Deuterostome Phylogeny and the Sister Group of the Chordates: Evidence from Molecules and Morphology J. McClintock Turbeville, Joseph R. Schulz, 1 and RudoljA. Raff Department
of Biology and Institute for Molecular and Cellular Biology, Indiana University
Complete coding regions of the 18s rRNA gene of an enteropneust hemichordate and an echinoid and ophiuroid echinoderm were obtained and aligned with 18s rRNA gene sequences of all major chordate clades and four outgroups. Gene sequences were analyzed to test morphological character phylogenies and to assess the strength of the signal. Maximum-parsimony analysis of the sequences fails to support a monophyletic Chordata; the urochordates form the sister taxon to the hemichordates, and together this clade plus the echinoderms forms the sister taxon to the cephalochordates plus craniates. Decay, bootstrap, and tree-length distribution analyses suggest that the signal for inference of deuterostome phylogeny is weak in this molecule. Parsimony analysis of morphological plus molecular characters supports both monophyly of echinoderms plus enteropneust hemichordates and a sister group relationship of this clade to chordates. Evolutionary parsimony does not support chordate monophyly. Neighbor-joining, Fitch-Margoliash, and maximum-likelihood analyses support a chordate lineage that is the sister group to an echinoderm-plus-hemichordate lineage. The results illustrate both the limitations of the 18s rRNA molecule alone for high-level phylogeny inference and the importance of considering both molecular and morphological data in phylogeny reconstruction. Introduction
The quest to find the closest relatives of the chordates has been one of the classic problems in systematic biology. Although embryological and anatomic characters support monophyly of a taxon that includes the Chordata, Hemichordata, and Echinodermata, the relationships among these deuterostomes remain controversial (fig. 1). However, the favored hypothesis inferred with morphological characters from extant taxa indicates that the echinoderms are the sister group of a hemichordate-plus-chordate clade (fig. 1). It is important to note that chaetognaths (phylum Chaetognatha) traditionally have been considered deuterostomes, but their relationships remain uncertain based on morphological and molecular data (Brusca and Brusca 1990, pp. 882, 889; Bone et al. 199 1; Telford and Holland 1993), and they are not included in this investigation. Hypotheses of deuterostome phylogeny have only recently been evaluated with analyses of gene sequence Key words: bilateria, chordates, deuterostomes, phylogeny, 18s rRNA.
morphology,
Address for correspondence and reprints: J. M. Turbeville, Department of Biology, University of Michigan, Ann Arbor, Michigan 48109-1048. 1. Present address: Department of Biology, University of California, San Diego, La Jolla. Mol. Bid. Evol. 11(4):648-655.
1994. 0 1994 by The University of Chicago. All rights reserved. 0737-4038/94/l lOS-0009$02.00
648
data. Analyses of partial 18s rRNA sequences (- 1,000 nucleotides) from echinoderms, a urochordate, a cephalochordate, and vertebrates by distance-matrix (Field et al. 1988) and parsimony (Patterson 1989) analyses did not support deuterostome monophyly. Analyses with evolutionary parsimony supported monophyly of the deuterostomes, although relationships among deuterostome taxa were not resolved (Lake 1990). A recent analysis (Holland et al. 199 1) compared -500 bp of the 18s rRNA genes of an enteropneust hemichordate, echinoderm, and two vertebrates. The analysis weakly supported a hemichordate-plus-vertebrate clade, but very few informative sites are present in the region sequenced. An integrated approach, combining morphological and molecular characters, has not been utilized to infer deuterostome phylogeny. We report here the first test of phylogenetic relationships of deuterostomes with complete coding regions of 18s rRNA genes from all major deuterostome phyla and include an assessment of deuterostome phylogeny with both molecular and morphological characters. Material and Methods
