Actinofibrils

Actinofibrils probably were modified epidermal scales that formed in place as part of the dorsal epidermis of the dactylopatagium. In order for this to occur, the stratum basale of the dorsal epidermis of the dactylopatagium would have been divided into broad germinative strips of fibril keratinocytes separated by narrow intervening strips of unspecialized keratinocytes, with the fibril keratinocytes synthesizing and accumulating hard keratin such as is found in claw sheathes and the unspecialized keratinocytes synthesizing and accumulating the softer keratins of unspecialized epidermis. An initial embryonic layer of fibril keratinocytes probably formed during the late embryonic development of the individual, and as the individual grew the fibril germinative strip increased in length and width. The added area of fibril keratinocytes might have been symmetrically distributed around the previous layer of fibril keratinocytes such that the thickest part was in the center of the actinofibril. However, it is also possible that the added keratinocyte area was not symmetrically distributed in much the same way that the development of the dermal shields on the carapace of box turtles (Terrapene; Zangerl, Reference Zangerl1969) and the plastron of mud turtles (Kinosternon; Mosimann, Reference Mosimann1956) is asymmetrical such that the embryonic shield is toward one edge of the dermal shield. If the added area was biased toward the posterolateral end of the germinative strip, then the thickest part of the actinofibril would be toward the anteromedial end of the germinative strip, which would be consistent with the fact that in the Zittel wing the raised longitudinal strips are tall and sharply defined in the more anterior and medial parts of the dactylopatagium and indistinct to absent near the trailing edge and wing tip, and the fact that thicker anteromedial ends would be better suited to the function of resisting compressive loads and redirecting tensile loads to the proximal wing phalanges.

Actinofibrils formed deep in the epidermis would not have been shed and replaced during the lifetime of the individual, so the rate of cell division in the stratum basale of the fibril germinative strip need only have been high enough to keep up with the growth of the individual and produce a sufficiently stiff actinofibril. The layer of unspecialized surface epidermis that covered over the actinofibrils might have been formed by having the stratum basale of the intervening strips spread laterally over the actinofibrils and merge with that of adjacent intervening strips to cover the actinofibrils and form a continuous sheet of unspecialized stratum basale that produced the surface epidermis or by having the developing keratinocytes from the intervening strips spread laterally and merge to cover the actinofibrils and form a continuous sheet of surface epidermis. The rate of cell division in the unspecialized stratum basale would have been high enough to produce a surface epidermis that could have been shed or lost through abrasion.

In Figure 3.3, actinofibrils are shown as having cross-sections equivalent to a plano-convex lens. Such a cross-section would result if the actinofibrils were built up through the addition of multiple planar layers of keratinocytes produced by a planar germinative strip as the animal grew from embryo to adult. The cross-section could also have approximated a biconvex lens or a plano-convex lens with a flat external surface if the germinative strip was concave rather than planar, or could conceivably have taken other shapes if the rate of addition of keratinocytes was not uniform across the width of the fibril germinative strip. Note that the gabled roof appearance of the calcitic fragment suggests that the germinative strip was V-shaped.

The widths and lengths of actinofibrils are variable across the dactylopatagium. In the central area posterior to WP2 and 3, actinofibrils were ~0.2 mm wide, whereas anteriorly, close behind the wing phalanges they were narrower as they also seem to be in the more medial parts of the dactylopatagium posterior to WP1. In the medialmost part of the dactylopatagium and in the adjacent lateral plagiopatagium the raised longitudinal strips are short, closely spaced, and less regular than elsewhere, which suggests that the actinofibrils were shorter and narrower than elsewhere. However, it is possible that that appearance is partly a result of the manner of preservation of the patagium, such that the actinofibrils may have been longer and more regularly arranged than the appearance of the raised strips suggests. Note that there may have been a limit to the permissible width of actinofibrils if the unspecialized stratum basale or the developing keratinocytes of the adjacent intervening strips were to merge and form a continuous sheet of surface epidermis covering the actinofibrils. In the middle of the dactylopatagium behind WP2 and 3 actinofibrils seem to be quite long and straight. A special case of varying width seems to be found in the actinofibrils that are adjacent to the prominent fold lines in the dactylopatagium. In the case of Fold Line C, the actinofibrils immediately on either side of the fold line are narrow and their raised longitudinal strips less prominent than most strips whereas the actinofibrils a short distance on either side of the fold line are of normal breadth and prominence.

While mapping the raised longitudinal strips of the Zittel wing I did not see any evidence that the grooves between them, and thus the actinofibrils that had occupied the grooves, were not continuous longitudinally. The only possible evidence of discontinuity was in places where a raised strip seemed to jog to one side; however, that might merely represent a local change in the width of the germinative strip rather than the posterolateral end of an anteromedial actinofibril abutting the anteromedial end of a more posterolateral actinofibril. Thus it is probable that individual germinative strips and actinofibrils extended unbroken from the wing spar to the trailing edge, which in the Zittel wing could be up to 125 mm long. It seems that there would be no difficulty and no disadvantage in having such long actinofibrils, though there probably also would be no disadvantage to having shorter ones if their ends were staggered.

