Biochemical Systematics and Ecology 28 (2000) 319}350
The interrelationships of Acanthomorph "shes:
A total evidence approach using molecular
and morphological data
E.O. Wiley!,",*, G. David Johnson", Walter Wheaton Dimmick!
!Natural History Museum and Department of Systematics and Ecology, University of Kansas, Lawrence,
KS 66045, USA
"Division of Fishes, National Museum of Natural History, Washington, DC 20560, USA
Received 30 November 1998; accepted 11 June 1999
Abstract
DNA sequence and morphological data were analyzed for specimens of twenty-"ve species of
acanthomorph "shes and two specimens representing the outgroups Aulopiformes and My-
ctophiformes. A 572 base-pair (bp) segment of the 12S ribosomal mitochondrial gene, 1112 bp
from three regions of the 28S ribosomal nuclear gene, and 38 morphological transformation
series were analyzed under the criterion of maximum parsimony. The total evidence analysis
resulted in a set of four most parsimonious trees. Relationships common to all trees are largely
congruent with the hypothesis articulated by Johnson and Patterson (1993. Bull. Mar. Sci. 52,
554}626). ( 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Phylogenetic analysis; Molecular systematics; Acanthomorpha; Ctenosquamata; Paracanthop-
terygii; Euacanthopterygii; Acanthopterygii; Smegmamorpha; Percomorpha
1. Introduction
The Acanthomorpha (Rosen, 1973), or so-called spiny-rayed "shes, are the crown
group of the major radiation of extant "shes, the Teleostei. With about 300 families
and over 14,000 species, they comprise the majority of living teleosts, exhibiting
remarkable morphological and ecological diversity. In habitat, they range from
*Corresponding author. Tel.: #1-785-864-4540; fax: #1-785-864-5335.
E-mail address: ewiley@eagle.cc.ukans.edu (E.O. Wiley)
0305-1978/00/$ - see front matter ( 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 3 0 5 - 1 9 7 8 ( 9 9 ) 0 0 0 6 9 - 1
Fig. 1. Outline drawings of some "sh groups used in this study: (a) Synodontidae, (b) Myctophidae,
(c) Lamprididae, (d) Polymixiidae, (e) Percopsidae, (f) Gadidae, (g) Batrachoididae, (h) Melamphaidae,
(i) Zeidae, (j) Berycidae, (k) Holocentridae, (l) Mugilidae, (m) Atherinidae, (n) Melanotaeniidae, (o) Be-
lonidae, (p) Mastacembelidae, (q) Centrarchidae, (r) Balistidae. (Drawings from J. S. Nelson, 1994. Fishes
of the World, Wiley-Interscience, New York. Used with permission.)
mountain streams to the abyssal depths of the ocean; some of their extensive diversity
in form is evident in Fig. 1, where a representative of some of the families included in
this study is shown in outline drawing.
The Acanthomorpha originated with Rosen's (1973) seminal paper on interrelation-
ships of higher euteleosteans, and although the group found general acceptance (e.g.,
Fink and Weitzman, 1982; Lauder and Liem, 1983; Fink, 1984), there was consider-
able ambiguity in the distribution of the "ve characters used by Rosen to diagnose it.
Rosen (1985) considered only two characters synapomorphic for acanthomorphs,
and Stiassny (1986) presented four additional ones. Johnson and Patterson (1993)
320 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
considered new and previously proposed characters and accepted seven as providing
valid support for monophyly of the Acanthomorpha. Rosen (1973) proposed that the
Myctophiformes (lantern"shes and relatives) are the sister group of the Acan-
thomorpha and that the Aulopiformes (lizard"shes and relatives) are the sister group
of these two together, a group he called the Ctenosquamata. Rosen (1985) challenged
this hypothesis, but it was subsequently corroborated by Johnson (1992) and has not
been challenged since.
Although the monophyly of the Acanthomorpha and its sister group relationships
with other higher teleost groups (Neoteleostei) are now well established, the relation-
ships among major lineages within the Acanthomorpha remain controversial. Joh-
nson and Patterson (1993) reviewed, compared and evaluated the major alternative
hypotheses (see their Figs. 2}4, 11, 18 and 19) and proposed a new hypothesis of
acanthomorph relationships (their Figs. 24 and 25). The purpose of our study is to test
the hypothesis of Johnson and Patterson (1993) using a combination of molecular and
morphological data.
The morphological transformation series (i.e., characters) are those used by Joh-
nson and Patterson (1993), and our strategy was to select taxa that would best allow
us to test their hypothesis and pertinent alternatives. The "nal selection, however, was
less than optimal, being constrained by the availability of tissue samples and the
relatively small number of species that could be collected within the limited time frame
of this study. As a consequence, our test of the Johnson and Patterson (1993)
hypothesis is largely based on di!erent acanthomorph taxa. The most notable
di!erences are the following: (1) exclusion of the most basal family of Lam-
pridiformes, Veliferidae, and substitution of Lampris; (2) inclusion of additional gen-
era of Paracanthopterygii to test the monophyly of the group; (3) inclusion of the
centrarchid Lepomis macrochirus to test the earlier hypothesis that Elassoma is most
closely related to centrarchids among the families included in our study. In addition,
we included genera of several major percomorph groups (e.g., Perciformes, Scor-
paeniformes, Pleuronectiformes, Tetraodontiformes) in order to provide some repres-
entation of the extensive diversity within that assemblage. Although we mention some
of the interesting rami"cations of their inclusion in our analysis, we do not suggest
that this is an attempt to explore percomorph intrarelationships. Such an investiga-
tion would obviously require much more comprehensive morphological data and
wider and more extensive taxonomic coverage.
2. Materials and methods
Most specimens used for acquisition of DNA sequence data were collected in the
"eld and immediately frozen in liquid nitrogen. Specimens used for morphological
comparisons were taken from existing museum collections. Species examined are
listed in Table 1.
