Mindell et al, 1999 by Felipe Elias

Syst. Biol. 48(1):138–152, 1999

Interordinal Relationships of Birds and Other Reptiles Based on Whole Mitochondrial Genomes D AVID P. MINDELL,1,4 MICHAEL D. SORENSON,2 DEREK E. DIMCHEFF,1 MASAMI HASEGAWA,3 JENNIFER C. AST,1 AND TAMAKI YURI1 Department of Biology and Museum of Zoology, University of Michigan, Ann Arbor, Michigan 48109, USA; E-mail: mindell@umich.edu (D.P.M), derekdim@umich.edu (D.E.D), jca@umich.edu (J.C.A.), komadori@umich.edu (T.Y.) 2 Department of Biology, Boston University, Boston, Massachusetts 02215, USA; E-mail: msoren@bio.bu.edu Institute of Statistical Mathematics, 4-6-7 Minami-Azabu, Minato-ku, Tokyo 106, Japan; E-mail: hasegawa@ism.ac.jp 1

Abstract.—Several different groups of birds have been proposed as being the oldest or earliest diverging extant lineage within the avian phylogenetic tree, particularly ratites (Struthioniformes), waterfowl (Anseriformes), and shorebirds (Charadriiformes). Dif culty in resolving this issue stems from several factors, including the relatively rapid radiation of primary (ordinal) bird lineages and the lack of characters from an extant outgroup for birds that is closely related to them by measure of time. To help resolve this question, we have sequenced entire mitochondrial genomes for ve birds (a rhea, a duck, a falcon, and two perching birds), one crocodilian, and one turtle. Maximum parsimony and maximum likelihood analyses of these new sequences together with published sequences (18 taxa total) yield the same optimal tree topology, in which a perching bird (Passeriformes) is sister to all the other bird taxa. A basal position for waterfowl among the bird study taxa is rejected by maximum likelihood analyses. However, neither the conventional view, in which ratites (including rhea) are basal to other birds, nor tree topologies with falcon or chicken basal among birds could be rejected in the same manner. In likelihood analyses of a subset of seven birds, alligator, and turtle (9 taxa total), we nd that increasing the number of parameters in the model shifts the optimal topology from one with a perching bird basal among birds to the conventional view with ratites diverging basally; moreover, likelihood scores for the two trees are not signi cantly different. Thus, although our largest set of taxa and characters supports a tree with perching birds diverging basally among birds, the position of this earliest divergence among birds appears unstable. Our analyses indicate a sister relationship between a waterfowl/chicken clade and ratites, relative to perching birds and falcon. We nd support for a sister relationship between turtles and a bird/crocodilian clade, and for rejecting both the Haemothermia hypothesis (birds and mammals as sister taxa) and the placement of turtles as basal within the phylogenetic tree for amniote animals. [amniote phylogeny; bird phylogeny; turtle phylogeny; mitochondrial genomes; Passeriformes; ratites; turtle phylogeny.]

The phylogenetic tree for extant birds is thought to consist of two primary groups: Paleognathae, which includes ratites (Struthioniformes: ostrich, rheas, emu, cassowaries, kiwis; known for their ightless condition) plus nine genera of tinamous (Tinamiformes), and Neognathae, which includes all other birds (e.g., Cracraft, 1981; Olson, 1985; Cracraft and Mindell, 1989; Sibley and Ahlquist, 1990). Some researchers have suggested that ratites 4 Address correspondence to David P. Mindell, Museum of Zoology, University of Michigan, 1109 Geddes Ave., Ann Arbor, Michigan 48109, USA.

evolved from ightless ancestors independently of other extant birds, based on analyses of palatal bones, plumage structure, or musculature (e.g., Lowe, 1935; Holmgren, 1955). The prevailing view, however, is that ratites are derived from ying ancestors and hence are secondarily ightless (e.g., DeBeer, 1956; Sibley and Ahlquist, 1972), and that they represent a basal offshoot among extant birds. 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 (ducks),

138

1999

MINDELL ET AL.—PHYLOGENY OF BIRDS AND REPTILES

Charadriiformes (shorebirds), Gaviiformes (loons), and Procellariiformes (tubenoses, such as albatrosses) (Olson, 1985, 1992; Noriega and Tambusi, 1995; Feduccia, 1996). Further, recent analyses based on 12 mitochondrial (mt) protein-coding genes and 2 rRNA genes combined (Mindell et al., 1997) and the mt cytb gene (H a¨ rlid et al., 1998) indicate a basal divergence for Passeriformes (perching birds). One of our primary objectives here is to test this unexpected hypothesis of Passeriformes as an early diverging avian lineage, using more taxa and more characters (including tRNA genes). Presuming the monophyly of Archosauria (birds and crocodilians as sister taxa), crocodilians are the most appropriate extant outgroup for rooting the avian tree. However, the bird–crocodilian divergence is estimated to be about 245 million years old (Benton, 1990), much earlier than the divergences among extant bird orders, which are estimated to be 55 to > 90 million years old (Feduccia, 1996; Cooper and Penny, 1997; Hedges et al., 1996; Kumar and Hedges, 1998; Waddell et al., 1999b). This long time span, prior to diversi cation among birds, makes it dif cult to nd 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 amniote outgroup taxa. Mindful of this problem and seeking to reduce potential rooting artifacts, we consider inclusion of both a turtle and a crocodilian an important step in resolving avian relationships. Inclusion of alligator and a turtle also provides the opportunity to address controversies regarding their placement relative to various other amniotes (Rieppel and deBraga, 1996; Kirsch and Mayer, 1998). Speci cally, we can test the Haemothermia hypothesis (Gardiner, 1982; Løvtrup, 1985) which posits that birds and mammals are sister taxa relative to reptiles, and the hypothesis that turtles are basal within the phylogenetic tree for amniotes (Gaffney, 1980). Although recent tests of these hypotheses (e.g., Gauthier et al., 1988; Caspers et al., 1996; Janke and Arnason, 1997) have not supported them, the hypotheses have yet to

139

be addressed concomitantly with data sets as large as complete mt genomes. METHODS Genomic DNA for six birds (greater rhea [Rhea americana] Struthioniformes; ostrich [Struthio camelus] Struthioniformes; redhead [Aythya americana] Anseriformes; peregrine falcon [Falco peregrinus] Falconiformes; village indigobird [Vidua chalybeata] oscine Passeriformes; grey-headed broadbill [Smithornis sharpei] suboscine Passeriformes), one turtle (Eastern painted turtle [Chrysemys picta]), and one crocodilian (American alligator [Alligator mississippiensis]) was isolated from muscle tissue and mtDNA was ampli ed using the polymerase chain reaction (PCR) and sequenced as described by Mindell et al. (1997). Long PCR products were generated with a rTth DNA polymerase-based XLPCR kit (Perkin-Elmer), gel-puri ed, and sequenced directly on an ABI 377 using the PCR primers and multiple internal primers (Sorenson et al., in press). Insertions of mtDNA into the nuclear genome have been documented in many taxa, and we have taken requisite precautions against inclusion of any former mt sequences in the nuclear genome (see Sorenson and Quinn, 1998). We examined all DNA sequence electropherograms for distinguishing features of nuclear copies, including: double peaks resulting from potential coampli cation of mtDNA and nuclear DNA sequences, unexpected insertions/deletions, frameshifts or stop codons, and mismatches in overlapping sequence for a given taxon from different ampli cation products. Features consistent with mt origin that we observe in our sequences are (1) presence of a conserved reading frame in proteincoding genes among all taxa, with decreasing rates of variability at third, rst, and second codon positions, respectively, and (2) absence of extra stop codons, frameshifts, or unusual amino acid substitutions. Further, we found no evidence of sequence changes yielding losses of known secondary structure for tRNA and rRNA genes that would indicate