As noted by Padian and Rayner (Reference Padian and Rayner1993) there are many places on the Zittel wing where a raised longitudinal strip was intercalated between two adjacent strips. These instances might represent the posterolateral end of an anteromedial actinofibril abutting the anteromedial ends of two more posterolateral actinofibrils so as to accommodate the radiating pattern of actinofibrils on the dactylopatagium while limiting the maximum width of individual actinofibrils. However, there is no evidence that such actinofibrils ended adjacent to one another and the instances could also represent the branching of one germinative strip into two. Frey et al. (Reference Frey, Tischlinger, Buchy and Martill2003) stated that in JME SOS 4784 actinofibrils bifurcated near the trailing edge but did not provide photographs or diagrams so it is not clear if this is what they were describing. Note that Padian and Rayner (Reference Padian and Rayner1993) also stated that some structural fibers appeared to bifurcate near the trailing edge of the dactylopatagium, but this was because they misinterpreted branching blood vessel traces as structural fibers.

Skin surface

The Marsh specimen preserves positive impressions of the dorsal and ventral surfaces of the dactylopatagium that show that the epidermis was smooth except for fine striae that are best observed under low angle illumination. The striae form a posterolaterally radiating pattern that loosely reflects the posterolaterally radiating pattern of the underlying actinofibrils. The sections of epidermis bounded by striae do not appear to have been heavily keratinized. This is shown by the distortions of their pattern associated with the oval depressions in the Marsh specimen’s left wing (Fig. 7.2), and so the surface epidermis seems to have formed a soft, compliant covering of the dorsal and ventral surfaces of the wings. Interestingly, the Marsh specimen seems to have been largely ignored by pterosaur workers because its preservation of traces of a smooth skin with only fine striae differed markedly from the raised longitudinal strips of the Zittel wing and so was deemed less informative.

Undulations and fold lines

The undulations in the Zittel wing indicate that the dactylopatagium was lax, which is to be expected because the wing was no longer attached to the trunk. The series of undulations behind WP2 and 3 probably resulted because the wingfinger was unloaded and flattened into the plane of the substrate, whereas when spread as in flight the curvature of the wingfinger and tension within the brachiopatagium would have taken up the slack, eliminating the undulations. Similarly, the undulations along the trailing edge behind WP1 in the Zittel wing and the left wing of the Marsh specimen would have flattened out when the brachiopatagium was under tension. The sharp edged creases of the Zittel wing, which are also evident in BSP 1907 I 37, indicate that the membrane was somewhat stiff and suggest that it had some natural camber, at least in the distal part of the dactylopatagium. The stiffness of the dactylopatagium is also evident in the deep furrows along the fold lines on the right wing of the Marsh specimen, which because the impression is a positive impression of the ventral surface of the patagium represent places where two sections of the dactylopatagium lying on the substrate resisted compression perpendicular to the long axes of the actinofibrils within the plane of the patagium and were pushed up into ridges along fold lines.

It is not clear how extensible Wellnhofer (Reference Wellnhofer1975, Reference Wellnhofer1987) thought the brachiopatagium was, but Padian and Rayner (Reference Padian and Rayner1993, fig. 15A) thought it was significantly extensible perpendicular to the long axes of actinofibrils and Bennett (Reference Bennett2000) thought it was as extensible as bat patagium perpendicular to the long axes of actinofibrils within the limits of a load-bearing collagen fiber network. However, the stiffness and resistance to compression perpendicular to the actinofibrils noted above plus the remarkable uniformity of the widths (~0.05 mm) of the raised longitudinal strips representing the strips of unspecialized epidermis between the actinofibrils indicate that the dactylopatagium was essentially inextensible. A significantly extensible membrane would not exhibit any stiffness, and it is probable that an extensible membrane, even a lax one, would exhibit some variation in the width of the intervening strips. The gentle undulations seem to represent the only significant shortening that could occur in the dactylopatagium. Thus, the dactylopatagium was somewhat stiff and inextensible.

