheim 1972; Yu et al. 2002, 2004; Yue et al. 2006).
Given the appendages’ elongation, their high-angle divergence from the body surface and considering the likely presence of basal proliferation and differentiation zones, any but a cylindrical base would be highly unusual considering the type of attachment known from elongated skin structures in extant amniotes—whether or not they are follicular (i.e. lizard frill scales, turkey beard bristles, feathers, and hair; see review on amniote appendages by Wu et al. 2004).
In the early phase of morphogenesis a filamentous primordium develops as a projection of the epidermis (Fig. 9a)—this process is similar to the development of the feather bud (Chuong et al. 2000; Widelitz et al. 2003). Thereafter the proliferation and differentiation zones of the epidermal collar and possibly a follicular structure are established.
Moulding of the later distal appendage section (Fig. 9b): analogous to barb and rachidial ridge formation in the feather follicle (Lucas and Stettenheim 1972; Yu et al. 2002, 2004) morphological features preserved in the fossil specimens as anterior flange, “middle axis”, “spine-like impressions”, and rugae develop within the differentiation zone of the epidermal collar. These structures are either part of the sculpture of an enveloping membrane or form below such an envelope in a distinct interior collar layer (hypothesis discussed in Voigt et al. 2009). A mechanism explaining how an appendage with a closed hull membrane can have a width exceeding the diameter of its constricted cylindrical base is the shaping of a longitudinal invagination as a step in the differentiation process (phases in Fig. 9b, c). The distal section unfolds when its formation is complete.
Reorganization in the differentiation zone and formation of the appendage’s transitional section (Fig. 9c). The shaping of an invagination in the distal differentiation zone of the epidermal collar ceases and thereby the basal cylindricity is inherited by the appendage. Meanwhile the differentiation mechanism leading to the corrugated appearance continues.
Moulding of the proximal appendage section (Fig. 9d). We find likely that the tripartition observed in the left and right plates of the fossil specimens is the product of the diagenetic flattening of a four-sector cylinder which forms after a specific change in the differentiation process. The anterior and posterior sectors plus either left or right sector give the appendage a tripartite appearance when being laterally compressed (Voigt et al. 2009). This interpretation is in agreement with the slight mismatch between left and right middle lobes in one specimen (FG 596/V/1). The formation of the tapering proximal end completes the appendage development.
Closer similarity of the proximal appendage section to the feather calamus was suggested for the first time by Sharov (1970) and subsequently revived in more recent reconstructions (Jones et al. 2000; Martin 2004). Accordingly the proximal rugae are not regarded as structures of the outward cylinder (as a continuation of the distal outward membranous envelope) but are impressions of distinct inner pulp chambers, separated by pulp caps. In feathers, these pulp caps mark the stepwise retraction of an evagination of dermal tissue, the papilla (Lucas and Stettenheim 1972; Yu et al. 2004). We find the shape and orientation of the rugae in some Longisquama specimens dissimilar to that of calamus internal structures, but regard a pulp-cap-like cyclic shaping of the proximal section reflected by its corrugation pattern as possible. Apart from the relatively frequent occurrence of isolated single appendage fossils we find no indication for cyclic shedding and renewal of the appendages.
We are aware of the questionable falsifiability of elaborate hypotheses about developmental processes in extinct animals and offer the following suggestions for hypothesis testing. The assumption of a distal-to-proximal growth mode would be questioned if the appendage proportions of a sufficiently large sample of specimens are found to show no clear indication of allometric growth. That a defined anterior–posterior axis existed could be disproved by a skeletal specimen with appendages attached whose features have a conflicting or non-regular orientation. On the basis of the hypothesis of a cylindrical proximal section and a complex internal structure we expect to find a range of distinct morphology, depending on sediment grain size and degree of compression. In fact we know different states of compression from Madygen shark egg capsules (Fischer et al. 2011) and await a similar series for the Longisquama appendages. Arguably advanced documentation techniques, for example high-definition surface scans of the different appendage reliefs, could help to discern structures which represent superficial features on either side of the appendage from those which possibly indicate former internal features enhanced by compression. Our assumption of distal appendage formation by longitudinal invagination implies that the width of the distal section should not account for more than the circumference of the proximal appendage section—in the event of a much greater distal width this hypothesis would be shown to be false. Furthermore, if experimentation on the morphogenesis of epidermal structures in recent amniotes reveals an alternative mechanism how sheetlike elongated appendages with a prominent middle axis relief can develop from a non-cylindrical primordium, our argument would be substantially weakened.
