Monday, July 31, 2006

Woodpecker wingbeats without comment

Of interest to some --

I did some counting and created the following graph, which I present with no discussion other than describing how I made it and a bit about statistical significance. If it is truncated or too small you can right-click or ctrl-click on it to open it in a new window:

The Pileated source videos are the ones posted by Cornell and by David Nolin. I found a total of seven of these useful, wth between 2 and 15 wingbeats resolvable. I defined the beginning of each wingbeat as maximum upward wing extension. I counted in whole frames, so the temporal resolution is 17-33 ms depending on the source footage. For Luneau, I took the "extended wing" interpretation of the frame in which the bird first appears to the left of the tree trunk, not the CLO's "folded wing" interpretation. Thus, that is counted as the zero time point, the start of the first wingbeat. Using the other interpretation would decrease the Luneau times by 33 ms or less. I used only the first 10 wingbeats of the Luneau video, as they are the ones that are agreed to be clearly discernable.

All Pileated comparison videos are of perched birds taking off, as is the Luneau video. Only three sequences show the Pileated in frame and on the wing for 10 full wingbeats. This small sample size precludes any meaningful statistical analysis as to whether the Luneau bird is an outlier or within the expected range of variation of the Pileated, a fact that must be borne in mind while examining these data. A rigorous analysis would need many more Pileated videos in which the bird remains on the wing for 10+ wingbeats, prefereably longer. Ideally there would be multiple videos of Pileateds that remain on the wing in full flight for the full three seconds or so that the Luneau bird does.

I have presented the data in this format (total elapsed time versus number of completed wingbeats) because it gives the best resolution of the individual birds and it is the least sensitive to judgement calls as to which frame represents the beginning of any particular wingbeat. A plot of individual wingbeat durations produces a confusing jumble of overlapping lines.

A note about comments -- please limit to direct discussion of the wingbeat data and its interpretation, without abrasive or hyperbolic language. No anonymous or pseudonymous comments. Thank you.


At 7:36 AM, Anonymous Anonymous said...

Excellent visualization of the data!

It does a good job of capturing a lot of pertinent info in a concise format that is both revealing and easy to understand. You might consider using line types (or point markers or labels) to indicate specific PIWO flights like Nolin's, and provide references to the sources below the graph.

It'd be great to see a larger sample represented this way.

I think knowing the situational details of the individual flights would be important to include in the analysis. Was the bird startled into takeoff? How close to the ground was the bird on takeoff? Was the bird flying over water? Etc.

Nice to see someone taking an analytical approach on addressing this important topic from the Luneau video debate.

At 1:49 PM, Blogger Bonsaibirder said...

Hi Bill,

Interesting analysis of the Luneau video.

I was particularly interested in your graphs of wingbeat frequency of the Luneau and Nolin birds.

From my naive perspective, the birds all start flapping at a similar rate and then the Nolin birds all slow down and the Luneau bird carries on at basically the same pace. I assume that is what the graph shows ?

I have said this elsewhere before but in my mind the biggest difference between the birds in the Nolin video and the Luneau bird is that the Nolin birds choose a landing site very quickly after taking off. Its been a while since i looked at either video but correct me if i am wrong - many of the Nolin birds head straight for a single tree and land there soon after taking off.

Assuming that their eyesight is prety good and they spot the tree almost immediately could this be the explanation for your graph? That is, the Nolin birds are in "landing" mode as soon as they see the tree. In contrast, the Luneau bird shows no signs of landing in the video and therefore is in escape flight throughout.

At 2:37 PM, Blogger Bill Pulliam said...

Bonsai --

Yes, I see the patterns the same way, and it well could be the Nolin birds' short-range flight that causes much of this difference. That is why I think we need many more Pileated videos, esxpecially ones where the birds remain on the wing longer, before being able to draw meaningful and robust conclusions.

I personally think the wigbeat "shape" is a more interesting difference than the wingbeat "rate." By this I mean the way the bird's wings are held during the downstrokes, as I discuss in the posting about the animated Luneau bird. I would really like to see more Pileated comparisons to get a better handle on that apparent difference.


Post a Comment

<< Home

Site Meter