140

SYSTEMATIC BIOLOGY

translocation to the nucleus and loss of function. Where possible, we have compared our sequences with homologs from GenBank for conspeci c individuals; in all such cases we found nearly identical matching. Individual gene sequence alignments for the new mt genomes and published vertebrate animal genomes were initiated with Clustal X (Thompson et al., 1997) and adjusted manually. Amino acids were used in alignments for the protein-coding genes. 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, using (1) protein-coding gene amino acids and nucleic acids, (2) tRNA genes, and (3) rRNA genes. For some analyses ND6 was analyzed separately from the 12 proteins encoded by the opposite strand, because of its distinctive amino acid and nucleotide base composition. 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 amino acid substitutions that required more nucleotide changes by using the PROTPARS weight matrix for amino acids (unless otherwise noted in the Discussion) and to more slowly accumulating rRNA changes by using rRNA transversions only. Equal weights for all tRNA characters were used, given that their average rate of change in the study taxa is slower then for rRNA, according to inferred numbers of all substitution types from preliminary MP analyses. Inclusion of rRNA transitions, however, did not change the tree topology in Figure 1. Maximum likelihood (ML) analyses for the same data is done with three different programs. PAUP* is used for nucleic acids only and enables use of models accounting for evolutionary rate heterogeneity across sites. PAML (Yang, 1997) is used for analyses of amino acids accounting for

VOL. 48

evolutionary rate heterogeneity across sites. Five different models of sequence evolution are used in various ML analyses (in order of increasing number of parameters): Jukes–Cantor (JC) (1969), assuming equal base composition and equal probability of change for all substitution types; Hasegawa–Kishino–Yano (HKY) (1985), accommodating unequal base composition and a transition:transversion ratio; the general time-reversible (GTR) model (Lanave et al., 1984; Tavar´e, 1986; Rodriguez et al., 1990), accommodating unequal base composition and different probabilities for each of six substitution types; GTR plus heterogeneous rates of change across sites following a discrete approximation of the gamma distribution ( G ) (Yang, 1994); and GTR plus G plus a proportion of sites assumed to be invariant (I) (Gu et al., 1995; Waddell and Penny, 1996). MOLPHY (Adachi and Hasegawa, 1996) is used for nucleic and amino acid analyses assuming homogeneous rates across sites. Substitution matrices, proportion of invariant sites, and G distribution shape parameters were estimated from the data sets being analyzed. Within MOLPHY we used the mtREV24-F model for amino acid substitution in ProtML; TotalML was used to sum likelihoods in combining the independent analyses of 12 protein-coding genes, ND6, and rRNA and tRNA genes. Also within MOLPHY, estimated bootstrap probabilities were determined by using the RELL (resampling of estimated log-likelihoods) method (Kishino et al., 1990). The new mt sequences reported and analyzed here have the following GenBank accession numbers: AF069423 (Eastern painted turtle), AF069428 (American alligator), AF069429-AF06943 1 (ostrich), AF090337 (redhead), AF090338 (peregrine falcon), AF090339 (greater rhea), AF090340 (grey-headed broadbill), and AF090341 (village indigobird). Our 15,898-bp Alligator mississippiensis and 10,714-bp Struthio camelus partial mt genome sequences are augmented as necessary in these analyses by sequences of conspeci c individuals from Janke and Arnason (1997) and Ha¨ rlid et al. (1997), respectively.

1999

MINDELL ET AL.—PHYLOGENY OF BIRDS AND REPTILES

141

FIGURE 1. Optimal phylogenetic hypothesis based on maximum parsimony (MP) analyses of all 13 mt proteincoding genes, 22 tRNA genes, and 2 rRNA genes combined, and for this same data set but excluding ND6, which differs in nucleotide base composition. Amino acids are used for protein-coding genes. The polytomy that includes Falco and Smithornis is the result of nding two equally short MP trees in analyses that excluded ND6: one in which they are sister taxa, and one in which the Smithornis divergence is basal to all birds except Vidua. From left to right, numbers by nodes denote bootstrap values for 100 replicates and support indices (Bremer, 1988) based on analyses without ND6. Branch lengths are proportional to the number of changes inferred. DNA sequence sources are Aythya, Rhea, Struthio, Falco, Vidua, Alligator, Chrysemys (present study, accession numbers in Methods section), Gallus (Desjardins and Morais, 1990), Struthio (H¨arlid et al., 1997), Alligator (Janke and Arnason, 1997), Balaenoptera (Arnason and Gullberg, 1993), Rhinoceros (Xu et al., 1996), Felis (Lopez et al., 1996), Homo (Anderson et al., 1981), Mus (Bibb et al., 1981), Didelphis (Janke et al., 1994), Macropus (Janke et al., 1997), Ornithorhynchus (Janke et al., 1996), and Xenopus (Roe et al., 1985).

RESULTS Description of the new complete mtDNA molecules for ve birds and Eastern painted turtle will be given elsewhere. This paper focuses on phylogenetic inference. We discovered a single extra nucleotide base in the mt ND3 gene of Rhea americana, Falco peregrinus, Aythya americana, and Chrysemys picta. This extra base appears not to be pro-

cessed during translation; if it is edited out or skipped, the downstream reading frame is maintained and amino acid sequence of the gene is conserved across diverse vertebrates, including those mentioned above (Mindell et al., 1998; see H a¨ rlid et al., 1998). The ND3 extra base, but not the ND3 gene, has been excluded from the proteincoding gene analyses reported here. We also

142

SYSTEMATIC BIOLOGY

VOL. 48

FIGURE 2. Optimal phylogenetic hypothesis based on maximum likelihood analysis of 13 mt protein-coding genes, 22 tRNA genes, and 2 rRNA genes combined, using the general time-reversible model and a discrete approximation to the gamma distribution to accommodate evolutionary rate heterogeneity across nucleotide sites. Protein-coding gene DNA rather than amino acids are used. Sequence sources are as listed in Figure 1.

discovered a novel mt gene order in Falco peregrinus and Smithornis sharpei,transposition of ND6 and the control region (Mindell et al., 1998), although this does not directly affect analyses of primary sequence reported here. MP analyses of 13 protein-coding gene amino acids, 2 rRNA genes, and 22 tRNA genes combined, using the PROTPARS matrix and transversions only for rRNA, yield the tree topology shown in Figure 1. This same topology was found in identical analyses with ND6 excluded, although in this case there was a second equally short tree indi-

cating a sister relationship for Smithornis and Falco. For this reason we show a polytomy in Figure 1. MP analysis with equal weighting of all characters for all genes combined differs from Figure 1 in placing Smithornis basal to all other birds, with subsequent divergence of Falco and then Vidua. MP analysis based on the protein amino acid sequences alone results in the same tree topology seen in Figure 1, with the polytomy resolved as Smithornis diverging just prior to Falco. MP analysis results for the tRNA genes match the combined analysis phylogeny (Fig. 1), except that the suboscine songbird