The complete coding regions of 18s genes of the sea urchin S’trongylocentrotus purpuratus (Echinodermata, Echinoidea), the brittle star Ophiopholis aculeata (Echinodermata, Ophiuroidea), and the acorn worm
Deuterostome
Phylogeny
649
ECHINODERMATA HEMICHORDATA
FIG. 1.-Hypotheses of deuterostome relationships. A, Traditional hypothesis. The Echinodermata is the sister taxon of a hemichordateplus-chordate clade (Maisey 1986; Schaeffer 1987). B, Hypothesis indicating monophyly of Hemichordata plus Echinodermata (see Grobben 1908; Schaeffer 1987, p. 190). C, Phylogeny indicating that the Echinodermata is the sister group of the Chordata (Jefferies 1986, p. 323). This phylogeny is dependent on the hypothesis that the extinct Stylophorans are not echinoderms but possess features of echinoderms and chordates. The systematic position of these fossils continues to be disputed (Jollie 1982; Jefferies 1986, pp. 345-358; Parsley 1988). Common to all hypotheses is chordate monophyly. Deuterostome morphological synapomorphies (among others) are as follows: anus forms near site of blastopore, and mouth forms secondarily elsewhere (Ax 1989; Schaeffer 1987); and protocoel pore(s) (Balser and Ruppert 1990). Only characters from extant taxa relevant to the conflicts are plotted and, except where indicated, are from Maisey (1986) and Schaeffer (1987). 1 = Axial complex (heart/glomerulus in hemichordates; Ruppert and Balser 1986; Balser and Ruppert 1990); 2 = gill clefts; 3 = dorsal hollow nerve cord with neuropore (Bateson 1884); 4 = neurenteric canal (Siewing 1969, p. 189; Schaeffer 1987); 5 = notochord (the hemichordate stomochord is commonly interpreted as a unique feature of the group, but homology of the stomochord and the chordate notochord is still disputed; see Balser and Ruppert 1990); 6 = postanal somatic tail; 7 = endostyle; 8 = vertebrate-like actins (Vandekerckhove and Weber 1984; Kovilur et al. 1993); 9 = equivalence of fate maps; IO = somites; and 11 = differentiation of neural tube into ependymal and synaptic layers. For additional synapomoprhies of the cephalochordates plus craniates and for craniate synapomorphies, see (Maisey 1986; Schaeffer 1987). Homology of the neural gland to Hatschek’s pit and craniate adenohypophysis is doubtful (Ruppert 1990) and was excluded. For disputed fossil characters defining nodes marked by an asterisk (*) in phylogeny C, see the work of Jefferies (1986, p. 323). A comprehensive comparison of phylogeny C would require inclusion of fossil characters and taxa; however, although critical reevaluation of the conflicting interpretation of the fossil characters is prerequisite and necessary, it is beyond the scope of this contribution. Blackened blocks represent characters defining the branch; and unblackened blocks represent hypothesized losses. Outgroups are not shown. For recent alternatives that include taxa for which sequences are not yet available, see the work of Brusca and Brusca (1990, pp. 872-873) and Schram (1 991).
Saccoglossus
kowalevskii (Hemichordata, Enteropneusta) were amplified by the polymerase chain reaction (PCR) from genomic DNA templates with eukaryoticspecific 18s primers (2-22 forward, S-ACCTGGTTGATCCTGCCA-3’, 1865- 1847 reverse, S-TGATCCATCTGCAGGTTC-3’; adapted from Medlin et al. 1988). The amplified genes were ligated into a T/A vector (Invitrogen) prior to sequencing both coding and noncoding strands. Direct sequencing of asymmetrically amplified PCR product was initially attempted for S. purpuratus, but sequencing the cloned gene was more consistent. In addition to the M 13 plasmid primers, five forward and six reverse internal primers were used,.whose sequences are as follows (numbers correspond to positions on the human sequence): 430-447, S-CGGAGARGGAGCCTGAGA-3’; 6 1 1-626,5’-GTGCCAGCMGCCGCGG-3’; 1024-1038, 5’-ATCAAGAACGAAAGT-3’; 1324-1338, 5’-GGTGGTGCATGGCCG-3’; 1488- 1503, 5’-CAGGTCTGTGATGCYC-3’; 445-429, 5’-TCAGGCTCCCTCTCCGG-3’; 632-6 15, 5’-GWATTACCGCGCGGCKGCTG-3’; 880-865, 5’-CCGAGGTCCTATTCCA-3’; 120 1- 1187, 5’-ATTCCTTTRAGTTTC-3’; 1503- 1488, 5’-GGGCATCACAGACCTG-3’; and 1708- 1692, 5’ACGGGCGGTGTGTRC-3’.
