Now that I've seen the published version of the recent Jones et al. Nature paper on relative leg lengths in non-avian dinosaurs and birds, and digested some of the commentary thereon, I can offer my spin on the various matters. One colleague privately characterized the rather heterogeneous group of authors as "injudicious." I think "ecumenical" might be a more appropriate choice of words. The co-authors are not of one mind about all of the issues raised by the paper, and the way it was written reflected necessary compromises among points of view. My first thought is that some of the comments about the paper strike me as a classic case of the tail wagging the proverbial dog. The title of the paper is: "Cursoriality in bipedal archosaurs", not something like "Why _Caudipteryx_ really isn't a non-avian theropod (with hugs and kisses to Alan Feduccia)". The thoughts about the phylogenetic affinities of said beast are, in my opinion, a minor part of the paper. I'm not trying to be disingenuous here: Nobody on this list will be astonished to learn that some of the authors of the paper lean toward the idea expressed in what was not the title of the paper, but that is NOT the main point that this article is all about. My involvement in this project stemmed from another study, which was actually finished earlier, but which has not yet appeared in print. This paper, co-authored with Gatesy, Holtz, Hutchinson, and Robinson, is about theropod locomotion, and should appear soon in the American Zoologist feather symposium. As I worked on it, I noticed that non-avian theropods (NATs) and ground birds differ in the length of the hindlimb relative to the length of digit III (a crude proxy for that part of the foot responsible for making footprints), with birds having a relatively longer leg length. However, if one takes into account the fact that most of the action in the avian leg is at the knee, rather than the hip, the "effective" limb length in NATs and ground birds is much the same. I thought this was interesting, but wondered what would happen if one used a better proxy for animal size than digit III length, such as body trunk length. As it happened, Terry Jones and John Ruben were interested in the same question, but in the context of the matter about which I was not being disingenuous in a preceding paragraph. We agreed to work together on this project, and the rest is history. Now, in my opinion there are two components to the Nature paper. One of these is the question of whether at least most bipedal non-avian dinosaurs (NADs) differ from ground birds in the length of the leg relative to the trunk (and how this might affect the matter of effective limb lengths). The second is how _Caudipteryx_ (and NATs thought to be close relatives of said beast) fit into the picture. I will consider these matters separately, because I do not believe that they must stand or fall together. Let us examine each of these in turn. 1) Relative leg lengths in NADs and ground birds. Tom Holtz criticized us for using skeletal reconstructions (in part) as our sources of data. Indeed, we are guilty. He mentions several theropod taxa which were probably not well enough preserved to permit good reconstructions of the trunk and hindlimb, and I thank him for correcting us. He also argues that we have introduced circularity into our argument by using "imaginary measurements which are themselves scaled by the artist from other known specimens." It is undeniably true that these kinds of data are not ideal. The question is, are they good enough for addressing the question posed? Granted that they are almost surely not exact, are they so inexact that this would cause the relationship between hindlimb length and trunk length to be indistinguishable in NADs and ground birds? In considering incomplete specimens, let me suggest that our argument can be circular, in the context of the question being asked, only if those incomplete specimens were reconstructed in such a manner as to make them them fit our preconceptions of what their limb length-trunk length relationships should be. In contrast, if, say, an incomplete specimen was given additional vertebrae in the reconstruction because a close relative is known to have had said vertebrae, that will constitute inexactness, but not circularity. Because all of the reconstructions were published prior to our study, I think that the latter situation is what applies here. If it had just been me, I would not have fitted a regression line to the points for extinct forms, for the very reason that some our data points were indeed rather fuzzy. But we can check the effects of this inexactness on our results. First, I redid the theropod regression, tossing out the points we shouldn't have used, on the basis of Tom's comments. Then I did it again, using only those taxa for which, as best I can tell, we can be reasonably confident that the limb length-trunk length ratio isn't too far from reality: Eoraptor, Coelophysis, Dilophosaurus, Sinraptor, Allosaurus, Compsognathus, Sinosauropteryx, Gorgosaurus, Gallimimus, Struthiomimus, and Deinonychus. Although the regression coefficients changed a bit,there was no major change in the results. I therefore think that this is a robust, real result: most bipedal and facultatively bipedal NADs had relatively shorter legs, compared with their body trunk lengths, than do ground birds, and this surely will be related to the way they moved. This result fits nicely with what was seen in my other paper, in which leg lengths were compared with digit III lengths. Of course, it may be that with further work, Tom or somebody will find data that falsify these conclusions of the two papers. If so, I will mourn the death of yet another beautiful hypothesis and move on. 2) Sinornithoides, Bambiraptor, and Caudipteryx: Having considered the dog, I turn now to the tail. This is where my co-authors and I could be on thinner ice, in my opinion (and that is just MY opinion, not that of my co-authors). We excluded S and B from our regression equation on the grounds that these might be juveniles, and we were trying to remove ontogeny as a variable (but note that in our paper we explicitly stated that we were doing this, and why, so give us good marks for candor). I argued that we should nonetheless plot the S and B points on our graphs, even if we didn't use them in our regression, but was outvoted (see paragraph one, above, "compromises"). Based on some work I did over the summer, however, I am not confident that S and B can be disregarded in this kind of analysis, even if they are based on juveniles. I looked at hindlimb length relative to trunk length in an ontogenetic series of alligators, from very small to quite large individuals (this study is to be published in the Journal of the Paleontological Society of Korea, part of a symposium volume). With increasing alligator size, limb length does indeed become proportionately less. Unfortunately, this trend is not very strong, becomes obvious only across the entire size range of gators. There is very little change between half-grown (in linear dimensions) and big gators. If the same was true of S and B, then unless these specimens were quite small individuals of their species, it is possible that their limb length:trunk length ratios may not have been that different from adults. On the other hand, it could be argued that because gators are aquatic quadrupeds, they might not show as marked ontogenetic changes in limb:trunk proportions as terrestrial bipedal archosaurs, but it would take another study to determine if this is the case. It is worth noting here that in the abstracts for last year's SVP meeting Foster and Chure reported that a young individual of _Allosaurus_ had a leg length, relative to ilium length, that was 33% longer than in adults, suggesting the possibility of an ontogenetic drop in leg length relative to body trunk length as well. However, these authors also indicated that the cursorial limb proportions of the juvenile are lost by the time the beasts are about half-grown. It is therefore possible some NADs may indeed have departed from what is seen in most others, and have been more like ground birds in their limb:trunk proportions--and, as Tom noted, it is mighty suspicious that the guys who are doing this are among the forms thought, on other grounds, to be closely related to birds (it's interesting, though, that Deinonychus and Velociraptor seem to be more like typical NADs in this regard). And if you want to stir the pot even more, try putting the Elzanowski and Pasko reconstruction of Archaeopteryx, and the Chiappe reconstruction of Confuciusornis, on the graph--they plot more like typical NADs than like ground birds. Go figure. To sum up, I still think it likely that Caudipteryx had a style of locomotion more like a ground bird than like most non-avian theropods. This could very well mean that it derived from volant ancestors, as we argued in our paper. But it could also be--again as we noted in our paper--that Caudipteryx (along with S and B?) was indeed a member of a plexus of NADs that had independently (note the comments in the preceding paragraph about Archaeopteryx and Confuciusornis) adopted a style of locomotion similar to what ground birds later employed. I don't have a firm opinion on this, and the Nature paper didn't, either (see, one last time, paragraph one, above, "compromises").
Description: Unknown data type