Tissue samples were stored at !703C until dissected. Approximately 0.1 g of tissue
was dissected and DNA was extracted from frozen tissue by standard chloro-
form/phenol methods (Maniatis et al., 1982). We used the Polymerase Chain Reaction
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 321
Table 1
Species used in the present study
Aulopiformes (Synodontidae)
Synodus intermedius, SYN. USNM uncat. (Tissue No. 311).
Synodus variegatus. USNM 315318
Myctophiformes (Myctophidae)
Hygophum hygomii, HYG. KU uncat. (Tissue No. 263).
Hygophum macrochir. AMNH 25019
Lampridiformes (Lamprididae)
Lampris guttatus, LAM. USNM 357482, MCZ 255173, SIO 82-70.
Polymixiiformes (Polymixiidae)
Polymixia japonica, POL. KU uncat. (Tissue No. 258).
Polymixia lowei. USNM 308378.
Paracanthopterygii sensu Patterson and Rosen (1989)
Percopsiformes
Percopsis omiscomaycus (Percopsidae), PERC. UAIC 11218.07; USNM179711.
Aphredoderus sayanus (Aphredoderidae), APH. UAIC 10015.11; USNM217374.
Gadiformes
Pollachius virens (Gadidae), PLL. KU Uncat. (Tissue No. 359), USNM 187248.
Merluccius bilinearis (Merlucciidae), MER. KU Uncat. (Tissue No. 367), USNM 239843.
Batrachoidiformes
Opsanus tau (Batrachoididae), OPS. KU 22948, USNM 118326.
Ophidiiformes
Petrotyx sanguineus (Ophidiidae), PEX. USNM 327557.
Stephanoberyciformes
Scopeloberyx robustus (Melamphaidae), SB. KU uncat. (Tissue No. 276), USNM 215774.
Zeiformes
Zeus faber (Zeidae), ZEU. AMS NI1090; USNM 307842.
Zenopsis nebulosus (Zeidae), ZNP. AMS NI 1099
Zenopsis conchifer FMNH 67090.
Beryciformes sensu Johnson and Patterson (1996)
Beryx sp.(Berycidae), BER. KU uncat. (Tissue No. 827).
Beryx splendens. AMNH 95743
Holocentrus coruscus (Holocentridae), HOL. USNM 327564.
Holocentrus vexillaris. USNM 269553.
Percomorpha sensu Johnson and Patterson
Smegmamorpha sensu Johnson and Patterson (1996)
Elassomatidae
Elassoma evergladii (Elassomatidae), EL. UAIC 10854.02
Elassoma zonatum.USNM 230627
Atherinomorpha
Melanotaenia splendens (Melanotaeniidae), MEL. KU 25191.
Melanotaenia nigricans. USNM 173746.
Atherinomorus stipes (Atherinidae), ATH. USNM uncat. (Tissue No. 326).
Strongylura notata (Belonidae), STR. USNM 327560.
Strongylura marina USNM 292769.
Gambusia aznis (Poeciliidae), GAM. KU uncat. (Tissue No. 831).
Gambusia vittata. USNM 206285
Mugilidae
Mugil curema (Mugilidae), MUG. UAIC 10853.09.
Mugil cephalus USNM 156159
322 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Table 1
Species used in the present study
Synbranchiformes
Mastacembelus sp. (Mastacembelidae), MAS. KU 22982, AMNH 42129.
`Highera Percomorpha sensu Johnson and Patterson (1996)
Perciformes
Lepomis macrochirus (Centrarchidae), LEP. KU 25193.
Lepomis marginatus. AMNH 79149.
Morone chrysops (Moronidae). MOR. KU 22901.
Morone americana. USNM 109851.
Pleuronectiformes
Bothus lunatus (Bothidae). BOT. USNM uncat. (Tissue No. 154).
B. ocellatus. USNM 273281.
Tetraodontiformes
Melichthys niger (Balistidae). USNM uncat. (Tissue No. 105), ANSP 109442.
Dactylopteriformes
Dactylopterus volitans (Dactylopteridae). USNM uncat. (Tissue No. 237), USNM 348833.
The structure of the classi"cation uses a listing convention (Wiley, 1981) for the implied phylogenetic tree.
Abbreviations are those used in the phylogenetic trees presented in Figs. 2 and 8. Both morphological and
DNA data were collected for some species. First catalogue number refers to specimens from which DNA
data were obtained. Second catalogue number(s) refer to specimens from which morphological data were
obtained. When morphological data was collected on a close relative, the species is listed below and the
catalogue number refers to a specimen for which morphological data were collected. Abbreviations follow,
Leviton et al. (1985).
(Saiki, 1990) to amplify selected gene regions from genomic extractions. The 12S
ribosomal mitochondrial gene region was ampli"ed using amplitaqTM DNA polymerase
from Perkin-Elmer Cetus corporation. Table 2 details the primers used in this study.
Ampli"ed products were loaded onto NuSieve GTGt (FMC) agarose gels and
electrophoresed at 85}90 V for approximately one hour. The target band was excised
from the gel and the DNA was recovered with QIAquickTM gel extraction kits
(Qiagen). The puri"ed PCR product was manually sequenced with the fmolTM DNA
sequencing system (Promega) using the primers indicated in Table 2.
