Friday, November 04, 2011

The Imperial Film

Note: Those who did not come here via Cyberthrush's blog might be interested in the debate between me and David Sibley on these matters in the comment section of this post.

Also note: My subsequent post elaborates more on the wingbeat rate and flight dynamics issues.

As should be widely known by now, the legendary film of a female Imperial Woodpecker shot in Mexico in the 1950s has finally been published and is available for all to see. Contrary to the conspiracy theories advances by Scott Crocker and others, this film does not in fact undermine the claim that the infamous video shot by David Luneau in Arkansas in 2004 is an Ivory-billed Woodpecker. In fact, even on first viewing, it would appear to support the Luneau et al Ivorybill claim.

The Lammertink et al paper covers the film itself, its history, and the present-day circumstances at the spot where it was filmed; I will not delve in to that here. Please read the article and view the video at the site I linked above if you have not already done so. My focus here will be on the impact this film has on analysis of the Luneau video, as it fills a major data gap that seriously impeded that analysis until now: the lack of any films of Ivory-billed Woodpeckers in flight for comparison. Of course, the Imperial Woodpecker is not the same species as our North American Ivorybill, but it is very closely related. I don't have the reference at hand, but wasn't there a recent genetic study that found that the genetic distance between the Imperial and the North American Ivorybill was about the same size as the distance between the Cuban and North Amercian Ivorybill subspecies? I believe it is fair and justifiable to use the Imperial as a close analog to what our Ivorybill would have looked like on film.

For the record, I had no advance access to this film. I saw it for the first time on the day of publication last month, same as everyone else. So I have not had months or years to examine it, nor were any of my earlier conclusions about the Luneau video influenced by having viewed the Imperial film, having spoken to anyone who had viewed it, or even having heard any credible rumors as to what it contained.

All of my previous analyses of the Luneau video can be reached from this link which is the same as the one in the right-hand sidebar on this blog. I will take this work as read and will not repeat it in detail, simply refering to my important conclusions and findings. Anyone who read my work could tell that I had very strong leanings towards an Ivorybill identification for this bird, and found multiple major and fatal flaws in the Sibley et al analyses that concluded that this video was consistent with a poorly-imaged Pileated Woodpecker. However, the lack of any suitable Ivorybill comparison material left unresolved dangling issues in several areas that have stopped me from proclaiming that this bird was definitively an Ivorybill. These issues fell into two areas: the launch sequence, and the flight mechanics (structure and pattern of wingbeats).

Launch behavior

First, to the launch sequence. This focuses on Infamous Frame 33 of the Luneau video, when a flash of white appears from behind a tree trunk just as the bird launches in to flight. The Cornell interpretation has been that this is the white shield on the back of the bird, and they used this to estimate the size of the bird. Sibley et al felt this flash showed the underside of a raised wing, and my own examinations of launch sequences of Pileated Woodpeckers led me to agree with this alternative interpretation. However, in contrast to Sibley et al, I found that the subsequent frame strongly suggested that the trailing edge of this underwing was white, not black, and that all the supposed black trailing edges they identified in the entire video were nothing but edge artifacts created by the digital imaging process. Hence even this alternative interpretation pointed to Ivorybill, not Pileated.

Now having seen the Imperial film, I have to say I may have been mistaken in disagreeing with Cornell's interpretation of this sequence. The behavior shown by the Imperial in the two launch sequences in the film is a very good match for the Cornell hypothesis and very different from that shown by typical Pileateds. Of course, as I mentioned before, both interpretations (when you stop misinterpreting imaging artifacts as plumage) come down on the Ivorybill side of the scale; but the Cornell interpretation is much stronger in both providing an unambiguous view of the "white shield" and in giving a size estimate that is outside of the Pileated range and right on for an Ivorybill. So this is a much heavier brick in the Ivorybill tray.

Flight mechanics: Flap rate

Discussions of the flap rate of the bird in the Luneau video generated a great deal of controversy. Its wingbeat frequency was higher than that shown in any of the Pileated Woodpecker comparison material available; in spite of claims of unpublished videos showing otherwise, there has still not been a video of a Pileated Woodpecker presented to the public in which the bird achieves and sustains such a rapid wingbeat rate. Of course, the real problem was that there was no film of an Ivorybill in flight to indicate whether or not this species actually possessed a rapid and sustained flap rate. Sibley and others argued that as a larger species, it should be expected to show a slower flap rate.