1999

MINDELL ET AL.—PHYLOGENY OF BIRDS AND REPTILES

is basal to all birds and the oscine songbird is basal to falcon/duck/chicken/ratite (Table 1, tree 6). MP analysis for rRNA transversions alone differs from the combined analysis tree (Fig. 1) only in switching positions for the oscine and suboscine songbirds. Separate MP analysis of all rRNA changes equally weighted yielded three equally parsimonious trees, the consensus indicating the suboscine songbird as sister to all birds and the oscine songbird as sister to the waterfowl. Focusing on the earliest divergence among the bird taxa, we conducted MP analyses of individual protein-coding gene amino acid sequences for all 18 taxa and found that 6 of 13 (46.1%) genes indicated either the oscine (Atp6, CO1, CO2, Cytb) or suboscine (ND2, ND4) songbird as diverging basally relative to the other birds. For the other protein-coding genes, the earliest diverging lineage is either unresolved (Atp8, ND3, ND4L, ND6), falcon (CO3), or chicken (ND1). ML analyses of DNA, using the GTR model with the discrete approximation of the G distribution; a proportion of invariant sites (I) estimated from the data for the protein-coding, rRNA, and tRNA genes combined nds the optimal tree topology in Figure 2. This is identical to Figure 1, but the polytomy is resolved by placing Smithornis and Falco as sister taxa. Further ML analyses based on the amino acid sequences, rRNA genes, and tRNA genes separately and in combination, both with and without assumption of a homogeneous rate across sites, are presented in Tables 1–4. Analyses for 12 protein-coding genes (all but ND6) accounting for rate heterogeneity across sites using a G distribution (Tables 2, 4) indicate the same relationships seen in Figures 1 and 2. The ML analyses of 13 amino acid sequences and of all genes summed (the “total” column), which assume rate homogeneity, differ in postulating a sister relationship for turtle and alligator (Table 3). Separate ML analyses of rRNA and tRNA genes assuming rate homogeneity agree with Figure 1 in placing birds and alligator as sister taxa (Table 3) but disagree in some of the relationships within birds (Table 1).

143

The highest log-likelihood score for a tree based on all genes in which duck is basal among birds is more than two standard deviations lower than that of the ML tree, allowing rejection of the hypothesis that Anseriformes represents the oldest (earliest diverging) extant avian lineage (Tables 1, 2). Basal positions for the ratites, galliform, or falconiform lineages cannot be rejected by the same criterion. However, the highestscoring ML trees in which their placement is basal are recovered in only 10.4%, 3.5%, and 7.9% of bootstrap estimates, respectively, based on analyses for all data sets combined that assume rate homogeneity (Table 1). Their positions are derived in MP analysis (Fig. 1), where bootstrap support is 87% or higher for all avian nodes, with one exception. We conducted a series of ML analyses for a subset of nine taxa (seven birds, one alligator, one turtle) to further assess the position and stability of the basal divergence among birds (Table 5). We calculated log-likelihood scores for two different avian topologies, with two different data sets, for each of ve different models of sequence evolution (in order of increasing number of parameters): JC, HKY, GTR, GTR + G , and GTR + G + I. We nd that increasing the number of parameters in the model marginally shifts support from a tree with a songbird diverging basally among birds to the conventional topology with ratites diverging basally (Table 5). Neither of these two topologies can be rejected, however, except in the simplest model (JC) for analysis of all mt genes plus all codon positions, which rejects the topology with ratites diverging basally. MP and ML analyses of the combined data sets, with MP based on protein gene amino acids and ML based on protein gene nucleotides, placed turtle as sister to archosaurs (birds/alligator) (Figs. 1, 2). ML combined data set analyses based on amino acids for the protein genes (Table 3, total column), rather than DNA (as in Fig. 1), show a slight preference for placing alligator and turtle as sister taxa. However, the differencein log-likelihood scores between the latter hypothesis and that in Figures 1 and 2 is only

Tree topology

(((((DUCK,CHIC),RATI),FALC),SONGs),SONG) (((((DUCK,RATI),CHIC),FALC),SONGs),SONG) ((((DUCK,CHIC),(RATI,FALC)),SONGs),SONG) ((((DUCK,CHIC),RATI),(FALC,SONGs)),SONG) ((((DUCK,(CHIC,RATI)),FALC),SONGs),SONG) (((((DUCK,CHIC),RATI),FALC),SONG),SONGs) ((((DUCK,RATI),CHIC),(FALC,SONGs)),SONG) (((DUCK,CHIC),(RATI,(FALC,SONGs))),SONG) (((((DUCK,RATI),CHIC),FALC),SONG),SONGs) ((((DUCK,CHIC),RATI),FALC),(SONGs,SONG)) (((DUCK,(CHIC,RATI)),(FALC,SONGs)),SONG) (((DUCK,CHIC),((FALC,SONGs),SONG)),RATI) (((DUCK,RATI),((FALC,SONGs),SONG)),CHIC) (((((DUCK,CHIC),RATI),SONG),FALC),SONGs) ((((DUCK,CHIC),RATI),(SONGs,SONG)),FALC) ((DUCK,(RATI,((FALC,SONGs),SONG))),CHIC) (((DUCK,CHIC),(FALC,(SONGs,SONG))),RATI) ((DUCK,(RATI,(FALC,(SONGs,SONG)))),CHIC) (DUCK,((CHIC,RATI),((FALC,SONGs),SONG)))

Tree no.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

–52.0 ± 21.1 (0.0000) –58.8 ± 22.8 (0.0000) –65.2 ± 22.9 (0.0010) –27.1 ± 16.4 (0.0000) –66.2 ± 22.3 (0.0000) –23.2 ± 23.2 (0.0002) –34.9 ± 17.9 (0.0000) –32.4 ± 17.0 (0.0001) –30.2 ± 24.7 (0.0003) –35.5 ± 21.8 (0.0000) –40.6 ± 17.6 (0.0000) –12.0 ± 15.7 (0.0297) –6.4 ± 9.4 (0.0372) –3.0 ± 22.4 (0.1300) –20.6 ± 22.4 (0.0132) ML (0.1992) –22.6 ± 21.2 (0.0079) –12.7 ± 13.6 (0.0366) –32.9 ± 13.8 (0.0000)

–1.0 ± 21.8 (0.1675) –6.9 ± 27.9 (0.1040) –17.6 ± 28.2 (0.0546) ML (0.1858) –14.8 ± 26.8 (0.0328) –37.3 ± 26.9 (0.0007) –5.1 ± 18.8 (0.1244) –15.9 ± 15.3 (0.0526) –41.4 ± 31.8 (0.0000) –22.8 ± 28.6 (0.0103) –12.3 ± 17.4 (0.0493) –45.8 ± 25.3 (0.0041) –41.8 ± 29.4 (0.0058) –62.8 ± 22.3 (0.0001) –24.4 ± 28.8 (0.0292) –56.8 ± 27.4 (0.0006) –33.1 ± 33.5 (0.0312) –40.4 ± 35.5 (0.0131) –92.9 ± 27.2 (0.0000)

2 rRNA genes

D lnL± SE(BP)

13 protein genes D lnL± SE(BP)