Results were visualized by autoradiography and scored by visual inspection. Se-
quencing both strands was not accomplished in favor of maximizing our data for
di!erent regions and species. Two sources of error are possible, random and system-
atic. We presume that random errors such as misreading gels would lower signal and
introduce spurious autapomorphies, but would not cause skewed results. We did
encounter some systematic errors in the form of stops and compressions. We do not
feel that these apparent artifacts have compromised our results because they usually
e!ected all or most taxa when they occurred, and thus were easy to identify. Missing
data values were used for ambiguous sites or for obvious compressions #anked by
readable sequence. Finally, selected sequences were compared against results obtained
using an automated sequencer (ABI 310 Genetic Analyzer) where complementry
strands were sequenced. Only minor di!erences were observed in these sequences
when compared to manual sequences and these were ascribed to errors in
interpretation of the manual gels. All sequences were deposited in Genbank
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 323
Table 2
Sequencing (S) and ampli"cation (A) primers used in this study
Name Sequence Strand Use
Mitochondrial 12S gene
12Sa! 5@AAACTGGGATTAGATACCCCACTA3@ Light A, S
12Sb! 5@AGGAGGGTGACGGGCGGTGTGT3@ Heavy A, S
12Sd" 5@GGGTTGGTAAATCTCGTGC3@ Light A, S
Nuclear 28S gene
28W# 5@CCTGTTGAGCTTGACTCTAGTCTG3@ A, S
28X# 5@GTGAATTCTGTTCACAATGATAGGAAGAGCC3@ A, S
28MM$ 5@AGCCAATCCTTATCCCGAAGTTACG3@ A, S
28DD# 5@GTCTTGAAACACGGACCAAGGAGTCT3@ A
28EE# 5@ATCCGCTAAGGAGTGTGTAACAACTCACC3@ S
28FF# 5@GGTGAGTTGTTACACACTCCTTAGCGGAT3@ S
!Modi"ed from Kocher et al. (1989).
"Modi"ed 503 primer of John Patton, Washington University. Position in the human genome is 972}810.
#Hillis and Dixon (1991).
$Modi"ed from Hillis and Dixon (1991).
(12S rDNA, AF149982}AF150008; 28! rDNA, AF153285}AF153311; 28ee rDNA,
AF150637}AF150663; 28mm rDNA, AF152115}AF152141; 28wx rDNA,
AF152142}AF152168)
Initially, sequences were aligned by manual inspection. Next, we consulted second-
ary structure models of the 12S gene (Van der Peer et al., 1994) and the 28S rDNA
gene for Xenopus (Clark et al., 1984) and produced heuristic models of the secondary
structure of applicable regions for Synodus, one of our outgroup genera. This
model was used to compare other species. Secondary structure, as indicated by
complementry strands forming ladders #anked by unpaired loops, was then used to
re"ne the initial alignments. The data were exported to MALIGN (Wheeler and
Gladstein, 1994) using an option that constrains the data for alignment among
presumed loop and stem regions. This alignment was inspected visually and adjusted
for problematic alignments while maintaining the loops and stems of the secondary
structure.
Stem and loop regions were examined for possible site saturation (multiple muta-
tions at a site) by plotting the number of mutations between pairs of taxa against the
Tamura-Nei genetic distance coe$cients generated from MEGA 1.01 (Kumar et al.,
1993). Thus, saturation studies were conducted on these classes of data separately.
Gene regions composed of sites for which only ambiguous homology statements
could be made due to the presence of gaps or saturation for all classes of substitutions
were removed from the analysis, as presented in Section 3, Results. Classes of
substitution that showed saturation were `screeneda from the analysis using step-
matrices, as presented in Section 3, Results. All remaining data were analyzed as
unordered and equally weighted characters.
Data were collected for 38 morphological transformation series. Dissected cleared-
and-stained specimens were examined using a Leitz dissecting microscope. Specimens
324 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
examined for morphological characters are di!erent from those from which DNA
data were obtained. The transformation series and associated characters listed below
are modi"ed from Johnson and Patterson (1993). One transformation series used
by Johnson and Patterson (1993) was omitted [`extrascapular unmodi"ed (0),
or enlarged and covering parietal (1)a] because it is a synapomorphy for
Stephanoberyciformes, a group represented by only a single species in the present
study. Character numbers presented below correspond to column numbers in Fig. 2.
1685. * Dorsal and anal "ns spines absent (0) or present (1).
1686. * Rostral cartilage absent (0) or present (1).
1687. * Medial caudal cartilages present (0) or absent (1).
1688. * Infracarinalis muscles joined (0) or separate (1).
1689. * Posttemporal loosely attached to epioccipital (0) or tightly attached (1).
1690. * Medial pelvic process ends in cartilage (0) or ends in bone (1).
1691. * First centrum unmodi"ed anteriorly (0), or with exoccipital facets (1).
1692. * First epineural dorsolateral (0), or in horizontal septum (1), or absent (2).
1693. * Posterior pelvic process ends in cartilage (0), or ends in bone (1).
1694. * Spina occipitalis absent (0), or present (1).
1695. * Anterior (3}6) epineurals originate on neural arch (0), or on centrum,
parapophysis, or rib (1), or are absent (2).
1696. * Epipleurals present (0), or absent (1).
1697. * Epicentral ligaments present on all or some of vertebrae 1}8 (0), or absent on
these vertebrae (1).
1698. * Distal parts of epineurals 2}5 dorsolateral (0), in horizontal septum (1), or
absent (2).
1699. * Pelvic "n spine absent (0), present with a symmetrical base (1), or present
with a complex base (2).
1700. * Pelvic radials in a continuous row (0), or either in a discontinuous row or
absent (1).
1701. * Anteromedial process of pelvic bone absent (0), or present (1).
1702. * Baudelot's ligament originates on "rst vertebra (0), on the basioccipital (1),
on the exoccipital (2), or is absent (3).
1703. * Dorsal "n originates behind the fourth neural spine, supraneurals present
(0), in front of the fourth neural spine, supraneurals present (1), in front of the
fourth neural spine, supraneurals absent (2), behind the fourth neural spine,
supraneurals absent (3), or anterior to the "rst neural spine, supraneurals
absent (4).
1704. * Epineurals on vertebrae 3}6 on neural arch, centrum, or parapophysis (0),
on rib (1), epineurals absent (2), or epineurals present but ribs absent (3).
1705. * Dorsal "n spines absent or without chain-link articulation (0), with chain-
link articulation (1), or spine-bearing radials with no distal radials (2).
1706. * Supraneurals end distally in cartilage (0), in bone (1), or supraneurals absent
(2).
1707. * Second ventral procurrent caudal ray unmodi"ed or absent (0), or shortened
proximally (1).
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 325
1708. * No ligament from shaft of postcleithrum to posterolateral corner of pelvic
girdle (0), ligament present (1), or no ligament and girdle secondarily dis-
placed posteriorly (2).