Because the dactylopatagium was somewhat stiff and inextensible, it was necessary to fold it in order to store it compactly. The dactylopatagium would have folded up somewhat like a traditional folding hand fan consisting of a sector of paper supported by slender slats pivoting around a single point. The Zittel wing, Marsh specimen, and other specimens such as BSP 1907 I 37 exhibit a consistent pattern of prominent fold lines, of which five have been identified by letters. In both specimens described here Fold Line B extends from roughly one-third of the distance along WP1 from its proximal end to the trailing edge behind the mid-point of WP2, Fold Line C extends from roughly the mid-point of WP1 to the trailing edge behind the second IP joint and in part cuts across the regular pattern of raised longitudinal strips in the Zittel wing and the pattern of striae that reflects the pattern of underlying actinofibrils in the Marsh specimen, Fold Line D extends from a point one-fifth along WP2 to the trailing edge behind the mid-point of WP3. and Fold Line E extends from a point three-fifths along WP3 to the trailing edge behind the distal third of WP4. The similarity of the positions and shapes of the folds and the deviations from the pattern of actinofibrils suggest that the folds were not merely the result of incidental folding of a uniform patagium but rather were genetically controlled lines of increased flexibility that enabled the otherwise somewhat stiff dactylopatagium to fold consistently and compactly. It is not clear how the dactylopatagium folded up along each of the fold lines, but the folding along Fold Line E of the right wing of the Marsh specimen indicates that the patagium posteromedial to the fold line folded up and over the patagium anterolateral to the line. Presumably there was another fold close behind Fold Line E that brought the patagium down and backward.

One reviewer seemed to think that a wing with a somewhat stiff and inextensible dactylopatagium would present a different appearance when folded as proposed here than a wing with an elastic membrane bearing widely spaced cylindrical fibers furled by contraction of the membrane, and furthermore suggested that no pterosaur specimen preserves evidence of the proposed fan folding. Bennett (Reference Bennett2000; fig. 3) discussed and illustrated the sort of folding proposed here, with the dactylopatagium folded into a narrow band alongside the wingfinger just as is seen in specimens such as BSP 1938 I 503 (Wellnhofer, Reference Wellnhofer1975, pl. 4, fig. 1). I find that the majority of specimens with soft tissue preservation of wing membranes present an appearance that is consistent with both fan folding of a stiff dactylopatagium and contractive furling of an elastic dactylopatagium; however, the right wing of the Marsh specimen presents an appearance that is consistent with fan folding but not consistent with contractive furling (Fig. 6.1). The outer part of the dactylopatagium is folded along Fold Line E such that the edge of the fold rather than the trailing edge of the dactylopatagium forms what looks like the trailing edge of the wing behind the middle third of WP4. I cannot conceive of a manner in which contractive furling could produce such an appearance.

It is possible that the idea of genetically controlled fold lines was arrived at independently by others: Monninger et al. (Reference Monninger, Frey, Elgin, Tischlinger, Sartori and Schneider2010, p. 52) commented in an abstract that actinofibrils were “the guiding structure for formation of ... folds.” However, it is not clear to me whether they meant merely that folds were parallel to the long axes of actinofibrils, which had been noted before (Padian and Rayner, Reference Padian and Rayner1993; Bennett, Reference Bennett2000) or if they meant that specific variation in the size and pattern of actinofibrils controlled the location of the folds.

Retrophalangeal wedge

Previous authors noted the retrophalangeal wedge immediately behind the wing spar, but interpreted it as only behind the wingfinger (Padian and Rayner, Reference Padian and Rayner1993; Tischlinger and Frey, Reference Tischlinger and Frey2010; Monninger et al., Reference Monninger, Frey and Tischlinger2012) whereas the Zittel wing shows that it extended proximally almost to the elbow, and the wedge’s arcing posteromedial margin on the Marsh specimen also suggests it extended behind the antebrachium. Various functions have been proposed for the retrophalangeal wedge including: (1) streamlining the transition between the thick wing spar and thin brachiopatagium (Padian and Rayner, Reference Padian and Rayner1993; Tischlinger and Frey, Reference Tischlinger and Frey2010; Monninger et al., Reference Monninger, Frey and Tischlinger2012), (2) containing generative tissue that produced the actinofibrils (Padian and Rayner, Reference Padian and Rayner1993), (3) anchoring the actinofibrils to the wing spar and transferring lift forces from the actinofibrils to the wing spar (Padian and Rayner, Reference Padian and Rayner1993; Tischlinger and Frey, Reference Tischlinger and Frey2010; Monninger et al., Reference Monninger, Frey and Tischlinger2012), and (4) reinforcing the IP joints and preventing their flexion (Monninger et al., Reference Monninger, Frey and Tischlinger2012).

Faced with the problem of explaining the development of cylindrical keratinous fibers on the ventral surface of the epidermis, Padian and Rayner (Reference Padian and Rayner1993) suggested that the retrophalangeal wedge contained the generative tissues from which the fibers grew like fingernails, posterolaterally in a radiating pattern to be abraded away at the trailing edge. This suggestion is not supported because as discussed above actinofibrils developed in place within the epidermis, and so no concentrated mass of generative tissue was necessary.

