Note to users. If you're seeing this message, it means that your browser cannot find this page's style/presentation instructions -- or possibly that you are using a browser that does not support current Web standards. Find out more about why this message is appearing, and what you can do to make your experience of our site the best it can be.
On the basis of new information from the 10th specimen of Archaeopteryx,Mayr et al. (Reports, 2 December 2005, p. 1483) suggested thatbirds, or avian flight, originated twice. We investigate thestatistical support for this phylogenetic hypothesis and showthat it is no better supported by available morphological characterdata than the hypothesis of a single avian origin.
1 Department of Earth Sciences, University of Bristol, Wills Memorial Building, Queens Road, Bristol, BS8 1RJ, UK. 2 Department of Zoology, Natural History Museum, Cromwell Road, London, SW7 5BD, UK. 3 Department of Palaeontology, Natural History Museum, Cromwell Road, London, SW7 5BD, UK.
* To whom correspondence should be addressed. E-mail: ian.corfe{at}bristol.ac.uk
In a recent report, Mayr et al. (1) provided information ona new specimen of Archaeopteryx and posited a new classificationof derived coelurosaurian theropods in which Aves (2) was eitherpolyphyletic or required expansion to include deinonychosaurs,a clade previously considered to be nonavian dinosaurs. Thisnovel phylogenetic hypothesis requires a complex pattern ofparallel gains and/or secondary losses of flight and other "avian"features in the theropod-bird lineage. Given the controversialnature of a hypothesis suggesting that either birds, or avianflight, originated twice, we reexamined the evidence for theirconclusions by determining support for the proposed phylogeny,carrying out statistical comparisons of the fit to the databetween their hypothesis and competing alternatives, and investigatingcharacters supporting the novel relationships suggested.
The new material (1) reveals osteological information unavailablefrom other Archaeopteryx specimens, permitting rescoring ofeight morphological characters within an existing charactermatrix (3) used to resolve coelurosaur interrelationships. Thisprincipally alters the systematic relationships of Confuciusornis,which is no longer recovered as the sister taxon to Archaeopteryxwithin a monophyletic Aves, but placed as the sister taxon toMicroraptor, within the dromaeosaurid clade.
We examined the relative support provided by the phylogeneticanalysis of Mayr et al. (1) for both monophyletic and polyphyleticAves using bootstrap proportions (4) and decay analysis (5).In general, support for the phylogeny of Mayr et al. is weak(Fig. 1), with that for the newly proposed clade Microraptor+Confuciusornisparticularly low: Bootstrap proportions indicate that a monophyleticAves (containing Archaeopteryx, Rahonavis, and Confuciusornis)is recovered in more of the bootstrap replicate data sets. Toexamine differences in fit to the data (1) of the competinghypotheses, we used an analysis (6) in which Aves was constrainedto be monophyletic. This recovered 768 constrained trees of600 steps, just one step longer than those from the unconstrainedanalysis (7). We then used the nonparametric Templeton test(8) to compare the fit of polyphyletic (unconstrained) and monophyletic(constrained) avian topologies to the data. The range of probabilityvalues obtained from pairwise comparisons (P = 0.819 to 0.853)indicates that the null hypothesis (9) cannot be rejected andthat there is insufficient data to choose among the two alternativephylogenetic hypotheses (10).
Fig. 1. Clade support on the strict consensus tree of (1). Bootstrap proportions are above the branch, decay index values below. The 50% majority rule bootstrap tree was poorly resolved, so the bipartition table was used to identify support for the relationships considered. Important results for the phylogeny presented in (1) include bootstrap proportions for Microraptor+Confuciusornis, 12.5%; for Archaeopteryx+Rahonavis, 16.9%; for a monophyletic Aves (Archaeopteryx+Rahonavis+Confuciusornis), 14.2%, which is higher than the grouping Microraptor+Confuciusornis; for Microraptor+Sinornithosaurus 16.1%, also higher than Microraptor+Confuciusornis. Decay indices are very low (1) for both Microraptor+Confuciusornis and Archaeopteryx+Rahonavis+Rahonavis. Average bootstrap proportion, 47%; average decay index, 1.93.
[View Larger Version of this Image (37K GIF file)]
We carried out additional analyses using more recent versionsof the Theropod Working Group's phylogenetic data matrix (11,12), modified by the codings suggested by Mayr et al. (1). Althoughthe modified analysis of (11) results in a polyphyletic Aves(13), use of a Templeton test to compare constrained (monophyleticAves) and unconstrained trees (14) indicates that there is insufficientevidence to reject a monophyletic Aves (P = 0.8084 to 0.8474).Recoding (12) did not affect the topology or number of mostparsimonious trees (MPTs) recovered; Confuciusornis and Archaeopteryxgroup together, whereas Rahonavis is recovered within Dromaeosauridaerather than Aves.
