S13.2: Molecular phylogeny of bird orders with crocodilian and chelonian outgroups

1Department of Biology and Museum of Zoology, University of Michigan, Ann Arbor, MI 48109, USA, fax 734 763 4080, email mindell@umich.edu; ²Boston University, Department of Biology, Boston, MA 02215, USA, email msoren@bio.bu.edu

The phylogenetic tree for extant birds is thought to consist of two groups: Paleognathae, which includes ratites (Struthioniformes; ostriches, rheas, emus, cassowaries, kiwis) and nine genera of tinamous (Tinamiformes), and Neognathae, which includes all other birds (Cracraft 1981; Olson 1985; Cracraft & Mindell 1989; Sibley & Ahlquist 1990). Some researchers have considered ratites to be primitive among birds on the basis of their palatal bones ever since the original description of this group by Huxley (1867). Relative antiquity for paleognaths is not supported by the fossil record, however, and other features, such as the trend toward flightlessness and large size, appear to be more recently derived (De Beer 1956). Other extant orders have been suggested as being among the earliest diverging birds, based on fossils attributed to that group (or its ancestors) predating fossils from other modern groups. These include Anseriformes, Charadriiformes, Gaviiformes (loons) and Procellariiformes (tubenoses, such as albatrosses) (Olson 1985; Olson 1992; Noriega & Tambusi 1995; Feduccia 1996).

Previous phylogenetic analyses for avian orders have been hampered by difficulties in rooting the avian tree. Presuming monophyly of Archosauria (birds and crocodilians as sister taxa), crocodilians are the most appropriate, extant outgroup. However, the bird-crocodilian divergence is estimated to be 245 million years old (Benton 1990), which is much earlier than the divergences among extant bird orders estimated to be >90 to 55 million years old (Feduccia 1996, Cooper & Penny 1997; Hedges & Kumar 1998). This long time span, prior to diversification among birds, makes it difficult to find characters variable enough to be informative of phylogeny within birds, yet conserved enough to be informative in comparisons between birds and crocodilians or any other reptile or mammal outgroups.

Genomic DNA for five birds (Rhea americana, Aythya americana, Falco peregrinus, Vidua chalybeata, and Smithornis sharpei), one turtle (Chrysemys picta), and one crocodilian (Alligator mississippiensis) was isolated from muscle tissue and mitochondrial DNA was amplified using the polymerase chain reaction (PCR). Long PCR products were generated with a rTth DNA polymerase-based XL-PCR kit (Perkin Elmer), gel-purified, and sequenced directly on an ABI 377 using the PCR primers and multiple internal primers (Sorenson et al. unpubl.). Insertions of mitochondrial DNA into the nuclear genome have been documented in many taxa, and we have taken requisite precautions against inclusion of any former mitochondrial sequences in the nuclear genome. We examined all DNA sequence electropherograms for distinguishing features of nuclear copies, including: double peaks resulting from potential coamplification of mitochondrial DNA and nuclear DNA sequences, unexpected insertions/deletions, frameshifts or stop codons, and mismatches in overlapping sequence for a given taxon from different amplification products. Features consistent with mitochondrial origin that we observe in our sequences are (i) presence of a conserved reading frame in protein-coding genes among all taxa, with decreasing rates of variability at third, first and second codon positions, respectively and (ii) absence of extra stop codons, frameshifts, or unusual amino acid substitutions. Further, we found no evidence of sequence change yielding loss of known secondary structure for tRNA and rRNA genes that would indicate translocation to the nucleus and loss of function.

