J Mol Evol (1992) 35:102-113
Journal of
Molecular Evolution
@ Springer-VerlagNewYorkInc. 1992
First publ. in: Journal of Molecular Evolution 35 (1992), pp. 102-113
Molecules, Fossils, and the Origin of Tetrapods
Axel Meyer and Sarah I. Dolven
Department of Ecology and Evolution, State University of New York, Stony Brook, NY 11794, USA
Summary.
Since the discovery of the coelacanth,
L a t i m e r i a c h a l u m n a e , more than 50 years ago, pa-
leontologists and comparative morphologists have
debated whether coelacanths or lungfishes, two
groups of lobe-finned fishes, are the closest living
relatives of land vertebrates (Tetrapoda). Previously, Meyer and Wilson (1990) determined partial
DNA sequences from two conservative mitochondrial genes and found support for a close relationship of lungfishes to tetrapods. We present additional D N A sequences from the 12S r R N A
mitochondrial gene for three species of the two lineages of lungfishes that were not represented in the
first study: Protopterus a n n e c t e n s a n d Protopterus
aethiopicus from Africa and N e o c e r a t o d u s forsteri
(kindly provided by B. Hedges and L. Maxson) from
Australia. This extended data set tends to group the
two lepidosirenid lungfish lineages (Lepidosiren and
Protopterus) with N e o c e r a t o d u s as their sister group.
All lungfishes seem to be more closely related to
tetrapods than the coelacanth is. This result appears
to rule out the possibility that the coelacanth lineage
gave rise to land vertebrates. The common ancestor
of lungfishes and tetrapods might have possessed
multiple morphological traits that are shared by
lungfishes and tetrapods [Meyer and Wilson (1990)
listed 14 such traits]. Those traits that seem to link
L a t i m e r i a and tetrapods are arguably due to convergent evolution or reversals and not to common
descent. In this way, the molecular tree facilitates
an evolutionary interpretation of the morphological
differences among the living forms. We recommended that the extinct groups of lobe-finned fishes
be placed onto the molecular tree that has lungfishes
and not the coelacanth more closely related to tetrapods. The placement of fossils would help to furOffprint requests to: Axel Meyer
ther interpret the sequence of morphological events
and innovations associated with the origin of tetrapods but appears to be problematic because the
quality of fossils is not always high enough, and
differences among paleontologists in the interpretation of the fossils have stood in the way of a consensus opinion for the branching order among lobefinned fishes. Marshall and Schultze (1992) criticized the morphological analysis presented by Meyer and Wilson (1990) and suggest that 13 of the 14
morphological traits that support the sister group
relationship of lungfishes and tetrapods are not
shared derived characters. Here we present further
alternative viewpoints to the ones of Marshall and
Schultze (1992) from the paleontological literature.
We argue that all available information (paleontological, neontological, and molecular data) and rigorous cladistic methodology should be used when
relating fossils and extant taxa in a phylogenetic
framework.
Key words:
Polymerase chain reaction - - 12S
rRNA -- Coelacanth -- L a t i m e r i a c h a l u m n a e -Ray-finned fishes -- Lungfishes -- L e p i d o s i r e n -Protopterus -- N e o c e r a t o d u s -- Conquest of land -Vertebrate phylogeny
Introduction
The origin of tetrapods is a problem of obvious
importance. To understand the evolutionary sequence of morphological events and innovations that
facilitated the conquest of land, it is essential to
determine the phylogenetic relationships among the
living as well as the extinct representatives of the
groups involved. The evolutionary relationships of
lungfishes to tetrapods have been in dispute since
Konstanzer Online-Publikations-System (KOPS)
URL: http://www.ub.uni-konstanz.de/kops/volltexte/2007/3669/
URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-36693
~
A
103
L, Lungfish
T, Tetrapods
C, Coelacanth
~
R, Ray-finned Fish
L, Lungfish
T, Tetrapods
C, Coelacanth
R, Ray-finned Fish
L, Lungfish
C, Coelacanth
T, Tetrapods
R, Ray-finned Fish
Fig. 1. Phylogenetic relationships of the major groups of living
bony fishes (Osteichthyes) to one another. A Tree relating the
ray-finned fishes (R: Actinopterygii) and the three groups of lobefinned fishes (Sarcopterygii): coelacanths (also called Actinistia:
C), lungfishes (Dipnoi: L), and tetrapods (T). In this tree x refers
to the ancestor of all groups and y to the common ancestor of
all lobe-finned fishes. The terminology: zL is the lungfish lineage,
yC is the coelacanth lineage, zT is the tetrapod lineage, and x R
is the ray-finned fish lineage. B A tree that relates tetrapods and
coelacanths more closely and places lungfish as the sister group
of all sarcopterygians in agreement with Schultze (1987). C Tree
relating coelacanths and lungfishes as the sister group of tetrapods. The morphological data presented in Table 2 of Meyer and
Wilson (1990) when analyzed with parsimony P A U P (Swofford
1991) favor tree A (TL = 30, CI = 0.73) over alternatives B (TL
= 43, CI = 0.51) and C (TL = 37, CI = 0.59).
their discovery more than 100 years ago (see Patterson 1980; Rosen et al. 1981 for historical perspectives on the problem); they have variously been
considered to be related to actinistians (i.e., coelacanths), tetrapods, actinopterygians (ray-finned fishes), or crossopterygians (an unnatural group combining coelacanths and rhipidistian fishes) (Fig. 1)
(reviewed by Forey 1988). Since the sensational discovery of the "living fossil," Latimeria chalumnae
(Smith 1939, 1953; reviewed in Thomson 1991) it
has become the favorite closest living relative of
tetrapods. Almost all biology texts (e.g., Romer 1966)
and the majority of publications in the primary literature seem to support coelacanths over lungfishes.
However, there is by no means universal agreement
in the vast literature on this point (e.g., Wahlert
1968; Lovtrup 1977; Wiley 1979; Forey 1980, 1987;
Rosen et al. 1981; Lagios 1982; Fritzsch 1987;
Northcutt 1987; Schultze 1987; reviewed by Forey
1988; Gee 1990). Without certainty about the
branching patterns linking the living groups (much
less the extinct groups), it remained difficult to develop a detailed model of how fishes conquered land.
The Contribution of Molecular Data
Meyer and Wilson (1990) sequenced 664 bp of two
slowly evolving mitochondrial genes (12S rRNA and
cytochrome b) from three species--a ray-finned fish
(Cichlasoma citrinellum), a South American lungfish (Lepidosiren paradoxa), and the coelacanth (L.
chalurnnae)--and compared them to the published
sequences of tetrapod mitochondrial D N A (mtDNA), e.g., for the frog Xenopus laevis (Roe et al.
1985). We concluded, with statistical confidence
(Felsenstein 1985) that the lungfishes and not the
coelacanth are the closest living relative oftetrapods
(Fig. 1A). Because paleontologists had not agreed
on the branching order among the living groups,
much less the extinct taxa, we regarded it as a worthwhile contribution to present a statistically supported branching pattern for the living groups derived from a molecular data set. Only after a best
estimate of branching pattern is achieved can we
hope to retrace character evolution and gain an improved understanding of the conquest of land.
Previous molecular analyses did not settle the
issue because they did not include a lungfish (Maeda
et al. 1984; Hillis and Dixon 1989). Since our initial
study (Meyer and Wilson 1990), additional molecular data have been published (Gorr et al. 1991;
Stock et al. 1991) but could not resolve the branching order because of the inadequacy of the molecule
or the methods of phylogenetic reconstruction chosen (Meyer and Wilson 1991; Sharp et al. 1991;
Stock and Swofford 1991).
