Sunday, February 22, 2009

Empiricism and theory

So, the experimentalist is giving a little in-house symposium one day, presenting some unexpected results that he had encountered. He puts his graph up on the screen, showing a sharp peak. Immediately the prominent theoretician interrupts him, saying "It is obvious what is happening, you would expect to see that peak in precisely that location." She then continues with a long derivation explaining the results. Before she finishes, though, the experimenter interrupts her and says "Oh, sorry, my mistake; the figure is upside down!" He corrects it, and the theoretician resumes, "It is obvious what is happening, you would expect to see that trough in precisely that location..."

Always keep this in mind when perusing theoretical armchair musings; my own as much as those of everyone else!

Wednesday, February 18, 2009

Something new in the Ivorybill world?

This has come up recently over at In response to discussions about things that go bump in the woods, "Fangsheath" posted a link to the Singer Tract recordings of real, for sure Ivorybills communicating with each other. Mixed in with the vocalizations are rapping sounds, which alas do not include any double raps** of the sort typical of other Campephilus species and described for the Ivorybill in many historical accounts. But what the posters in that forum have discussed is that the ordinary-sounding raps actually might be occurring in a consistent, repeated pattern, rather like a very slow-motion drum. Take a listen for yourself (the audio clip runs about 10 minutes; it consists of the same 5-minute sequence played twice):

This may seem silly and unnecessary, but I remind everyone ONCE AGAIN that these are for sure, real, unquestionable Ivorybill sounds recorded in the 1930s by people who were watching the birds make the sounds; I'm not just going off the deep end babbling about odd noises of unknown origin in some random tract of woods.

The thought here is that what initially sound like innocuous series of raps actually seem to follow a consistent, repeated pattern, a short sequence that in at least one occurrence accelerates: Rap...Rap...Rap...Rap..Rap..Rap. Remember that these birds are at an active nest, so these raps are communication, not foraging.

Examining it in detail reveals the following:

A total of 9 rap series, ranging from 3 to 11 raps. The median rap spacing ranges between series from 0.13 to 0.29 s, but within an individual series the range is much smaller, about 10-20% between the longest interval and the shortest one. In every case but one the series accelerates from the first to the last interval (by about 10-20% of course). In the one exception, the series accelerates for the first 4 intervals, then decelerates for the final 2. These numbers are just from my first pass through, I will not be surprised if a more detailed analysis changed them in detail but not in overall patterns. This to me is actually pretty suggestive of a patterned display, not just casual rapping. I leave to others the semantic distinction as to whether this constitutes a "drum" or some other term.

Very interesting possibility that this might be a specific, stereotyped pattern and not just banging on trees. If so, and if any of us have actually been hearing Ivorybills banging in the woods in recent years, one might expect that this display should be heard as well, perhaps more often than the double rap.

Anyone wanna reanalyze a few gazillion hours of ARU data?

**It is always said that this recording includes no double raps; however, what ARE the more muffled sounds at 121.5 and 124.1 sec, then? These are part of what sounds like a more distant rap sequence responding to the closer rap series. They sound to me like double raps with a fairly long interval, about 0.18 sec between the raps, more like the Florida ARU sounds than the Arkansas ARU sounds and not like what was historically described as a very short interval between the raps. Of course this spacing is in the range heard in the longer series, so perhaps these might better be considered "series" of just 2 raps, which might be distinct from the true and undocumented "double rap." Interesting... even with such limited hard data, there still might be more things to be discovered in it!

Sunday, February 15, 2009

Unfortunate timing

Right now in the middle of the Great Backyard Bird Count, it seems eBird has gone on the fritz. The message says they will be down much of today and tomorrow. Alas, this seems to mean that hundreds of people who will click the eBird link off the GBBC page to see what it is about will just get an error message. I can't imagine this outage was planned; it must just be bad luck. Still it cuts out one of eBird's big potential recruiting times, and WE NEED MORE OBSERVERS! It is also the final day of the Rusty Blackbird Blitz, though I suspect those observers are more established will be likely to check back later and try again.

Tuesday, February 10, 2009

From an Undisclosed Location...

Steve Sheridan's 2007 photo of what appears to be a large white-backed woodpecker in a cypress slough has finally been made public. If you have not yet been there, you should go look at the original website first, then come back here to read my discussion. I'm only going to focus on this one color image for now ("Mystery Bird #3"); there has been far more activity than this in the area, including kents, double knocks, sightings, and reconyx images that, though still inconclusive, are more intriguing than any I have seen from other sites (and I have seen plenty...).

