Sunday, June 10, 2012

Feathers vs Membranes

A recent discussion arose on the Dinosaur Mailing List that included some questions regarding the relative merits of membrane wings and feathered wings, mostly in the context of pterosaurs vs birds. In that spirit, I thought I'd give a little rundown of the relative advantages/costs of each type of vertebrate wing.

Avian Wings
Birds are the only flying vertebrates to use keratinized, dermal projections (i.e. feathers) to form their wings.  Feathers have the distinct advantage of being potentially separate vortex-generating surfaces, meaning that a bird can split its wing up into separate airfoils, thereby greatly changing its lift and drag profile as required (Videler, 2005).  Tip slots are the most obvious example of this mechanism, whereby the tip of the wing is split into several separate wingtips by spreading the primary feathers of the distal wing.  The alula, which lies along the leading edge of a bird’s wing, and is controlled by digit I, is another example of a semi-independent foil unit (Pennycuick, 1989; Videler, 2005).  The splayed primaries of a slotted avian wingtip passively twist nose-down at high angles of attack (and therefore at high lift coefficients), and this feather twist reduces the local angle of attack at the distal end of slotted avian wings, preventing them from stalling (Pennycuick, 2008).  Slotted avian wingtips may therefore be nearly "unstallable", though this does not prevent the overall wing from stalling (Pennycuick, pers comm.).  Feathered wings can also be reduced in span without an accompanying problem of slack and flutter – the feathers that form the contour of the wing simply slide over one another to accommodate the change in surface area.  Despite these advantages, feathers have some costs as wing components, as compared to membranous wings.  Feathered wings are relatively heavy (Prange et al., 1979) and cannot be tensed and stretched like a membrane wing (which has ramifications for cambering). Theoretically, avian wings should not be able to produce maximum lift coefficients as high as an optimized membrane wing (Cunningham, pers comm.), but experimental data to determine if transient, maximum lift coefficients actually differ significantly between bats and birds are not yet available (Hedenstrom et al., 2009).

Chiropteran Wings
Bats have a wing surface formed primarily by a membrane stretched across the hand, antebrachium, brachium, and body down to the ankle.  Unlike birds, which have a limited number of muscles that produce the flapping stroke (two, primarily: m. pectoralis minor and m. pectoralis major), bats have as many as 17 muscles involved in the flight stroke (Hermanson and Altenbach, 1983; Neuweiler, 2000; Hedenstrom et al., 2009).  The membranous wings of bats are expected to have a steeper lift slope than the stiffer, less compliant wings of birds (Song et al., 2008).  This results from the passive cambering under aerodynamic load that occurs in a compliant wing: as lift force increases, the wing passively stretches and bows upwards, producing more camber, and thereby further increasing the lift coefficient and total lift.  While there are some advantages for a flying animal in having such a passive system, bats presumably must mediate this effect with the many small muscles (and fingers) in their wings – tensing the wings actively while under fluid load will mediate the amount of camber that develops.  This would be important to mediate drag and stall, though no empirical data currently exist to indicate exactly how bats respond to passive cambering.  The work by Song et al. (2008) also indicates that compliant, membrane wings achieve greater maximum lift coefficients than rigid wings, but data have yet to be collected demonstrating that this holds in vivo for bats and birds.  Compared to birds, the distal wing spar in bats is quite compliant (Swartz and Middleton, 2008).

Pterosaur Wings
The structure and efficiency of pterosaur wings is obviously not known in as much detail as those of birds or bats, for the simple reason that no living representatives of pterosaurs are available for study.  However, soft tissue preservation in pterosaurs does give some critical information about their wing morphology, and the overall shape and structure of the wing can be used (along with first principles from aerodynamics) to estimate efficiency and performance.

It is known from specimens preserving soft tissue impressions that pterosaur wings were soft tissue structures, apparently composed of skin, muscle, and stiffening fibers called actinofibrils, though the exact nature and structure of actinofibrils has been the topic of much debate (Wellnhofer 1987; Pennycuick 1988; Padian and Rayner 1993; Bennett 2000; Peters 2002; Tischlinger and Frey 2002).  Associated vasculature is also visible in some specimens, especially with UV illumination (Tischlinger and Frey, 2002).  Recent work on the holotype of Jeholopterus ningchengensis (IVPPV12705) seems to confirm that the actinofibrils were stiffening fibers, imbedded within the wing, with multiple layers (Kellner et al., 2009).  The actinofibrils were longer and more organized in the distal part of pterosaur wings than in the proximal portion of the wing, which may have implications for the compliance of the wing going from distal to more proximal sections.  The inboard portion of the wing (proximal to the elbow) is called the mesopatatgium, and was typified by a small number of actinofibrils with lower organization, which would have made this part of the wing more compliant than the outboard wing.