N o r m a r k et al. (1991) collected additional
104
mtDNA sequences for a shorter piece ofcytochrome
b [and inferred 97 amino acids (AA)] than was gathered by Meyer and Wilson (1990), for several fishes
and for another lungfish lineage, the African Protopterus spec. They further determined mtDNA sequences coding for up to 159 AA of the mitochondrial cytochrome oxidase I and up to 73 AA for the
mitochondrial cytochrome oxidase II genes in several fishes. The DNA sequences of the coelacanth,
Latimeria, or any lungfish were not determined by
them for these two additional genes, therefore only
their new cytochrome b sequences for Protopterus
are relevant to the issue at hand. Normark et al.'s
(1991) cytochrome b data support the lungfish-tetrapod sister group relationship by a bootstrap value
of 51%. Reasons for Normark et al.'s (1991) lower
level of support than the one previously reported
by Meyer and Wilson (1990) are several. Normark
et al. (1991) sequenced a smaller piece of cytochrome b; they did not sequence the 12S gene and
did not include it into their analysis; they hence had
fewer phylogenetically informative sites than Meyer
and Wilson (1990) to investigate the coelacanthlungfish-tetrapod question. We (Meyer and Wilson
1990) found the 12S rRNA gene to contain a higher
density of phylogenetically informative sites (17 of
240) than the cytochrome b gene (16 of 360) for the
taxa included in our analysis. The number of phylogenetically informative sites will partly depend on
the number of taxa included in the analysis. Normark et al. (1991) included a larger number of taxa
in their analysis than we (Meyer and Wilson 1990)
did. Sanderson and Donoghue (1989) showed that
homoplasy increases, and the consistency index of
data decreases, with the number of taxa in an analysis. This is probably because the probability of
character-state changes increases with the total
number of branches on a tree, given that there are
only four states in DNA sequences. Normark et al.
(1991) presented support for the lungfish-tetrapod
relationship and confirmed our finding that mtDNA
can be used for "deep" phylogenetic questions, not
only for population-level work and relationships
among closely related species.
The molecular support for the lungfish-tetrapod
sister group relationship suggested that their common stem might be longer than had been believed
based on paleontological data. If this finding is correct, it could mean that the common lineage of lungfishes and tetrapods is longer or, alternatively, suggests a closer link between coelacanths and
actinopterygian fossils than previously thought. This
difference between molecular data and current paleontological understanding could be resolved in
several ways but will probably have to await further
discovery or reinterpretation of Devonian sarcopterygian fossils.
Based on our molecular tree, Meyer and Wilson
(1990) addressed the question of whether the character states of morphological traits in the extant taxa
under consideration represent shared derived traits
(synapomorphies) or homoplasies. Meyer and Wilson (1990) concluded that 14 of 22 phylogenetically
informative characters taken from the literature
(compiled by Prof. Peter Forey) also support the
branching pattern based on our molecular results
(Fig. l). In agreement with the parsimony principle
these 14 traits were interpreted to be shared derived
characters of lungfish and tetrapods that arose along
the yz lineage (Fig. 1A), whereas the other eight
characters required more than one character change
to fit the most parsimonious molecular and morphological tree.
Limitations of Molecular Data
Because molecular data can only be collected from
living groups, Meyer and Wilson (1990, p. 363) postulated that the extinct groups ofsarcopterygian fishes should be fit in the phylogeny that links lungfishes
as the closest living relatives of tetrapods to the
exclusion of Latimeria.
Only once a most parsimonious phylogeny of living as well as extinct groups is well established, can
characters be traced back along that cladogram. Because much debate among paleontologists and comparative biologists had not led to a generally accepted phylogeny, we did not feel confident to extend
our molecular phylogeny into the uncharted waters
of extinct fishes. Morphological character states of
the lungfish-tetrapod clade interpreted (by Meyer
and Wilson 1990) to be synapomorphies among the
living lobe-finned fishes may turn out to be homoplasious if the fossils are placed onto the molecular
branching pattern.
The contribution of Marshall and Schultze (1992)
is welcome in this regard. They stress the importance of fossils in the study of the origin of vertebrates and question Meyer and Wilson's (1990) interpretation of 14 traits as synapomorphies between
lungfish and tetrapods, arguing that 13 of these traits
were incorrectly interpreted by us. Before the criticisms of Marshall and Schultze (1992) are addressed, new molecular data and further analyses
for all the lungfish lineages that were not represented
in the first study (Protopterus from Africa and Neoceratodus from Australia) will be presented.
Materials and M e t h o d s
Partial D N A sequences o f the m i t o c h o n d r i a l 12S r R N A gene
were d e t e r m i n e d f r o m the M i d a s cichlid fish [C. citrinellum,
105
Lepidosiren
.... AGG•ACTACTAGCACAAGCTAAAAACTCAAAGGACTTGG•GGTGCCTCAAACCCACCTAGAGGAGCCTGTCCTAAAACCGATAATC
Protopterusl
CGCCAGGAACTACAAGCCCAAGCTTAAAACCCAAAGGACTTGG•GGTGCCTCACACCCACCTAGAGGAGCCTGTTCTAGAACCGATAATC
Protopterus2
CcCCAGGAACTACAAG•CCAGGCTTAAAACCCAAAGGACTTGGCGGTGCCTCACACCCACCTAGAGGAGCCTGTTCTAGAACCGATAATC
Neoceratodus
....................
Frog
CGCCAG-AACTACGAGCCTAAGCTTAAAACcCAAAGGA•TTGGCGGTGCTCCAAACCCACCTAGAGGAGcCTGTTCTGTAATCGATACC•
AGCTTAAAACCCAAAGGACTTGGCGGTGCCTCACACCCACCTAGAGGAGCCTGTTCTAGAACCGATAATC
Latimeria
..... GGAACAACAAGCCACAGCTTAAAACTCAAAGGACTTGGCGGTGCTTCATACCc•C-TAGAGGAGCCTGTTCTAGAACCGATAAAC
Ray-Finned
..... GGAACTACGAGCATTAGCTTAAAACCCAAAGGACTTGGCGGTGCTTTAGACCCCCCTAGAGGAGCCTGTTcTAGAACCGATAATC
Lepidosiren
CACGTTTTACCTAACCACTTTTTGCCAATTCAGCCTATATACCGCCGTCGCCAGCCAACCCCGTGAGAGAAATAAAGTAGGCAAAATTAA
Protopterusl
CACGTTTTACCCAACC-TTCCCTGGCATTTCAGCCTATATAcCGCCGT•GCCAGCCAACCCCCTGAGGCCCACT-AGTTGGCAAAATAGA
Protopterus2
CACGTTTTACC•AACC-TTCCCTGGCATTTCAGCCTATATA•CGCCGTCGCCAG•CAACCCCCTGAGGCCCACT-AGTTGGCAAATAGA
Neoeeratodus
CACGTTAAACCTCACCGCTTCTTGCCACTACCGTCTATATACCACCGTCGCCAGCTTACCC•GTGAGGGTGAAAAAGTAAGCACAATTGG
Frog
CT•GCTAAACCTCACCACTTCTTGCCAAACCCGCCTATATACCACCGTCGCCAGCCCACCTCGTGAGAGATTCTTAGTAGGCTTAATGAT
Latimeria
Ray-Finned
CCCGATCAACCTCAACCACACTTGCTATTTCAGCCTATATACCGCCGTCGCCAGCCCACCCTGTGAAGGAAATACAATGGGCAAAAATAA
CCCGTTAAACCTCACCCTCCCTTGTCATCCCCGcCTATATACCGCCGTCGTCAGCTTACCCTGTGAAGGCACTATAGTAAGCAAAACTAG
Lepidosiren
CTTAGTTAAATACGTCAGGTCGAGGTGTAGCATATGAAGTGGGAAGAGATGGGCTACATTTTCTTG
Protopterusl
Protopterus2
TAGCATCTAACACGTCAGGTCGAGGTGTAGCACATGGGAGGG/k~-GAAATGGGCTACAT2E'2CT/~a~
....