1709. * Uncinate process present on "rst epibranchial and interarcual cartilage
absent (0), uncinate process present and interarcual cartilage present (1),
uncinate process absent or not articulating with second pharyngobranchial
and interarcual cartilage absent (2).
1710. * Second ural centrum distinct (0), or fused with "rst preural centrum#"rst
ural centrum (PU1#U1) (1).
1711. * Six hypurals (0), or "ve or fewer hypurals (1).
1712. * Pelvics with seven or more rays (0), or with six or fewer rays (1).
1713. * Transforming ctenoid scales absent (0), or present (1).
1714. * One or more free pelvic radials (0), or no free pelvic radials (1).
1715. * All or some epineurals above horizontal septum (0), all in horizontal septum
(1), or two or fewer epineurals (2).
1716. * Principal caudal "n rays 19 (0), 18 (1), or 17 or fewer (2).
1717. * Distal and proximal ceratohyals separated by cartilage (0), sutured (1), or
sutured with a dorsal prong (2).
1718. * Orbitosphenoid present (0) or absent (1).
1719. * Pelvic bones loosely attached or overlapping medially (0), with broad
median contact (1), or sutured (2).
1720. * First epineural on neural arch or absent (0), or on transverse process (1).
1721. * Jakubowski's organ absent (0), or present (1).
1722. * Parahypural articulating with "rst preural centrum (0), or truncated prox-
imally (1)
Morphological and DNA data were combined to form a total evidence matrix.
Morphological data were analyzed as unordered and equally weighted characters.
Data analysis was carried out under the principle of maximum parsimony
using PAUP 3.1.1 (Swo!ord, 1993) and PAUP*4.0d64. First, we performed a
total evidence analysis in order to generate a phylogenetic hypothesis of relationship.
This was carried out using a heuristic search with 20 random starting trees in an
e!ort to avoid local parsimony optima (Maddison, 1991). Second, we performed
three additional series of analyses using the same options, one each including the 12S
rDNA and 28S rDNA data respectively and another including only morphological
data.
We used three criteria to evaluate tree support for our total evidence analysis. First,
we examined the support associated with each node. Three classes of transformation
series were identi"ed based on the consistency index: unique and unreversed (ci"1.0),
intermediate (0.5(ci(1.0), and low (ci(0.5). Second, we performed a branch
support analysis (Bremer, 1994) using the strict consensus tree that resulted from the
total evidence analysis to check the robustness of the internal nodes that were
common to each of the most parsimonious trees using the program Treerot (Sorenson,
1996). Third, we performed bootstrap analyses (Felsenstein, 1985) using the bootstrap
option of PAUP using 100 replicates.
326 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 2. See p. 339 for caption.
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 327
Fig. 2. Continued.
328 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 2. Continued.
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 329
Fig. 2. Continued.
330 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 2. Continued.
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 331
Fig. 2. Continued.
332 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 2. Continued.
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 333
Fig. 2. Continued.
334 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 2. Continued.
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 335
Fig. 2. Continued.
336 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 2. Continued.
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 337
Fig. 2. Continued.
338 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 2. Aligned DNA sequences and morphological data for 25 acanthomorph and two outgroup species.
Taxon abbreviations are in Table 1.
3. Results
A total of 1722 transformation series (TS) for 25 acanthomorph and two outgroup
taxa were obtained after alignment of the DNA sequences and integration of the
morphological and molecular data (Fig. 2). There were a greater number of variable
12S rDNA sites (251 of 572 bp, 33% total) than 28S rDNA sites (178 of 1112 bp, 16%
total).
Alignments of the 12S rDNA sequence are shown in Fig. 2, TS 1-572. Stem regions
comprise the following transformation series: 1}4, 11}14, 18}20, 40}45, 52}56, 80}84,
100}101, 106}107, 109}114, 123}125, 132}135, 140}143, 145}146, 150}151, 155}158,
168}171, 175}178, 180}182, 194}197, 215}218, 220}223, 225}227, 235}239, 243}246,
250}253, 257}259, 262}274, 283}286, 289}298, 302}305, 314}317, 326}332, 346}362,
367}368, 373}375, 380}382, 395}396, 421}423, 425}427, 431}438, 443}450, 453}463,
469}472, 479}482, 501}503, 518}520, 536}540, 546}550, and 565}569. Inspection of
loop regions lead us to discriminate between two classes of loops: (1) conserved loop
regions and (2) non-conserved loop regions. Conserved loop regions are character-
ized by few gaps and relatively unambiguous alignments. These comprise TS 5}10,
15}17, 46}51, 57}79, 85}99, 102}105, 126}131, 136}139, 147}149, 152}154, 159}167,
172}174, 183}193, 219, 228}234, 240}242, 247}249, 254}256, 260}261, 275}282,
287}288, 299}301, 306}313, 318}325, 363}366, 369}372, 376}379, 388}394, 424,
428}430, 439}442, 451}452, 464}468, 489}494, 507}511, 527}535, 541}545, 551}564,
and 570}573. Non-conserved loop regions are characterized by large gaps and
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 339
Fig. 3. 12S rDNA saturation study I. Pair-wise Tamura-Nei distances plotted against number of mutations
for 25 species of acanthomorph "shes and two outgroups for loop regions. Above: Variable loop transistion
(TS) and transversion (TV) plots. Middle: Conserved loop transversions. Bottom: Conserved loop
transitions plots for arginine}guanine (AG) and cytosine}thymine (CT). A second-order polynominial trend
line is "tted to the data in each plot.
ambiguous alignments. These comprised TS 21}39, 108, 115}122, 126}131, 144, 179,
198}214, 224, 333}345, 383}387, 397}421, 473}478, 483}488, 495}500, 512}517, and
521}526.