This image has been circulating confidentially for a long time; it has been reviewed by State and Federal agencies and an array of external reviewers, many of whom are far more prominent than I. Obviously this image is not the "smoking gun" that proves beyond a reasonable doubt to all people involved that the Ivory-billed Woodpecker was still flying free and alive in 2007. But neither is it easy to just dismiss out of hand. Some of us have been scratching our heads in secrecy about this shot for well over a year; it's good to finally have it out in the open!

To summarize the story of the image, it is part of a long sequence of photos that were shot not of the mystery bird, but of the Pileated Woodpecker that is seen front and center in the full image. The photographer was not aware of the presence of the second bird until after the fact; it appears only in that one shot. No trace of it has been found in any of the other images. The images were not rapid-fire; many seconds elapsed between adjacent shots. For now I am not revealing the location; there is concern about disurbance to the site, where there are ongoing searches by private citizens and official agency crews. This info may be revealed in the near future; that is not my call to make. I will just leave it as having been taken in the United States within the documented historical range of the Ivory-billed Woodpecker.

After initial excitement, a substantial proportion of those who have reviewed this image have ultimately come down on a fairly negative, or at best indecisive, stance towards the possibility that it might represent anything other than a Pileated Woodpecker. No one, including me, has found any indications at all that the image is fabricated; indeed it makes very little sense that someone would go through all the trouble to fake an image so carefully, and yet still leave it so frustratingly ambiguous. There are two big questions that immediately leap out about it:

1. Is the "white shield" really part of the bird, or is it a trick of light or intervening vegetation?
2. Where is the "white suspender" on the bird's back? Or, 2a: Could a white back stripe be hidden by the wing?

Conversely, attempting to reconcile the bird with a Pileated, even if you allow for the possibility of an anomalous white shield, still runs in to two other basic questions:

3. Can a Pileated's neck "look like that;" i.e. show such a narrow white stripe and such a broad expanse of black between the white stripes?
4. Can a Pileated's crest look like that, so large and bright and bushy?

Jumping ahead to a preview of my answers, before I actually present the reasoning behind them:

1. The white shield is part of the bird.
2. A dorsal stripe could easily be obscured by the bird's wing.
3. It does not seem that a Pileated's neck can look like that.
4. Yes, a Pileated's crest can look like that.

So again spoiling my punchline, there are difficulties with either identification. Other than the fact that the Ivorybill is such a phenomenally rare bird, that species is the easiest to reconcile with all the features in the image.

Now, to why I come to these answers:

1. Is the "white shield" real?

This has been one of the biggest items of discussion, and it is the point where I think that most reviewers have gotten it wrong. Many have decided that it is either a trick of glare off of the bird's back or some brightness produced by something in the foreground vegetation in front of the bird. Given the flat, shadow- and glare-free lighting in the rest of the image, the first option (glare off the bird's back) does not seem reasonable. Plus, the sharp line between the black and the white does not seem consistent with the glare hypothesis. Hence, I'll focus on the more serious possibility, that it is caused by something between the bird and the camera. Fortunately, we don't just have the image with the bird; we have a whole sequence of images both before and after the one shot in which the mystery bird appeared. I singled out three images: the shot with the bird in it (numbered 2), and the immediately previous and subsequent shots (numbered 1 and 3, of course). I have made two animations that blink different combinations of these images to try to resolve the matter of where this "white shield" is actually located. The images from which these animations were constructed are all copyright 2007 by Steve Sheridan; I have used them here with permission.

NOTE: These are not embedded videos or flash animations. The thumbnails link directly to .mov (quicktime) files. For the best effect, you should right click (or control click if using a one button mouse), download the linked file to your computer (not the thumbnail image, but the linked file). Then open the .mov file in your own media player and set it to play as a continuous loop so you will get the "blink" effect. This may seem inconvenient, but we're doing science here, not entertainment. Streaming media always results in image degradation; I want to preserve detail, not obscure it, even at the cost of convenience.

The first loop is a rapid blink of shots 1 and 3, the birdless frames. In between these two shots, the camera position shifted slightly, so the rapid blink gives a 3-dimensional effect that makes the various layers of depth in the image jump apart from each other:

Watching this, most people will see an actual 3D illusion; some people will not see the illusion directly but you can still resolve the depth dimension by looking at the differential amounts of parallax as the images jump back and forth.