TAACATCTAACACGTCAGGTCGAGGTGTAGCACATGAGAAGGAA-GAAATGGGCTACATTTTCTGA
.... TAGAA-CACGGA
Neoceratodus
Frog
TACAAC•AAAAA•GTCA•GT•GAGGTGTAGCGAATGAAGCGGAAA•AAAT•GGCTACATTTTCTA--TAAAAGAATA•TACGAAA•AC•C
Latimeria
Ray-Finned
AAAAATTAAAAACGTCAGGTCGAGGTGTAGCAAATGAGATGGGAAGAAATGGGCTACATTTTCTAAATA•TAGAATATTACGAAAAAA•A
TAAAACC•AAAA•G•CAGGTCGAGGT•TA•CATATGA•AGGGAAAGAAATG•GCTACATTCCcTA•-CTTTAG•GAA-CACG•ATAAT••
Lepidosiren
AAATGAAA-- TAAAGCTG-GAAGGT--
Protopterusl
Protopterus2
CCATGAAA- -TTGGGGTTTTAAGCTG GATTTAGTAGTAAGAAAA .................
.............................................................
Neoceratodus
TAATGAAA- - -CAAAGTCAGAAGGTGGATTTAGCAGTAAGAAAAAC
Frog
CTATGAAACCAGAT
Latimeria
CAGC GAAA -C CTGTAC TTTGAAGGAG GATTTAGCAGTAAAAGGG GAATAGAGAGC C C C TCT
TAAT GAAAA- -GTACATTAGAAGGAG GATTTAGCAGTAAG CAAAA-ATAGAGGCA ......
Ray-Finned
.... TAGAATA-TACGAATAGC-T
TAGAAAA-CACGGAC-~k-CC
......
~TTCATCAA~AcGT~AGGTCAA~GT~TAGCATATGAAGTGGGAAGAAATGGGCTACATTTTCTATACCTTAGAATA~AACGAAAGATCT
GTTTAGAAGAAA ......................
CAAGAATATTTTTCT
C G A G A A A A G G CG G A T T T A G C A G T A A A G A G A A A C A A G A G A G T T C
CTCT
Fig. 2. Sequences of parts of the mitochondrial 12S rRNA gene.
The sequences shown correspond about to positions 2547-2875
for 12S rRNA in the frog mtDNA sequence (Roe et al. 1985).
Dashes indicate proposed indels (and at the end missing data).
Protopterus 1 is P. annectens and 2 is P. aethiopicus. At the
positions at which deletions or additions occurred, alternative
alignments cannot be ruled out completely. The asterisk at position 245 indicates the position up to which data were included
in the phylogenetic analysis; to the right of it the alignment is
less certain and data were excluded from the analysis.
Regrettably, three nucleotides of the cytochrome b sequence
from the ray-finned fish Cichlasoma citrinellum were reported
incorrectly in Fig. 2 of Meyer and Wilson (1990); they appear
correctly in Meyer et al. (1990). These nucleotides are 1, 4, and
28; instead of C, A, C they should be A, C, G. Corrected nucleotides 1 and 4 remain phylogenetically uninformative, the correction of nucleotide 28 however adds another (the 17th in cytochrome b) phylogenetically informative site in support of the
lungfish-tetrapod sister group relationship.
Actinopterygii (ray-finned fish)], the three lineages of lungfishes
[L. paradoxa (South America), Protopterus annectens and Protopterus aethiopicus (Africa)] and the coelacanth (L. chalumnae).
The partial 12S sequence of the Australian lungfish Neoceratodus
forsteri was kindly provided by B. Hedges and L. Maxson. DNA
was extracted, amplified via the polymerase chain reaction (PCR),
and directly sequenced as described by Kocher et al. (1989) and
Meyer et al. (1990) using tissues from frozen specimens. The data
from some of these species had been previously published (Meyer
and Wilson 1990); for this study the sequences for two additional
genera of lungfishes (Protopterus and Neoceratodus) were added.
A shorter 12S sequence of Neoceratodus than reported here had
been determined from a formalin-fixed specimen by A.M.
Results and Discussion
Amplifications and Direct Sequencing. Amplifications were
done in 25/xl of Tris (67 mM, pH 8.8) containing 2 mM MgClz,
1 mM of each dNTP, 1/zM of each primer, template DNA (101000 ng), and Taq polymerase (1.25 units, Perkin-Elmer-Cetus).
The primer sequences for the PCR amplification of the partial
12S rRNA are L1091 (5'-AAAAAGCTTCAAACTGGGATTAGATACCCCACTAT-3') and H1478 (5'-TGACTGCAGAGGGTGACGGGCGGTGTGT-3')(Kocher et al. 1989). L
and H refer to the light and heavy strands, respectively, and the
numbers refer to the 3' position of the primers in human mtDNA
(Anderson et al. 1981).
F i g u r e 2 shows the a l i g n e d D N A o f the p a r t i a l (up
to 325 bp) 12S r R N A s e q u e n c e s . T h e a l i g n m e n t o f
the r R N A s e q u e n c e s is o b v i o u s for the first 245 b a s e
p o s i t i o n s . A f t e r t h a t the a l i g n m e n t b e c o m e s a m b i g u o u s so t h a t p h y l o g e n e t i c a n a l y s i s was c o n f i n e d to
the first 245 bp. I n this s e g m e n t , p o s i t i o n s v a r i e d
b y base s u b s t i t u t i o n s a n d single b a s e - p a i r a d d i t i o n s
or d e l e t i o n s . T h i r t y - f i v e v a r i a b l e p o s i t i o n s i n v o l v e d
o n l y t r a n s i t i o n a l changes, 52 also t r a n s v e r s i o n s , a n d
6 a d d i t i o n s or d e l e t i o n s were f o u n d .
A m o n g closely r e l a t e d species t r a n s i t i o n s u s u a l l y
o c c u r m o r e f r e q u e n t l y t h a n t r a n s v e r s i o n s . T h e r e were
six differences o b s e r v e d b e t w e e n the t w o A f r i c a n
lungfish species, o n l y o n e o f w h i c h was a t r a n s v e r sion. T h e r e were 40 differences b e t w e e n the S o u t h
A m e r i c a n a n d the A f r i c a n lungfishes, 19 o f w h i c h
were t r a n s v e r s i o n s . T h e f r e q u e n c y o f t r a n s v e r s i o n s
was h i g h e r t h a n t h a t o f t r a n s i t i o n s w h e n lungfishes
were c o m p a r e d to the o t h e r groups, u n d e r s c o r i n g
the a n t i q u i t y o f the split b e t w e e n the lineages. I n
106
Lepidosiren
Protopterus
annectens
Protopterus
aethiopicus
Neoceratodus
Frog
Latimeria
Ray-Finned Fish
Fig. 3. Phylogenetictree using transversion-parsimonyanalysis PAUP (version 3.0r/31, Swofford1991). Only the first 245bp aligned positions of 12S rRNA were used in the analysis; the
ray-finnedfish was declaredthe outgroup. One shortest tree (TL
= 73 steps) was found. Numbers indicate percentagebootstrap
values of 1000 replicates(Felsenstein 1985).
accordance with other studies (e.g., Miyamoto and
Boyle 1989; Mindell and Honeycutt 1990), attention in the phylogenetic analysis was hence confined
to transversions.