Deep homology of Longisquama appendages, feathers, and other types of elongated skin appendages?
Apart from the skin appendages of Longisquama further types of elongated integumentary structure have been found in fossil diapsid reptiles: Appendages in the form of thin filaments or bristles are present in coelurosaurs (Chen et al. 1998; Xu et al. 1999a, b; Currie and Chen 2001; Xu and Norell 2004; Norell and Xu 2005), the ornithischian dinosaurs Psittacosaurus and Tianyulong (Mayr et al. 2002; Zheng et al. 2009), and pterosaurs (Bakhurina and Unwin 1995; Frey and Martill 1999; Ji and Yuan 2002; Wang et al. 2002; Kellner et al. 2010). Furthermore vaned feathers (Ji et al. 1998; Norell et al. 2002; Xu et al. 2003) and multiple-branched appendages without vanes (Currie and Chen 2001; Xu et al. 2001) occur in certain non-avian coelurosaur groups. Considering their close relationship to birds, the elongated unbranched and branched filamentous appendages of non-avian coelurosaurs have been interpreted as “primitive feathers” and precursor structures of non-avian and avian vaned feathers. This interpretation is based on a series of hypothetical evolutionary stages introduced by Prum (1999), which he proposed under consideration of developmental mechanisms in feathers of extant birds: Beginning with an unbranched cylindrical feather (stage I), differentiation leads to specialized vaned feather types, for example the remiges of birds (stage Va). After assignment of coelurosaur appendage fossils to Prum’s vane-less primordial feather types (stages I–III), the origin of feathers and some of the hypothetic stages have been linked to nodes of the phylogenetic tree of higher theropods (Sereno 1999; Prum and Brush 2002; Xu 2006).
In accordance with his feather-evolution model, Prum’s principal criterion for appendages which shall be referred to as “feathers” is that they are homologous with avian feathers in their arising from a follicle (i.e. a cylindrical invagination of the epidermis; Prum 1999; Prum and Brush 2002). If the elongated appendages of ornithischians, pterosaurs, and Longisquama do not fulfil this criterion, i.e. if they did not grow from a follicle or acquired a follicle independently (as in the case of mammalian hair), their homology with avian feathers on a lower structural level, e.g. a thin filamentous outgrowth without a basal invagination, remains a possibility. As discussed by Bonde and Christiansen (2003), Xu (2004), and Xu et al. (2009) the presence of elongated filamentous or more complex appendages could be the synapomorphy of a clade including dinosaurs and pterosaurs. Even though they might have been widespread within this clade, only under exceptional taphonomic conditions are the appendages fossilized—as in the Early Cretaceous Jehol Group (Zhou et al. 2003). Could the Middle Triassic or early Late Triassic diapsid Longisquama be an early member of such a clade?
Given the state of preservation of the only partial skeleton, none of the yet proposed assignments of Longisquama to non-dinosaurian archosaurs (Sharov 1970; Jones et al. 2000; Martin 2004; Frances and Pourtless 2009), prolacertiform archosauromorphs (Peters 2000; but see Hone and Benton 2007), lepidosauromorphs (Unwin et al. 2000), or a basal diapsid clade including coelurosauravids and drepanosaurids (Senter 2004; but see Renesto and Binelli 2006; Renesto et al. 2010) seems to be well supported by osteological data or can be ruled out with certainty. Unlike Sharov (1970) and some later authors we observed no clear indications for antorbital or mandibular fenestrae in the holotype specimen and thus, on that basis, cannot support assignment to archosaurs. In the place of the alleged antorbital fenestrae on the left slab of the holotype, which have been the basis for far-reaching interpretations by Peters (2000) and Frances and Pourtless (2009), no skull bone substance is present according to our observation and the sediment relief seems to be of limited informativeness with regard to skull features.