The Imperial isn't an Ivorybill, but it is very closely related and even larger. As discussed by Lammertink et al, it shows a very fast flap rate, considerably faster than what is typical for a Pileated. In the first 5 wingbeats of the second flight sequence, it achieves 8 flaps/second, nearly matching the Luneau bird.

I think we can safely conclude now that the Ivorybill had a fast, sustained flap rate, significantly faster than that of the Pileated, and likely extremely close to what the Luneau bird shows.

Flight mechanics: Wingbeat patterns

The flight of a Pileated Woodpecker after launch typically consists of a short series (3 to 7) of fairly rapid wingbeats, then a break into bounding flight where individual or paired wingbeats are separated by bounds (periods when the wings are folded in to the body and the bird functions as a ballistic missile). Sustained cruising flight of a Pileated generally continues this pattern of discrete flaps separated by bounds: flap....flap,flap....flap....flap....flap...

The flight of the Imperial contrasts sharply with this. In the first flight sequence the bird executes 7 wingbeats without a break before it moves out of sight. In the second it completes 15 steady wingbeats then executes a bound. After the bound, it resumes steady flight with 5 more uninterrupted flaps before it disappears from view. In the final sequence, the launch is not visible. When the bird appears in frame it is in steady flight, and completes at least 12 wingbeats before closing up into a bound and "cannon-balling" until the end of the film. One could describe the flight of this bird more as "flapapapapapapapapapapapapap.....flapflapflapflapflap"

The Luneau bird is in view for 4 seconds from launch to its final diappearance. It passes behind trees intermittently during this time. It is in continuous view shortly after launch for about 1.2 seconds, during which it completes 10 uninterrupted flaps without a hint of bounding flight. For the rest of the video, every time it reappears between the trees it appears to be engaged in steady wingbeats without bounds. There is no point at which the wingbeats appear to have stopped or even be separated by brief bounds. The bird completes 23 visible wingbeats during these 4 seconds; if you watch in slow motion and count along with the rhythm there seem to be no breaks, and interpolating behind the trees yields about 27 flaps in this time. This is a far better match for the Imperial than for a Pileated.

Flight mechanics: Wing movements

The film is shot at 24 frames per second (fps) which means that there are only a few frames (typically 3-4) representing each wingbeat. Hence, in any given wingbeat there are only a few wing postures shown. However, each flight sequence shows multiple wingbeats, and the adjacent flaps are often viewed from a similar angle and occur at about the same rate. However, as the flaps and camera shutter are never in perfect synch, each successive wingbeat is imaged at different point in the flap cycle. So, one can composite several adjacent wingbeats to create an approximate representation of what a single wingbeat would have looked like had it been shot at a higher frame rate. This is similar to the manner in which films of rotating objects (vehicle tires, fans, etc.) often will appear to show the object spinning in slow motion, or in reverse. I have constructed these composite wingbeats for two of the three flight sequences in the Imperial film, with interesting results. The second sequence yields the most attractive result, whith three successive flaps combined in to one:

Note that the bird jumps around in the frame, and the background does all sorts of distracting things. But if you concentrate on the bird, this yields what looks like a pretty fair representation of a singe flap filmed at 72 fps (by a very shaky hand).

The third flight sequence yields an interesting composite (again three separate flaps combined for a simulated frame rate of 72 fps):

In this case we see the bird almost directly from behind.

The first flap sequence does not yield a good composite. The bird traveled far across the frame, and the flaps were almost in synch with the shutter so there is less of an effect of each frame filling in the gaps from the others to create a smooth wing motion. And for purposes of comparison with the Luneau video, this side-on viewing angle is perhaps the least informative of the three.

Note that both of these composite sequences show the bird in cruising flight, not during the initial takeoff and acceleration. The salient feature notable here is the relatively shallow nature of the wingbeats compared to Pileated flight. The wings are not fully opened on the downstroke until they are at about a 45 degree angle, and they are draw in rapidly at the bottom of the stroke without being fully extended downwards. This is of course exactly what is shown by the Luneau bird and is not typical of Pileated flight.