A parsimony analysis (exhaustive search option)
(Swofford 1991) with the ray-finned fish as an outgroup was conducted. A single shortest tree was
found (tree length = 73, Fig. 3). This tree links the
three lepidosirenid lungfishes Protopterus and Lepidosiren; their sister group is the Australian lungfish
Neoceratodus. The lungfish clade is the sister group
to the frog. The coelacanth was found to be the sister
group to the lungfish-tetrapod clade. The inclusion
of three species of two other lineages o f lungfishes
does not seem to contradict the earlier result of Meyer and Wilson (1990): lungfishes and not the coelacanth seem to be the closest living relatives of
tetrapods.
The additional 12S rRNA sequences support the
finding, based on morphology (e.g., Marshall 1987;
Schultze and Campbell 1987), that the two groups
of lepidosirenid lungfishes (Protopterus and Lepidosiren) are more closely related to each other than
to the Australian lungfish, Neoceratodus (Fig. 3).
Furthermore, the new data seem to confirm (albeit
weakly) that the lungfishes are a monophyletic group
and are more closely related to the frog than is the
coelacanth.
A bootstrap analysis (Felsenstein 1985) with 1000
replications using PA UP version 3.0r/31 (Swofford
1991) favored the above-mentioned findings: (1) the
joining of the lepidosirenid lungfishes and (2) the
sister group relationship of the frog and lungfishes
to the exclusion o f the coelacanth (Fig. 3). However,
the bootstrap values did not reach very high levels
(Felsenstein 1985), making these results somewhat
tentative. Obviously, the addition o f two species o f
lungfish, while maintaining the overall pattern of
relations, lowered the consistency index, as expected
(Sanderson and Donoghue 1989).
More data will be needed to further test this phylogenetic hypothesis. We have already sequenced
portions of the nuclear 18S rRNA gene and several
mitochondrial genes in some of these taxa (A. Meyer, unpublished). It is not clear whether these genes
will be able to resolve the pattern of relations, however. The 18S gene evolves too slowly and sporadically (see also Stock et al. 1991), and the 16S rRNA
gene evolves too rapidly and shows much length
variation that hinders optimal alignments. Longer
portions o f the mitochondrial 12S rRNA gene and
sequences of nuclear and mitochondrial proteincoding genes seem to hold the greatest promise for
providing data that might resolve this question with
statistical confidence.
Extinct and Extant Lobe-Finned Fishes and the
Origin of Tetrapods
The fossil record of actinopterygians and all groups
of sarcopterygians goes back to a narrow window
(about 30 million years wide) that dates back approximately 400 million years (Benton 1990). The
virtually simultaneous occurrence of all the groups
involved makes the question of the origin of tetrapods hard to answer based on the sequence o f appearance of fossils.
Without a doubt, fossils are important in the reconstruction of phylogenies; this point has been made
in an apt manner before (Gauthier et al. 1988; Donoghue et al. 1989). Fossils should even help to resolve the relationships among living taxa, because
some fossils are likely to possess combinations of
characters that are not present in any living taxon.
Nevertheless, there is no denying that fossils have
shortcomings as well. Most of the fossils are only
available from incomplete skeletons and do not carry the same amount of information that can be found
in recent forms. More characters for a cladistic analysis will always be found in recent taxa than in extinct ones. This fact (and others) has led to the observation that, in practice, fossils have not made
much difference in the determination o f branching
patterns (Patterson 1981).
Often the fossil record is difficult to interpret, and
the same features can be viewed differently by different researchers. Miles ( 1977) and Schultze (1987)
both place much importance on the presence or ab-
107
sence of cosmine and intracranial joints in sarcopterygian fishes, yet they reach radically different conclusions about the relationships of the groups under
consideration and the evolution of these two characters. The history oftetrapods is particularly onerous because, as Miles (1977, p. 315) states, "no
phylogeny of vertebrates can be perfectly parsimonious because of parallel and convergent evolution."
A well-corroborated phylogeny is needed first in order to ask whether the morphological traits that were
interpreted as synapomorphies of the lungfish-tetrapod clade really are synapomorphies or whether
they represent homoptasies.
Marshall and Schultze's Critique of Meyer and
Wilson (1990)
Critique on Molecular Data and Analysis
Marshall and Schultze (1992) reanalyzed the molecular data and confirmed that our data are robust
and do not contain any obvious biases and that the
phylogenetic analysis is sound. They confirm that
the most parsimonious interpretation of the molecular data links lungfishes with tetrapods as sister
groups to the exclusion of the coelacanth. Our data
have also been reanalyzed and our conclusions confirmed by Forey (1991 a). Also, the new sequences
do not seem to contain any obvious base compositional biases that might influence phylogenetic inference (analysis not shown).
Critique on Morphological Analysis
and Inferences
Based on the branching pattern established by our
molecular data, we mapped morphological traits
onto our phylogenetic tree (Meyer and Wilson 1990).
Some of these traits were from soft tissues that are
not preserved in fossils. The data were restricted to
phylogenetically informative morphological traits
present in all four extant groups under consideration, as they were the only ones that were available
for molecular analysis.
Following the parsimony principle, Meyer and
Wilson (1990) interpreted the inferred sharing of
derived morphological characters in lungfish and
tetrapods to mean that they are synapomorphies
that arose (or were lost) along the yz lineage (Fig.
1A). Of course, once the fossils are placed onto the
tree, increasing the number of branches on the tree,
this interpretation might have to be modified according to fossil evidence. Unfortunately, no consensus on the position of fossil groups that would
have allowed us to securely incorporate paleontological data could be extracted from the paleontological literature. As indicated in our paper, the data
in Table 2, compiled by P. Forey, were based exclusively on published sources (Rosen et al. 1981;
Lagios 1982; Forey 1987; Fritzsch 1987; Northcutt
1987; Schultze 1987). It is therefore not entirely
obvious why Marshall and Schultze (1992) question
us rather than the other eight authors from the primary literature upon which the data in Table 2 of
Meyer and Wilson (1990) are based. Before we address Marshall and Schultze's (1992) critique on
Meyer and Wilson's (1990) interpretation of the
morphology of the living lobe-finned fishes trait by
trait, we will first discuss a few general points.
The Identity of the Closet Living Relative
of Tetrapods
Marshall and Schultze (1992) believe that Meyer
and Wilson (1990) were misled in their interpretation of the morphology of living groups because they
did not include fossils in their analysis. Several paleontologists believe that lungfish are the sister group
of tetrapods even when fossils are considered (e.g.,
Rosen et al. 1981; Forey 1980; Gardiner 1980, 1984).
Meyer and Wilson (1990) were aware that other
paleontologists are skeptical of these results (e.g.,
Holmes 1985; Schultze 1987). This debate is still
continuing among paleontologists.
Marshall and Schultze (1992, Fig. 3) present a
branching pattern of living forms and fossils that is
consistent with our molecular finding by placing
lungfishes and not the coelacanth as the living sister
group of tetrapods. Their favored phylogenies (Fig.
2C and D and Fig. 3 in their paper), however, include various fossil groups that separate lungfishes
and tetrapods. They use this phylogeny (their Fig.
2C; Fig. 4B here) in their assertion that 13 out of
14 characters most parsimoniously interpreted to
be synapomorphies by Meyer and Wilson (1990)
are homoplasious. Obviously, for their interpretation to be correct (1) the branching pattern must be
correct, and (2) the interpretation of character states
in the fossils must be unambiguous. Neither seems
to be established with a high degree of certainty.