Notwithstanding the probable convergent nature of the alleged feather-like features observed in the complexly differentiated appendages of Longisquama, we find it likely that their similarity to vaned feathers is based on a homologous inventory of developmental processes, genetic control mechanisms, and structural preconditions of the skin, e.g. the capability to form specialized keratins. Since the 1990s various examples of genetic homologies and homologous developmental mechanisms underlying the formation of morphologically and phylogenetically disparate structures and organs have been described on the basis of evolutionary developmental biology studies and referred to as “deep homologies” (Shubin et al. 1997, 2009; Scotland 2010). Deep homology has also been discussed for skin ossifications that occur convergently in many fossil and extant tetrapod groups (Main et al. 2005; Hill 2006; Vickaryous and Hall 2008; Vickaryous and Sire 2009) and similarly we propose that deep homology played a role in the evolution of elongated skin appendages within the Diapsida.
Considering the problematic falsifiability of such claims in long extinct groups we formulate our deep homology hypothesis as follows: “Vaned feathers and Longisquama appendages share a set of homologous developmental mechanisms and of structural modifications of the skin not yet present in the last common ancestor of lepidosaurs and archosaurs.” Accordingly we regard the predisposition of the skin to form simple filamentous and more complex elongated appendages as the synapomorphy of an archosauromorph clade including Longisquama, pterosaurs, ornithischians, and coelurosaurs (Fig. 10). These groups might have acquired their elongated skin structures independently through convergent evolution. Our deep homology hypothesis could be shown to be false in the following ways:
Longisquama is convincingly excluded from the Archosauromorpha on the basis of a revision of its skeletal morphology;
appendages of similar complexity and feather-like differentiation occur outside the Archosauromorpha;
Longisquama is placed outside the archosaur crown-group and the skin of extant crocodiles is shown to be no more bird-like than lepidosaur skin with respect to its genetic, cellular, structural or developmental properties.
In fact several studies on the development of reptilian skin describe similarities in the embryonic epidermis of crocodylians and birds that are not shared by squamates, in particular beta-keratins of a similar type (Sawyer and Knapp 2003; Sawyer et al. 2003a; Alibardi et al. 2006; Alibardi and Toni 2008; Dalla Valle et al. 2009), and our assumption that Longisquama is more closely related to archosaurs than to lepidosaurs was inspired by these results. Sawyer and Knapp (2003) compared the embryonic development of avian scutate and crocodilian scales with that of the embryonic feather filament. They found that in these embryonic integuments similar cell populations, which can give rise to functional structural beta keratin elements, are present. The authors conclude that the ability to develop such structural elements—as in the case of elongated and highly differentiated skin appendages—was a plesiomorphic feature of the common ancestor of birds and crocodilians. Furthermore, the results of Dalla Valle et al. (2009) suggest that beta keratins of the avian beak, scale, and claw are compositionally and structurally closer to those of the crocodilian epidermis than to feather keratins. The authors concluded that feather keratins may have evolved early in the history of archosauromorphs before the divergence of birds and crocodilians. In conflict with such an early origin another recent study by Greenwold and Sawyer (2011) provides molecular data that suggest a Late Triassic to Early Jurassic divergence of feather keratins and feather-like keratins from other types of avian keratins, demonstrating that the discussion of feather keratin origins on the basis of molecular and developmental data is not yet closed.
