Article Contentsp. 599p. 600p. 601p. 602p. 603p. 604p. 605p. 606p. 607p. 608p. 609p. 610p. 611p. 612p. 613Issue Table of ContentsSystematic Biology, Vol. 51, No. 4 (Aug., 2002), pp. 541-672Front MatterThe Effects of Topological Inaccuracy in Evolutionary Trees on the Phylogenetic Comparative Method of Independent Contrasts [pp. 541 - 553]Analysis of Color Spectra in Comparative Evolutionary Studies: Molecular Phylogeny and Habitat Adaptation in the St. Vincent Anole (Anolis trinitatis) [pp. 554 - 569]Perils of Paralogy: Using HSP70 Genes for Inferring Organismal Phylogenies [pp. 570 - 587]Increased Taxon Sampling Greatly Reduces Phylogenetic Error [pp. 588 - 598]A Genomic Schism in Birds Revealed by Phylogenetic Analysis of DNA Strings [pp. 599 - 613]Points of ViewExamining Basal Avian Divergences with Mitochondrial Sequences: Model Complexity, Taxon Sampling, and Sequence Length [pp. 614 - 625]The Utility of the Incongruence Length Difference Test [pp. 625 - 637]Phylogeny Reconstruction and Functional Constraints in Organellar Genomes: Plastid atpB and rbcL Sequences versus Animal Mitochondrion [pp. 638 - 647]Units in Biogeography [pp. 648 - 652]Resolution of a Supertree/Supermatrix Paradox [pp. 652 - 664]Increased Taxon Sampling Is Advantageous for Phylogenetic Inference [pp. 664 - 671]Back MatterSyst. Biol. 51(4):599-613, 2002 DOI: 10.1080/10635150290102285 A Genomic Schism in Birds Revealed by Phylogenetic Analysis of DNA Strings SCOTT V. EDWARDS,1 BERNARD FERTIL,2 ALAIN GIRON,2 AND PATRICK J. DESCHAVANNE2 Department of Zoology and Burke Museum, University of Washington, Box 351800, Seattle, Washington 98195, USA; E-mail: [email protected] 2INSERM U 494, 91 bd de l'Hopital, 75634 Paris, France Abstract.-The molecular systematics of vertebrates has been based entirely on alignments of primary structures of macromolecules; however, higher order features of DNA sequences not used in tradi- tional studies also contain valuable phylogenetic information. Recent molecular data sets conflict over the phylogenetic placement of flightless birds (ratites - paleognaths), but placement of this clade criti- cally influences interpretation of character change in birds. To help resolve this issue, we applied a new bioinformatics approach to the largest molecular data set currently available. We distilled nearly one megabase (1 million base pairs) of heterogeneous avian genomic DNA from 20 birds and an alligator into genomic signatures, defined as the complete set of frequencies of short sequence motifs (strings), thereby providing a way to directly compare higher order features of nonhomologous DNA sequences. Phylogenetic analysis and principal component analysis of the signatures strongly support the tradi- tional hypothesis of basal ratites and monophyly of the nonratite birds (neognaths) and imply that ratite genomes are linguistically primitive within birds, despite their base compositional similarity to neognath genomes. Our analyses show further that the phylogenetic signal of genomic signatures are strongest among deep splits within vertebrates. Despite clear problems with phylogenetic analysis of genomic signatures, our study raises intriguing issues about the biological and genomic differences that fundamentally differentiate paleognaths and neognaths. [Bioinformatics; CpG island; genomics; isochore; ratite.] The phylogenetic analysis of the primary structure (sequence) of DNA has matured in recent years to encompass a wide vari- ety of techniques, including incorporation of secondary structures of RNA and pro- teins to improve alignment and tree build- ing (Suyama et al., 1997; Schoniger and von Haeseler, 1999). Most of these methods rely on or produce alignments of primary structures for assigning homology to in- dividual sites prior to or during phyloge- netic analysis (Mindell and Meyer, 2001). It is less well appreciated that homol- ogy exists in DNA sequences at organi- zational levels higher than the individ- ual DNA site and that "nonhomologous" DNA sequences that are not alignable by normal criteria can also contain phyloge- netic information of use to systematists (Karlin and Burge, 1995; Karlin et al., 1997; Schneider, 1997). For example, distant re- lationships among proteins whose primary structures are unalignable can sometimes be found by examining secondary structures (Bullock et al., 1996; Matsuo et al., 1996) or hydrophobicity profiles (Leunissen and de Jong, 1986; Naylor et al., 1995; Ladunga and Smith, 1997). Here, we explore this idea with particular reference to the phylogenetic posi- tion of the major clade of flightless birds, the ratites, and the possibility that homology ex- ists at higher order levels in DNA sequences captured in the particular DNA strings of avian species. Our analysis also suggests how large-scale bioinformatics analysis can inform phylogenetic analysis of major clades and provide new insights into genome evo- lution in birds and their relatives. Previous molecular studies of higher level relationships in birds and in vertebrates generally have gleaned information from character states of homologous sites ob- servable in primary alignable sequences of macromolecules (e.g., Cooper and Penny, 1997; Groth and Barrowclough, 1999; van Tuinen et al., 2000). However, the vast majority of DNA sequences in the databases come from studies on diverse taxa and di- verse nonhomologous genes that cannot be aligned with one another. This largest source of DNA sequence data is likely to contain information of use to phylogeneticists. In ad- dition to the information found in character states of aligned DNA sites, global sequence features and characteristics of higher order DNA sequence structure are known to 599SYSTEMATIC BIOLOGY provide some phylogenetic information, particularly at very deep phylogenetic levels such as among microbial lineages (Nussinov, 1984; Beutler et al., 1989; Pietrokovski et al., 1990; Burge et al., 1992; Karlin and Ladunga, 1994; Konopka, 1994; Karlin et al., 1997; Abella et al., 1999). However, the utility of such global sequence features for vertebrate systematics is unclear. Such higher order information presumably arises from species-specific differences in genome dynamics such as genome-wide patterns of mutation, DNA repair, and selection at the molecular level. Genomic signatures (Karlin and Burge, 1995; Deschavanne et al., 1999) are