Tuesday, February 21, 2006

Updated Luneau video analysis

Cornell's detailed Luneau video analysis is now available here. Cornell's summary lists nine "diagnostic traits" that lead them to believe that the Luneau bird is an Ivory-billed Woodpecker.

I disagree with Cornell on all nine points--I see nothing in that video that is inconsistent with a completely normal Pileated Woodpecker.

Below, I've listed each of Cornell's points, followed by my opposing viewpoint:

(1) The underwing pattern in flight consistently appears largely white, giving the appearance of having black wingtips but lacking any black along the rear, or trailing edge.
In many frames, both underwings do appear to have trailing black edges.

Here is an example:


Note that in Cornell's paper, they state: "With these distances and light conditions, bleeding tends to exaggerate the apparent extent of white in the wings." I think this explains the faintness of the trailing black underwing edges.
(2) The upperwing pattern in flight consistently shows a broad, white trailing edge, with no frames demonstrating the conspicuous dark rear border to be expected of normal Pileated Woodpeckers.
I think we are seeing the bird's underwings, not the upperwings. The pictures here show that a fleeing, normal Pileated shows a lot of flashing white underwing, both when the wings are above horizontal and when the wings are below horizontal.

In support of their point #2, they inexplicably show us a video of a Franklin's Gull, flying away in a leisurely fashion. This is supposed to "prove" that a fleeing Pileated's dorsal wing surface should be prominently visible when viewed from behind.

This comparison makes no sense to me. Instead of admitting that their stiff-winged Pileated re-enactment model was seriously flawed, they went far afield to find a species where a stiff-winged model might have been more reasonable.
(3) The wings are longer relative to the body diameter than in Pileated Woodpecker and consistent with the wing shape of Ivory-billed Woodpecker.
I'm completely unconvinced that you can accurately measure wing length, much less body diameter, from this low-quality video.
(4) Reenactment of the scene using life-sized, realistically painted, dynamically flapping models produced images remarkably similar to those of the Luneau video using the Ivory-billed Woodpecker model, and images clearly identifiable as Pileated Woodpecker using a model of that species.
Cornell posted this picture of the reenactment models:

These models are fatally flawed in that their stiff, board-like wings simply can't be made to fly like the Luneau bird. (In addition, the smooth surfaces of the models may reflect light differently than the feathers of an actual bird.)
(5) The wingbeat frequency is 8.6 beats per second, which is almost identical to that recorded for Ivory-billed Woodpecker (as documented by one acoustic record from 1935). The wing-beat frequencies of Pileated Woodpecker are not known to exceed 7.5 beats per second, and more typically range between 3 and 6 beats per second.
The above argument is countered by information from a comment here:
----
This really is unconvincing. They record a wing-beat hz at 8.6 and compare this to their records of 2-4 beats level flight "and 4-7.5 beats for short periods during hasty departures (n = 5)". "Moreover, experts who have studied Pileated Woodpecker flight using video analysis timed the fastest departures at 7 beats per sec (Tobalske 1996, personal communication). Thus, wingbeat frequency of the woodpecker in the Luneau video is faster than any recorded Pileated Woodpecker".

A sample size of 5 is nowhere near big enough to rule out a Pileated being able to do 8.6, whilst the Tobalske reference (paper published in Auk) is one study where "nonmaneuvering flights with approximately no change in mean altitude were included in the sample". The paper specifically did not look at flapping flight of fleeing woodpeckers. Thus the sample size of fleeing Pileated flights is tiny.

To then try and suggest that this ties up well with the sound recording of an Ivory-bills wingbeats (sample size = 1) is equally flimsy.
----
Also note that the Luneau bird's wingbeat frequency slows significantly after the initial flaps.