Padian and Rayner (Reference Padian and Rayner1993) viewed the brachiopatagium as essentially cantilevered behind the wing spar with the actinofibrils transferring lift forces anteromedially to the wing spar. Such a structure would require that the actinofibrils were anchored to the wing spar in such a way that the lift forces could be transferred to the spar, and Padian and Rayner (Reference Padian and Rayner1993, fig. 12) suggested that local tension in the wedge was important in the force transference. The suggestion that the wedge was involved in anchoring the actinofibrils and transferring aerodynamic forces to the wing spar is not supported because actinofibrils formed as part of the epidermis would be securely attached to the underlying dermis, which in turn could have been securely attached to the periosteum of the bones of the wing spar, so no broad wedge of tissue would be necessary to transmit forces. However, a planar rather than interwoven array of actinofibrils alone or as part of composite of actinofibrils and an elastic membrane would not have been stiff enough to transmit lift forces. McGowan (Reference McGowan1991) noted that cylindrical actinofibrils 0.05 mm in diameter would be too slender to resist bending forces, and though the actual actinofibrils were broader than previously thought, they were probably little thicker and little better suited to resist bending out of the plane of the patagium. This is clearly shown by the presence of the Zittel wing’s undulations and fold lines; a patagium flexible enough to undulate and fold compactly would not be stiff enough to transfer lift forces as Padian and Rayner (Reference Padian and Rayner1993) proposed. The dactylopatagium must have been tensioned between the plagiopatagium medial to it and the wingfinger anterior to it, and lift forces must have been transferred medially and anteriorly by tension in the collagen fiber layer. Note that other authors who have viewed the dactylopatagium as self-supporting and self-cambering with actinofibrils transmitting lift forces (Schaller, Reference Schaller1985; Frey et al., Reference Frey, Tischlinger, Krüger and Hone2007; Tischlinger and Frey, Reference Tischlinger and Frey2010; Monninger et al., Reference Monninger, Frey and Tischlinger2012) have not provided evidence or argumentation that actinofibrils were stiff enough to resist bending and transfer lift forces to the wing spar or that the layer of collagen fibers did not transfer such loads to the wing spar and body by tension. Frey et al. (Reference Frey, Tischlinger, Krüger and Hone2007) suggested that intrinsic muscle tissue ventral to the actinofibrils contracted to bend the actinofibrils to camber the dactylopatagium, but did not provide any evidence that there were muscle fibers associated with the actinofibrils or collagen fibers.

The suggestion that the retrophalangeal wedge consisted of dense fibrous connective tissue that reinforced the IP joints and prevented them from flexing is not supported because the wedge extended proximally behind the MCP joint of digit IV, because the wedge was as well developed behind the middle of the wing phalanges as behind the IP joints, and because such reinforcement was unnecessary. Monninger et al. (Reference Monninger, Frey and Tischlinger2012) stated that there is no evidence of strong IP ligaments, which may be true in the case of Rhamphorhynchus because of the small size and immaturity of most specimens, but Bennett (Reference Bennett2000) described prominent IP ligament attachment scars in mature individuals of Pteranodon so there is evidence that pterosaurs had strong IP ligaments in their wingfingers. Even without that evidence, the extant phylogenetic bracket (Witmer, Reference Witmer1995) informs us that pterosaurs had IP ligaments. Pterosaurs, like their extant relatives (e.g., lepidosaurs and birds), exhibit the osteological correlates of synovial joints in their wingfingers (e.g., expanded articular ends with complementary surfaces, different bone texture on the articular ends indicating that they were covered by articular cartilages), and so it can be concluded that the synovial joints of pterosaurs included a fibrous joint capsule with parts thickened into ligaments to hold the bones together and control their movements. Moreover, IP ligament attachment scars on most pterosaur wing phalanges would be relatively smaller and less prominent than those of their non-volant relatives because of the large radius of the joint surfaces and the extremely limited movement allowed by the joint. In the case of a typical ginglymoid IP joint, the joint surfaces have a small radius and allow considerable flexion and extension, and so the IP ligaments must attach to a small area around the center of joint rotation on the medial and lateral sides of the proximal phalanx, resulting in prominent attachment scars. In all pterosaurs except anurognathids (Bennett, Reference Bennett2001, Reference Bennett2007), the joint surfaces have a large radius and allowed almost no flexion and extension, and so the ligaments need not have attached to a small area but rather could attach to broad areas resulting in less prominent scars. Lastly, dense fibrous connective tissue though strong in tension is not suited to resist compression, and so a band of dense fibrous connective tissue (or even cartilage) behind the IP joints would not be suited to preventing flexion of the joints.

Having rejected other proposed functions of the retrophalangeal wedge, only a streamlining function remains. The retrophalangeal wedge would have streamlined the airflow over the wing, reducing drag (Palmer, Reference Palmer2010). Note that the wedge is widest behind the carpus and MCP joint where the wing spar is thickest, and tapers out proximally where streamlining of the elbow might be more difficult because the wedge would be on the extensor side of the joint and perhaps because it would be well behind the leading edge of the wing where streamlining might be less important. It is possible that there was also a streamlining structure medial to the elbow, but if so the Zittel wing does not preserve any trace of it.