Sequences were aligned using the SeqApp manual aligner for Macintosh (D. Gilbert, Indiana University). Alignments were refined, where possible, by using a secondary-structure model of Xenopus laevis (Atmadja et al. 1984). Analyses were limited to “reliably” aligned regions (Turbeville et al. 1992; m-65%-70% similarity and greater; see Hillis and Dixon 199 1). Data were analyzed using the maximum-parsimony program of PAUP 3.0s and 3.1.1 (D. Swofford, Smithsonian Institution), the DNADIST, FITCH, DNAML, and NEIGHBOR programs of Phylip 3.4 and 3.5~ (J. Felsenstein, University of Washington), and the evolutionary-parsimony program (Lake 1987) as implemented in the PAUP package. Analyses with evolutionary parsimony were carried out as described by Turbeville et al. ( 199 1, 1992). For parsimony analyses, the HEURISTIC search was used with random stepwise addition. Gaps were treated as missing data in all analyses. The Kimura correction to estimated distances was employed in distance analyses. One hundred bootstrap replicates were performed for maximum-likelihood analysis, and 500 were run for all other analyses. The tree-length distribution test was carried as described by Hillis and Huelsenbeck ( 1992).
650
Turbeville et al.
Results
Maximum parsimony (MP) analyses of 18s sequences alone do not support monophyly of the Chordata, a result that is at variance with strong morphological evidence (fig. 1). The urochordates form the sister taxon to the hemichordate, and together this clade plus the echinoderms form the sister taxon to the cephalochordates plus craniates (fig. 2A). Support is weak for this arrangement, as indicated by bootstrap percentages and decay values (fig. 2A). The poorly supported artifactual placement of the long-branched urochordates may be the result of their sequences converging with the most divergent nonchordate deuterostome, the enteropneust hemichordate Saccoglossus. Chordata was monophyletic in only 16% of 500 bootstrap replicates. A monophyletic Chordata is found as three of nine equally parsimonious trees that require four additional substitutions. Two correspond to morphological phylogeny B (fig. 1). Morphological hypotheses A and C (fig. 1) are among 45 equally parsimonious trees saved at 11 extra substitutions. Evaluation of 100 trees saved that are within 1% of the length of the shortest tree suggests further that signal for reliable phylogeny inference among the major deuterostome taxa is weak in this molecule (fig. 2B). The tree-length distribution test (Hillis and Huelsenbeck 1992) was utilized to assess the strength of the signal in these sequence data. According to Hillis and Huelsenbeck ( 1992), data sets containing phylogenetic
signal result in left-skewed tree-length distributions, and those containing random noise result in nearly symmetrical distributions. Tree-length distribution of 100,000 trees including all taxa is significantly left skewed (gl = -0.794; fig. 3A). However, when all strongly supported clades are constrained (those with decay indices of five or greater and with bootstrap values 88% and greater), the distribution of the remaining trees is nearly symmetrical (gl = 0.031; fig. 3B). A similar result is obtained if only a single representative of each clade is included in the analysis. Thus, this test suggests that the signal is limited across the major deuterostome taxa. Topological variation was observed when individual ingroup or outgroup taxa were removed or substituted and analyses were rerun. For example, the hemichordate-plus-urochordate clade was found as the sister group to the cephalochordate-plus-craniate clade when Placopecten and Eurypelma alone were used as outgroups. When the short-branched (as inferred from the tree with all taxa) echinoderm Strongylocentrotus was removed, two of three equally parsimonious trees were equivalent to morphological hypothesis B (fig. 1). The other tree is identical to the shortest tree found including all taxa (fig. 2A). Thus, topological instability is apparently not related solely to the longest-branched taxa (Styela, Herdmania, and Myxine), and their removal has no effect on the placement of the other taxa. When the 1 molecular and 11 morphological characters listed in figure 1 are combined with the sequence-
3 G 1 9
XWlOpJS sebastolobus
SqUallJS Lathha
Myhe .
Bmnchioskma CEPHALOCHORDATA
2 a
0
2
Petromyzon 8 Brandliostofna 0 Stvh 4 Hwdman& S8ccoglossu8 Ophlophollr Stmngylocontrotus Placopecten Limicolaria
FIG. 2.-Phylogenies inferred by parsimony analyses of 18s rRNA gene sequences. A, Minimal length tree found by analyzing all reliably aligned positions. The HEURISTIC search was employed with random stepwise addition. Two hundred replicates were run. There are 304 phylogenetically informative (parsimony) sites of 1,63 1 analyzed. Tree length is 1,155 and consistency index is 0.656. If uninformative sites are excluded, tree length is 876 and consistency index is 0.547. Numbers above nodes are bootstrap percentiles for 500 replicates. Numbers in boldface (i.e., “decay indices”) indicate the number of extra steps, within five, at which the clade descending from that node is no longer supported (see Bremer 1988; Donoghue et al. 1992). Branch lengths are proportional to the number of nucleotide substitutions. Scale bar = 50 substitutions. B,Strict consensus of 100 trees within 1% length of the shortest (885~step) tree.