Moreover, the interpretation of character evolution
of some morphological traits is quite different if
their Fig. 2D is used as the "correct" phylogeny
rather than their Fig. 2C.
Schultze has not always believed that the branching patterns presented in Fig. 2C and D or Fig. 3 of
Marshall and Schultze (1992) or as Fig. 4B in this
paper are the most parsimonious ones. Schultze
(1987, p. 39) states"The dipnoans are not the closest
sister group of tetrapods, independently if living
forms only are considered, or fossil forms included."
He maintains that a sister group relationship of dipnoans and tetrapods is "cladistically inappropriate"
and that "the hypothetical common ancestor of the
tetrapods and dipnoans is the common ancestor of
all sarcopterygians" (p. 71). Figure 4A shows the
branching pattern favored by Schultze in 1987. He
108
Tetrapods
links the coelacanths and the extinct porolepiforms
and osteolepiforms into the group crossopterygians
and places lungfishes at the base of all lobe-finned
fishes. Schultze's analysis includes morphological
characters from fossils but excludes soft tissue characters because "they are unavailable in fossils" (p.
69).
To the best of our knowledge, Schultze has not
changed his interpretation of the fossil record, which
he believes to indicate that coelacanths and not
lungfish are the closest living relatives of tetrapods.
Still, in 1991 (p. 191) Schultze states "actinistians
can be considered as the closest living relatives of
tetrapods." A critique by Schultze of Meyer and Wilson's (1990) character-state assignments with Latimeria as the closest living relative oftetrapods would
have seemed to be more in line with Schultze's opinions. It would be illuminating to know explicitly
which new data or which change in interpretation
of old data led Schultze to such a drastic change of
opinion regarding the relationships of sarcopterygians.
The Phylogenetic Position of Diabolepis and
Other Crucial Fossils
Diabolepis speratus (Chang and Yu 1984) is a species of fossil fish described on the "oasis of two fairly
complete skulls, five anterior cranial portions, flagmentary tooth plates, and lower jaw rami. It is heralded to be one of the earliest lungfish fossils found
so far. Marshall and Schultze (1992) place Diabolepis as the sister group to lungfishes in their Fig.
2C. Much of their argument about the homology of
morphological traits depends on the reliability of
the sister group relationship and character diagnosis
of Diabolepis and lungfishes.
Some uncertainty surrounds the phylogenefic position of this species with respect to fossils and recent
lungfishes: Schultze and Campbell (1987) state that
the fossil is not well enough known "for definite
statements about its relationship" (p. 25) and that
(p. 37) all characters listed for Diabolepis as common with dipnoans "are questionable." Furthermore, they say "the evidence for considering Diabolepis as the sistergroup of the Dipnoi is weak" (p.
36). Campbell and Barwick (1987) suggest that
"'Diabolepis should be regarded as a modified 'crossopterygian' that shares no unique derived characters with primitive dipnoans" (p. 128). In 1991
Schultze (p. 191) still seems to be doubtful of the
close relationship of Diabolepis with lungfishes but
favors rather the view that Diabolepis more closely
resembles primitive porolepiforms (like Youngolepis) than lungfishes.
Diabolepis and lungfishes are believed to be sister
groups by Forey (1987). However, Forey differs in
some aspects of the interpretation on the signifi-
Osteolepiforms
Porolepiforms
Coelacanth
Lungfish
Ray-finned Fish
Tetrapods
Osteolepiforms
Porolepiforms
Diabo/epis
Lungfish
Coelacanth
Ray-finned Fish
,/,.0 T, Tetrapods
•
?
Osteolepiforms
• Porolepiforms
o
.ay-f,°oed Fish
Fig. 4. A Phylogenetic relationship among the major sarcopterygian taxa according to Schultze (1987). Extinct groups under
consideration (osteolepiforms, poroleptiforms, and Diabolepis)
are marked with filled dots. B Phylogenetic relationship among
the major sarcopterygian taxa according to Maisey (1986) used
by Marshall and Schultze (1992) to argue that most of the 14
characters that have been interpreted to by synapomorphies of
the lungfish-tetrapod clade (by Meyer and Wilson 1990) to be
convergences. Stippled line indicates Maisey's (1986) suggestion
(p. 232 and Fig. 12) that porolepiforms are not a monophyletic
group and whose placement is uncertain in relation to Diabolepis
and lungfishes. Note the inconsistencies between A and B. The
position of the coelacanth is highlighted. C Diagram pointing out
that the placement of the fossil groups is not certain.
cance of Diabolepis from Schultze (1987); he points
out that it also shows an external nasal opening very
close to the jaw and no external intracranial joint,
which are features that Schultze and Arsenault (1985)
109
use to link Panderichthys (an osteolepiform fish) and
primitive amphibians. Forey (1987, p. 84) states
"To this point in the essay I have discussed the
introduction of lungfish fossils that have extended
our knowledge of morphological variation to the
extent that they have embraced characters formerly
thought to be restricted to 'rhipidistians' and tetrapods. These fossils have therefore reduced the effectiveness of the traditional argument that places
'rhipidistians' as the sistergroup of tetrapods."
However, this placement also means that some of
the characters listed by Rosen et al. (1981) " m a y
have arisen independently" (Forey 1987, p. 89).
Maisey (1986) reviewed Diabolepis and several other fossils believed to be primitive dipnoans and suggested that several of these fossil forms are more
closely related to extant dipnoans than Diabolepis
and suggests that (p. 232) " 'porolepiforms' are really a paraphylefic group of 'stem dipnoans.' " It
appears that the discussion regarding the phylogenetic position of Diabolepis is not settled.
The exact assignment of fossils to lineages of early
lungfishes is problematic (Miles 1977; Campbell and
Barwick 1987; Marshall 1987). Marshall (1987)
conducted the first cladistic study of this group that
included a data matrix and used numerical analyses
of traits. The results of his analysis differed in several
respects from earlier studies. He brought his and the
earlier results into agreement by abandoning some
of the features of his cladogram, because he states
"for theoretical and practical reasons, parsimony is
not a good criterion for choosing between the possible phylogenies of dipnoans" (Marshall 1987, p.
151).
In the last 10-15 years a few new important fossils
have been discovered that triggered some paleontologists to change their view about both the relationships of the living groups and also the question
of the monophyly of the extinct group Rhipidistia.
It is now widely believed that the Rhipidistia (formerly believed to be composed of porolepiforms
and osteolepiforms) is not a natural group (Maisey
1986; Forey 1987). Porolepiforms and osteolepiforms had been closely allied to tetrapods by some
paleontologists (e.g., Schultze 1987). Other paleontologists have linked lungfishes as the sister group
to tetrapods to the exclusion of all extinct groups
(e.g., Forey 1980, 1987; Gardiner 1980; Rosen et
al. 1981). Panderichthys and Elpistostege are believed by some paleontologists to be intermediate
between osteolepiform rhipidistians and tetrapods
(e.g., Vorobjeva 1980; Schultze and Arsenault 1985).
Without the firmly established relationships of
fossil and living forms it seems premature to try to
trace single characters, because the interpretation of
traits as synapomorphies (shared derived) or symplesiomorphies (shared primitive) or homoplasies
depends on the branching pattern and has to be
made post priori (though decisions about what character states are homologous are made a priori by
character analysis). The decision about whether a
particular character is due to recent common ancestry or convergent evolution can only be made in
a phylogenetic context that takes all available information into consideration.
Trait-by-Trait Response to Marshall and
Schultze's Critique
Quite often, a different interpretation and many
opinions regarding the same fossil can be found in
paleontology. Whether Marshall and Schultze (1992)
or other paleontologists are correct in their reading
of the fossils has to be decided in paleontological
circles. We will simply furnish citations to provide
alternative interpretations of character states. In the
end we remain unconvinced that the placement of
the fossils by Marshall and Schultze (1992) is strongly supported by a numerical cladistic analysis (Fig.