The lack of evidence for a second row of appendages (as noted by Unwin and Benton 2001) and the difficulty of envisaging a stable airfoil emerging from the mostly narrow sticklike and punctually attached appendages render the hypothesis of an aerodynamic function unlikely (Stephan 2003; Voigt et al. 2009). Alternative uses, for example insulation, protective mimicry, and sexual display, have been discussed (Unwin et al. 2000). As suggested by Stephan (2003) the series of dorsal appendages in Longisquama could be a functional analogue to the dorsal crests of elongated scales in extant squamates. In iguanian lizards complex display behaviour involving dorsal crest displays occurs (Watkins 1998; Ord et al. 2002). The hypothesis that integumentary structures of high complexity evolved in the context of display is not unreasonable if the mating systems and display behaviour were likewise highly derived.
Given the type of preservation, arrangement of skin impressions in the holotype specimen, and lack of similar morphology among plants we reject the hypothesis that the imprints of the dorsal skin appendages of Longisquama are in fact fossilized plant organs. The appendages’ probable tubular base and structural complexity, apparent from their differentiation along several body axes, can be explained by formation in an elaborate developmental process that began with a filamentous appendage primordium. Their later development was marked by continuous elongation and shaping of the wide distal and narrow proximal sections in subsequent developmental stages. In agreement with some similarities in the skin keratin structure and skin development of extant crocodilians and birds which are not shared by extant squamates, we consider the ability to form simple filamentous and more complex elongated skin appendages as the synapomorphy of an archosauromorph clade including Longisquama, crocodylians, pterosaurs, non-avian dinosaurs, and birds. Rather than indicating homology of the structures themselves, the similarity of Longisquama’s appendages and vaned feathers might be the consequence of deep homology. The evolution of complex sexual display behaviour might have been the functional background for the appendages’ high level of differentiation.
This work was supported by the State of Saxony (scholarship to M. B.) and by the German Research Foundation (DFG II—VO 1466/1-1). We are grateful to Evgenii N. Kurochkin and Vladimir R. Alifanov for access to the type material in Moscow in February 2007, to Jan Fischer, Daniel Krause, and Robert Georgi for discussion and for their support during the 2007 field season, to Ilja Kogan for his help with translation of Russian publications and reports, and to Christian Foth, Nicholas Fraser, and Jörg W. Schneider for their comments on earlier versions of the manuscript.
- Alibardi, L., L.W. Knapp, and R.H. Sawyer. 2006. Beta-keratin localization in developing alligator scale and feathers in relation to the development and evolution of feathers. Journal of Submicroscopic Cytology and Pathology 38: 175–192.Google Scholar
- Bakhurina, N.N., and D.M. Unwin. 1995. A preliminary report on the evidence of “hair” in Sordes pilosus, an Upper Jurassic pterosaur from Middle Asia. In Sixth symposium on mesozoic terrestrial ecosystems and biota, short papers, ed. A. Sun, and Y. Wang, 79–82. Beijing: China Ocean Press.Google Scholar
- Bonde, N., and P. Christiansen. 2003. The detailed anatomy of Rhamphorhynchus: axial pneumaticity and its implications. In Evolution and paleobiology of pterosaurs, ed. E. Buffetaut, and J.-M. Mazin. London: The Geological Society. (Special Publications 217: 217–232).Google Scholar