Deuterostome Phylogeny
65 1
(ML) trees are similar in major respects to neighborjoining trees (fig. 6C and D and table 1). A chordate lineage is weakly supported, as suggested by bootstrap results (66% with FM and 45% with ML). Both FM and ML find a poorly supported echinoderm-plus-enteropneust hemichordate group (47% with FM and 58% with ML) as the sister group of chordates (fig. 6C and D). Tree lengths
FIG. 3.-Tree-length distributions. A, Distribution of a random sample of 100,000 trees including all taxa. B,Distribution of the remaining 3 15 trees when all clades with decay indices >5 or bootstrap values >88% are removed with the constraint option of PAUP 3.1.1. The EXHAUSTIVE search was employed.
data matrix, the shortest tree indicates a monophyletic Chordata that is the sister group to an echinoderm-plushemichordate clade (fig. 4). This tree is equivalent to one of nine suboptimal trees of length 880, when the rRNA data alone are used. The phylogeny inferred with the combined data set is equivalent to phylogeny B (fig. 1). The combined analysis also indicates where the signal is weak, since only five chordate morphological synapomorphies are required to find a monophyletic Chordata. Evolutionary parsimony does not find support at the 5% significance level for a monophyletic Chordata; support for inclusion of the urochordates in a chordate clade is not significant (fig. 5). The method favors a hemichordate-plus-echinoderm clade, .but not significantly (fig. 5). Neighbor joining (NJ) supports a chordate lineage, whose sister taxon is a Saccoglossus-plus-echinoderm group. It also indicates that the cephalochordate is the closest chordate relative of the craniates. Support for these relationships as assessed by bootstrapping is weak (fig. 6A). This method failed to find a monophyletic Chordata when mollusks alone were included as outgroups and the longest branched ingroup taxon was deleted; the position of the urochordates is unresolved (fig. 6B). Fitch-Margoliash (FM) and maximum-likelihood
Discussion The results of these analyses raise cautions regarding reliability of phylogenies reconstructed with rRNA gene sequences. Taxa chosen for the ingroup and outgroup may radically alter the topology, regardless of length, as previously recognized (see Swofford and Olsen 1990; Turbeville et al. 199 1; Hillis and Huelsenbeck 1992). The variation in topology observed in this study when particular taxa, long-branched or otherwise, are deleted and the analyses are rerun suggests that there is conflict in the existing signal. Artifactual grouping of taxa with long branches (see Felsenstein 1978) is also a recurrent problem. Increasing the number of both ingroup and outgroup taxa could potentially reduce these problems, because their inclusion might reveal previously undetected homoplasy (Sanderson 1990; Swofford and Olsen 1990). Another serious pitfall of sequence comparisons concerns the inference of positional homology, or alignment (see Swofford and Olsen 1990; Hillis and Dixon 199 1). Although we have limited our analyses to,regions
Branchiostoma
S=Wllo-u~ HEMICHORDATA Ophlopholls ECHINODERMATA Strongylocentrotu8
1 1 E”~~‘ 1 ma ;;;?$@?I”
Ten&ii
MOLLUSCA
ARTHROPODA
FIG. 4.-Strict consensus of two equally parsimonious trees found in a combined parsimony analysis of 18s rDNA sequence data, the morphological and single molecular character listed in fig. 1. All characters were weighted equally. A monophyletic Chordata is found. Numbers above nodes are bootstrap percentiles. Numbers in boldface below nodes are clade decay values. Length of trees is 1,172 (893 when uninformative sites are ignored). The overall confidence index is 0.656 and. when uninformative sites are excluded.I 0.549.