Individual gene alignments for the seven new mitochondrial genomes and published genomes from two additional birds (Gallus gallus, Struthio camelus), eight mammals and one amphibian (Xenopus laevis) were initiated with Clustal X (Thompson et al. 1994) and adjusted manually. Appropriate secondary structure models were used in alignment of tRNA and rRNA genes. tRNA loops and other ambiguous alignment regions were excluded from analyses. We conducted phylogenetic analyses separately and in combination on (i) protein-coding gene amino acids and nucleic acids, (ii) tRNA genes, and (iii) rRNA genes. ND6 was analysed separately from the 12 proteins encoded by the opposite strand due to its distinctive amino acid and base composition. We have discovered a novel mitochondrial gene order in Falco peregrinus and Smithornis sharpei involving translocation of ND6 (Mindell et al. unpubl.), however, this does not directly affect analyses of primary sequence reported here. Heuristic maximum parsimony (MP) analyses for amino acids and nucleotides were conducted with 100 replicate searches and random addition of taxa using PAUP* (4d63) (Swofford 1998). Greater weight was given to more slowly evolving characters by using the PROTPARS weight matrix for amino acids, and using rRNA transversions only. Equal weights for all tRNA characters were used as their rate of change in the study taxa is slower then for rRNA, based on inferred numbers of all substitution types from preliminary MP analyses. Inclusion of rRNA transitions, however, had no effect on the optimal topology.

Maximum likelihood (ML) analyses accounting for rate heterogeneity with estimated proportions of invariant sites excluded were conducted using PAUP*. Transversion:transition ratios and proportion of invariant sites were estimated for each data set by taking an initial MP tree and using the ‘Describe Trees’ feature with ML options for the attributes mentioned set to ‘estimate’. ML analyses assuming rate homogeneity across sites were conducted using the NucML (HKY model), ProtML (mtREV24-F model), and TotalML programs in the MOLPHY package (Adachi & Hasegawa 1996). TotalML was used to combine the independent analyses of 12 protein, ND6, rRNA and tRNA genes. To allow comparison of all possible trees for birds in ML analyses with MOLPHY, non-avian outgroup taxa were constrained to the phylogenetic positions in our optimal topology based on MP and ML accounting for site rate heterogeneity analyses. Replicate comparisons constraining alligator and turtle as sisters and having trichotomies for whale/rhino/cat and for placental/marsupial/monotreme mammals had no effect on the optimal TotalML topology for birds. Puzzle (3.1) (Strimmer & von Haeseler 1996) was used in additional ML analyses using a Gamma-distribution in accounting for site rate heterogeneity and generally supported results obtained from the other analyses. In light of apparent constraints on the evolution of functionally important hydrophobic amino acids (Naylor & Brown 1997), and potential for their convergent similarity, we also analysed amino acids synonymising isoleucine, leucine and valine, however, the phylogenetic results were not altered.

Both weighted maximum parsimony (MP) and maximum likelihood (ML) analyses for protein, tRNA and rRNA gene sequences combined yielded the same optimal tree. This tree indicates a basal position for an oscine songbird and more recently derived positions for Anseriformes and ratites, two lineages previously hypothesized to be basal among birds. This tree supports sister relationships for rhea and ostrich (ratites) and for representatives of Anseriformes and Galliformes, congruent with previous molecular and morphological analyses (Sibley & Ahlquist 1990; Mindell et al. 1997; Livezey 1997). The optimal tree further indicates a sister relationship between the ratites and the Anseriformes/Galliformes clade. This is incongruent with the traditional view in which the Anseriformes/Galliformes group is sister to all other neognaths (including Passeriformes and Falconiformes). Results of ML analyses based on both rate heterogeneity and rate homogeneity across sites indicate the same avian phylogeny. The highest log-likelihood score for a tree in which Anseriformes is basal among birds is more than two standard errors lower than that of the ML tree, allowing rejection of the hypothesis that Anseriformes represents the oldest extant avian lineage (Kishino & Hasegawa 1989). Basal positions for the ratites, Galliformes, or Falconiformes cannot be rejected by the same criterion; however, based on all data sets combined, the highest scoring ML trees in which their placement is basal, are recovered in only 10.4%, 3.5% and 7.9% of bootstrap replicates, respectively and their positions are derived in MP analyses. MP bootstrap support is 87% or more for all avian nodes, with one exception.