4C).
When Marshall and Schultze's (1992) revised
character states and fossil taxa are included in a
numerical cladistic analysis, the phylogeny that they
claim they believe in (Fig. 4B) is not attained (data
not shown). One has to assume that they have unlisted characters that support the branching pattern
of Fig. 4B and outweigh the characters listed by
Meyer and Wilson (1990) or their own reassigned
character states for traits 1-14.
Trait 1: Internal Nostrils. Marshall and Schultze
(1992) suggest that the internal nares evolved independently in lungfishes and tetrapods, because
these structures are not homologous and not present
in the putative sister group of lungfishes, Diabolepis.
The homology of these structures between lungfishes
and tetrapods had been questioned earlier by Holmes
(1985) and Schultze (1987). Yet, Forey (1980, 1987)
agrees with the other three authors of Rosen et al.
( 1981) in their interpretation of the internal nostrils
as homologous structures (and synapomorphies) of
lungfishes and tetrapods. Forey (1987) discusses the
evidence for interpretation of the choanae of lungfishes and Diabolepis and tetrapods as a synapomorphy or parallelism and tentatively concludes (p.
87) "At present therefore I accept the choana as an
homology specifying a group lungfishes and tetrapods."
Trait 2: Palate Fused with Neurocranium. Whether or not the fusion of the palate to the neurocranium
is a synapomorphy of lungfish and tetrapods or a
convergently evolved trait in these clades depends
on the assumed phylogeny of the groups. Marshall
and Schultze (1992) imply that the phylogeny (Fig.
110
4B) is generally accepted. But Forey (1987), using a
larger data set (he includes evidence from living
taxa) than Schultze (1987), arrives at a drastically
different phylogeny: one that links lungfish as sister
group of tetrapods to the exclusion of all osteolepiforms.
when branching orders are to be established. Once
a most parsimonious solution is found, a posteriori
character evolution can be traced back and a priori
assumptions about character states being synapomorphies of a clade potentially can be revised based
on this branching pattern.
Trait 3: Glottis. Marshall and Schultze (1992) assert that we only considered the morphology of a
cichlid fish, as a representative of ray-finned fishes,
when we assigned a character state to this trait. That
is not so; the vast majority of all ray-finned fishes
do not possess lungs. Primitive actinopterygian fishes like Amia, Lepisosteus, and Polypterus do posses
lungs, whose homology to lungs oftetrapods is questionable. If lungs are present, some mechanism of
closing the trachea must also be present. However,
that mechanism and its ontogenetic derivation
(therefore homology) can be quite different in different groups of organisms [e.g., the reptile and amphibian condition is quite distinct (Carl Gans, personal communication)]. Most importantly, the glottis
and epiglottis are likely to be cartilaginous structures
that would not be preserved in the fossil record and
therefore would not be accessible to paleontological
investigation. The repeated argument of Marshall
and Schultze (1992) pointing out supposed differences between the two living groups of lungfish
(Neoceratodus and lepidosirenid lungfishes) simply
implies that in one of these lineages a trait has been
lost or gained but does not detract from the fact that
lungfishes are considered to be a natural group (Marshall 1987; Schultze and Campbell 1987).
Trait 5: Autopalatine Bone. Marshall and Schultze
(1992) believe that this particular bone must have
been present in Diabolepis, the putative sister group
of lungfish, although that actual bone has not been
found in the fossil (Chang and Yu 1984).
Trait 4: Pharyngobranchial Gill Arch Elements. The
pharyngobranchial gill arch elements are a set of
bones that hold the gills necessary for aquatic respiration. These elements are reduced in recent lungfish and tetrapods. Most currently, Coates and Clark
(1991) revised the ideas about the evolution of this
trait (and trait 7, the hyomandibular bone) based
on the fossil Acanthostega. Acanthostega, one of the
earliest tetrapods known, has a full set of these elements. Its gill arch elements "resemble those of a
Devonian lungfish such as Chirodipterus rather than
the proximally narrow, bipartite ceratohyal and large
hyomandibular bone of the Devonian osteolepiform
Eusthenopteron'" (Coates and Clark 1991, p. 234).
These findings seem to support a sister group relationship between lungfishes and tetrapods to the exclusion of the osteolepiform Eusthenopteron (contra
Schultze 1987). Such a phylogeny is not in concordance with the phylogeny shown in Fig. 2C and D
of Marshall and Schultze (1992). Coates and Clark
(1991) present other traits that link osteolepiforms
with tetrapods to the exclusion of lungfishes, demonstrating that all evidence needs to be considered
Trait 6: Depressor Mandibulae Muscle. This is a
soft-tissue trait that is not available for study in
fossil fishes. Bemis (1987) reports on the presence
or absence of several muscles in the two groups of
living lungfishes. It is not obvious how the presence
of these muscles should make lepidosirenid lungfishes more terrestrially adapted than Neoceratodus.
Of course, we are aware that lepidosirenid lungfishes
are more adapted for terrestrial life than Neoceratodus but for reasons other than the depressor mandibulae. It is not apparant how the arguments of
Marshall and Schultze (1992) that recent lungfish
lineages differ in their adaptations to terrestrial life
(given that both authors have argued that lungfishes
are monophyletic) should have any bearing on the
disputed interpretation of the cladistic status of the
morphological traits.
Trait 7: Free Hyomandibular Bone. Marshall and
Schultze's (1992) interpretation seems to differ from
the findings of Coates and Clark (1991) and Rosen
et al. (1981) (see above: trait 4).
Trait 8: Ethmoid Commissure Sensory Canal The
argument of Marshall and Schultze (1992)--repeated loss of his structure--would seem correct if the
phylogeny (Fig. 4B) was shown to be supported by
a thorough cladistic analysis.
Traits 9-12: Saccus Vasculosus of Pituitary Gland,
Pars Nervosa of Pituitary Gland, Truncus Arteriosus
of Heart, and Divided Auricle of Heart. None of these
traits are preserved in fossils, and therefore character states of these traits in fossils are not available
for study. Marshall and Schultze (1992) state that
trait 10 is absent in Neoceratodus. It is not clear
whether Lagios (1982) investigated the presence or
absence of trait 10 in Neoceratodus. Burggren and
Johansen (1987) studied traits 11 and 12 in detail,
and small differences in the degree to which these
traits are expressed exist. Lacking evidence to the
contrary from fossils, all four traits currently have
to be interpreted as synapomorphies of a lungfishtetrapod sister group.
111
Traits 13 and 14: Limbs with More Than Four
Mesomeres and Pelvic Girdles Joined. Marshall and
Schultze (1992) dismiss these two traits as having
evolved in parallel rather than being shared derived
characters. Their argument again depends on the
assumed phylogeny. Rosen et al. (1981) and Forey
(1987) presented the morphological data on these
two traits; their interpretation views them as synapomorphies of lungfish and tetrapods.
Further, Marshall and Schultze (1992) state that
"The fossil record indicates that air breathing, and
its associated physiological adaptations, arose independently in lungfish and tetrapods." They seem
to include terrestrial locomotion as an associated
physiological adaptation of air breathing because
they dismiss traits 13 and 14 as parallelism based
on their claim that air breathing "and its associated
physiological adaptations" evolved independently
in lungfish and tetrapods. It would seem that air
breathing and terrestrial locomotion are two independent sets of characters. An air-breathing fish does
not need to be adapted for terrestrial locomotion.