652
Turbeville et al. Xenopus Latimeria sebastolobus squaius Myxine Petromyzon BraflChiO6t~
XLatimeda sebastolokrs
Sty& Herdmania
tiww Myxine PelKMnyron Branch&stoma
)pz0.*4(
Saccoglorsur
>--
Ophiopholla Strongybcontrotur
PLacoQedm LimkAaria
Saccoglor8ur
(
Ophlophdis Strongyiocontrotur
EUrypelM
Tenekio
A
B
FIG.5.-Representative analyses with evolutionary parsimony. A, Test of urochordate relationships. The favored tree from combining results for 6 craniates and the cephalochordate X 2 urochordates X hemichordate x 2 echinoderms (28 quartets) is shown. The P value for the expected tree (linking urochordates and craniates and the cephalochordate) is 0.72. The P value for the third alternative is 0.69. B, Test of deuterostome relationships. The favored tree from a combination of 6 craniates and the cephalochordate X hemichordate X 2 echinoderms X the 2 mollusks and 2 arthropods (56 quartets) is shown. The favored tree links echinoderms and hemichordates, although not significantly. The P values for trees linking the hemichordate and chordates and the echinoderms and chordates are P = 0.6 and P = 0.38, respectively.
of the sequence that are well conserved (Hillis and Dixon 199 1), it is probable that there are improperly aligned positions contributing to inferred noise (homoplasy), in addition to homologous sites that have incurred multiple hits. It should be noted that somewhat different topologies resulted when a computer-generated alignment (Clustal V; Higgins et al. 1992) was analyzed; only FM and ML found a chordate lineage (authors’ unpublished observations). Nonindependent changes may also lead to incorrect phylogenies, although this may be difficult to assess (see Wheeler and Honeycutt 1988; Hillis and Dixon 199 1; Dixon and Hillis 1993). Substitution bias also has been implicated in incorrect phylogeny inference (Lockhart et al. 1992; Marshall 1992). Our preliminary observations (authors’ unpublished data) reveal no obvious bias that might account for the placement of the urochordates outside the chordates by parsimony analysis. The expectation has been that gene phylogenies would allow reliable inference of the phylogenetic position of organisms that cannot be placed on the basis of morphological comparisons. But because optimal trees reconstructed from sequence data alone often conflict with well-supported morphological phylogenies (e.g., chordate monophyly), it will be essential to compare phylogenies obtained from several appropriate genes. The additional information present in complete 18s rDNA sequences does not provide significantly greater resolving power for inference of deuterostome relationships than do previously published partial sequences. The data still contain minimal signal and sufficient noise
to prevent unambiguous resolution of the conflicting hypotheses. It is significant that one of the hypotheses (fig. 1B) is supported by NJ, FM, and MP when morphological and molecular data are combined (ML differs only in the placement of the urochordates) (fig. 4 and table 1). The phylogeny inferred from total evidence by MP may be considered the current best combined hypothesis of deuterostome relationships. However, the combined analysis of molecular and morphological data, although justifiable (IUuge 1989; Donoghue and Sanderson 1992; Jones et al. 1993), remains controversial (Swofford 199 1; de Queiroz 1993). A major concern is that that the different data sets may not be homogeneous (Bull et al. 1993). The limitations of sequence data previously discussed also illustrate the value of initially analyzing sequence data apart from other characters, as these problems might otherwise go undetected (Swofford 199 1). If the phylogeny inferred from total data is confirmed, it will suggest that the axial complex (heart-glomerulus) of hemichordates and echinoderms is a synapomorphy (see Balser and Ruppert 1990) of the clade and that two complex morphological characters-gill slits and dorsal hollow nerve cord-shared by enteropneust hemichordates and chordates are homoplasies rather than shared-derived homologues (see Schaeffer 1987; fig. 1B and table 1). One interpretation of character evolution in the combined hypothesis is that the ancestral deuterostome possessed gill slits and a dorsal hollow nerve cord and that these features were lost in the stem lineage of the echinoderms (fig. 1B). If the combined
Deuterostome
Phylogeny
653
1 5 5
f
Petru4nyzorl Branchiostoma
c
UROCHORDATA
_ -
0
,_j-sstyela Herdmania Saazoglossus Ophlophotls Strongyloconbotus P-m
ARTHROPODA
Limicdaria
1 B
A
1 YpJfw_ 100
- 07 57
PWOmyZOfl
98
bhwia
sebastolobus Gqal~
47
Myxine
Myxine
66.