Using MacClade (15), we examined the distribution of characters(16) on the strict consensus trees of Mayr et al. (1) and Hwanget al. (3). Of the five characters (17) that unambiguously supportMicroraptor+Confuciusornis within Dromaeosauridae, the scoringof character 111, a separate or fused scapula and coracoid,is controversial (18). Of the two characters (19) uniting Archaeopteryx+Confuciusornisin a monophyletic Aves in (3), co-ossification of the metatarsals—character166 (20)—is also problematic (21). After considerationof other specimens (22), we reverted to the scorings of (3)for Archaeopteryx for characters 111 and 166 but kept otherrescorings as in (1). This results in avian monophyly with identicaltrees, tree lengths, and strict consensus topology to (3). Recodingof just two disputable characters in Archaeopteryx is sufficientto explain the hypothesis of avian polyphyly presented by Mayret al. (1).
The new Archaeopteryx specimen provides valuable informationon the morphology of basal birds and the relationships of taxaacross the theropod-bird transition. However, Templeton testsand bootstrap analyses indicate that the hypothesis of a polyphyleticAves is no better supported by available data than that of amonophyletic Aves. That alternative codings of Archaeopteryxfor two controversial characters shift the resulting phylogenetichypothesis between a monophyletic and polyphyletic Aves emphasizesthe lack of robustness. We conclude that statistical supportfor the novel hypothesis of Mayr et al. (1) is weak and thatthere is little current consensus as to the relationships betweenArchaeopteryx, Rahonavis, and Confuciusornis within Coelurosauria(Fig. 2), complicating attempts to trace the sequence of characteracquisitions during the origin of flight. In noting this, wehope to draw attention to the need for further work on coelurosauriananatomy and phylogeny.
Fig. 2. The relationships of the three avian taxa considered in (1) are subject to consider able uncertainty. Examining six recent phylogenies, all possible combinations of relationship are seen across either a monophyletic Aves or within Paraves (other taxa are not shown). (A) Hwang et al., 2002 (3). (B) Hwang et al., 2004 (11). (C) Chiappe, 2002 (26). (D) Forster et al., 1998 (27). (E) Makovicky et al., 2005 (12). (F) Mayr et al., 2005 (1). Four of the analyses (A, B, E, and F) are based on different iterations of the same basic data matrix. (C) and (D) are independently derived but may share characters (though the latter does not include Confuciusornis).
[View Larger Version of this Image (17K GIF file)]
2. Aves as currently conceived contains the Mesozoic "birds" Archaeopteryx, Rahonavis, and Confuciusornis but excludes deinonychosaurian theropods (23).[Abstract/Free Full Text]
3. S. H. Hwang, M. A. Norell, H. Gao, Am. Mus. Novitat.3381, 1 (2002). [CrossRef]
4. J. Felsenstein, Evol. Int. J. Org. Evol.39, 783 (1985).
6. D. L. Swofford, PAUP* Version 4.0b10 (Sinauer Associates, Sunderland, 2002).
7. The analyses of (1) and (3) used NONA (24) and collapsed branches with a minimum length of zero, which can result in trees that are not of minimal length. This was replicated in PAUP using the setting "Collapse branches if minimum length is zero (amb-)." Filtering these trees in PAUP for the best score retains only the MPTs. For example, the analysis of (1) constrained with a monophyletic Aves recovered 206,550 MPTs. Condensing by collapsing branches with a minimum length of zero returned 768 trees; filtering these by best score resulted in 120 MPTs. The resulting strict consensus was topologically similar to that of (1), other than a decrease in resolution within the clade Troodontidae. Condensed tree numbers are reported in the text for comparison with the original results of (1) and (3), but the condensed and filtered MPTs are used for Templeton test analyses.
8. A. R. Templeton, Evol. Int. J. Org. Evol.37, 221 (1983). [CrossRef]
9. The null hypothesis of the Templeton test is that differences between the trees in their fit to the data are no greater than expected from random sampling error.
10. Constraining Microraptor and Confuciusornis together in an analysis of the data presented by Hwang et al. (3) resulted in 136 constrained MPTs of step-length 603 after condensing and filtering. The range of probability values (P = 0.491 to 0.568) obtained from a Templeton test comparing these trees with the 36 condensed and filtered trees from unconstrained analysis of (3) indicates that for the original data set, unmodified by the codingssuggested in (1), there is insufficient data to reject the alternative hypothesis of avian polyphyly.