A fish adapted for terrestrial locomotion does not
have to be adapted for air breathing. A priori there
is no reason to think that these character complexes
should be coupled in any way.
Toward a Resolution of the Origin of Tetrapods
The contribution of molecular data is its ability to
identify the lungfishes and not the coelacanths as
the living sister group oftetrapods, a result that had
not been agreed upon based on morphological data,
although it had been suggested before (reviewed in
Forey 1988). This conclusion (the branching pattern
of Meyer and Wilson 1990) based on molecules may
lead to an increased understanding of the morphological traits that might have preadapted the common ancestor of lungfishes and tetrapods to life on
land. Obviously, the estimate of the common morphology of the ancestor of tetrapods would have
looked differently had we found that the coelacanth
and not the lungfishes is the closest living relative
of the tetrapods. Once the attachment of the extinct
lobe-finned fishes in a tree (Fig. 1A or 4C) on lineages (e.g., zT, zL, xy, or yz) is achieved, morphological changes can be ordered and reinterpreted
cladistically. Knowledge of the sequence of morphological changes involved in the colonization of
land might in this way be refined.
Character modification along a lineage may happen in all clades. To have legs is one of the character
states that defines tetrapods. However, to have lost
legs secondarily (as in snakes) does not make a difference in the assignment of snakes as tetrapods, as
there are several other characteristics of snakes that
clearly make them tetrapods. Similarly, even highly
modified extant lungfishes can still be placed in the
lungfish-tetrapod clade based on several morphological traits shared by all members of that clade.
We argue that such traits are not always present in
the fossils in question and that interpretations of
fossils are contradictory. Hence support for particular trees based on paleontological data alone tend
to be weak. Only paleontologists can decide, however, what common ancestors might have looked
like, based on the interpretation of sister group relationships of living and extinct forms.
Traits will sort out in a way such that at the
branch points ofa phylogenetic tree the sister groups
will end up with different assortments of novelties
and plesiomorphic traits. Tetrapods did not pass
through a point of "lungfishness." Both the lungfishes and tetrapod sister groups are derived from
some sort of sarcopterygian fish with some traits
that both lungfishes and tetrapods still have. This
sorting out of traits along the yz (Fig. 1A) lineage
will occur all along that common lungfish-tetrapod
lineage. If extinct groups did branch off this lineage,
they will have some features that are characteristic
of later lungfish and others shared with later tetrapods.
Phylogenetic trees are only statistical statements
about genealogical relationships that are hopefully
based on as many homologous characters as possible
analyzed in a fashion that is logically acceptable
(e.g., Cloutier 1991; Forey 1991b). Because characters can evolve repeatedly and be lost along a
particular lineage, not all character states present at
the terminal taxa of a tree will represent true synapomorphies but may be homoplasies. It is in this
light that Meyer and Wilson (1990) made an assessment about the potential morphology of the
common ancestor of lungfish and tetrapods, an ancestor they did not share with the coelacanth.
Some paleontologists explicitly ignore neontological data. Schultze's (1987) reasoning for this
practice is that many traits present (e.g., soft traits
that do not fossilize) in the living representatives
are not found in the fossils or are likely to be modified from the ancestral lineages. This paleontological perspective seems to throw out most of the information that is obtainable. It seems unfounded to
categorically ignore neontological data; all data-paleontological, neontological, and molecular-should be collected and analyzed in a manner that
is open to criticism and checks.
A common practice in the paleontological literature is to simply list shared traits (presumed to be
synapomorphies) on cladograms and neglect to conduct numerical analyses. Of course, only a rigorous
numerical cladistic analysis including an outgroup
(and explicit presentation of data in a matrix) will
be able to identify which of the shared traits are
112
derived. To establish a phylogeny for the extant
lobe-finned fishes and fossils will require a careful
cladistic analysis of all relevant taxa. This cladistic
analysis should use all available forms of data including characters (soft anatomy and molecules)
from the living members of lobe-finned fishes. It
would seem an inappropriate practice to present arguments about relationships on a trait-by-trait basis
without a final cladistic analysis that reveals synapomorphies and homoplasies. Without the presentation of a data matrix, a discussion is futile. Phylogenetic statements without a clear presentation of
character states for all taxa under consideration are
subjective statements that cannot be criticized and
discussed.
Previously, Panchen and Smithson (1987) discussed many of the points made here. (1) Cladistic
methodology is necessary for a solution to this problem. (2) Fossils will rarely be able to overturn a
phylogeny that includes many more traits from living groups. (3) The interpretation of similarities in
structures, if homologous, will have to be interpreted either as synapomorphic or as homoplasious;
this decision depends on the most parsimonious
phylogenetic estimate. (4) Many of the conclusions
(about character state evolution but probably not
branching order) depend on the study of early lungfish fossils, and whether or not these fossils can be
assigned with any certainty to belong to the lungfish
lineage or to other lineages.
We agree with Schultze (1991, p. 111) who states,
"we will see a continuing discussion over the relationships of sarcopterygians in the future."
Acknowledgments. For discussion and comments we thank Mike
Bell, Max Hecht, Ellen Prager, David Reznick, Robert Sokal,
David Wake, and Paul Wilson. Our manuscript was also critically
read and improved by Charles Marshall and Hans-Peter Schultze.
Allan C. Wilson provided support (through NSF and NIH), advice, inspiration, and insights during the initial part of this work.
Tissues were kindly provided by David Reznick (Lepidosiren),
Robert Murphy (Latimeria), and William Bemis and Richard
Sage (Protopterus). Blair Hedges and Linda Maxson determined
the sequence for Neoceratodus reported here and kindly provided
it for this paper. This work received partial support (to A.M.)
from the Systematic Biology Section of NSF (BSR-9107838 and
BSR-9119867). The sequences reported have been deposited in
GenBank (accession nos. M87532-M87537).