Bmnchiostoma
4!3 loo
100f-SW
Herdmania Branchiioma
Strongylocwtrotus
1
4
C
D
FIG. B.-Distance and likelihood trees. A, Phylogeny inferred by neighbor-joining analyses of 18s sequences. Branch lengths are proportional to the number of nucleotide substitutions. B,Neighbor-joining tree resulting after the longest-branch ingroup (Myxine) and arthropod outgroups were removed. C, Fitch-Margoliash tree. D,Maximum-likelihood tree. Scale bars = 20 substitutions/l,000 nucleotide positions. Numbers at nodes are bootstrap percentiles; there were 500 replicates for A and C and 100 for D.
hypothesis is confirmed, it will also strengthen the necessity of reevaluating the extinct stylophorans. Jefferies (1986, pp. 19 l-238) interprets these fossils as chordates possessing a calcite skeleton and asymmetry/dexiothetism (reduction of structures on the right side). These two characters occur in echinoderms and are interpreted by Jefferies ( 1986, p. 323; also see Cripps 199 1; fig. 1C) as synapomorphies of an echinoderm-plus-chordate clade. Other paleontologists interpret stylophorans as extinct echinoderms (see Parsley 1988; Sprinkle 1992), a hypothesis in accord with the provisional results presented here. An exhaustive evaluation of character evolution will require a strongly supported hypothesis of deuterostomes relationships, and we advocate further testing of deuterostome phylogeny. This will require additional molecular data from all deuterostome taxa, including pterobranch hemichordates, in combination with a
reassessment of deuterostome phological characters.
relationships
with mor-
Sequence Availability New sequences are available in GenBank. Accession numbers are L28054, L28055, and L28056 for Saccoglossus, Ophiopholis, and Strongylocentrotus, respectively. The alignment is available from EMBL (accession number DS 169 14) by anonymous FTP or E-mail and from the senior author (J.M.T.) via E-mail: “
[email protected]”. Note added in proof: While this paper was in press, an assessment of deuterostome phylogeny with complete 18s rDNA sequences appeared (Wada and Satoh 1994). The results of this study, which employed NJ and ML and included additional sequences, are only partly congruent with those presented here. A chordate lineage was not found.
654
Turbeville et al.
Table 1 Summary of Molecular and Morphological Data Analyses
BRUSCA,
R. C., and G. J. BRUSCA. 1990. Invertebrates. Sinauer, Sunderland, Mass.
BULL, J. J., J. P. HUELSENBECK, C. W. CUNNINGHAM, D. L. MORPHOLOGICAL PHYLOGENY~
ANALYSIS~
A
B
MPC...... MP”....... EPd. ...... NJ. ....... FM. ....... ML. ......
-
+
-
+ + +
C
-
“MPC, Maximum parsimony analysis of molecular plus morphological characters. Other analyses with sequencesalone: MP, maximum parsimony; EP, evolutionary parsimony; NJ, neighbor joining, FM, Fitch-Margoliash; ML, maximumlikelihood.Withinthe Chordata,MPC, NJ, andFM linkcephalochordatesandcraniates.ML linksurochordatesandcraniates. bA plus sign (+) indicatesthat topology was found; and a minus sign (-) indicatesthat it was not found. CH = Chordata; HE = Hemichordata;and EC = Echinodermata. ‘See text andfig. 2. dSee text.