Various tissues might have formed the retrophalangeal wedge; however, most pterosaurs, including Rhamphorhynchus (Bonde and Christensen, Reference Bonde and Christiansen2003), exhibit extensive skeletal pneumaticity, which was presumably evolved to displace marrow and lighten the skeleton. Therefore, it is unlikely that the wedge would have been formed of dense connective tissues that would add significantly to the mass of the wing if a lighter alternative were available. It is probable that the wedge was an extra-skeletal pneumatic feature. If, as reconstructed here, the dorsal and ventral surfaces of the wedge were in contact with the dorsal and ventral dermis (Fig. 9.2), then the wedge could be a pneumatic diverticulum bounded by thin epithelia and connected to the intra-skeletal pneumatic spaces. It could deflate and collapse when the wing was folded and inflate when tension in the dactylopatagium pulled the dorsal and ventral skins taut. The facts that the Zittel wing preserves a separate fold line in the retrophalangeal wedge and that there are places where raised longitudinal strips can be seen in gaps in the wedge are consistent with the wedge being a pneumatic diverticulum bounded by thin epithelia.

It is possible that the unusual wing phalanx cross-sections seen in Rhamphorhynchus (Fig. 9.2) and Nesodactylus with streamlined anterodorsal and ventral surfaces but markedly concave posterior surfaces reflect the presence of the retrophalangeal wedge. It is conceivable that other rhamphorhynchoid pterosaurs that did not have such wing phalanx cross-sections did not have pneumatized retrophalangeal wedges behind their wing phalanges. However, it is unlikely that Rhamphorhynchus and Nesodactylus were the only taxa with intrapatagial pneumatized structures to streamline the wing cross-section.

Vascular supply to the brachiopatagium

Frey et al. (Reference Frey, Tischlinger, Buchy and Martill2003) described and illustrated vasculature within the dactylopatagium of Rhamphorhynchus as consisting of one large vessel subparallel to the wing phalanges that gave rise to smaller branches and loops that in turn sent off small branches, and noted that the Zittel wing preserved some vessel traces visible under UV illumination. The pattern is similar to that supplying the small intestine in humans, in which a large superior mesenteric artery gives rise to multiple smaller intestinal arteries interconnected by looping arcades that in turn send off still smaller vasa recta, and the positive impressions of the small vessels along the trailing edge correspond to the vasa recta. Note that if the Zittel wing originally preserved physical traces of the large vessel and loops in addition to those visible under UV illumination, they would have been ventral to the actinofibril layer and lost when the counterpart was chipped off.

There is no evidence that the dactylopatagium contained any muscle tissues and no reason to think that it did. Rather the dactylopatagium seems to have consisted of little more than two layers of skin and only enough hypodermis containing vasculature and nerves needed to support those two layers. Therefore, the dactylopatagium probably was rather inactive metabolically, such that it is unlikely to have required high levels of perfusion to supply its tissues with oxygen and nutrients. The presence of large vessels suggests that the patagium was important in thermoregulation (Frey et al., Reference Frey, Tischlinger, Krüger and Hone2007), in which case it could have been used to lose heat by radiation and convection or absorb heat for warming after a night’s cooling. Note that it is not clear whether the vessel traces of JME SOS 4784, NHMW 1998z0077/0001, and the Zittel wing represent arteries, veins, or artery-vein pairs, and so there is no evidence of separation of the arterial and venous supplies, which would be necessary to prevent countercurrent heat exchange if the dactylopatagium was to be used for thermoregulation.

Plagiopatagium

The Zittel wing presents a plagiopatagium that may not be complete, but certainly is contracted considerably from its extent in flight. That contraction is indirect evidence that the plagiopatagium was extensible such that it would have behaved much as bat patagium does, extending spanwise and chordwise when appropriately tensed and contracting so as to be stored compactly when relaxed. The Zittel wing provides no direct evidence as to the structure of the plagiopatagium except for the fine lineations that are parallel to the trailing edge. The fine lineations are smaller than the raised longitudinal strips associated with actinofibrils and their orientation is inconsistent with keratinous elements that would resist longitudinal compression. Rather their orientation is consistent with collagen and/or elastin fibers that would bear tensile loads within the patagium. The Marsh specimen provides no information about the plagiopatagium except for the indirect evidence that there are essentially no traces of the plagiopatagium despite well-preserved traces of the brachiopatagia and the tail vane, which suggests that the dactylopatagium was a soft tissue that was not more resistant to decay than most of the body’s soft tissues.

Trailing edge tendon?