(6) White plumage on the back is visible on the retreating bird as it begins to gain altitude. Ivory-billed Woodpecker has white on the back; Pileated Woodpecker has entirely black back.
I think Cornell completely misinterpreted the position of the bird; my explanation is here.
(7) The dorsal view of the right wing as it begins to unfold shows a triangle of white that matches in size and position the white on the folded wing of an Ivory-billed Woodpecker beginning to launch into flight.
I think there are at least two problems with this argument.

1. Below is the inset sketch from Cornell's paper, followed by an Arthur Allen (1935) photo of a perched Ivory-bill. On the folded wing, note that the white patch on the inset sketch is much more extensive than on the Allen photo; also note that the shape of the white patch differs between the sketch and the photo.






2. Let's carefully consider the Luneau bird's movements during the 10-frame "launch sequence". I would argue that Cornell's "folded wing" theory is an extremely poor fit for the contents of those 10 frames.

For this discussion, let's call field 33.3 (the point of maximum wing flash) Frame 0. Note that at about 60 frames per second, the spacing between frames is about .0166 seconds. We know that at Frame -2, the bird's body is completely hidden by the tree; by Frame 7, the bird's right wing appears on the right side of the tree--the bird is in full flight, evidently well away from the tree, with the right wing near the top of the upstroke.


Under Cornell's theory, in response to the approaching canoe, the Luneau bird lunges powerfully to its right (our left), flashing a lot of folded wing within two frames (.033 seconds), but ending with its head still fully behind the tree (as pictured in Fitzpatrick's inset sketch above). The bird then immediately checks its momentum and lunges powerfully to its left (our right), with its wing still almost entirely folded by Frame 1.

Now we need to consider what the bird was doing behind the tree, between Frames 1 and 7. I would argue that according to Cornell's "folded wing" theory, the bird is now caught "between flaps". The Luneau bird takes about 7 frames per wingbeat. When we see the right wing held high in Frame 7, is that the first or second time that the wing was in that position? If it's the first, why did the wing open so slowly, and how could the bird possibly have lunged so powerfully? If it is the second, how did the Luneau bird complete a wingbeat cycle within about 3-4 frames? (Note that if the wing was just opening in Frame 1, it likely wouldn't be fully raised until Frame 3 or Frame 4, leaving it only 3-4 frames to complete one full wingbeat cycle by Frame 7).

I think the folded wing theory makes no sense. I think Jackson's "fully opened wing" theory is a much better fit. Under that theory, in response to the approaching canoe, the bird doesn't need to lunge violently right and left--it simply launches as normal, raising its wings and completing one wingbeat cycle in the normal 7 frames.

Below is a blown-up section of Field 33.3 from Cornell's paper; I think it is more correctly interpreted as the underwing of a normal Pileated Woodpecker as shown.

Note that under my interpretation, the black border to the left of the white blob is the black trailing edge of a normal Pileated's underwing.


Cornell rejects the "vertically held underwing" hypothesis for two reasons.

First, they claim that the "tail of the woodpecker is still pointed down vertically in Field 33.3, meaning that the body of the woodpecker is perched vertically at this moment". I'm unconvinced that the woodpecker is perched vertically in Field 33.3--for one thing, the black smudge that Cornell interprets as the bird's tail is still visible on the tree even after the bird has flown away. A black smudge does appear to the left of the tree in Fields 50 and 66.7--this appears consistent with a Pileated's tail, extending out and to our left as the bird launches from the far side of the tree, then turns to its left to flee.

Second, they attempt to rule out a Pileated's underwing via a precise calculation of the width/length ratio of the white patch in Field 33.3. I think this ratio cannot be accurately calculated from such a low quality video, especially since they admit that "...bleeding tends to exaggerate the apparent extent of white on the wings".
You can read more about Field 33.3 here.