SWOFFORD, and P. J. WADDELL. 1993. Partitioning and combining data in phylogenetic analysis. Syst. Biol. 42 : 384397. CRIPPS, A. P. 199 1. A cladistic analysis of the cornutes (stem cornutes). Zool. J. Linnean Sot. 102:333-336. DE QUEIROZ, A. 1993. For consensus (sometimes). Syst. Biol. 42:368-372. DIXON, M. T., and D. M. HILLIS. 1993. Ribosomal RNA secondary structure: compensatory mutations and implications for phylogenetic analysis. Mol. Biol. Evol. 10:256-267. DONOGHUE, M. J., R. G. OLMSTEAD, J. F. SMITH, and J. D. PALMER. 1992. Phylogenetic relationships of Dipsacales based on rbcL sequences. Ann. MO. Bot. Gardens 79:333345. DONOGHUE, M. J., and M. J. SANDERSON. 1992. The suitability of molecular and morphological evidence in reconstructing plant phylogeny. Pp. 340-368 in P. S. SOLTIS, D. E. SOLTIS, and J. J. DOYLE, eds. Molecular systematics of plants. Chapman & Hall, New York and London. FELSENSTEIN, J. 1978. Cases in which parsimony or compatibility methods will be positively misleading. Syst. Zool. 27: 401-410. FIELD, K. G., G. J. OLSEN, D. J. LANE, S. J. GIOVANNONI,
Acknowledgments
We thank Jessica Kissinger, John Logsdon, and Gavin Naylor for helpful suggestions and Greg Wray for Ophiopholis sperm samples. This research was supported by NSF grant BSR-88 18044. LITERATURE CITED ATMADJA, J., R. BRIMACOMBE,
and B. H. E. MADEN. 1984. experimental determination of secondary structural elements, and locations of methyl groups in the secondary structure model. Nucleic Acids Res. 12:2649-2667. AX, P. 1989. Basic phylogenetic systematization of the Metazoa. Pp. 229-245 in B. FERNHOLM, K. BREMER, and H. JORNVALL, eds. The hierarchy of life. Elsevier, Amsterdam. BALSER, E. J., and E. E. RUPPERT. 1990. Structure, ultrastructure, and function of the preoral heart-kidney in Succoglossus kowalevskii (Hemichordata, Enteropneusta) including new data on the stomochord. Acta Zool. Stockh. 71:235249. BATESON, W. 1884. The early stages in the development of Balanoglossus (sp. incert.). Q. J. Microsc. Sci. 24:208-236. BONE, Q., H. KAPP, and A. C. PIERROT-BULTS. 199 1. Introduction and relationships of the group. Pp. l-4 in Q. BONE, H. KAPP, and A. C. PIERROT-BULTS, eds. The biology of chaetognaths. Oxford University Press, Oxford, New York, and Tokyo. BREMER, K. 1988. The limits of amino-acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42:795803. Xenupus luevis 18s ribosomal RNA:
M. T. GHISELIN, E. C. RAFF, N. R. PACE,
and R. A. RAFF. 1988. Molecular phylogeny of the animal kingdom. Science 239:748-753. GROBBEN, K. 1908. Die systematische Einteilung des Tierreiches. Verh. Zool. Bot. Ges. Wien 58:49 l-5 11. HIGGINS, D. G., A. J. BLEASBY, and R. FIJCHS. 1992. Clustal V-improved software for multiple sequence alignmnet. Comput. Appl. Biosci. 8: 189- 19 1. HILLIS, D. M., and M. T. DIXON. 199 1. Ribosomal DNA: molecular evolution and phylogenetic inference. Q. Rev. Biol. 66:4 1 l-453. HILLIS, D. M., and J. P. HUELSENBECK. 1992. Signal, noise, and reliability in molecular phylogenetic analyses. J. Hered. 83:189-195. HOLLAND, P. W. H., A. M. HACKER, and N. A. WILLIAMS. 199 1. A molecular analysis of the phylogenetic affinities of Succoglossus cumbrensis brensis Brambell & Cole (Hemichordata). Philos. Trans. R. Sot. Lond. Biol. 332: 185- 189. JEFFERIES, R. P. S. 1986. Ancestry of the vertebrates. British Museum, London. JOLLIE, M. 1982. What are the ‘Calcichordata’? and the larger question of the origin of chordates. J. Linnean Sot. Zool. 75:167-188. JONES, T. R., A. G. KLUGE, and A. J. WOLF. 1993. When theories and methodologies clash: a phylogenetic reanalysis of the North American ambystomatid salamanders (Caudata: Ambystomatidae). Syst. Biol. 42:92- 102. KLUGE, A. G. 1989. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Syst. Zool. 38:7-25. KOVILUR, S., J. W. JACOBSON, R. L. BEACH, W. R. JEFFERY,
and C. R. TOMLINSON. 1993. Evolution of the chordate muscle actin gene. J. Mol. Evol. 36:361-368.