Bramwell and Whitfield (Reference Bramwell and Whitfield1974, p. 543–544) modeled pterosaur wings with and without a collagenous load-bearing trailing tendon and preferred the model without, whereas Pennycuick (Reference Pennycuick1988) suggested that a trailing edge tendon was present. Padian and Rayner (Reference Padian and Rayner1993) argued against a trailing edge tendon on the grounds that there is no evidence for one, that the tip of the dactylopatagium was rounded whereas a load-bearing tendon would follow a straight line to the tip, and that a tendon was unnecessary. In addition, Bennett (Reference Bennett2000) noted the curved fourth wing phalanges of some pterosaurs (e.g., Pteranodon) could not have borne the tensile loads that a trailing edge tendon would apply.

Recently, Monninger et al. (Reference Monninger, Frey and Tischlinger2012) described what they interpreted as a trailing edge tendon in a specimen of Rhamphorhynchus. They reconstructed it as extending from wing tip to ankle and suggested that it stabilized the trailing edge and contributed to camber control, but they did not explain how it would have functioned. I do not doubt that some pterosaur specimens preserve linear trailing edge features of some sort; however, the mere presence of a linear trailing edge feature in the brachiopatagium does not indicate that the feature was a discrete load-bearing structure. The actinofibril layer, the collagen fiber layer of the dorsal dermis, and the hypodermis each must have had a posterior margin, and the dorsal and ventral epidermis and dermis must have been continuous with one another, respectively, most of which need not have been coincident with each other or the posterior margin of the dactylopatagium, and all of these could have produced a linear trailing edge feature in a fossil preserving patagial soft tissues.

A load-bearing trailing edge tendon would be compatible with a wing in which the patagium was subjected to chordwise tension between the wing spar and tendon and the tendon rather than the patagium bore spanwise tension. However, it is incompatible with the interpretation presented here of a somewhat stiff inextensible dactylopatagium and would place significant limitations on the shape and control of a tenopatagial plagiopatagium. It makes more sense to interpret the plagiopatagium as carrying spanwise tensile loads by way of collagen, elastin, and intrinsic muscle fibers spread across the chord of the plagiopatagium, which would provide greater control of the area and three dimensional shape of the plagiopatagium. Proponents of a load-bearing trailing edge tendon should provide evidence or argumentation for the presence and properties of the tendon and against alternative interpretations, and explain how it might have functioned within the context of the pterosaur patagium and flight.

Wing planform and flight

Rhamphorhynchus has largely escaped the attentions of aerodynamicists, who prefer to concentrate on the largest of pterosaurs, and there have been few reconstructions of the wing planform of Rhamphorhynchus (Fig. 9.1). Wellnhofer (Reference Wellnhofer1975) described the wing planform as narrow and scythe-like and illustrated it with the trailing edge following a smooth arc from the wing tip to near the hip joint. Padian and Rayner (Reference Padian and Rayner1993) did not illustrate the planform, but presented schematic illustrations that showed a narrow and scythe-like wing similar to that illustrated by Wellnhofer. Bennett (Reference Bennett2000) reconstructed the planform with the trailing edge following an S-curve from the wing tip to the proximal tibia based on the assumption that the dactylopatagium was extensible perpendicular to actinofibrils and with the patagium extended such that the length of the medialmost actinofibrils determined its chord posterior to the MCP joint. Here the planform is reconstructed based on the interpretation that the dactylopatagium was essentially inextensible and that the trailing edge of the plagiopatagium followed an arc of essentially constant radius to attach near the ankle as has been shown by Elgin et al. (Reference Elgin, Hone and Frey2011). The planform of the plagiopatagium is quite broad chordwise whereas that of the dactylopatagium is significantly narrower than previously thought. Note that there is no evidence that the trailing edge followed an arc of essentially constant radius, and if the trailing edge arc had a small radius behind the proximal antebrachium and increasing radii laterally to the dactylopatagium and posteriorly to the ankle, then the plagiopatagium would have a narrower chord and less area. In that case, control of the posterior part of the plagiopatagium by rotation of the femur about its axis for pitch control could have been more or less independent of the control of the area and shape of the rest of the plagiopatagium.