References
Anderson S, Bankier AT, Barrell BG, de Bruijn MHL, Coulson
AR, Drouin J, Eperon IC, Nierlich DP, Roe BA, Sanger F,
Schreier PH, Smith AJH, Staden R, Young IG (1981) Sequence and organization of the human mitochondrial gehome. Nature 290:457-465
Bemis WE (1987) Feeding systems of living dipnoi: anatomy
and function. J Morphol Suppl 1:249-275
Benton MJ (1990) Phylogeny of the major tetrapod groups:
morphological data and divergence dates. J Mol Evo130:409424
Burggren WW, Johansen K (1987) Circulation and respiration
in lungfishes (Dipnoi). J Morphol Suppl 1:217-236
Campbell KSW, Barwick RE (1987) Paleozoic lungfishes--a
review. J Morphol Suppl 1:93-131
Chang MM, Yu X (1984) Structure and phylogenetic significance of Diabolichthys speratus gen. et sp. nov., a new dipnoan-like form from the Lower Devonian of eastern Yunnan,
China. Proc Linn Soc NSW 107:171-184
CloutierR (1991) Patterns, trends, and rates ofevolution within
the Actinistia. In: Musick JA, Bruton MN, Balon EK (eds)
The biology of Latimeria chalumnae and evolution of coelacanths. Kluwer, Hingham MA, pp 23-58
Coates MI, Clark JA (1991) Fish-like gills and breathing in the
earliest known tetrapod. Nature 352:234-236
Donoghue MJ, Doyle JA, Gauthier J, Kluge AG, Rowe T (1989)
The importance of fossils in phylogenetic reconstruction. Annu
Rev Ecol Syst 20:431-460
Felsenstein J (1985) Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783-791
Forey PL (1980) Latimeria: a paradoxical fish. Proc R Soc Lond
B 208:369-308
Forey PL (1987) Relationships of lungfishes. J Morphol Suppl
1:75-91
Forey P1 (1988) Golden jubilee for the coelacanth Latimeria
chalumnae. Nature 336:727-732
Forey PL (1991a) Blood lines of the coelacanth. Nature 351:
347-348
Forey PL (1991b) Latimeria chalumnae and its pedigree. In:
Musick JA, Bruton MN, Balon EK (eds) The biology o f Latimeria chalumnae and evolution of coelacanths. Kluwer,
Hingham MA, pp 75-97
FritzschB (1987) Inner ear o f the coelacanth fish Latimeria has
tetrapod affinities. Nature 327:153-154
Gardiner BG (1980) Tetrapod ancestry: a reappraisal. In:
Panchen AL (ed) The terrestrial environment and the origin
of land vertebrates. Academic Press, New York, pp 177-185
GardinerBG (1984) The relationships ofthe palaeomiscid fishes, a review based on new specimens o f M i m i a and Moythomasia from the Upper Devonian of western Australia. Bull
Br Mus Nat Hist Geol 37:173--428
Gauthier J, Kluge AG, Rowe T (1988) Amniote phylogeny and
the importance of fossils. Cladistics 4:105-209
Gee H (1990) Fossil fishes and fashion. Nature 348:194-195
Gorr T, Kleinschmidt T, Fricke H (1991) Close tetrapod relationships of the coelacanth Latimeria indicated by haemoglobin sequences. Nature 351:394-397
Hedges BS, Moberg KD, LR Maxson (1990) Tetrapod phylogeny inferred from 18S and 28S ribosomal RNA sequences
and a review of the evidence for amniote relationships. Mol
Biol Evol 7:607-633
Hillis DM, Dixon MT (1989) Vertebrate phylogeny: evidence
from 28S ribosomal DNA sequences. In: Fernholm B, Bremer
K, Jornvall H (eds) The hierarchy of life. Elsevier, Amsterdam, pp 355-367
HolmesEB (1985) Are lungfishes the sister group oftetrapods?
Biol J Linn Soc 25:379-397
Kocher TD, Thomas WK, Meyer A, Edwards SV, Paabo S, VillablancaFX, WilsonAC (1989) Dynamicsofmitochondrial
DNA evolution in animals: amplification and sequencing with
conserved primers. Proc Natl Acad Sci USA 86:6196-6200
Lagios MD (1982) Latimeria and the Chondrichthyes as sister
taxa. Copeia 1982:942-948
Lovtrup S (1977) The phylogeny ofvertebrata. Wiley, London
Maeda N, Zhu D, Fitch WM (1984) Amino acid sequences of
lower vertebrate parvalbumins and their evolution. Mol Biol
Evol 1:473-488
113
Maisey JG (1986) Heads and tails: a chordate phylogeny. Cladistics 2:201-256
Marshall CR (1987) Lungfish: phylogeny and parsimony. J
Morphol Suppl 1:151-162
MarshallCR, SchultzeH-P (1992) Relative importance of molecular, neontological and paleontological data in understanding the biology of the vertebrate invasion of land. J Mol Evol
34:93-101
Meyer A, Wilson AC (1990) Origin oftetrapods inferred from
their mitochondrial DNA affiliation to lungfish. J Mol Evol
31:359-364
MeyerA, WilsonAC (1991) Coelacanth's relationships. Nature
353:219
MeyerA, KocherTD, BasasibwakiP, WilsonAC (1990) Monophyletic origin of Lake Victoria cichlid fishes suggested by
mitochondrial DNA sequences. Nature 347:550-553
MilesRS (1977) Dipnoan (lungfish) skulls and the relationships
of the group: a study based on new species from the Devonian
of Australia. Zool J Linn Soc 61:1-328
Mindell DP, Honeycutt RL (1990) Ribosomal RNA in vertebrates: evolution and phylogenetic applications. Annu Rev
Ecol Syst 21:541-566
Miyamoto MM, Boyle SM (1989) The potential importance of
mitochondrial DNA sequence data to eutherian mammal
phylogeny. In: Fernholm B, Bremer K, Jornvall H (eds) The
hierarchy of life. Elsevier, Amsterdam, pp 437-450
Normark BB, McCune AR, Harrison RG (1991) Phylogenetic
relationships of neopterygian fishes, inferred from mitochondrial DNA sequences. Mol Biol Evol 8:819-834
NorthcuttRG (1987) Lungfish neural characters and their bearing on sarcopterygian phylogeny. J Morphol Suppl 1:277-297
Panchen AL, Smithson TR (1987) Character diagnosis, fossils
and the origin of tetrapods. Biol Rev 62:341-438
Patterson C (1980) Origin of tetrapods: historical introduction
to the problem. In: Panchen AL (ed) The terrestrial environment and the origin of land vertebrates. Academic Press, New
York, pp 159-175
Patterson C (1981) Significance of fossils in determining evolutionary relationships. Annu Rev Ecol Syst 12:195-223
Roe BA, Ma D-P, Wilson RK, Wong JF-H (1985) The complete
nucleotide sequence of the Xenopus laevis mitochondrial genome. J Biol Chem 260:9759-9774
RomerAS (1966) Vertebrate paleontology, ed3. University of
Chicago Press, Chicago
Rosen DE, Forey PL, Gardiner BG, Patterson C (1981) Lungfishes, tetrapods, paleontology, and plesiomorphy. Bull Am
Mus Nat Hist 167:159-276
Sanderson MJ, Donoghue MJ (1989) Patterns of variation in
levels of homoplasy. Evolution 43:1781-1795
Schultze H-P (1987) Dipnoans as sarcopterygians. J Morphol
Supp] 1:39-74
SchultzeH-P (1991) CT scan reconstruction ofthe palate region
of Latimeria chalumnae. In: Musick JA, Bruton MN, Balon
EK (eds) The biology ofLatimeria ehalumnae and evolution
of coelacanths. Kluwer, Hingham MA, pp 183-192
Schultze H-P, Arsenault M (1985) The panderichthyid fish E1pistostege: a close relative of tetrapods? Palaeontology 28:
293-309
Schultze H-P, Campbell KSW (1987) Characterization of the
Dipnoi, a monophyletic group. J Morphol Suppl 1:25-37
Sharp PM, Lloyd AT, Higgins DG (1991) Coelacanth's relationships. Nature 353:218-219
Smith JLB (1939) A surviving fish of the order Actinistia. Nature 143:455-456
Smith JLB (1953) The second coelacanth. Nature 171:99-101
Stock D, Moberg KD, Maxson LR, Whitt GS (1991) A phylogenetic analysis of the 18S ribosomal RNA sequence of the
coelacanth Latimeria chalumnae. In: Musick JA, Bruton MN,
Balon EK (eds) The biology of Latirneria chalumnae and
evolution of coelacanths. Kluwer, Hingham MA, pp 99-117
StockDW, SwoffordDL (1991) Coelacanth's relationships. Nature 353:217-218
Swofford DL (1991) PAUP: phylogenetic analysis using parsimony, version 3.0rB1. Illinois Natural History Survey,
Champaign IL
Thomson KS (1991) Living fossil: the story of the coelacanth.
Norton, New York
Vorobjeva EI (1980) Observations on two rhipidistian fishes
from the Upper Devonian of Lode, Latvia (Lode). Zool J
Linn Soc 70:191-201
Wahlert G von (1968) Latimeria und die Geschichte der Wirbeltiere. Fischer, Stuttgart, West Germany
Wiley EO (1979) Ventral gill arch muscles and the interrelationships of gnathostomes, with a new classification of the
Vertebrata. Zool J Linn Soc 67:149-179
Received November 11, 1991/Revised and accepted April 6,
1992