11. S. H. Hwang, M. A. Norell, Q. Ji, K. Gao, J. Syst. Pal.2, 13 (2004).
12. P. J. Makovicky, S. Apesteguia, F. L. Agnolin, Nature437 1007 (2005). [CrossRef] [Medline]
13. Analysis of the modified data set resulted in 864 trees of 627 steps, compared with the 2592 trees of 629 steps from the original analysis presented in (11). The strict consensus resembles that of (1) in positioning Confuciusornis as the sister taxon to Microraptor instead of in the monophyletic Aves from the unmodified data set. Constraining Aves as monophyletic on this modified data set returns 1727 constrained trees of 628 steps.
14. Because of the large number of trees, each individual tree was not considered; rather, every 10th of the 160 trees was compared to the 280 trees.
15. D. R. Maddison, W. P. Maddison, MacClade 4: Analysis of phylogeny and character evolution (Sinauer Associates, Sunderland, MA, 2003), version 4.06.
17. The five characters are 111, 123, 164, 165, and 171. Characters 123, 164, and 165 are uncontroversial. However, character 171, a reversed first toe, although absent in the new specimen, is not a definite feature of the Archaeopterygidae as a whole. The London and Eichstätt specimens offer conflicting evidence.
18. Character 111 was rescored by Mayr et al. (1) from fused (1) to (?) in Archaeopteryx and optimized as separate (0); it may, however, be fused in late ontogeny in Archaeopteryx (22). Considering this evidence and reverting to the scoring of Hwang et al. (3) for this character, but keeping all other recodings as suggested by Mayr et al. (1), results in 528 trees of 600 steps; the strict consensus has a polytomy between Archaeopteryx, Rahonavis, Confuciusornis, Microraptor, Troodontidae, and Dromaeosauridae, plus decreased resolution within Troodontidae. The coding of this homoplastic character (it requires 5 steps on the strict consensus tree of (1) and has a low consistency index value of 0.2) clearly affects the phylogenetic hypotheses inferred from the data.
19. Characters 165 and 166. Character 165, distal tarsal separation, changes from supporting Archaeopteryx+Confuciusornis in (3) to supporting Microraptor+Confuciusornis in (1), but the derived state (1), distal tarsals fused to metatarsals, is present in all three taxa.
20. Character 166. Metatarsals not co-ossified (0), or co-ossification of metatarsals begins proximally (1) or distally (2).
21. Character 166 was recoded in (1) from (1) to (0); (22) indicates that, although there is no evidence for proximally beginning metatarsal co-ossification, the metatarsals may have been superficially or incompletely co-ossified in some specimens. The character as currently conceived cannot represent this information. Reverting to state (1) as coded in (3) for character 166, but keeping the other recodings of (1), results in the same trees and strict consensus as when carried out for character 111 (18). Both characters 111 and 166 are osteological fusion characters, and as such are potentially prone to ontogenetic and intraspecific variation, leading to difficulties in coding observed variation in a phylogenetically meaningful way.
22. A. Elzanowski, in Mesozoic Birds: Above the Heads of Dinosaurs, L. M. Chiappe, L. M. Witmer, Eds. (Univ. of California Press, Berkeley, 2002), chap. 6.
23. K. Padian, in The Dinosauria, D. B. Weishampel, P. Dodson, H. Osmolska, Eds. (Univ. California Press, Berkeley, 2 ed., 2004), chap. 11.
24. P. Goloboff, NONA (NO NAME), ver. 2. Published by the author, Tucumán, Argentina, 1999.
25. J. M. Clark, M. A. Norell, P. J. Makovicky, in Mesozoic Birds: Above the Heads of Dinosaurs, L. M. Chiappe, L. M. Witmer, Eds. (Univ. California Press, Berkeley, 2002), chap. 3.
26. L. M. Chiappe, in Mesozoic Birds: Above the Heads of Dinosaurs, L. M. Chiappe, L. M. Witmer, Eds. (Univ. California Press, Berkeley, 2002), chap. 20.
27. C. A. Forster, S. D. Sampson, L. M. Chiappe, D. W. Krause, Science279, 1915 (1998).[Abstract/Free Full Text]
28. We thank P. Barrett, M. Benton, S. Braddy, L. Säilä, M. Wilkinson, and two anonymous referees for comments on a previous version of this paper.
Received for publication 2 June 2006. Accepted for publication 3 August 2006.
The editors suggest the following Related Resources on Science sites:
In Science Magazine
TECHNICAL COMMENTS
Gerald Mayr and D. Stefan Peters (1 September 2006) Science313 (5791), 1238c.
[DOI: 10.1126/science.1130964] |Abstract »|Full Text »|PDF »