- Cowen, R. 1981. Homonyms of Podopteryx. Journal of Paleontology 55: 483.Google Scholar
- Dobruskina, I.A. 1994. Triassic floras of Eurasia. Österreichische Akademie der Wissenschaften. Schriftenreihe der Erdwissenschaftlichen Kommissionen 10: 1–422.Google Scholar
- Dobruskina, I.A. 1995. Keuper (Triassic) flora from Middle Asia (Madygen, Southern Fergana). New Mexico Museum of Natural History and Science Bulletin 5: 1–49.Google Scholar
- Evans, S.E. 1988. The early history and relationships of Diapsida. In The phylogeny and classification of tetrapods, volume 1: Amphibians, reptiles, birds, ed. M.J. Benton, 221–260. Oxford: Clarendon Press.Google Scholar
- Fastovsky, D.E., and D.B. Weishampel. 2004. The evolution and extinction of the dinosaurs, 485. Cambridge: Cambridge University Press.Google Scholar
- Fraser, N.C. 2006. Dawn of the Dinosaurs: Life in the triassic, 328. Bloomington and Indianapolis: Indiana University Press.Google Scholar
- Frey, E., and D.M. Martill. 1999. Soft tissue preservation in a specimen of Pterodactylus kochi from the Upper Jurassic of Germany. Neues Jahrbuch für Geologie und Paläontologie Abhandlungen 210: 421–441.Google Scholar
- Gans, C., I. Darevskii, and L.P. Tatarinov. 1987. Sharovipteryx, a reptilian glider? Paleobiology 13: 415–426.Google Scholar
- Haubold, H., and E. Buffetaut. 1987. A new interpretation of Longisquama insignis, an enigmatic reptile from the Upper Triassic of Central Asia. Comptes Rendus de l’Académie des Sciences Série 2(305): 65–70.Google Scholar
- Ivakhnenko, M.F. 1978. Urodelans from the Triassic and Jurassic of Soviet Central Asia. Paleontological Journal 1978: 362–368. (In Russian).Google Scholar
- Ji, Q., and C. Yuan. 2002. Discovery of two kinds of protofeathered pterosaurs in the Mesozoic Daohugou Biota in the Ningcheng Region and its stratigraphic and biologic significances. Geological Review 48: 221–224.Google Scholar
- Jones, T.D., J.A. Ruben, P.F.A. Maderson, and L.D. Martin. 2001. Longisquama fossil and feather morphology. Science 291: 1901–1902.Google Scholar
- Kogan, I., K. Schöneberger, J. Fischer, and S. Voigt. 2009. A nearly complete skeleton of Saurichthys orientalis (Pisces, Actinopterygii) from the Madygen Formation (Middle to Late Triassic, Kyrgyzstan, Central Asia) - preliminary results. Freiberger Forschungshefte C 532: 139–152.Google Scholar
- Lucas, A.M., and P.R. Stettenheim. 1972. Avian anatomy: Integument. Washington, DC: US Department of Agriculture.Google Scholar
- Martin, L.D. 2004. A basal archosaurian origin of birds. Acta Zoologica Sinica 50: 978–990.Google Scholar
- Paul, G.S. 2001. Dinosaurs of the air: the evolution and loss of flight in dinosaurs and birds. Baltimore and London: John Hopkins University Press.Google Scholar
- Peters, D. 2000. A re-examination of four prolacertiforms with implications for pterosaur phylogenesis. Rivista Italiana di Paleontologia e Stratigrafia 106: 293–336.Google Scholar
- Prum, R.O., and A.H. Brush. 2002. The evolutionary origin and diversification of feathers. Quarterly Review of Biology 77: 261–295.Google Scholar
- Reisz, R.R., and H.-D. Sues. 2000. Palaeontology: The ‘feathers’ of Longisquama. Nature 408: 428.Google Scholar
- Renesto, S., and G. Binelli. 2006. Vallesaurus cenensis Wild 1991, a drepanosaurid (Reptilia, Diapsida) from the Late Triassic of northern Italy. Rivista Italiana di Paleontologia e Stratigrafia 112: 77–94.Google Scholar
- Renesto, S., J.A. Spielmann, S.G. Lucas, and G.T. Spagnoli. 2010. The taxonomy and paleobiology of the Late Triassic (Carnian-Norian: Adamanian-Apachean) drepanosaurs (Diapsida: Archosauromorpha: Drepanosauromorpha). New Mexico Museum of Natural History and Science Bulletin 46: 1–81.Google Scholar