The Falconiformes and suboscine songbird nodes are unresolved due to our finding of two MP trees, one in which they are sisters and one in which the suboscine songbird is basal to all birds except the oscine songbird. Appearance of Passeriformes as non-monophyletic is unexpected, as previous morphological (Raikow 1982) and molecular (Sibley & Ahlquist 1990; Mindell et al. 1997) analyses support monophyly. MP analysis constrained to maintain monophyly of Passeriformes (Vidua and Smithornis as sister taxa) yielded two trees, each 22 steps longer than the MP tree. In one, Passeriformes is basal to all other birds, and in the other Falconiformes is basal to all other birds. Exclusion of either the oscine or the suboscine songbirds in alternative MP analyses yield single shortest trees with the included passeriform basal to all other birds. Exclusion of the falconiform species yields a single MP tree with Vidua basal among birds. Thus, neither constraint to passeriform monophyly, nor changes in inclusion for the two passeriforms and one falconiform, alters hypothesized relationships for Anseriformes, Galliformes or ratites, and provides no support for them as being basal among birds. We are skeptical of passeriform non-monophyly given instability of the suboscine songbird node and the requirement imposed for convergent evolution of shared traits including an aegithognathous palate, bundled spermatozoa with coiled heads, features of syringeal structure, and unique hind-limb and foot musculatures (Raikow 1982). Our finding does indicate, however, a relatively early divergence between the oscine and suboscine songbird groups.

MP and ML analyses of protein, tRNA and rRNA genes as three separate data sets agree with the combined analyses with only a few exceptions. Both MP and ML analyses based on protein amino acid sequences only yield the combined analysis tree for avian relationships. MP and ML analyses for tRNA genes also match the combined analysis phylogeny except the suboscine songbird is basal to all birds and the oscine is basal to all birds except the suboscine. MP analyses and ML analyses incorporating rate heterogeneity across sites for rRNA alone differ from the combined analysis tree only in switching positions for the oscine and suboscine songbirds.

There has been much debate regarding the age of extant avian orders and whether only one (Feduccia 1996) or many (Cooper & Penny 1997; Hedges & Kumar 1998) extant forms arose during the Cretaceous and survived the extinction events marking its end 65 million years ago. Extant orders of birds purported to be represented by Cretaceous fossils include Anseriformes (waterfowl; Presbyornis), Gaviiformes (loons; Neogaeornis), Charadriiformes (shorebirds), and Procellariformes (tubenoses). If the fossils are correctly attributed, phylogenetic analyses indicating that some other avian lineage is basal to any of these four implies a divergence event older than the fossils. In this light, rejection of a basal position for Anseriformes further supports the antiquity of birds, Passeriforms in this case, and their diversification prior to the end of the Cretaceous. Finally, we point out that our current study taxa cover only a subset of extant bird diversity (five of about 24 commonly recognized orders), and future analyses including data from more birds, particularly shorebirds, loons and tubenoses, as well as inclusion of lizards and snakes will help resolve whether other avian taxa are basal to Passeriformes, as well as whether turtles are sister to Archosaurs, or to a larger group including Archosaurs and lizards.

We thank Derek Dimcheff, Tamaki Yuri and Jennifer Ast for assistance in the laboratory and comments on the paper. We thank David Swofford for permission to use a pre-release version of PAUP* (4d63) and Masami Hasegawa for advice and assistance with MOLPHY as well as comments on the paper. Robert B. Payne kindly provided Smithornis sharpei and Vidua chalybeata tissue samples. This work was supported by an NSF grant to DPM. MDS was supported by an NSF grant to R. B. Payne.

Adachi, J. & Hasegawa, M. 1996. MOLPHY ver. 2.3: programs for molecular phylogenetics based on maximum likelihood. Computer Science Monographs, No. 28. Tokyo, Institute of Statistical Mathematics.