I find four distinct depth layers: The bright sky in the background; the tree trunk; the branches and foliage in front of the tree trunk; and a very close hazy semi-transparent blur from vegetation right in front of the camera. The critical thing to note here is that all of the bright areas are background sky; there is no suggestion at all of anything bright in the foreground layers. They appear to consist solely of green foliage, dark twigs, and open spaces that allow the background to be seen.

In the next loop, I have added frame 2 (the bird) in the sequence, slowed the blink, and added cross-fade to smooth the transition between frames:

To my eyes, when combined with the information gotten from the first 3D loop, there's really not any doubt that the "white shield" is attached to the bird and behind the intervening foliage. It is larger in extent than the patches of sky in the birdless frames; the foreground twigs appear to be visible crossing in front of it; and it appears to block some of the dark tree trunk that is visible in the birdless shots. Going back and forth between the two loops, I see every indication that we have a white-shielded bird and no indications that the white is from any other source than the bird. I also does appear that the black spot to the lower left of the shield is part of the bird as well; if so it would almost surely be a wingtip as it seems unlikely any other part of the bird (such as the tail) could be in that spot and be consistent with the rest of the bird's visible posture.

2. Where is the back stripe?

Excellent question. The curve of the bird's wing is visible right about the spot where we would expect the dorsal stripe to be, so we can hypothesize that the dorsal stripe could be there but hidden by a wing that is cocked a bit dorsally; perhaps in preparation for flight. This is of course nothing but an ad hoc "just so" story; but it's still worth investigating whether it is possible. Unfortunately there are not many photos of living perched Ivorybills; all that do exist show dorsal stripes plainly visible where they ought to be. There are many more specimen mounts and photos of those mounts; Cyberthrush posted a collection of these images (and yet he swears to me he had no idea of the existence of the mystery photo when he did so). There are suggestions in those photos that the dorsal stripe could at least be somewhat obscured, but it's not entirely convincing.

Other than these limited comparisons, we do have many photos of other common species of woodpeckers that have similar dorsal stripes. I have gone through many still images of these, especially Lineated and Pale-billed Woodpeckers; and framed-through several videos of perched birds in motion of these species that are available online (google and youtube searches will turn up examples). A good collection of images of other Campephilus species can be found here. Search for "pale" in that gallery and you will get dozens of shots of Pale-billed Woodpeckers. A rather large proportion of the shots of perched Pale-billeds viewed from the side show little or no dorsal stripe; even more so when the bird is higher than the camera. So the answer here is that yes it does seem to be possible, perhaps even likely, that a perched ivorybill in that pose could have its dorsal stripe hidden behind the wing.

3. Can a Pileated's neck look like that?

It appears to me that the answer to this one is "no." I have looked frame-by-frame through multiple videos of Pileateds (youtube will provide you with many examples), and at many still photos. I have not seen a single shot where a Pileated shows such a broad expanse of black on the back of the neck bordering such a narrow neck stripe. On the other hand, this neck pattern is seen quite often in shots of Campephilus woodpeckers that have neck stripes.

4. Can a Pileated's crest look like that?

This seems to be a "yes." In my framing through videos I saw many cases of a Pileated showing what looked like a large bushy crest, due to blurring of various kinds and especially when the bill is pointed up. Usually the bill is also visible in these cases. A related matter is why doesn't the mystery bird show a black crown if it is an Ivorybill? Well, the red on the nape of a male Ivorybill extends quite far down, much farther than on a Pileated. So, for an Ivorybill to show a crest like that it could be looking straight away from the camera, not tipping its head up to expose the black crown. So the "big bushy red crest" is indeterminate; either species could look like that.

Another issues that has been raised is, why doesn't the bird's back appear any darker than that of the Pileated (Ivorybills have darker, glossier backs than Pileateds)? However, in the quick blink animation you can see that there is some very close foliage completely out of focus casting a greenish translucent haze over a large area that includes the spot where the bird appeared. Pixel sampling of the white and black on the bird indicates a lot of green in both. Conversely, the Pileated in the same frame appears to be in the clear, and its black tends more towards the reddish-brown. Different lighting, different circumstances, different intervening space; I don't think you can really compare these colors between birds. As a revealing exercise, if you take the original image, showing both birds, and use photoshop or some similar application to increase the color saturation to psychadelic levels, you get an interesting result. The black and white parts of the mystery bird almost all go quite green; the black on the foreground Pileated goes to black speckled with bits of red and green. Intriguingly, the upper part of the mystery bird's black back does not go green; it becomes a deep blue-black color. Fascinating.