Flight mechanics: Bowed wings

One of the salient features I found in the Luneau video that was inconsistent with Pileated flight was the manner in which the bird bows its wings downward during the downstroke. This is not apparent in the two composite sequences of the Imperial flight shown above. However, it does appear earlier in the Imperial flight sequences, most prominently in sequence 2 while the bird is accelerating and banking. A final composite shows the bird in mid-downstroke on 6 successive wingbeats:

The flap rate of the Imperial at this time was almost exactly 8/s, so the wings are imaged in nearly the exact same part of the cycle in each frame. Note that the bow increases as the bird banks, then flattens out when cruising flight is achieved. The wing bowing never looks as prominent as is shown by the Luneau bird, but it is present nonetheless. It also appears to be present in the side views of the first Imperial launch sequence in frames 654, 658, and 663. It does not show at all in the third flight sequence, which is missing the initial acceleration and also shows the slowest flap rate of the three.

The naked eye view

The frame-by-frame analyses are revealing about the behavor of the bird, but they do not necessarily reflect directly on what a birder in the field sees when a large black and white woodpecker flies off a tree and vanishes in to the forest. To give a suggestion of what this might all look like to the naked eye, I created a mash-up of 5 segments of fleeing large woodpeckers. I used the first two Imperial flight sequences, the Luneau Video, and two of David Nolin's Pileated videos. For the Pileateds I deliberately chose the two most Luneau-like of them, showing a bird executing rapid flaps and sustaining steady flight for the longest time before breaking in to bounding flight. The modern digital videos are deinterlaced at 30 fps (alternate deinterlaced frames); the Imperial film has been adjusted from 24 to 30 fps by duplicating every 4th frame (standard film-to-video conversion procedure). You likely will need to play this video repeatedly to take it in and start to get a sense of the comparisons between the birds. The mash-up is not intended to show detailed wing movements, etc; it is deliberately shrunken so you just get the quick flash impression, as though you were startled in the woods by stumbling upon one of these birds.

First to the marked similarities. All the birds explode from the tree in a barrage of rapidly flashing black-and-white. The differences in flap rate are not really discernible initially. And all take off like a chiropteran out of Hades. The Pileateds' breaks from steady flight to rapid bounding flight are noticeable at the end of their sequences on repeat viewing, but it is not a dramatic shift. Remember that I deliberately chose the two most Luneau-like (and also Campephilus-like) videos I had on hand for this comparison. The real contrasts to the naked eye would probably happen later in the sequence, when the slow bounding flaps of a Pileated would contrast sharply with the sustained rapid flaps of a big Campephilus (perhaps broken by well-defined bounds); but lacking a longer Imperial movie this is not knowable for certain. Overall I would say that in a place where a large Campephilus was known to occur alongside Pileated Woodpeckers, then a practiced eye could learn to distinguish the two species from a rapid naked-eye glimpse as they explode off trees and vanish into the forest based on flight gestault. Hopefully this situation will arise again somewhere in the future; at the present there is nowhere on earth that matches this description.


As I mentioned before, I have in the past generally hedged my declarations about the identity of the bird in the Luneau video, stopping just short of proclaiming what species the bird is rather than what I felt it was most likely to be. Enough of that.

The bird in the video shot by David Luneau in Arkansas in 2004 is an Ivory-billed Woodpecker, Campephilus principalis. I have no residual doubts, either personal or scientific. This species did not go extinct in 1945 or 1984, it persisted at least through 2004 and is quite likely still extant as I type these words. All the anomalies identified a priori (without foreknowledge of the Imperial film) in the flight of the Luneau bird that did not appear to line up with a Pileated Woodpecker do line up with the documented flight of an Imperial Woodpecker, the Ivorybill's closest relative. Combine this with the fact that there are no other discernable features of the bird that are inconsistent with the Ivorybill ID and several that do not match a Pileated (again, when you stop trying to make plumage markings out of image artifacts), and the case is closed.

I will echo what I said in some of my other Ivorybill posts. This creature is not a ghost, a fantasy, or a yeti. It is a bird. A very rare bird that is very difficult to locate and relocate. But it is a bird nonetheless, one that is probably alive and pecking on a tree somewhere within a few hundred kilometers of where I sit right now. Birders, ornithologists, and conservationists should treat it as such.

If any of these video links fail to function, please post a comment to let me know this so I can address the issue. I work over dialup so it is often hard for me to evaluate these things myself. Thanks!


At 9:38 PM, Anonymous Anonymous said...

Good work. Thank you for addressing this. Wish you could publish...

What do you make of Mike Collins' claim (if I understand it correctly) that Cornell's interpretation of the launch violated laws of physics? I can't point to where his claim was, but he was certainly dogmatic about it.