DeuterostomePhylogeny 655 LAKE, J. A. 1987. A rate-independent technique for analysis of nucleic acid sequences: evolutionary parsimony. Mol. Biol. Evol. 4: 167- 19 1. . 1990. Origin of the Metazoa. Proc. Natl. Acad. Sci. USA 87:763-766. LOCKHART,P. J., C. J. HOWE, D. A. BRYANT,T. J. BEANLAND, and A. W. D. LARKUM. 1992. Substitutional bias confounds inference of cyanelle origins from sequence data. J. Mol. Evol. 34:153-162. MAISEY, J. G. 1986. Heads and tails: a chordate phylogeny. Cladistics 2:201-256. MARSHALL,C. R. 1992. Substitution bias, weighted parsimony, and amniote phylogeny as inferred from 18s rRNA sequences. Mol. Biol. Evol. 9:370-373. MEDLIN, L., H. J. ELWOOD, S. STICKEL, and M. L. SOGIN. 1988. The characterization of enzymatically amplified eukaryotic 16S-like rRNA-coding regions. Gene 7 1:49 l-499. PARSLEY, R. L. 1988. Feeding and respiratory strategies in Stylophora. Pp. 347-361 in C. R. C. PAUL and A. B. SMITH, eds. Echinoderm phylogeny and evolutionary biology. Clarendon, Oxford. PATTERSON,C. 1989. Phylogenetic relations of major groups: conclusions and prospects. Pp. 47 l-488 in B. FERNHOLM, K. BREMER,and H. JORNVALL,eds. The hierarchy of life. Elsevier, Amsterdam. RUPPERT, E. E. 1990. Structure, ultrastructure and function of the neural gland complex of Ascidia interrupta (Chordata, Ascidiacea): clarification of hypotheses regarding the evolution of the vertebrate anterior pituitary. Acta Zool. Stockh. 71:135-149. RUPPERT, E. E., and E. J. BALSER. 1986. Nephridia in the larvae of hemichordates and echinoderms. Biol. Bull, 17 1: 188-196. SANDERSON,M. J. 1990. Estimating rates of speciation and evolution: a bias due to homoplasy. Cladistics 6:387-391. SCHAEFFER,B. 1987. Deuterostome monophyly and phylogeny. Evol. Biol. 21: 179-235. SCHRAM,F. R. 199 1. Cladistic analysis of metazoan phyla and the placement of fossil problematica. Pp. 35-46 in A. M. SIMONETTAand S. CONWAYMORRIS, eds. The early evolution of Metazoa and significance of problematic taxa. Cambridge University Press, Cambridge.
SIEWING, R. 1969. Lehrbuch
der vergleichenden Entwicklungsgeschichte der Tiere. Paul Parey, Hamburg. SPRINKLE,J. 1992. Radiation of Echinodermata. Pp. 375-398 in J. H. LIPPS and P. W. SIGNOR, eds. Origin and early evolution of the Metazoa. Plenum, New York. SWOFFORD,D. L. 199 1. When are phylogeny estimates from molecular and morphological data incongruent? Pp. 295333 in M. M. MIYAMOTOand J. CRACRAFT, eds. Phylogenetic analysis of DNA sequences. Oxford University Press, Oxford. SWOFFORD,D. L., and G. J. OLSEN. 1990. Phylogeny reconstruction. Pp. 41 l-501 in D. M. HILLIS and C. MORITZ, eds. Molecular Systematics. Sinauer, Sunderland, Mass. TELFORD, M. J., and P. W. H. HOLLAND. 1993. The phylogenetic affinities of chaetognaths: a molecular analysis. Mol. Biol. Evol. 10:660-676. TURBEVILLE,J. M., K. G. FIELD, and R. A. RAFF. 1992. Phylogenetic position of phylum Nemertini, inferred from 18s rRNA gene sequences: molecular data as a test of morphological charcter homology. Mol. Biol. Evol. 9:235-249. TURBEVILLE,J. M., D. M. PFEIFER, K. G. FIELD, and R. A. RAFF. 199 1. The phylogenetic status of arthropods, as inferred from 18s rRNA sequences. Mol. Biol. Evol. 8:669686. VANDEKERCKHOVE, J., and K. WEBER. 1984. Chordate muscle actins differ distinctly from invertebrate muscle actins: the evolution of different vertebrate muscle actins. J. Mol. Biol. 179:391-413. WADA, H., and N. SATOH. 1994. Details of evolutionary history from invertebrates to vertebrates, as deduced from sequences of 18s rDNA. Proc. Natl. Acad. Sci. USA 9 1: 180 l1804. WHEELER, W. C., and R. L. HONEYCUTT. 1988. Paired sequence difference in ribosomal RNAs: Evolutionary and phylogenetic implications. Mol. Biol. Evol. 5:90-96. SIMON EASTEAL, reviewing Received
January
Accepted
February
10, 1994 3, 1994
editor