(8) The distance between the wrist area and the tip of the tail (32-36 cm, as measured when the bird begins to take flight) is comparable to known measurements of Ivory-billed Woodpecker and considerably larger than even the largest Pileated Woodpecker we measured.
See (7) above. I think Cornell's wrist-to-tailtip measurement is meaningless, since they've completely misinterpreted the position of the bird's right wing.
(9) Only 20 seconds before the woodpecker flees, a bird with the size and color pattern of an Ivory-billed Woodpecker was perched within 3 m of the site from which the woodpecker took flight.
There are many candidate "six-pixel Ivory-bills" at various places in the Luneau video, and I think that all of them are just artifacts of an unfocused camera. Here are some of them:


According to Jerome Jackson, John Fitzpatrick admitted that the "six-pixel bird" in Cornell's paper is a likely a branch stub.
================================================
Some other notes and observations:

A) Regarding flight style:
Note that James Tanner said "In flight the Ivory-bill looks surprisingly like a Pintail".

In my opinion, the Luneau bird's flight style is nothing like a Pintail; in fact, the Luneau bird's flight style is completely consistent with a Pileated Woodpecker.

Compare the Luneau bird's flight style with this known Pileated (Cornell's Clip 11), and then compare it to this Yellow-billed Pintail video.

(Note: The Pintails are not in powered flight throughout the entire video. However, the video shows a few flaps of powered flight as the birds fly to the left against a headwind, and it shows more powered flight as the birds turn to the right and fly away from the camera.)

Below, I've sketched the wingbeat cycle of a fleeing Pileated, with the downstroke on the left and the upstroke on the right. Note the acute angle of the wings at the top of the upstroke, and note how far the wingtips extend downward on the downstroke. On the upstroke, note how the wingtips are pulled closer in toward the body.

In contrast, note that a Pintail's wings in powered flight are held relatively flat, with the wingtips held fairly equidistant from the body at all times. Note also that the wingbeat is quite "shallow", with the wingtips never positioned very high or very low relative to a horizontal position.



B) If the Luneau bird is an Ivory-bill, then where is the black underwing line? Note that Cornell provides this intentionally-blurred Ivory-bill picture, but you can still see the black underwing line:



(Yes, I know that the "white bleeding" argument has been used to explain the absence of the black underwing line in the Luneau video. The black trailing edge of a Pileated's underwing seems to be visible in many frames showing the flying bird, and it also appears to be present in Cornell's figure S1 [see point 7 above]; however, in no frame do I see any hint of a black underwing line.)

C) Here's what Tim Gallagher says about the Luneau video in "The Grail Bird", pages 224 and 225 (the bold font is mine):
In the blown-up film, I could see what appeared to be a large bird with a black-crested head and a white bill peering out from behind a tupelo...I was completely floored. Virtually all of the ivory-bill's major field marks were there, albeit fuzzy.
Gallagher is listed as one of the author's of Cornell's online Luneau video analysis. Which frame(s) show the black-crested head and white bill?

D) Questions about the Luneau audio are here.

E) You may be interested in some more Luneau video analysis sent in by readers--a composite picture here, an analysis of the Luneau bird's underwing here, and some "red dot analysis" here.

10 comments:

Anonymous said...

A truly outstanding analysis.

In the blown-up film, I could see what appeared to be a large bird with a black-crested head and a white bill peering out from behind a tupelo...I was completely floored. Virtually all of the ivory-bill's major field marks were there, albeit fuzzy.

This quote shows how Cornell can see all of this convincing evidence in the video when other experts can't. It is a case of people seeing what they want to see.

D.Stadnyckyj said...

"In the blown-up film, I could see what appeared to be a large bird with a black-crested head and a white bill peering out from behind a tupelo...I was completely floored. Virtually all of the ivory-bill's major field marks were there, albeit fuzzy."

I see a head/crest also but no bill which would be a very telling field mark.

I don't see an IBWO though. I see a Pileated.

This is what I don't understand. If Mr.Gallagher sees this also than why is there so much confusion in the birds' orientation and whether we're viewing a raised or folded wing?

Anonymous said...

Well I have heard that the video looks better in the "movie" version.
But sharper? Anyway, the wing-shape
looks Pileated. I can see from other
videotapes of blurred Pileated's that it doesn't take too much blurring to lose the red crest completely and make it seem black, or smudged. Blown up version or not blown up. The sooner this video disappears from the Cornell "proof" pile, the better they will look.