Plots of number of mutations versus Tamura-Nei Distance for the three classes of
12S rDNA data are shown in Figs. 3 and 4. We conclude that (1) both transitions and
340 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 4. 12S rDNA saturation study II. Pair-wise Tamura-Nei distances plotted against number of muta-
tions for 25 species of acanthomorph "shes and two outgroups for stem regions. Left: Total transversions
(TV). Middle: Arginine}guanine transitions (TS). Right: Cytosine}thymine transitions. A second-order
polynominial trend line is "tted to the data in each plot.
transversions are saturated in the non-conserved loop regions, (2) both classes of
transitions are saturated in the conserved loop regions, and (3) A}G transitions are
saturated in the stems. Thus, nonconserved loops were eliminated from further
analysis and step matrices were employed that screened classes of saturated mutations
from further analysis in conserved loop and stem regions.
Data from four noncontiguous regions of the 28S rDNA gene were collected (Fig.
2). From 5@ to 3@ they were: 28! (TS 573}831), 28ee (TS 832}1052), 28mm (TS
1053}1269), and 28wx (TS 1270}1684: see Hillis and Dixon, 1991). Alignment of the
28S rDNA was trivial. However, one region (TS 1371}1378) was largely unreadable
and was eliminated from subsequent analyses. Stems comprised the following TS:
592}594, 597}601, 604}609, 611}612, 616}617, 619}627, 632}634, 655}662, 672}679,
682}691, 695}698, 705}708, 710}714, 720}724, 726}729, 734}736, 743}745, 749}754,
756}757, 759}761, 770}778, 794}796, 798}807, 815}819, 821}823, 827}828, 840}851,
854}861, 863}866, 868}871, 873}876, 892}903, 909}918, 920}925, 927}928, 935}936,
938}939, 942}943, 948}949, 951}952, 954}955, 959}969, 971}972, 997}1002,
1010}1015, 1017}1019, 1024}1041, 1053}1054, 1059}1063, 1065}1072, 1076}1079,
1081}1082, 1085}1089, 1098}1100, 1108}1110, 1112}1120, 1122}1123, 1134}1136,
1138}1139, 1142}1148, 1163}1171, 1174}1176, 1200}1205, 1207}1211, 1234}1243,
1249}1252, 1254}1257, 1260}1265, 1280}1284, 1286}1303, 1307}1311, 1315}1331,
1337}1349, 1361}1369, 1378}1386, 1390}1400, 1403}1420, 1423}1426, 1430}1441,
1453}1460, 1465}1481, 1489}1497, 1513}1521, 1524}1526, 1529}1532, 1540}1543,
1546}1553, 1557}1561, 1572}1582, 1593}1602, 1611}1615, 1619}1624, 1638}1643,
1661}1666, and 1680}1684. Loops comprised the remaining tranformation series.
Three classes of mutations were investigated for saturation, AG and CT transitions
and total transversions (Fig. 5). Based on plots of Tamura}Nei distances versus
number of mutations, we conclude that 28S rDNA stems are saturated for A}G
transitions while the loops are saturated for C}T transitions. These classes of muta-
tions were screened from further analysis through the use of step matrices.
Variation among taxa for the 38 morphological transformation series is shown in
Fig. 2. They comprise transformation series 1686}1722.
A series of parsimony analyses was performed, both on the total evidence matrix
and partitioned subsets of the data. In each case, Hygophum was designated the
operational sister group of Acanthomorpha and Synodus was designated the second
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 341
Fig. 5. 28S rDNA saturation study. Pair-wise Tamura-Nei distances and number of muitations for 25
species of acanthomorph "shes and two outgroups. Left side, top to bottom: Total loop transversions (TV),
arginine}guanine (AG) loop transitions (TS), and cytocine}thymine (CT) loop transitions. Right side, top to
bottom: Total stem transversions, AG stem transitions, and CT stem transitions. A second-order poly-
nominial trend line is "tted to the data in each plot.
outgroup as per previous hypotheses of higher teleost relationships. Summary data
for each analysis is shown in Table 3.
The total evidence analysis found four most parsimonious trees of 977 steps. A strict
consensus tree was generated that re#ects the 22 putative monophyletic groupings
342 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Table 3
Summary of Tree `statisticsa and characters for trees found in this study
Analysis
Total evidence 12S rDNA 28S rDNA Morphology
Parameter
No. trees 4 6 23 137
Tree length 997 steps 580 steps 259 steps 116 steps
Ensemble CI 0.4724 0.4310 0.6371 0.4828
Ensemble HI 0.6422 0.5690 0.3629 0.5172
Ensemble RI 0.4605 0.6846 0.5251 0.7297
Ensemble RC 0.2176 0.3878 0.3572 0.3523
No. TS 1722 572 1112 38
TS excluded 136 124 12 0
TS informative 256 142 76 38
TS uninformative 213 109 104 0
Note: Excluded, informative, and uninformative refer to the number of characters in each class of data.
common to each of the most parsimonious trees (Fig. 6). A summary of branch
lengths, number and quality of synapomorphies, Bremer support values and boot-
strap values for each node appearing on the strict consensus tree is shown in Table 4.
A summary of previously named groups that appear on the strict consensus tree is
shown in Table 5. The four equally parsimonious trees di!ered primarly in their
interpretations of the phylogenetic positions of Elassoma and Mastacembelus, as
shown in the four subtrees in Fig. 7a}d.
Analysis of the 12S rDNA data alone resulted in six equally parsimonious
trees whose strict consensus is shown in Fig. 8a. Analysis of the 28S rDNA
data resulted in 23 equally parsimonious trees (consensus in Fig. 8b) while
that of the morphological data resulted in 113 equally parsimonious trees (consensus
in Fig. 8c).