A final issue that came up is the size of the bird. One of the searchers reported measurements of the distance from the camera to both birds. If these measurements are taken at face value, the "mystery bird" appears to be too small to even be a Pileated, much less an Ivorybill. Rather than following this line of reasoning to its surreal conclusions (the bird is an inconceivably contorted Red-headed Woodpecker?? This seems utterly impossible), it is simplest to question the accuracy of the distance measurements. They were made with a laser range-finding device intended for golfing. The odds seem very good that it could have been thrown off by intervening vegetation and produced inaccurate measurements, especially for the distant bird through many layers of forest. There is also the possibility of mistaking the tree on which the bird was perched. Given these problems, until a physical measurement is made on the ground with actual tape measures, I think we should just forget about the distance numbers. What is clear from the photo, with no need for other information, is that the bird is a robust-bodied woodpecker with a long, narrow neck and a large red crest. This gives us only two species to consider.

ADDED 3/10/09: Many people online are trying to make this bird into a Red-headed Woodpecker, in spite of the fact that this is a completely untenable ID. Even if accurate distance and size measurements confirm the bird's small size, I'm not sure that is the route to use to explain the mystery bird. The neck on a Red-headed Woodpecker is red, not black. There is no black or white on the neck at all to get stretched and distorted in to a white stripe on black background. The red goes down, uninterrupted, to the base of the nape and on to the upper breast. Those "white straps" on the shoulders are breast feathers protruding from under the folded wing. They will not stretch into a neck stripe no matter how hard you pull. Same with the black on the back; that is the bird's mantle and you can't stretch that into a neck either. Even if the white stripe is something other than plumage on the mystery bird, the neck is still quite plainly black, not red. So ignore the size and shape issues, stretch a Red-headed's neck until just before it snaps, and you still won't get a black neck. The mystery bird is not a Red-headed Woodpecker.

My conclusion? Well, that hardly matters. Everyone who sees this image will form their own conclusion regardless of what I might think. I've simply spelled out some of the important features I find in the image, which I hope you gentle readers will find helpful in forming your own conclusions. I will summarize, than in my judgement in order for the bird to be an Ivorybill we have to accept two "unlikelies:" the hidden dorsal stripe and the very existence of the bird in this location. For it to be a Pileated we have to accept one "unlikely" (the white shield; a Pileated with a white shield has still never been documented by photo or specimen) and one "seemingly impossible" (the configuration of the neck stripe). Of course some would rate the existence of the Ivorybill also as a "seemingly impossible" not just an "unlikely," in which case they will find the evidence tied. Conversely, one could conclude that the existence of a white-backed Pileated is equally dubious, and rate that as "seemingly impossible" as well. Which means, once again, it may come down to your own preexisting and personal beliefs about the possibility that the Ivorybill could still exist at all. What I will find most interesting is not the conclusions of the regular disputants; rather, I will be curious to hear the feelings of the great masses of experienced birders who have been following all this much more quietly and with far less rigid opinions than we loudmouths.

Animations and other content copyright 2009 William M. Pulliam; source materials copyright 2007 Steve Sheridan and used with permission.

Monday, February 09, 2009

Summary of my Luneau postings

For referencing and archiving, here are links to the posts in which I laid out my own independent analysis of the controversial video of a large woodpecker filmed by David Luneau:

1. Flight mechanics
Wingbeat rate
Pileated flight mechanics
Luneau bird flight mechanics
Estimated distances and angles in the video
Animation of frame-by-frame interpretation of first 0.6 second of video

2. Black-and-white patterns: Plumage versus digital artifacts
Introduction: reality versus artifacts
Detailed frame-by-frame examination
Quick look at four key frames

3. Responses to other analyses published online
Response to Martin Collinson's paper
Response to Louis Bevier's website
Answers to Sibley et al's five questions

4. Comparisons to Imperial Woodpecker film published in 2011
Imperial Woodpecker versus Luneau bird
Furthr analysis of wingbeat frequency issues

Also relevant is my general discussion of video interpretation as a bird identification tool. It was triggered by one of Mike Collins' videos, but includes points of more general applicability.

As these video postings have gotten buried deep in my archives I thought I should gather the links together for reference. I'll add a sidebar link to this post so that they will remain easily accessibly in case any future historians might delve into this bizarre episode in the history of birding and field ornithology.

Site Meter