One reviewer objected that the wing reconstruction proposed here could not function in flight because wing area could not be reduced “to less than 50% of the fully extended area” whereas bats and birds can do so. Padian (Reference Padian1983a, Reference Padian1983b, Reference Padian1985, Reference Padian1987a, Reference Padian1987b, Reference Padian1991) argued against bat-like interpretations of pterosaurs as erroneous, but the bird-like interpretation of pterosaurs advocated in their place has proved to be similarly erroneous (e.g., pterosaurs were plantigrade quadrupeds [Lockley et al., Reference Lockley, Logue, Moratalla, Hunt, Schultz and Robinson1995; Bennett, Reference Bennett1997a, Reference Bennettb; Unwin, Reference Unwin1997] with the hindlimb fully involved in the wing [Elgin et al., Reference Elgin, Hone and Frey2011]). Pterosaurs were probably neither bat-like nor bird-like and differed from bats and birds as much as bats and birds differ from one another; therefore, there is no reason to think that pterosaur wings must have been functionally similar to bat and bird wings. It is true that many birds can reduce wing area in flight; however, some birds (e.g., diomedeids, trochilids, spheniscids) do not reduce wing area significantly, and neither do insects including large (e.g., Titanus, Ornithoptera, the extinct Meganeuropsis [Beckemeyer and Hall, Reference Beckemeyer and Hall2007]) and heavy species (e.g., Goliathus). Likewise, man-made ornithopters ranging from von Holst’s (Reference Holst1957) model of Rhamphorhynchus and MacCready’s model of Quetzalcoatlus (Brooks et al., Reference Brooks, MacCready, Lissaman and Morgan1985; Cowley, Reference Cowley1986) to Festo’s Smartbird gull-like ornithopter (Mackenzie, Reference Mackenzie2012) have demonstrated that the ability to alter wing area is not necessary for flapping flight. Thus, the reviewer’s objection is baseless.

Although the ability to reduce wing area is not necessary for successful flapping flight, the wing reconstruction proposed here would allow wing area to be altered. Contraction of intrinsic muscle fibers arranged subperpendicular to the trailing edge in the plagiopatagium could pull the trailing edge anteriomedially (large arrow in Fig. 9) reducing its chord and area. In addition, as discussed by Bennett (Reference Bennett2000, p. 281–282), flexion of the shoulder, elbow, and wrist would reduce wingspan, and contraction of collagen, elastin, and intrinsic muscle fibers in the plagiopatagium could reduce its area. It would also be possible to maintain tension on the medial margin of the dactylopatagium so as to keep it spread even if the plagiopatagium was lax and passively contracted. The only control envisioned by Bennett (Reference Bennett2000) that would not be possible with the new interpretation would be decreasing the area of the dactylopatagium in flight.

Other pterosaurs

Both Wellnhofer (Reference Wellnhofer1987) and Pennycuick (Reference Pennycuick1988) noted linear features in the wings of the Vienna Pterodactylus (NHMW 1975/1756/0000). Wellnhofer interpreted them as similar to what he thought was preserved on the Zittel wing, i.e., widely spaced cylindrical structures embedded within a bat-like extensible membrane. Pennycuick (Reference Pennycuick1988, p. 308) noted that the surface of the wing impression was different from that of the surrounding matrix and suggested it was covered by a layer of encrusting calcite. Based on my examinations, the Vienna Pterodactylus exhibits closely spaced broad actinofibrils similar to those of the Zittel wing preserved by way of calcification with dendritic deposits of manganese and iron oxides in the intervening strips (Bennett, Reference Bennett2013b); however, a detailed description of the wings of the Vienna Pterodactylus is beyond the scope of this paper and will be undertaken elsewhere.

The wings of Sordes pilosus described and illustrated by Unwin and Bakhurina (Reference Unwin and Bakhurina1994) included long, straight, closely packed fibers in the dactylopatagium that are clearly actinofibrils. Their preservation is consistent with the interpretation of actinofibrils as closely spaced broad keratinous structures in a somewhat stiff and inextensible dactylopatagium. Unwin and Bakhurina (Reference Unwin and Bakhurina1994, fig. 2c) did not provide measurements, but based on their figure the actinofibrils, though somewhat variable in width, were ~0.11 mm wide and spaced 6–7 per mm. Sordes had an estimated wingspan of 65 cm (Unwin and Bakhurina, Reference Unwin and Bakhurina1994), but if scaled up to the size of the Zittel wing the actinofibrils would be ~0.17 mm wide and spaced 3.8–4.5 per mm, quite close to the measurements of the Zittel wing’s actinofibrils. Note that advocates of widely spaced cylindrical actinofibrils on or in an extensible elastic patagium might interpret the close spacing of the actinofibrils as evidence of the contraction of the elastic membrane in order to furl the wing; however, it is also consistent with the interpretation presented here, and the actinofibrils are much wider (>2 times absolute width and >3 times relative width) than the Zittel wing’s raised longitudinal strips that some may still wish to interpret as actinofibrils. Unwin and Bakhurina (Reference Unwin and Bakhurina1994) did not describe fibers of any sort in the plagiopatagium, but described shorter fibers from the uropatagium, which they interpreted as having unraveled in some cases. They did not make it clear whether they interpreted the short fibers as homologous with the actinofibrils preserved in the lateral dactylopatagium of Sordes. Bennett (Reference Bennett2000) argued that the short fibers were not homologous with the actinofibrils of the dactylopatagium and suggested that they were elastic fibers, but I now wonder if they might not be the branching integumentary structures that Kellner et al. (Reference Kellner, Wang, Tischlinger, Campos, Hone and Xi2010) termed pycnofibers.