4. Discussion
We expected that a combination of mitochondrial and nuclear ribosomal genes
would provide a strong data base from which we could evaluate acanthomorph
relationships. This expectation was only partly met. Saturation studies indicate that
many of the regions of the 12S ribosomal gene useful in studies at lower taxonomic
levels (Wiley et al., 1998; Tang et al., 1999) are saturated, especially in loop regions
where only transversions could be used and then only for the `conserved loops.a The
28S rDNA gene fragments were very conservative. Many of the few sites that do vary
in the 28s rDNA regions we studied also show signs of transition saturation and those
that do not are not of particular help in evaluating hypotheses unless used in a total
evidence context. Given these observations, we might conclude that the DNA data
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 343
Fig. 6. A strict consensus tree of four equally parsimonious trees for 25 acanthomorphs under the
constraint that Synodus and Hygophum are sequential outgroups. Numbers above terminal branches are
branch lengths. Internode labels designate putative monophyletic groups. Tree summaries are shown in
Tables 3 and 4.
played a relatively small role in the total evidence analysis. But, the morphological
data, analyzed separately, did little better (Fig. 8c). Thus, we "nd it a remarkable
consequence of character interactions within a total evidence analysis that we ob-
tained any interpretable results at all, much less results that corroborate many of the
conclusions of Johnson and Patterson (1993).
344 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Table 4
Distribution of synapomorphies on strict consensus of four most parsimonious trees for acanthomorph
"shes (Fig. 6)
Node bl ci"1 ci?0.49 ci(0.5 Bremer Bootstrap
1 13 6 1 6 8 100
2 15 1 3 11 6 84
3 7 2 2 3 1 61
4 17 5 4 8 9 92
5 4 0 1 3 1 51
6 15 2 4 9 7 94
7 27 3 10 14 19 100
8 21 7 2 12 12 100
9 14 0 2 12 5 85
10 5 0 1 4 1 (50
11 6 0 2 4 1 (50
12 5 0 0 5 1 (50
13 13 1 3 9 5 61
14 7 0 0 7 1 (50
15 2}6 0 1}3 1}4 1 (50
16 6}8 0 1}2 5}6 3 58
17 5}6 0 0 5}6 2 (50
18 8}9 1 1 6}7 3 76
19 11}12 0 1 10}11 5 86
20 9}11 1 1}2 7}9 4 65
21 11}13 1 6 4}6 2 (50
22 11}12 1 3 7}8 4 58
Nodes with variable numbers re#ect di!erences in support among the most-parsimonious trees for
a particular internode. Abbreviations: bl, branch length; ci"1, total unique and unreversed synapomor-
pies; ci?0.49 and ci(0.5, total synapomorphies with consistency indices greater or lesser than the value
indicated; Bremer, the Bremer support value; Bootstrap, the bootstrap value.
Table 5
Previously recognized groups that appear on the strict consensus tree (Fig. 6)
Node Clade Previously recognized by
1 Acanthomorpha Rosen (1973)
2 Euacanthomorpha Johnson and Patterson (1993)
3 Holacanthopterygii Johnson and Patterson (1993)
4 Percopsiformes Berg (1947)
7 Gadiformes Berg (1947)
8 Zeidae Starks (1898)
14 Percomorpha Johnson and Patterson (1993)
16 unnamed Stiassney (1993)
18 Atherines Dyer and Cherno! (1996)
19 Cyprinodontea Dyer and Cherno! (1996)
20 Perciformes Berg (1947)
22 Beryciformes Johnson and Patterson (1993)
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 345
Fig. 7. Partial topologies of the four equally parsimonious trees found in this study. More basal topological
relationships are identical although nodal support might vary. Black bars are unambiguous synapomor-
phies (ci"1). Striped bars are synapomorphies with intermediate ci values (ci?0.49). The total number of
synapomorphies with low ci-values (ci(0.5) is shown in brackets beside the white bars.
As noted, our selection of taxa di!ers substantially from that of Johnson and
Patterson (1993), and these di!erences undoubtably account for some of the incongru-
ence between their results and ours. We included several groups of Paracanthop-
terygii which Johnson and Patterson (1993) had represented only by the basal genus
Percopsis, and, we included more nominal percomorphs in our formal analysis.
Furthermore, Johnson and Patterson's (1993) discussion of morphological character
variation is more extensive than ours and their taxon selection for the formal analysis
346 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
Fig. 8. Results from partitioned data sets. (a) The 12S rDNA data. A strict consensus tree of six equally
parsimonious trees (tree length 580 steps, CI"0.4310, HI"0.5690, RI"0.3878, RC"0.1671). (b) The
28S rDNA data. A strict consensus tree of 23 equally parsimonious trees (tree length 259 steps, CI"0.6371,
HI"0.3629, RI"0.5607, RC"0.3572). (c) Morphology data. Strict consensus of 137 equally parsimoni-
ous trees (tree length 116 steps, CI"0.4828, HI"0.5172, RI"0.7297, RC"0.3523).
was made from those taxa thought to be basal members of groups while our selection
was made on the basis of available frozen material. Thus, our analysis is best viewed as
an independent test of the morphological data (`nonoptimala taxon selection), and an
expansion of the parsimony analysis in terms of certain groups (paracanthop-
terygians, percomorphs) and characters (molecular data).
The overall topology of the strict consensus tree corroborates the basal position of
Lampridiformes despite the fact that Lampris exhibits several characters that are
interpreted as convergent similarities with higher acanthomorphs. These are TS 1696,
character 1 (1696}1), 1700}1, 1704}1, 1711}1, 1714}1. In contrast, Velifer has the
outgroup conditions (character `0a for each TS: Johnson and Patterson, 1993).
Euacanthomorph monophyly is strongly corroborated (Bremer Support, b"6), but
Polymixia is only weakly corroborated as a basal euacanthopmorh (b"1) because
many of the potential synapomorphies of Euacanthomorpha are convergent in
Lampris. Support would have been stronger if Velifer was used as a representative
lampridiform "sh. Strong to moderate corroboration for the monophyly of
many long-recognized taxa was expected and found (Tables 4 and 5). These include
Percopsiformes (Percopsis and Aphredoderus), Gadiformes (Pollachius and Merluc-
cius), Zeidae (Zeus and Zenopsis), Beryciformes s.s. (Holocentrus and Beryx),
Mugilomorpha#Atherinomorpha, and Perciformes (Lepomis and Morone).
