Supplemental Data

Full TextFlight of the Dodo
Beth Shapiro, Dean Sibthorpe, Andrew Rambaut, Jeremy Austin, Graham M. Wragg, Olaf R. P. Bininda-Emonds, Patricia L. M. Lee, and Alan Cooper

Supplementary Material


Feather samples were obtained from extant taxa, while tissue (single toe pad) and bone samples were obtained from museum specimens of rare or extinct taxa (details available on request). We obtained 1cm2 of cortical bone from a tarsal of the Oxford University Museum of Natural History dodo specimen and from a solitaire femur found in the Caverne Bambara cave system on Rodrigues Island. The samples were cleaned and DNA was extracted using organic solvents and standard ancient DNA techniques (1). Samples from museum skins were rinsed with 0.5M EDTA to remove chemicals used for preservation purposes (e.g. alum and arsenic) before extraction. PCR amplifications were performed as in (2). Primer sequences are available from the authors on request. Sequences have been deposited in GenBank (Accession numbers AF483281-AF483353).

Phylogenetic analyses:

Likelihood ratio tests were used to determine the simplest ML model that could not be rejected in favor of a more complex model (3). Using PAUP v4.0b8 (4), full-heuristic searches were performed with starting trees generated by 10 randomly seeded stepwise addition sequences. We used a general time reversible (GTR) likelihood model (six substitution types) incorporating rate heterogeneity (8 rate categories) and a proportion of invariant sites. Branch swapping was performed twice using TBR, with re-estimation of parameters. Parameter values are available from the authors on request. Of the 1409 characters used in the analysis, 588 (41.7%) were variable and 495 of these (35.1%) were phylogenetically informative. Tests for congruence within 12S and cytochrome b revealed no conflict, and the combined data were used for the remainder of the analyses. Because of the advanced state of degradation of the dodo specimen, we were unable to sequence approximately 200 bp of cytochrome b and 150 bp in 12S. Analyses excluding the 350 sites missing from the dodo resulted in no difference in tree topology, although when they were excluded the bootstrap value on the dodo/solitaire node increased to 94%.

ML bootstrap consensus values were generated using parameters as for the main analysis, but with starting trees generated by neighbor joining. Bayesian posterior probabilities were estimated for each branch of the tree using a MCMC approach (5). Four chains were run, starting from random trees, for 1,000,000 generations sampling trees every 500 generations. The first 40,000 generations were discarded to allow the chains to converge. The Bayesian posterior probabilities are then approximated by the frequency that a particular bipartition of taxa appears within the set of 1920 trees.

Six avian families had previously been suggested to have close associations with the Columbiformes (6): Pterocliidae (sandgrouse), Gruidae (cranes), Turnicidae (buttonquails), Laridae (gulls, skuas, etc), Coliidae (mousebirds), Psittidae (parrots). We obtained 12S and cytochrome b sequences for 18 genera in these families from Genbank, and generated the same data from a sandgrouse, Pterocles. The ML analyses consistently showed the Columbidae to be most closely related to the gulls (Laridae). While the support for this arrangement was moderate, it was recovered under a wide variety of parameters. While a shorebird-pigeon relationship has often been raised (6, 7) this appears to be the first molecular evidence in support of the relationship. Consequently, we used two species of Laridae (Great Skua, Catharacta skua, U76807, U76772 and Parasitic Jaeger, Stercorarius parasiticus, U76826, U76766) as outgroups in the Columbiform phylogenetic analyses.

Molecular Clock Estimation:

The program QDATE was used to estimate divergence times as in (8) with a simple modification for cases in which only a single fossil calibration point is available. The pigeon fossil record is too poor to provide an internal calibration point for the molecular rate calculations, so the well-dated divergence between penguins and Procellariformes (58 Mya, 9), as well as the less constrained chicken/guinea fowl split (40 Mya, 9) were used as external calibrations in separate analyses. The quartets consisted of the calibration pair and either the solitaire/dodo or solitaire/Nicobar pigeon. Divergence date estimates were similar using either calibration (within the 95% CI), as were additional molecular clock analyses using non-parametric rate smoothing (10) with the penguin/albatross calibration point. Dates given in the text were generated using the penguin/albatross calibration and no rate smoothing.

1. Cooper, A., Poinar, H. Science289, 1139 (2000).
2. Cooper, A. et al.Nature409, 704-707 (2001).
3. Kishino, J., Hasegawem, L. J. Mol. Evol. 29, 170 (1989).
4. Swofford, D.L. PAUP* Phylogenetic Analysis Using Parsimony (*and Other Methods) Sinaur, Sunderland, MA (1998).
5. Huelsenbeck, J.P., Ronquist, F. Neilsen, R., Bollback, J.P. Science in press.
6. Sibley, C.G., Ahlquist, J.E. Phylogeny and classification of birds: A study in molecular evolution (Yale University Press, New Haven and London, 1990).
7. Cracraft, J. The Auk98, 681 (1981).
8. Rambaut, A., Bromham, L. Mol. Biol. Evol.15, 442 (1999).
9. Cooper, A., Penny, D. Science275, 1109 (1997).
10. Sanderson, M.J. Mol. Biol. Evol.14, 1218 (1997).