Benton, M.J. 1990. Phylogeny of the major tetrapod groups: morphological data and divergence dates. Journal of Molecular Evolution 30: 409-424.

Cooper, A. & Penny, D. 1997. Mass survival of birds across the Cretaceous-Tertiary boundary: molecular evidence. Science 275: 1109-1113.

Cracraft, J. 1981. Toward a phylogenetic classification of birds of the world (class Aves). The Auk 98: 681-714.

Cracraft, J., & Mindell, D.P. 1989. The early history of modern birds: A comparison of molecular and morphological evidence. In Fernholm, B. Bremer, K. & Jörnvall, H. (eds) The Hierarchy of Life. Amsterdam, Elsevier Science Publishers: 389-403.

De Beer, G. 1956. The evolution of ratites. Bulletin British Museum (Natural History) 4:59-70.

Feduccia, A. 1996. The Origin and Evolution of Birds. New Haven, Yale Univ. Press.

Hedges B.H. & Kumar, S. 1998. A molecular timescale for vertebrate evolution. Nature 392: 917-920.

Huxley, T.H. 1867. On the classification of birds and on the taxonomic value of the modifications of certain of the cranial bones observable in that class. Proc. Zool. Soc. London 1867: 415-472.

Kishino, H. & Hasegawa, M. 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. Journal of Molecular Evolution 29: 170-179.

Livezey, B. C. 1997. A phylogenetic analysis of basal Anseriformes, the fossil Presbyornis, and the interordinal relationships of waterfowl. Zool. J. Linn. Soc. 121: 361-428.

Mindell, D.P., Sorenson, M.D., Huddleston, C.J., Miranda, H.C., Jr., Knight, A., Sawchuk, S.J., & Yuri, T. Phylogenetic relationships among and within select avian orders based on mitochondrial DNA. In Mindell, D. P. (ed) Avian Molecular Evolution and Systematics. San Diego, Academic Press: 213-247

Mindell, D.P., Sorenson, M.D., & Dimcheff, D.E. Unpubl. in review. Multiple independent origins of mitochondrial gene order.

Naylor, G.J.P. & Brown, W.M. 1997. Structural biology and phylogenetic estimation. Nature 388: 527-528.

Noriega, J.I. & Tambusi, P. 1995. A late Cretaceous Presbyornithidae (Aves: Anseriformes) from Vega Island, Antarctic Peninsula: paleobiogeographic implications. Ameghiniana 32: 57-61.

Olson, S.L. 1985. The fossil record in birds. In Farner, D. S., King, J. R. & Parkes, K. C. (eds) Avian Biology. Academic Press: 79-238

Olson, S.L. 1992. Neogaeornis wetzeli Lambrecht: a Cretaceous loon from Chile. J. Vertebrate Paleontology 12: 122-124.

Raikow, R.J. 1982. Monophyly of the Passeriformes: test of a phylogenetic hypothesis. The Auk 99: 431-445.

Sibley, C.G., & Ahlquist, J.A. 1990. Phylogeny and classification of birds: A study in molecular evolution. New Haven, Yale Univ. Press.

Sorenson, M.D., Ast, J.C., Dimcheff, D.E., Yuri, T., & Mindell, D.P. Unpubl. in review. Primers for a PCR-based approach to mitochondrial genome sequencing in birds and other vertebrates.

Strimmer, K. & von Haeseler, A. 1996. Quartet puzzling: a quartet maximum-likelihood method for reconstructing tree topologies. Molecular Biology and Evolution 13: 964-969.

Swofford, D.L. 1998. PAUP: Phylogenetic analysis using parsimony (and other methods), ver. 4.0. Sunderland, Sinauer Associates

Thompson, J.D., Higgins, D.G. & Gibson, T.J. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice. Nucleic Acids Research 22: 4673-4680.