Kellner et al. (Reference Kellner, Wang, Tischlinger, Campos, Hone and Xi2010) identified two types of fibers in the patagia of Jeholopterus, and suggested that one and possibly both were actinofibrils. They suggested that actinofibrils could shorten or expand to provide flexibility and that such actinofibrils would not be incompatible with Padian and Rayner’s (Reference Padian and Rayner1993) load-bearing element interpretation or my spreading element interpretation (Bennett, Reference Bennett2000); however, extensible actinofibrils are incompatible with the latter two interpretations. Unfortunately, Kellner et al. (Reference Kellner, Wang, Tischlinger, Campos, Hone and Xi2010) did not provide illustrations or diagrams of fiber orientation and distribution, and they made no attempt to determine whether one or both of the two fiber types was homologous with the actinofibrils of the Zittel wing, JME SOS 4784 (Frey et al., Reference Frey, Tischlinger, Buchy and Martill2003), or the Vienna Pterodactylus, so their description adds little to our knowledge of pterosaur wings. However, based on their description it is probable that the short Type A fibers that were interpreted as actinofibrils are not actinofibrils but rather are collagenous tensile fibers like the longitudinal strings Frey et al. (Reference Frey, Tischlinger, Buchy and Martill2003) noted in JME SOS 4784, whereas the Type B fibers that were interpreted as possibly also being actinofibrils are keratinous actinofibrils. Note, however, that because anurognathids are the likely sister group to all other pterosaurs and possess wingfinger IP joints capable of flexion and extension unlike all other pterosaurs (Bennett, Reference Bennett2007), it is possible that the structure of the patagium of Jeholopterus and other anurognathids was distinctly different from that of other pterosaurs.

Bennett’s (Reference Bennett2000) spreading elements interpretation

Bennett (Reference Bennett2000) accepted that actinofibrils were widely spaced cylindrical structures on an elastic patagium and argued that they resisted longitudinal compression to prevent chordwise narrowing and redirected spanwise tension anteriorly to reduce loads on the distal phalanges. The present interpretation that the dactylopatagium did not include an extensible elastic membrane that would tend to narrow chordwise when under spanwise tension appears to negate the importance of preventing chordwise narrowing, but it is the presence of the closely spaced actinofibrils that would have made the dactylopatagium inextensible and determine its chord and area. The proposed function of redirecting spanwise tension from the plagiopatagium into more chordwise tension on the proximal wing phalanges would still be important for reducing the loads applied to the distal phalanges and allowing slender distal phalanges to bear the loads placed on them.

One reviewer simultaneously expressed doubt that keratinous actinofibrils would function to resist compressive loads and acceptance that they would function to redirect tensile loads anteriorly to proximal wing phalanges. Those views are incompatible because actinofibrils could not have redirected tension without resisting longitudinal compression. Bennett (Reference Bennett2000) discussed and illustrated this with what were termed strut and cable systems including struts representing actinofibrils. If actinofibrils could not resist any compression, they would when loaded in compression shorten until the lines of tension ran straight between their attachments to the body wall and hindlimb medially and the wing phalanges laterally, and so would not redirect tension at all. If actinofibrils could resist compression, they would not shorten significantly when loaded in compression and the lines of tension between their attachments to the body wall and hindlimb medially and the wing phalanges laterally would be bent by the actinofibrils, and the tension would be redirected to proximal wing phalanges rather than borne on distal ones. Perhaps the reviewer took my statements that actinofibrils would resist compression to suggest that they were subjected to and could resist compressive forces other than or exceeding those that tension within a tensed dactylopatagium would subject them to. However, I have never suggested that the role of keratinous actinofibrils was in resisting the sort of compressive loads that would be placed on them if one were to snatch a Rhamphorhynchus out of the air, put one’s thumb on the midshaft of WP1 and one’s index and middle fingers on the trailing edge behind WP1, and squeeze. In such a situation, the longitudinal compressive forces would exceed those which could be resisted by actinofibril and collagen fiber composite, the actinofibrils would bend, and the plane of the dactylopatagium would wrinkle and fold. However, Bennett (Reference Bennett2000) argued that in the context of the roughly planar dactylopatagium loaded in tension as in flight, actinofibrils would resist longitudinal compression and would be prevented from bending within the plane of the patagium by the collagen fibers of the dorsal dermis. If actinofibrils consisted of collagen or elastin rather than keratin, if they could “shorten or expand and therefore provide more flexibility to the wing membrane” as Kellner et al. (Reference Kellner, Wang, Tischlinger, Campos, Hone and Xi2010) suggested, then they would not have been able to resist longitudinal compression, would not redirect tension in the patagium, and would not prevent chordwise narrowing of the patagium under spanwise tension, and there would no explanation for their radiating pattern in the dactylopatagium.