–17.3 ± 7.4 (0.0000) –28.6 ± 11.3 (0.0000) –47.1 ± 13.0 (0.0000) –18.9 ± 12.6 (0.0000) –27.2 ± 11.7 (0.0000) ML (0.2905) –32.9 ± 15.4 (0.0000) –46.7 ± 17.0 (0.0000) –10.6 ± 8.1 (0.0199) –14.0 ± 8.5 (0.0044) –30.9 ± 15.7 (0.0000) –11.4 ± 18.9 (0.0881) –35.6 ± 20.4 (0.0000) –2.2 ± 7.1 (0.1547) –20.5 ± 11.0 (0.0003) –35.9 ± 19.5 (0.0000) –12.6 ± 16.8 (0.0468) –37.7 ± 17.2 (0.0000) –35.0 ± 20.3 (0.0000)

22 tRNA genes D lnL± SE(BP)

–24.3 ± 31.2 (0.0189) –48.2 ± 37.8 (0.0037) –84.0 ± 38.6 (0.0001) ML (0.3031) –62.2 ± 36.8 (0.0002) –14.6 ± 35.5 (0.1260) –26.9 ± 30.1 (0.0270) –49.1 ± 28.5 (0.0041) –36.2 ± 41.1 (0.0250) –26.4 ± 36.9 (0.0137) –37.9 ± 29.3 (0.0082) –23.2 ± 35.3 (0.0656) –37.8 ± 37.0 (0.0354) –22.1 ± 32.4 (0.0580) –19.6 ± 38.1 (0.0789) –46.7 ± 33.6 (0.0096) –22.3 ± 43.0 (0.1041) –44.8 ± 41.7 (0.0229) –114.9 ± 36.7 (0.0000)

Total

D lnL± SE(BP)

TABLE 1. Difference in log-likelihood scores (D lnL) for alternative avian phylogenetic hypotheses. Numbers in columns denote difference in log-likelihood scores between the maximum-likelihood (ML) tree and the alternative tree (row heading), followed by the standard error and bootstrap proportions (BP; Kishino and Hasegawa, 1989) for 104 RELL bootstrap estimates among the 945 possible rooted trees for relationships among six bird taxa, using ProtML for 13 mt protein genes, NucML for 2 rRNA genes and 22 tRNA genes, and TotalML for analysis of all the aforementioned genes combined (Adachi and Hasegawa, 1996). Only trees with BP > 1% for either 13 proteins or total, the ML tree for the rRNAs, and the highest scoring tree with DUCK basal (tree 19) are shown. Rhea americana and Struthio camelus (RATI) were assumed to be sister taxa, based on strong support in all other analyses, and relationships for the nonavian taxa were constrained as in Figure 1 plus two sh (Cyprinus carpio, Crossostoma lacustre) were included as an outgroup.

144 SYSTEMATIC BIOLOGY VOL. 48

1999

145

MINDELL ET AL.—PHYLOGENY OF BIRDS AND REPTILES

TABLE 2. Difference in log-likelihood scores for alternative avian phylogenetic hypotheses. Numbers and comparisons are as in Table 1. The rst column of log-likelihood difference scores is based on an assumption of evolutionary rate homogeneity across amino acid sites with use of ProtML in the MOLPHY program, and the second column of log-likelihood difference scores accommodates rate heterogeneity among sites through use of a G distribution with use of the AAML routine in the PAML program. 12 proteins D lnL± SE Tree no.

Tree topology

Equal rates across sites

Unequal rates across sites

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

–1.4± 20.7 –6.6± 27.2 –20.1 ± 27.1 ML –18.4 ± 26.1 –34.7 ± 25.8 –4.5± 17.8 –19.2 ± 14.4 –38.0 ± 31.0 –21.3 ± 27.3 –15.8 ± 16.3 –44.4 ± 24.1 –39.7 ± 28.8 –59.8 ± 21.0 –24.0 ± 27.6 –56.6 ± 26.6 –33.7 ± 32.3 –42.4 ± 34.5 –92.9 ± 26.3

ML –13.3± 11.9 –12.4 ± 10.2 –1.1± 12.4 –21.6 ± 10.1 –19.9 ± 9.0 –16.5 ± 17.3 –14.6 ± 16.4 –32.6 ± 14.8 –7.3± 11.7 –24.3 ± 16.1 –21.2 ± 21.0 –41.5 ± 23.3 –39.6 ± 14.5 –7.8± 15.4 –37.0 ± 22.1 –5.1± 18.9 –19.7 ± 20.2 –58.1 ± 22.1

0.1 (Table 3; compare trees 1 and 3). ML analyses of 12 amino acid sequences accounting for rate heterogeneity across sites show a slight preference for the turtle as sister to archosaurs (Table 4) in agreement with Figures 1 and 2. There was no support (i.e., recovery in RELL bootstrap samples) for placement of turtle as basal to other amniotes or for the Haemothermia hypothesis postulating birds and mammals as sister taxa; these topologies are rejected on the basis of the log-likelihood score differences (Tables 3 and 4). Neither of these latter two hypotheses was ever recovered in RELL bootstrap samples (Table 3). To allow comparison of all possible trees for the seven birds in ML analyses with MOLPHY (Tables 1, 2), the nonavian outgroup taxa were constrained to the phylogenetic positions shown in Figure 1. Replicate comparisons constraining alligator and turtle as sisters and hav-

ing trichotomies for whale/rhino/cat and for placental/marsupial/monotreme mammals had no effect on the optimal TotalML topology for birds. Similarly, relationships among birds and among mammals were constrained to the Figure 1 topology in the Table 3 and Table 4 comparisons. Replicate comparisons constraining birds and mammals to several alternative topologies, including the traditional avian tree with ratite diverging basally, also provided no support for the Haemothermia hypothesis or for turtles being basal among amniotes. In light of apparent constraints on the evolution of functionally important hydrophobic amino acids (Naylor et al., 1995), and potential for their convergent similarity, we also analyzed amino acids with MP and ML by scoring isoleucine, leucine, and valine as the same state; however, the phylogenetic results shown in Figures 1 and 2 were not altered.

1 2 3 4 5

Tree no.

(((BIRD,ALIG),TURT),MAMM) (((BIRD,TURT),ALIG),MAMM) ((BIRD,(ALIG,TURT)),MAMM) (((BIRD,ALIG),MAMM),TURT) ((BIRD,MAMM),(ALIG,TURT))

Tree topology ML (0.5174) –24.7± 10.8 (0.0005) –1.0± 14.6 (0.4712) –40.0± 11.8 (0.0000) –39.5± 20.0 (0.0000)

–5.3± 15.4 (0.3368) –22.0± 12.8 (0.0076) ML (0.6460) –88.5± 28.7 (0.0000) –93.3± 22.7 (0.0000)

2 rRNA genes

D lnL± SE(BP)

13 proteins genes D lnL± SE(BP)

ML (0.5279) –1.5± 6.0 (0.3416) –4.2± 5.1 (0.0809) –30.9± 10.5 (0.0000) –37.0± 12.0 (0.0000)

22 tRNA genes D lnL± SE(BP)

–0.1± 15.4 (0.4832) –43.0± 17.8 (0.0002) ML (0.5166) –154.2± 32.8 (0.0000) –164.5± 32.5 (0.0000)

Total

D lnL± SE(BP)

TABLE 3. Difference in log-likelihood scores for alternative amniote phylogenetic hypotheses. Numbers and comparisons are as in Table 1 for alternative trees for four taxa (birds, alligator, turtle, and mammals), with relationships within birds and mammals constrained as in Fig. 1. Of the 15 possible trees, those shown are either recovered at least once in 104 RELL bootstrap estimates for the combined data sets analysis, or represent the best tree (highest likelihood) with the turtle divergence basal among amniotes (tree 4), or represent the best tree denoting the Haemothermia hypothesis (tree 5). The latter two hypotheses were not recovered in any bootstrap replications.