The outer portion of the wing, which was likely less compliant the mesopatagium, is termed the actinopatagium (Kellner et al., 2009).   Because pterosaurs had membrane wings, they could presumably generate high lift coefficients, but exactly how high depends on certain assumptions regarding their material properties and morphology (pteroid mobility and membrane shape being two of these factors).

Now, for some punchlines...
Based on the structural information above, we might expect the following regarding pterosaurs and birds:

- Pterosaurs would have a base advantage in terms of maneuverability and slow flight competency.

- Pterosaurs would also have had an advantage in terms of soaring capability and efficiency

- Pterosaurs would have been better suited to the evolution of large sizes (though this was affected more by differences in takeoff - see earlier posts about pterosaur launch).

- Birds will perform a bit better as mid-sized, broad-winged morphs (because they can use slotted wing tips and span reduction).

- Birds would have an advantage in steep climb-out after takeoff at small body sizes (because they can work with shorter wings and engage them earlier).  This might pre-dispose them to burst launch morphologies/ecologies.

Interestingly enough, the fossil record as we currently know it seems to back up all of these expectations.  For example, the only vertebrates that seem to have been adapted to dedicated sustained aerial hawking in the Mesozoic were the anurognathid pterosaurs.  Large soaring morphs in the Mesozoic were dominated by pterosaurs, also.  On the other hand, mid-sized arboreal forms in the Cretaceous were largely avian.

Full references for all of the above literature is available upon request.  I'll post the full refs here as soon as I have a chance, but just email me in the meantime if need be (currently traveling in Boston).


  1. Mike,

    A few observations on this:

    Cl max. In the absence of flaps and other high lift devices (which birds may have...) Cl max is a function of camber and not much else, so if birds had high camber wings, they would have high Cl max. I guess the point about membrane wings is that it is possible to vary camber much more than in bird wings so in practice bats (and pterosaurs) can achieve higher Cl max. Also, as Song and others have shown, a flexible membrane wing can generate higher Cl max due to the stretching, which in effect allows it to have a range of camber values as angle of attack changes.

    Re stall in birds, I suspect that the lifting covert feathers are also influential in delaying stall and making it 'softer".

    Now, re the pterosaur punchlines, I do not see why they all follow so directly. If pterosaurs had a higher Cl max capability then they would indeed be able to fly more slowly (as per my 2010 paper) but I am less clear about why they are necessarily better at soaring. The minimum sink rate is crucial to staying up and that is not a direct function of Cl max capability. It also has to do with wing loading - for thermal soaring (but not dynamic soaring of course) a low wing loading gives nice "floaty" flight.

    1. Hey Colin, thanks for writing in.

      Yes, Cl max will be mostly a function of camber, but the anatomy of avian wings prevents them from having the sort of camber that membrane wings can achieve. Furthermore, birds have comparatively little camber control, so if the wings were highly cambered they'd essentially be "stuck" at a high Cl.

      Good thought on the covert lifting. Birds have a few other anti-stall devices, one of them being that they can vary the angle of attack along the wing quite a bit without having to worry about flutter much.

      With regards to pterosaurs, I am just talking about "base values" as it were - it's not say that every pterosaur would be better at soaring than every bird. If pterosaurs had higher Cl max capability then the slow flight capacity and larger maximum size components follow relatively directly.

      Soaring is indeed trickier, but pterosaurs do seem to have some advantages there. For one thing, they tended to have higher aspect ratio wings than birds. That is an advantage for loitering flight because it reduces min sink, and it's a huge advantage for rapid oceanic soaring. Birds close this gap in loitering flight a bit by using tip slots, but tip slots are only really effective over a fairly narrow range of speeds.

      In loitering flight, a high Cl max can be important because it allows a tighter turning radius, which means the animal can stay in a smaller thermal if need be (though this will come at some loss because the min sink rate rarely occurs at Cl max). Obviously the min sink rate is crucial to successful loitering flight, particularly in thermal systems, but I don't see any evidence that the wing loadings of pterosaurs were likely to be appreciably higher than in birds. In fact, using some of the broader-winged reconstructions, the average WL for a pterosaurs comes out a bit lower than for a bird of similar mass. It's also worth noting that a low wing loading is really only good for thermal climbing; rapid travel between thermals is improved by higher wing loading so cross-country performance overall requires a balance. Many of the living birds that are loitering specialists don't actually have great cross-country ability.

      In the end, loitering flight might be a bit of a wash between birds and pterosaurs, but pterosaur morphology would seem to provide an advantage in nearly all other forms of soaring, so I have to tentatively give them the winner card on soaring flight, at least for now.

  2. Huh, I got here after searching "membrane vs. feathers" on Google. Really enjoyed reading this. Looks like I'm gonna go down this rabbit hole tonight :D