Some general features of our topology are similar to the Johnson and Patterson
(1993) hypothesis and di!erent from those of previous authors, but these must be
viewed with caution because low branch support values are associated with many
nodes. Zeids (Zeus and Zenopsis) appear basally as hypothesized by Johnson and
Patterson (1993), not near the apex as they would be expected to group given previous
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 347
hypotheses that they are related to beryciforms and percomorphs (Rosen, 1973,
1985). Scopeloberyx is basal to Beryciformes s.s. and these groups are basal to
Percomorpha (Johnson and Patterson, 1993), contrary to most previous authors
(see section Introduction). Smegmamorpha (Johnson and Patterson, 1993) is allied
with `higher percomorphsa (i.e. perciforms, scorpaeniforms, tetraodontiformes, and
pleuronectiforms) rather than being basal to a beryciform#percomorph clade
(Rosen, 1973).
One major feature of our topology di!ers from that of Johnson and Patterson
(1993). Zeids appear more basal than Scopeloberyx. This may be the result of not
analyzing a stephanoberycid, since melamphaids such as Scopeloberyx share a number
of derived characters with acanthopterygian "shes that are not shared by stepah-
noberycids (see data matrix of Johnson and Patterson's (1993; p. 619)).
Morphological evidence for the monophyly of Paracanthopterygii (summarized by
Patterson and Rosen, 1989) is tenuous at best (Gill, 1997). Paracanthopterygii appears
polyphyletic on the strict consensus tree, with Percopsiformes and Gadiformes ap-
pearing more basally and the ophidiiform (Petrotyx) and batrachoidiform (Opsanus)
grouping within Percomorpha. This result questions the monophyly of the group. It
also questions the intrarelationships of its component members, since Patterson and
Rosen (1989) hypothesized that gadiforms are closely related to batrachoidiforms,
with ophidiiforms occupying a more basal position.
Finally, some of our results were unexpected. One of the more strongly corrobor-
ated groupings allies zeids with gadiforms, a hypothesis no one has proposed.
A poorly corroborated node aligns Petrotyx and Opsanus with the pleuronectiform
Bothus. This may be partly explained by our taxon selection because the most
basal pleuronectiform, Psetoddes, has some of the synapomorphies of higher
percomorphs that are lacking in Bothus. Other suspect and poorly supported nodes
include the placement of Mugil within Atherinomorpha rather than as the sister group
of a monophyletic Atherinomorpha, the grouping of Dactylopterus and Melichthys,
and the basal position of this group relative to the smegmamorph and perciform
"shes.
The newest and most controversial group of acanthomorphs, Smegmamorpha,
does not appear as a monophyletic group in the strict consensus tree (Fig. 6). One of
the most parsimonious trees does contain a monophyletic Smegmamorpha (Fig. 7a)
and none of the alternative hypotheses (Fig. 7b}d) contain alternative hypotheses that
are as strong in character support. However, we await a stronger test using more taxa
and di!erent genes to determine if independent molecular evidence can provide
stronger corroboration or refutation for this clade. It is notewrothy that Elassoma did
not group with its supposed centrarchid relative Lepomis. Although our analysis does
not resolve the precise relationships of Elassoma, it suggests that Elassoma is not
a centrarchid.
The results obtained in this study do not support several alternative hypotheses of
relationships proposed by other authors. Holocentrids (Holocentrus) do not appear as
the sister group of `higher percomorphsa (Lepomis, Morone, Dactylopterus, Melichthys,
Bothus in our sample) as proposed by Stiassny and Moore (1992). No paracanthop-
terygian appears to be closely related to atherinomorphs (Parenti, 1993). Other
348 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350
hypotheses not formally analyzed by Johnson and Patterson (1993) are corroborated.
These include Stiassny's (1993) hypothesis that mullets are related to atherinomorphs
and Dyer and Cherno!'s (1996) hypothesis that rainbow"shes (Melanotaenia) and
silversides (Atherinomorus) form a monophyletic group rather than being paraphyletic
relative to other antherinomorphs (Strongylura and Gambusia).
Acknowledgements
The bulk of this project was supported by a grant from the National Science
Foundation (DEB 93178812). The University of Kansas Research Development Fund
is gratefully acknowledged for purchasing the ABI 310 Genetic analyzer used in the
later part of the study. Field collecting was supported by DEB 93178812 and by grants
from the Smithsonian Institution Caribbean Coral Reef Ecosystems program and the
National Geographic Society. We thank the Ministry of Lands, Survey, and Natural
Resources, Kingdom of Tonga, the US Peace Corps, Tonga, and the sta! of Carrie
Bow Cay, Belize, for making our "eld work possible. Many thanks to our colleagues
for helping us collect specimens. Rick Mayden (University of Alabama) provided
Aphredoderus, Elassoma, and Mugil specimens. Jim Craddock (Woods Hole Oceano-
graphic Institute) and Karsten Hartel (Harvard University Museum of Comparative
Zoology) provided Hygophum and Scopeloberyx specimens. Thanks to the National
Marine Fisheries Service, Woods Hole, for providing a berth for Kate Shaw on two
survey voyages and to Kate for her skill and hard work in collecting specimens.
Finally, thanks to Judy Bevan, Mike Grose, and Ben Hou for their laboratory
assistance.
References
Berg, L. 1947. Classi"cation of Fishes Both Recent and Fossil. J. W. Edwards, Ann Arbor, Michigan, 517pp.
Bremer, K., 1988. The limits of amino acid sequence data in angiosperm phylogenetic reconstruction.
Evolution 42, 795}803.
Bremer, K., 1994. Branch support and tree stability. Cladistics 10, 295}304.
Clark, C.G., Tagve, B.W., Gerbi, S.A., 1984. Xenopus laevis 285 ribosomal RNA: a secondary structure
model and its evolutionary and functional implications. Nucleic Acids Res 12 (15), 6197}6220.
Dyer, B.S., Cherno!, B., 1996. Phylogenetic relationships among atheriniform "shes (Teleostei:
Atherinomorpha). Zool. J. Linnean Soc. London B 117, 1}69.
Felsenstein, J., 1985, Con"dence limits on phylogenies: An approach using the bootstrap. Evolution 39,
783}791.