146 SYSTEMATIC BIOLOGY VOL. 48

1999

MINDELL ET AL.—PHYLOGENY OF BIRDS AND REPTILES

147

TABLE 4. Difference in log-likelihood scores for alternative amniote phylogenetic hypotheses. Numbers and comparisons are as in Table 3. The two columns for log-likelihood difference scores differ in being based on an assumption of evolutionary rate homogeneity across amino acid sites in the rst column (using MOLPHY), and accounting for rate heterogeneity through use of a G distribution in the second column (using PAML).

Tree no. 1 2 3 4 5

12 proteins D lnL± SE

Tree topology

Equal rates across sites

Unequal rates across sites

(((BIRD,ALIG),TURT),MAMM) (((BIRD,TURT),ALIG),MAMM) ((BIRD,(ALIG,TURT)),MAMM) (((BIRD,ALIG),MAMM),TURT) ((BIRD,MAMM),(ALIG,TURT))

–5.0 ± 15.3 –24.5 ± 12.8 ML –75.9 ± 28.3 –82.3 ± 21.7

ML –6.8 ± 6.1 –2.1 ± 7.2 –48.8 ± 15.9 –54.1 ± 18.1

Our hypothesis supports a sister relationship for marsupials and monotremes exclusive of placental mammals, consistent with aspects of the Marsupionta hypothesis of Gregory (1947), based on both MP and ML analyses (Figs. 1 and 2). Because our analyses include new species of birds and a turtle as well as an alternative set of placental mammals, they provide a complementary test relative to earlier mt analyses of Janke et al. (1996, 1997). The Marsupionta hypothesis is discussed in more detail in Kirsch and Mayer (1998) and Waddell et al. (1999a).

D ISCUSSION Avian Phylogeny We found broad agreement in our MP and ML analyses of the combined character data set (Figs. 1 and 2; Tables 1, 3 “total” columns), with the single exception among the 18 study taxa regarding relative placement of the falcon. We consider the tree topology in Figure 1, and supported by Figure 2 and Tables 2 and 4 (“unequal rates” columns), as our most highly corroborated hypothesis of phylogeny. This hypoth-

TABLE 5. Comparison of negative log-likelihood scores between the traditional avian topology (ratites diverging basally; see Table 2, tree 17) and the ML topology (Fig. 2) with a passeriform diverging basally for 7 birds, alligator, and turtle. Scores are based on DNA for all protein-coding, tRNA, and rRNA mt genes combined. For the 13 protein-coding genes, all codon positions are used in (A), codon positions 1 and 2 only in (B). The more likely tree score is shown in bold.a Model

Ratites basal

Passerine basal

P (KH test)

(A) All mt genes, all codon positions JC 92138.888 HKY 90208.395 GTR 88413.175 GTR, G 84037.006 GTR, G , I 84007.120

91983.517 90115.622 88339.667 84043.361 84013.320

.0029 .0833 .1035 .7453 .7447

(B) All mt genes, codon positions 1 and 2 only JC 50878.874 HKY 49324.641 GTR 49011.176 GTR, G 46934.232 GTR, G , I 46912.820

50804.306 49296.073 48985.306 46955.098 46932.342

.1064 .5020 .5260 .2601 .2793

a Model abbreviations: JC = Jukes–Cantor; HKY = Hasegawa–Kishino–Yano; GTR = general time-reversible model; G = discrete approximation of the gamma distribution; I = a proportion of sites assumed to be invariant among taxa. P (KH test) = probability of getting a more extreme difference in scores for the two trees under the null hypothesis of there being no difference between them.

148

SYSTEMATIC BIOLOGY

esis indicates a basal divergence for an oscine songbird and more recent divergences for duck and ratites, two lineages previously hypothesized to be basal among birds. This tree supports sister relationships for rhea and ostrich (ratites) and for duck and chicken, congruent with previous molecular and morphological analyses (Sibley and Ahlquist, 1990; Caspers et al., 1997; Livezey, 1997; Mindell et al., 1997). The optimal tree further indicates a sister relationship between the ratites and the duck/chicken clade. This is incongruent with the traditional view in which the duck/chicken group is sister to all other (nonratite) birds, including songbirds and falcon. A basal divergence for the passeriform lineage relative to the other birds, as indicated in earlier analyses based on 12 mt proteincoding genes and 2 rRNA genes combined (Mindell et al., 1997) and the mt cytb gene (H a¨ rlid et al., 1997, 1998), remains the most unexpected and, potentially, most consequential result, because it requires a reassessment of the oldest relationships and divergences among extant avian lineages. Waddell et al. (1999b) nd that analyses using LogDet transformation distances with invariant sites removed (seeking to account for both base composition differences among taxa and unequal substitution rates at different sites) also support a basal divergence for a passerine among other birds. However, the traditional avian topology, which has the earliest divergence separating ratites (and Tinamiformes) from all other birds, cannot be rejected. Further, one can obtain the traditional topology (Table 1, tree 17) by attaching the root for birds to ratites instead of a songbird (as in Table 1, tree 1). Considering ML analyses with the G distribution model (Table 2), we nd that the best (tree 1) and the second best (tree 4) trees indicate basal divergence for a passerine. However, the third best tree has a basal divergence for ratites (tree 17), and its loglikelihood is lower than the best tree by only 5.1 (± 18.9), whereas the difference with the homogeneous rate model is 33.7 (± 32.3) (also Table 2). This notable reduction of the log-likelihood difference between the ratitebasal tree and the ML passerine-basal tree,

VOL. 48

resulting from introduction of the G distribution accounting for rate heterogeneity across sites, might indicate an artifact in which passerines and the sister taxon to birds (alligator) are attracted. However, Waddell et al. (1999b) found no signi cant departure from rate constancy among birds [including a passerine (Vidua)] and an alligator, based on a log-likelihood ratio test for amino acid change, nor did they nd significant differences in relative rate tests among pairs of birds for those same amino acid sequences. Further, there is a tendency for ML analyses based on more realistic models to give smaller and less signi cant loglikelihood difference scores for alternative trees in comparison with analyses based on less realistic models (e.g., Hasegawa and Adachi, 1996; Cao et al., 1998). Thus, although our current analyses indicate a basal divergence for a songbird, we cannot rule out the possibility that this is an artifact. Other support for the passerine lineage as a basal avian divergence comes from analyses of mt cytb alone, in which the traditional topology with ratites diverging basally among birds could be rejected on the basis of ML analyses using the Kishino– Hasegawa test (H a¨ rlid et al., 1998). We also found a basal divergence for a passerine based on mt cytb alone for our smaller number of study taxa. However, analyses, and presumably rejection of alternative hypotheses, based on a single gene can be particularly misleading. Some of the unexpectedness of a basal position for Passeriformes is due to the lack of fossil forms prior to about 25 million years ago (e.g., Mourer-Chauvir´e et al., 1989). This lack of older fossils is consistent, however, with the reduced frequency of preservation of nonaquatic, smaller birds with lighter and more fragile bones, and with a potential Gondwanan origin for the passeriform lineage—subject to a relative shortage of known fossil beds and fossil surveys in the relevant areas. Appearance of Passeriformes as nonmonophyletic is also unexpected, given the previous morphological (Raikow, 1982) and molecular (Sibley and Ahlquist, 1990; Mindell et al., 1997) analyses that support monophyly. MP analysis for