At 4:34 AM, Blogger Bill Pulliam said...

Re: Collins; as I recall it had to do with the apparent acceleration needed to pivot the whole body as rapidly as would be necessary to match their hypothesized launch sequence. I am probably not remembering this right, but I think it had something to do with how many times higher than g (gravity) would be required. But there's no special problem with accelerating body parts many times faster than g to my mind; we do it all the time with our fingers, arms, and legs, and even our whole bodies when we leap into the air, and we are much more massive than a woodpecker. There was probably something more to it that I am forgetting or did not catch, however. The fact that the real Imperial moves in much the same way and at the same speed as the inferred Ivorybill, I take that as evidence that there's nothing physically or anatomically wrong with the hypothesis.

Re: publishing, thanks, but as a successful escapee from Academia, that is one of the last things I am interested in doing. Besides, the Ivorybill is the third rail of ornithology and conservation biology, and I suspect this has as much as anything else to do with why it took so long to get this Imperial paper published! Considering the slander Science endured for the original paper, not many other editors likely want to touch this topic.

At 5:22 AM, Blogger fangsheath said...

It is worth noting that even before they begin to exhibit very distinct "bounds," the Nolin pileateds exhibit close-winged pauses, often only 1 or 2 frames in length at first. It is this kind of interruption in the rhythm that is most noticeably lacking in the Luneau video and imperial film. I believe this contributed to the historical characterization of ivory-bill flight as "ducklike" - when not in a bound, the wingbeats cycle very steadily.

At 5:45 AM, Blogger Bill Pulliam said...

Definitely; I was considering these pauses to be "mini-bounds," really. After the first 3-7 flaps, Pileateds indert a pause after most wingbeats. As the transition from takeoff to cruising completes in a Pileated the pauses grow into longer bounds, yielding the classic view we have all seen of a Pileated covering a long distance with just single widely-spaced flaps.

And indeed there are none of these pauses evident in either the Luneau or Imperial flights; though Luneau gets harder to judge towards the end and none of the Imperial flights are visible for as long as it takes the Luneau bird to completely vanish into the woods (or out of the frame; hard to tell for sure which it did).

At 10:56 AM, Anonymous Anonymous said...

Impressive analysis!

At 3:08 AM, Blogger cyberthrush said...

"...within a few hundred kilometers of where I sit right now"

Bill, (depending how one defines "few") just curious if that comment was meant as a reference to certain parts of western TN. or simply a more general allusion to Arkansas's Big Woods (or possibly even other areas)?

At 5:12 AM, Blogger Bill Pulliam said...

A general comment meant to cover the entire Lower Mississippi Valley region. If there was one in Arkansas in 2004, there could be one just about anywhere in this huge interconnected web of riparian forests in 2011. "Few" is of course deliberately ambiguous, because I in fact have no idea where they would actually be within this region at the present. Of course, there are all sorts of extremely rare birds within a "few" hundred kilometers of me right now that will never get seen by any birder, that's just part of the large numbers - small probabilities game!

At 9:18 PM, Anonymous Anonymous said...

Bill, I was a bit amused by how David Sibley took you to task on Cyberthrush's blog.

David Sibley wrote: "Bill is apparently unaware of Louis Bevier's website which has a detailed discussion of wingbeat rates showing that Pileateds can flap very quickly on takeoff, more quickly than the Imperial, and can keep it up just as long."

Can keep it up just as long? So here is what Lammertink et al. reported for the Imperial: "In the second launch (starting at frame 1,761), closer to the camera and perhaps an escape flight, the flap rate is 7.7 s–1 (up to 8.3 s–1) through wing-flap 8, and 7.3 s–1 (up to 7.9 s–1) through wing-flap 14." Note 7.3 to 7.9 s-1 through wing-flap 14.

At Bird Forum in June 2008, Louis Bevier conceded that dave_in_michigan's interpretation was correct (, with the PIWO in Bevier's yet-to-be-released video exhibiting 8.8 Hz in the first 5 cycles over 0.6 sec, 7.6 Hz in the next 4 cycles over 0.5 sec, and 6.0 Hz in the next 3 cycles over 0.5 sec. So even though Bevier claimed the cumulative was 7.5 Hz over 12 cycles, it had slowed to 6.0 Hz after the first 9 cycles. This is quite the departure from Sibley's understanding of Bevier's sleight-of-hand claim.


Post a Comment

<< Home

Site Meter