- Rietschel, S. 1985. Feathers and wings of Archaeopteryx, and the question of her flight ability. In The beginnings of birds, ed. M.K. Hecht, J.H. Ostrom, G. Viohl, and P. Wellnhofer, 251–260. Eichstätt: Freunde des Jura-Museums.Google Scholar
- Sharov, A.G. 1966. Unique finds of reptiles from Mesozoic deposits of Central Asia. Bulletin of Moscow Society of Naturalists, Geological Section (Bjulleten’ Moskovskogo Obscestva Ispytatelej Prirody, Otdel geologiceskij) 41(2): 145–146. (in Russian).Google Scholar
- Sharov, A.G. 1970. An unusual reptile from the Lower Triassic of Fergana. Paleontological Journal 1970: 112–116. (In Russian).Google Scholar
- Sharov, A.G. 1971. New flying reptiles from the Mesozoic of Kazakhstan and Kyrgyzstan. Transactions of the Palaeontological Institute (Trudy Paleontologicheskogo Instituta Akademiya Nauk SSSR) 130: 104–113. (in Russian).Google Scholar
- Shcherbakov, D.E. 2002. The 270 million year history of Auchenorrhyncha (Homoptera). Denisia 4: 29–36.Google Scholar
- Shcherbakov, D.E. 2008a. Madygen, Triassic Lagerstätte number one, before and after Sharov. Alavesia 2: 113–124.Google Scholar
- Shcherbakov, D.E. 2008b. Insect recovery after the Permian/Triassic crisis. Alavesia 2: 125–131.Google Scholar
- Shcherbakov, D.E. 2011. New and little-known families of Hemiptera Cicadomorpha from the Triassic of Central Asia—Early analogs of treehoppers and planthoppers. Zootaxa 2836: 1–26.Google Scholar
- Sixtel, T.A. 1960. Stratigraphy of the continental deposits of the Upper Permian and Triassic of Central Asia. Transactions of the Tashkent State University V. I. Lenin (Trudy Tashkentskogo Gosudarstvennogo Universiteta Imeni V. I. Lenina) 176: 1–146. (in Russian).Google Scholar
- Sixtel, T. A. 1962. Flora of the Late Permian and Early Triassic in Southern Fergana. Stratigrafia i paleontologia Uzbekistana i sopredelnych rayonov, kniga 1: 272–414 (Tashkent) (in Russian).Google Scholar
- Sytchevskaya, E.K. 1999. Freshwater fish fauna from the Triassic of Northern Asia. In Mesozoic fishes 2: Systematics and fossil record, ed. G. Arratia, and H.-P. Schultze, 445–468. Munich: Verlag Dr. Friedrich Pfeil.Google Scholar
- Tatarinov, L.P. 2005. A new cynodont (Reptilia, Theriodontia) from the Madygen formation (Triassic) of Fergana. Kyrgyzstan. Paleontological Journal 39(2): 192–198.Google Scholar
- Unwin, D.M., V.R. Alifanov, and M.J. Benton. 2000. Enigmatic small reptiles from the Middle-Late Triassic of Kirgizstan. In The age of dinosaurs in Russia and Mongolia, ed. M.J. Benton, M.A. Shishkin, D.M. Unwin, and E.N. Kurochkin, 177–186. Cambridge: Cambridge University Press.Google Scholar
- Unwin, D.M., and M.J. Benton. 2001. Longisquama fossil and feather morphology. Science 291: 1900–1901.Google Scholar
- Voigt, S., H. Haubold, S. Meng, D. Krause, J. Buchantschenko, K. Ruckwied, and A.E. Götz. 2006. The Madygen lagerstätte: A contribution to the geology and palaeontology of the Madygen Formation (Middle to Upper Triassic, SW Kyrgyzstan, Central Asia). Hallesches Jahrbuch für Geowissenschaften B 22: 85–119.Google Scholar
- Vorobyeva, E.I. 1967. A Triassic Ceratodus from Southern Fergana and some notes on the systematic and phylogeny of the family Ceratodontidae. Paleontological Journal 1967: 80–87. (in Russian).Google Scholar
- Watkins, G.G. 1998. Function of a secondary sexual ornament: The crest in the South American iguanian lizard Microlophus occipitalis (Peters, Tropiduridae). Herpetologica 54: 161–169.Google Scholar