1999

MINDELL ET AL.—PHYLOGENY OF BIRDS AND REPTILES

12 protein-coding genes and all tRNA and rRNA genes constrained to maintain monophyly of Passeriformes (Vidua and Smithornis as sister taxa) yielded two trees, each 22 steps longer than the MP tree based on the same data set in Figure 1. In one, Passeriformes is basal to all other birds; in the other, falcon is basal to all other birds. Exclusion of either the oscine or the suboscine songbirds in alternative MP analyses yields single shortest trees, with the included passeriform placed as basal to all other birds, as seen for Vidua in Figure 1. Exclusion of falcon yields a single MP tree with Vidua basal, also as in Figure 1. Thus, neither constraint to passeriform monophyly, nor changes in inclusion for the two songbirds and falcon, alter hypothesized relationships for duck, chicken, or ratites, or provide support for them as being basal among birds. We are skeptical of passeriform nonmonophyly, 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 nding does suggest, however, a relatively early divergence between the oscine and suboscine songbird lineages. Implications of Early Divergent Passeriformes Hypothesis There has been much debate regarding the age of extant avian orders and whether only one (Feduccia, 1996) or many (Hedges et al., 1996; Cooper and Penny, 1997) extant lineages arose during the Cretaceous and survived the extinction events marking the end of that period 65 million years ago. Extant orders of birds purported to be represented by late Cretaceous or early Paleocene fossils include Anseriformes (waterfowl; Presbyornis), Gaviiformes (loons; Neogaeornis), Charadriiformes (shorebirds), and Procellariformes (tubenoses). If the fossils are correctly attributed to these taxa, phylogenetic analyses indicating that some other avian lineage is basal to any of these four would imply a divergence event older than the fos-

149

sils. In this light, rejection of a basal position for Anseriformes (Tables 1, 2) further supports the antiquity of birds—the passeriform and falconiform or struthioniform (ratite) lineages in this case—and their diversi cation prior to the end of the Cretaceous. The avian order Passeriformes constitutes about 59% of all extant bird species, and the cause of this species-richness has been the subject of some debate (e.g., Raikow, 1986; Feduccia, 1996). Resolution of its phylogenetic position will help determine whether and to what degree arguments other than age are needed to explain passeriform diversity relative to that of other extant avian lineages. Our current study taxa cover only a subset of the extant bird diversity (5 of about 24 commonly recognized orders), and future analyses that include data from more birds, particularly shorebirds, loons, and tubenoses, as well as data from lizards and snakes will help resolve whether other avian taxa are basal to those considered here. Phylogenetic Placement of Turtles Our analyses provide an assessment of (1) the Haemothermia hypothesis (Gardiner, 1982), which posits that birds and mammals are sister taxa relative to the reptiles, including turtles, and (2) the hypothesis that turtles diverge basally within the tree for amniotes (Gaffney, 1980). We are able to reject both these hypotheses, given the mt gene partitions and all mt genes in combination (Tables 3 and 4, trees 4 and 5), by using the Kishino and Hasegawa test. Our ML analyses that assume rate homogeneity in Tables 3 and 4 (based on proteins and on all genes combined) indicate a sister relationship for alligator and turtle. However, our ML analyses that accommodate rate heterogeneity across sites (Fig. 2 and Table 4) and our MP and ML analysis in Figures 1 and 2 all agree in positing birds and alligator as sisters (archosaur monophyly), with turtle as sister to them. The conventional view, based on morphology, is that turtles are sister to an archosaur (birds and crocodilians) plus lepidosaur (lizards, snakes, and tuatara) clade. Our ndings are consistent with the conventional view, but do not replicate that

150

VOL. 48

SYSTEMATIC BIOLOGY

view, given our lack of a representative lepidosaur. The conventional view of turtles as sister to an archosaur/lepidosaur clade is based solely on interpretation of turtles’ lack of temporal fenestration (the anapsid condition) as a homologous trait shared with the extinct Palaeozoic anapsids. Archosaurs and lepidosaurs are characterized as having two temporal fenestrations (the diapsid condition), at least primitively if not currently. Rieppel and deBraga (1996) broadened the perspective on turtle af nities by including Mesozoic and extant taxa in new morphological analyses that support turtles as belonging within the diapsid group and sister to the lepidosaurs. Four recent molecular analyses support the diapsid af nities of turtles, but place them as sister to archosaurs (Zardoya and Meyer, 1998: based on mt rRNA; Platz and Conlon, 1997: based on pancreatic polypeptide amino acid sequences), or to alligator (Mannen et al., 1997: based on DNA for lactate dehydrogenase A; Kirsch and Mayer, 1998: based on DNA hybridization), or to birds (Mannen et al., 1997: based on DNA for lactate dehydrogenase B) rather than as sister to lepidosaurs, as hypothesized by Rieppel and deBraga (1996). Though we note an af nity between turtle and alligator in our mtDNA analyses using ML that assumes rate homogeneity, turtle as sister to archosaurs is better supported in our analyses that account for rate heterogeneity. Analyses extending Table 4, with the topology for birds constrained to the traditional view of ratites diverging basally from other birds, still found the highest scoring tree accounting for rate heterogeneity to be that placing turtles as sister to archosaurs. Utility of Mitochondrial Genomes Analyses of whole mt genomes for vertebrates have provided several controversial hypotheses, including a sister relationship for marsupials and monotremes, exclusive of placental mammals (Janke et al., 1996, 1997; Figs. 1, 2 in this study), and the nding that divergence between Amniota and Osteichthyes predates divergences among lung sh and all other bony sh (Rasmussen et al., 1998; but see Cao et al., 1998). Such

hypotheses contrast sharply with conventional views and lead to questions about the phylogenetic utility of mtDNA. We note, however, that inaccurate phylogenetic estimates based on molecular sequences stem from poor understanding and failure to incorporate knowledge of molecular evolutionary processes rather than a lack of historical information (Mindell and Thacker, 1996; Naylor and Brown, 1998). We have sought to consider potential biases in our analyses and to reduce them through the use of weighted maximum parsimony analyses of both nucleotide and amino acid sequences and the use of ML analyses with model parameters accommodating base composition heterogeneity as well as among-sites and among-taxa rate heterogeneity. Despite the large number of characters available in mt genomes, we recognize current limitations on taxon sampling, particularly in Reptilia inclusive of birds. We look to (1) improvements in understanding sequence evolution and (2) incorporation of additional taxa and data sets in further testing the hypotheses presented here. ACKNOWLEDGMENTS We thank Axel Meyer, Peter Waddell, Rafa Zardoya, and two anonymous reviewers for valuable comments on drafts of the manuscript. We thank David Swofford for giving us permission to use a prerelease version of PAUP* (4d63). Robert B. Payne kindly provided Smithornis sharpei and Vidua chalybeata tissue samples. This work was supported by a NSF grant to D.P.M. and supplementary funds from the University of Michigan Museum of Zoology Swales and Fargo Funds. M.D.S. was supported by a NSF grant to R. B. Payne.