Paul Sutera, New Paltz, NY

Anonymous said...

Many points in this hinge on the claim that as the bird is flying away, we are not seeing its back and upperwing.

I have been studying the woodpeckers around my house and orchard as they fly away from me when flushed off tree trunks. When they are headed away, above eye level and rising slightly at a pretty small angle, I see their backs. The flickers show white rumps, the sapsucker shows pale back and white wing patches, the downies show white backs. Most of what I see on these fleeing woodpeckers is upperwing and back. These arguments that this video is showing mostly the underside of a pileated even when it is heading off into the distance just don't seem to fly. Without this, a lot of the rest falls apart.

Anonymous said...

These arguments that this video is showing mostly the underside of a pileated even when it is heading off into the distance just don't seem to fly. Without this, a lot of the rest falls apart.

Did you actually read through the whole post and study the links, including the ones that show, counter-intuitively, that Pileateds flying away above the horizontal plane still show large flashes of white from the underwing?

In key frames, we may or may not be seeing the bird's back, which may or may not be white, depending on the frame, and even if white, the white may or may not be caused by digital artifacts or vegetation, and even if there are white feathers, they may or may not be white dorsal stripes.

It seems to me that Cornell's claims are crumbling fast.

Anonymous said...

Yes I read the whole thing and Cornell's whole thing and much of the rest of what has been written here. I'm talking about the longest part of the video when the bird is flying away not the first few bits where it has just launched from the tree. I look at real life reenactments when real woodpeckers flush and fly away the same way. I see a bit of underwing here and there but most of what I see is back and upperwing that is by far the most obvious thing. Go look at more real birds not just videos and see what you see.

Anonymous said...

I believe in apples to apples where possible. Since the "best" parts of Cornell's video matches, very closely, known Pileated video, in my mind that trumps Cornell's theories on what an Ivory-Bill might look like if out of focus and pixelatd.

How do you explain, for example, that there is no hint of the black stripe of the underwing of the Ivory-Bill in ANY frame?

Did you see the photos of fleeing Pileateds that were showing wide flashes of white underwing at the top, middle and bottom of the wingbeat?

Somebody, somewhere, needs to get some decent photographic proof.

Anonymous said...

With regard to the woodpeckers in the orchard, I would suggest that the Nelson analysis in no way precludes the possibility that some dorsal surface show to the camera. I'm doubtful that significant dorsal surfaces do show, but even if they do, Nelson's analysis simply states that the white seen in the video can much more reasonably be explained as underwing than upperwing. I agree. I've suggested here before that if top-wing is going to show at all, it will show on the up-stroke. I'll add to that a suggestion that if top-wing is going to show at all, it will show on the secondaries. The white in the Luneau video is on the downstoke and heavily in the primaries. This is underwing. I agree with the suggestion to go out in the field and take a look. Take a look a primaries on the downstroke. They will be underwing, as I believe is all of the white in the Luneau video.

Anonymous said...

Tom,

I am just interested to know how one can call the frame 33.3 the underside of an open Pileated wing when there is absolutely NO black trailing edge when you look at that frame from the video. There is black above but it doesn't continue down behind the white. If you also look at Cornell's view of that image, of the perched birds wing, where they use the specimen placed behind the tree, it's an exact match. Including the little notch in the top of the white.

Thanks, P. James

Anonymous said...

If you also look at Cornell's view of that image, of the perched birds wing, where they use the specimen placed behind the tree, it's an exact match.

You mean in Cornell's rebuttal to Sibley's rebuttal? Where they have the bird hanging on with one foot, perched sideways on the tree?

When you're looking at blurry black and white blobs, you can manipulate them and "see" whatever you want, which has been readily proven. If it appears to be a Pileated if you look at it a certain way (which you can) it almost certainly is.