Fink, W.L., 1984. Basal euteleosts: Relationships. In: Moser, H.G. (Ed.), Ontogeny and Systematics of
Fishes. American Society Ichthyologists and Herpetologists Spec. Publ. Allen Press, Lawrence, Kansas,
pp. 202}206.
Fink, W.L., Weitzman, S.H., 1982. Relationships of the stomiiform "shes (Teleostei), with a description of
Diplophos. Bull. Mus. Comp. Zool. 150, 31}93.
Gill, A.C., 1997. Comments on an intercalar path for the glossopharyngeal (Cranial IX) nerve as
a synapomorphy of the Paracanthopterygii and on the phylogenetic position of the Gobiesocidae
(Teleostei: Acanthomorpha). Copeia 1997 (4), 1022}1029.
Hillis, D.M., Dixon, M.T., 1991. Ribosomal DNA: molecular evolution and phylogenetic inference. Q. Rev.
Biol. 66, 411}453.
E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350 349
Johnson, G.D., 1992. Monophyly of the euteleostean clades-Neoteleostei, Eurypterygii, and Ctenos-
quamata. Copeia (1) 1992, 8}25.
Johnson, G.D., Patterson, C., 1993. Percomorph phylogeny: a survey of acanthomorph characters and
a new proposal. Bull. Mar. Sci. 52, 554}626.
Johnson, G.D., Patterson, C., 1996. Relationships of lower euteleostean "shes. In: Stiassny, M.L.J., Parenti,
L.R., Johnson, G.D. (Eds.), The Interrelationships of Fishes. Academic Press, San Diego, California, pp.
251}331.
Kumar, S., Tamura, K., Nei, M., 1993. MEGA: Molecular Evolutionary Genetics Analysis (1.01). The
Pennsylvania State University, University Park, PA.
Kocher, T.T., Meyer, A., Edwards, S., Paabo, S., Villablanca, F., Wilson, A., 1989. Dynamics of mitochon-
drial DNA evolution in animals: Ampli"cation and sequencing with conserved primers. Proc. Natl.
Acad. Sci. USA 86, 6196}6200.
Lauder, G.V., Leim, K.F., 1983. The evolution and interrelationships of the actinopterygian "shes. Bull.
Comp. Mus. Zool. 150, 95}197.
Leviton, A.E., Gibbs, R.H., Heal, E., Dawson, C.E., 1985. Standards in hertelology and Ichthyology: Part I.
Standard symbolic codes for institutional resource collections in herpetology and Ichthyology. Copeia
1985, 802}832.
Maddison, D.R., 1991. The discovery and importance of multiple islands of most-parsimonious trees. Syst.
Zool. 40, 315}328.
Maniatis, T., Fristch, E.F., Sambrook, J., 1982. Molecular Cloning. A Laboratory Manual. Cold Spring
Harbor Laboratory, Cold Spring, NY.
Parenti, L.R., 1993. Relationships of atherinomorph "shes (Teleostei). Bull. Mar. Sci. 52, 170}196.
Patterson, C., Rosen, D.E., 1989. The Paracanthopterygii revisited: order and disorder. In: Cohen, D.M.
(Ed.), Papers on the Systematics of Gadiform Fishes. Natural History Museum of Los Angeles County,
Los Angeles, California, pp. 5}36.
Rosen, D.E., 1973. Interrelationships of higher euteleostean "shes. In: Greenwood, P.H., Miles, S., Patter-
son, C. (Eds.), Interrelationships of Fishes. J. Linn. Soc. (London) 53 Supplement 1. Academic Press,
New York, pp. 397}513.
Rosen, D.E., 1985. An essay on euteleostean classi"cation. Am. Mus. Novit. 2827, 1}57.
Saiki, R.K., 1990. Ampli"cation of genomic DNA. In: Innis, M.A., Gelfand, D.H., Sninsky, J.J., White, T.J.
(Eds.), PCR Protocols. A Guide to Methods and Applications. Academic Press, San Diego, California,
pp. 13}20.
Sorenson, M.D., 1996. TreeRot. University of Michigan, Ann Arbor, MI.
Starks, E.C., 1898. The osteology and relationships of the family Zeidae. Proc. U. S. Nat. Mus. 21, 469}476.
Stiassny, M.L.J., 1986. The limits and relationships of the acanthomorph "shes. J. Zool. London B 1,
411}460.
Stiassny, M.L.J., 1993. What are grey mullets? Bull. Mar. Sci. 52, 197}219.
Stiassny, M.L.J., Moore, J.A., 1992. A review of the pelvic girdle of acanthomorph "shes, with comments on
hypotheses of acanthomorph interrelationships. Zool. J. Linnean Soc. London 104, 209}242.
Swo!ord, D.L., 1993. PAUP: Phylogenetic Analysis Using Parsimony. Illinois Nat. Hist. Survey,
Champaign, IL.
Tang, K.L., Berendzen, P.B., Wiley, E.O., Morrissey, J.F., Winterbottom, R., Johnson, G.D., 1999. The
phylogenetic relationships of the suborder Acanthuroidei (Teleostei: Perciformes) based on molecular
and morphological evidence. Mol. Phylogenetics Evol. 11, 415}425.
Van der Peer, Y.I., Van den Broeck, De Rijk, P., de Wachter, R., 1994. Data base on the structure of small
subunit RNA. Nucl. Acids. Res. 22 (17), 3488}3494.
Wheeler, W., Gladstein, D., 1994. MALIGN 1.99. American Museum of Natrual History, New York.
Wiley, E.O., 1981. Phylogenetics. The Theory and Practice of Phylogenetic Systematics. Wiley-Interscience,
New York, 439pp.
Wiley, E.O., Johnson, G.D., Dimmick, W.W., 1998. The phylogenetic relationships of lampridiform "shes
(Lampridiformes, Acanthomorpha), based on a total evidence analysis of Morphological and Molecular
data. Mol. Phylogenetics Evol. 10 (3), 417}425.
350 E.O. Wiley et al. / Biochemical Systematics and Ecology 28 (2000) 319}350