REFERENCES ADACHI, J., AND M. HASEGAWA . 1996. MOLPHY, version 2.3: Programs for molecular phylogenetics based on maximum likelihood. Comput. Sci. Monogr. No. 28. Institute of Statistical Mathematics, Tokyo. ANDERSON , S., A. T. BANKIER, B. G. BARRELL , M. H. L. DE BRUIJN , A. R. COULSON , J. DROUIN , I. C. E PERON , D. P. NIERLICH , B. A. ROE , F. SANGER , P. H. SCHREIER , A. L. H. SMITH , R. STADEN , AND I. G. YOUNG . 1981. Sequence and organization of the human mitochondrial genome. Nature 290:457–464. ARNASON , U., AND A. GULLBERG . 1993. Comparison between the complete mtDNA sequences of the blue and the n whale, two species that can hybridize in nature. J. Mol. Evol. 37:407–417. BENTON , M. J. 1990. Phylogeny of the major tetrapod groups: Morphological data and divergence dates. J. Mol. Evol. 30:409–424.

1999

MINDELL ET AL.—PHYLOGENY OF BIRDS AND REPTILES

BIBB, M. J., R. A. VAN ETTEN , C. T. WRIGHT, M. W. WALBERG, AND D. A. CLAYTON . 1981. Sequence and gene organization of mouse mitochondrial DNA. Cell 26:167–180. BREMER , K. 1988. The limits of amino-acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42:795–803. CAO , Y., A. J ANKE , P. J. WADDELL , M. WESTERMAN , O. ¨ TAKENAKA , S. MURATA , N. OKADA, S. PA¨ ABO , AND M. HASEGAWA . 1998. Con ict amongst individual mitochondrial proteins in resolving the phylogeny of eutherian orders. J. Mol. Evol. 47:307–322. CAO , Y., P. J. WADDELL , N. OKADA, AND M. HASEGAWA . 1998. The complete mitochondrial DNA sequence of the shark Mustelus manazo: Evaluating rooting contradictions with living bony vertebrates. Mol. Biol. Evol. 15:1637–1646. CASPERS, G.-J., D. U. DE WEERD , J. WATTEL , AND W. W. DE JONG. 1997. a -Crystallin sequences support a Galliform/Anseriform clade. Mol. Phylogenet. Evol. 7:185–188. CASPERS, G. J., G. J. REINDERS , J. A. LEUNISSEN , J. WATTEL , AND W. W. DE JONG. 1996. Protein sequences indicate that turtles branched off from the amniote tree after mammals. J. Mol. Evol. 42:580–586. COOPER, A., AND D. PENNY . 1997. Mass survival of birds across the Cretaceous–Tertiary boundary: Molecular evidence. Science 275:1109 –1113. CRACRAFT , J. 1981. Toward a phylogenetic classi cation of birds of the world (class Aves). Auk 98:681– 714. CRACRAFT , J., AND D. P. MINDELL . 1989. The early history of modern birds: A comparison of molecular and morphological evidence. Pages 389–403 in The hierarchy of life (B. Fernholm, K. Bremer, and H. J o¨ rnvall, eds.). Elsevier Science Publishers, Amsterdam. DE B EER , G. R. 1956. The evolution of ratites. Bull. Brit. Mus. (Nat. Hist.) Zool. 4:57–76. DESJARDINS , P., AND R. MORAIS. 1990. Sequence and gene organization of the chicken mitochondrial genome. J. Mol. Biol. 212:599–634. FEDUCCIA , A. 1996. The origin and evolution of birds. Yale Univ. Press, New Haven, Connecticut. GAFFNEY, E. S. 1980. Pages 593–610 in The terrestrial environment and the origin of the land vertebrates (A. L. Panchen, ed.). Academic Press, New York. GARDINER, B. G. 1982. Tetrapod classi cation. Zool. J. Linn. Soc. 74:207–232. GAUTHIER , J., A. G. KLUGE , T. ROWE. 1988. Amniote phylogeny and the importance of fossils. Cladistics 4:105–209. GREGORY , W. K. 1947. The monotremes and the palimpsest theory. Bull. Am. Mus. Nat. Hist. 88:1–52. GU, X., Y.-X. FU , AND W.-H. LI. 1995. Maximum likelihood estimation of the heterogeneity of substitution rate among nucleotide sites. Mol. Biol. Evol. 12:546– 557. ¨ HARLID , A., A. JANKE, U. ARNASON . 1997. The mtDNA sequence of the ostrich and the divergence between paleognathous and neognathous birds. Mol. Biol. Evol. 14:754–761. ¨ HARLID , A., A. J ANKE , U. ARNASON . 1998. The complete mitochondrial genome of Rhea americana and early avian divergences. J. Mol. Evol. 46:669–679.

151

HASEGAWA , M., AND J. ADACHI. 1996. Phylogenetic position of cetaceans relative to artiodactyls: Reanalysis of mitochondrial and nuclear sequences. Mol. Biol. Evol. 13:710–717. HASEGAWA , M., H. KISHINO, AND T. YANO. 1985. Dating of the human–ape splitting by a molecular clock of mitochondrial DNA. J. Mol. Evol. 21:160–174. HEDGES , S. B., P. H. PARKER, C. G. SIBLEY, AND S. KUMAR. 1996. Continental breakup and the ordinal diversi cation of birds and mammals. Nature 381:226–229. HOLMGREN , N. 1955. Studies on the phylogeny of birds. Acta Zool. 36:243–328. JANKE , A., AND U. ARNASON . 1997. The complete mitochondrial genome of Alligator mississippiensis and the separation between recent Archosauria (birds and crocodiles). Mol. Biol. Evol. 14:1266–1272. JANKE , A., G. FELDMAIER -FUCHS, W. K. THOMAS , A. VON ¨ HAESELER , AND S. PA¨ ABO . 1994. The marsupial mitochondrial genome and the evolution of placental mammals. Genetics 137:243–256. JANKE , A., N. J. GEMMELL , G. FELDMAIER -FUCHS, A. ¨ ABO ¨ VON HAESLER , AND S. PA . 1996. The mitochondrial genome of a monotreme—the platypus (Ornithorhynchus anatinus). J. Mol. Evol. 42:153–159. JANKE , A., X. XU, AND U. ARNASON . 1997. The complete mitochondrial genome of the wallaroo (Macropus robustus) and the phylogenetic relationship among Monotremata, Marsupialia and Eutheria. Proc. Natl. Acad. Sci. USA 94:1276–1281. JUKES, T. H., AND C. R. CANTOR . 1969. Evolution of protein molecules. Pages 21–132 in Mammalian protein metabolism (H. N. Munro, ed.). Academic Press, New York. KIRSCH , J. A. W., AND G. C. MAYER . 1998. The platypus is not a rodent: DNA hybridization, amniote phylogeny and the palimpsest theory. Phil. Trans. R. Soc. Lond. B 353:1221 –1237. KISHINO, H., AND M. HASEGAWA . 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29, 170–179. KISHINO, H., T. MIYATA , AND M. HASEGAWA . 1990. Maximum likelihood inference of protein phylogeny and the origin of chloroplasts. J. Mol. Evol. 30:151–160. KUMAR, S. AND S. B. HEDGES . 1998. A molecular time scale for vertebrate evolution. Nature 392:917–920. LANAVE , C., G. PREPARATA , C. SACCONE, AND G. SERIO. 1984. A new method for calculatin g evolutionary substitution rates. J. Mol. Evol. 20:86–93. 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. LOPEZ, J. V., M. CULVER , S. CEVARIO , AND S. J. O’BRIEN . 1996. Complete nucleotide sequences of the domestic cat (Felis catus) mitochondrial genome and a transposed mtDNA repeat (Numt) in the nuclear genome. Genomics 33:229–246. LØVTRUP, S. 1985. On the classi cation of the taxon Tetrapoda. Syst. Zool. 34:463–470. LOWE, P. R. 1935. On the relationship of the Struthiones to the dinosaurs and to the rest of the

152

SYSTEMATIC BIOLOGY

avian class, with special reference to the position of Archaeopteryx. Ibis 77:398–432. MANNEN , H., S. C.-M. TSOI , J. S. KRUSHKAL , W.-H. LI, AND S. S.-L. LI. 1997. The cDNA cloning and molecular evolution of reptile and pigeon lactate dehydrogenase isozymes. Mol. Biol. Evol. 14:1081–1087. MINDELL , D. P., SORENSON , M. D., AND DIMCHEFF, D. E. 1998. Multiple independent origins of mitochondrial gene order in birds. 1998. Proc. Natl. Acad. Sci. USA 95:10693 –10697. MINDELL , D. P., SORENSON , M. D., AND DIMCHEFF, D. E. 1998. An extra nucleotide is not translated in mitochondrial ND3 of some birds and turtles. Mol. Biol. Evol. 15:1568–1571. MINDELL , D. P., SORENSON , M. D., HUDDLESTON , C. J., MIRANDA , H. C., JR., KNIGHT, A., SAWCHUK, S. J., AND YURI , T. 1997. Phylogenetic relationships among and within select avian orders based on mitochondrial DNA. Pages 213–247 in Avian molecular evolution and systematics (D. P. Mindell, ed.). Academic Press, San Diego. MINDELL , D. P., AND C. E. THACKER. 1996. Rates of molecular evolution: Phylogenetic issues and applications. Annu. Rev. Ecol. Syst. 27:279–303. MOURER-CHAUVIR E´ , C., C. HUGUENEY , AND P. J ONET. 1989. Découverte de Passeriformes dans l’Oligocene supérieur de France. C. R. Acad. Sci. Ser. 2A 309:843–849. NAYLOR , G. J. P., AND W. M. BROWN . 1998. Amphioxus mitochondrial DNA, chordate phylogeny, and the limits of inference based on comparisons of sequences. Syst. Biol. 47:61–76. NAYLOR , G. J. P., T. M. COLLINS , AND W. M. BROWN . 1995. Hydrophobicity and phylogeny. Nature 373:565–566. NORIEGA, J. I., AND C. P. TAMBUSSI . 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 of birds. Pages 79– 238 in Avian biology, (D. S. Farner, J. R. King, and K. C. Parkes, eds.). Academic Press, New York. OLSON , S. L. 1992. Neogaeornis wetzeli Lambrecht: A Cretaceous loon from Chile. J. Vertebr. Paleontol. 12:122–124. PLATZ, J. E., AND J. M. CONLAN . 1997. . . . and turn back again. Nature 389:246. RAIKOW, R. J. 1982. Monophyly of the Passeriformes: Test of a phylogenetic hypothesis. Auk 99:431–445. RAIKOW, R. J. 1986. Why are there so many kinds of passerine birds? Syst. Zool. 35:255–259. RASMUSSEN , A., A. JANKE , AND U. ARNASON . 1998. The mitochondrial DNA molecule of the hag sh (Myxine glutinosa) and vertebrate phylogeny. J. Mol. Evol. 46:382–388. RIEPPEL, O., AND M. DE BRAGA . 1996. Turtles as diapsid reptiles. Nature 384:453–455. RODRÍGUEZ , F., J. L. OLIVER, A. MARÍN, AND J. R. MEDINA . 1990. The general stochastic model of nucleotide substitution. J. Theor. Biol. 142:485–501. ROE , B. A, D.-P. MA , R. K. WILSON , AND J. F.-H. W ONG. 1985. The complete nucleotide sequence

VOL. 48

of the Xenopus laevis mitochondrial genome. J. Biol. Chem. 260:9759 –9774. SIBLEY, C. G., AND J. E. AHLQUIST . 1972. A comparative study of the egg-white proteins of non-passerine birds. Bull. Peabody Mus. Nat. Hist. 39:1–276. SIBLEY, C. G., AND J. E. AHLQUIST . 1990. Phylogeny and classi cation of birds: A study in molecular evolution. Yale Univ. Press, New Haven, Connecticut. SORENSON , M. D., AND T. W. QUINN . 1998. Numts: A challenge for avian systematics and population biology. Auk. 115:214–221. SORENSON , M. D., AST , J. C., DIMCHEFF, D. E., YURI, T., AND MINDELL , D. P. In press. Primers for a PCRbased approach to mitochondrial genome sequencing in birds and other vertebrates. Mol. Phylogenet. Evol. SWOFFORD, D. L. 1998. PAUP*: Phylogenetic analysis using parsimony (*and other methods), ver. 4.0. Sinauer, Sunderland, Massachusetts. TAVARE´ , S. 1986. Some probabilistic and statistical problems on the analysis of DNA sequences. Lect. Math. Life Sci. 17:57–86. THOMPSON , J. D., T. J. GIBSON , F. PLEWNIAK, J. J EANMOUGIN , AND D. G. HIGGINS. 1997. The Clustal X windows interface: Flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 24:4876–4882. WADDELL , P. J., Y. CAO , AND M. HASEGAWA . 1999 Using novel phylogenetic methods to evaluate mammalian mtDNA, including amino acid-invarian t sites-LogDet plus site stripping, to detect internal con icts in the data, with special reference to the position of hedgehog, armadillo, and elephant. Syst. Biol. 48:31–53 (this issue). WADDELL , P. J., Y. CAO, M. HASEGAWA , AND D. P. MINDELL . 1999b. Assessing the Cretaceous superordinal divergence times within birds and placental mammals using whole mitochondrial protein sequences and an extended statistical framework. Syst. Biol. 48:119–137 (this issue). WADDELL , P. J., AND D. PENNY. 1996. Evolutionary trees of apes and humans from DNA sequences. Pages 53–73 in Handbook of human symbolic evolution (A. J. Lock, and C. R. Peters, eds.). Oxford Univ. Press, Oxford, England. XU, S., A. J ANKE , AND U. ARNASON . 1996. The complete mitochondrial DNA sequences of the greater Indian rhinoceros, Rhinoceros unicornis, and the phylogenetic relationship among Carnivora, Perissodactyla, and Artiodactyla (+ Cetacea). Mol. Biol. Evol. 13:1167– 1173. YANG , Z. 1994. Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: Approximate methods. J. Mol. Evol. 39:306–314. YANG , Z. 1997. PAML: A program package for phylogenetic analysis by maximum likelihood. CABIOS 13:555–556. ZARDOYA , R. AND A. MEYER. 1998. Complete mitochondrial genome suggests diapsid af nities of turtles. Proc. Natl. Acad. Sci. USA 95:14226 –14231. Received 21 July 1998; accepted 17 September 1998 Associate Editor: P. Waddell