? 2003 The Society for the Study of Evolution. All rights reserved.
Evolution, 57(6), 2003, pp. 1374–1386
THE DYNAMICS OF MALE BROODING, MATING PATTERNS, AND SEX ROLES IN
PIPEFISHES AND SEAHORSES (FAMILY SYNGNATHIDAE)
ANTHONY B. WILSON,1,2INGRID AHNESJO¨,3AMANDA C. J. VINCENT,4,5AND AXEL MEYER1,6
1Department of Biology, University of Konstanz, 78457 Konstanz, Germany
3Department of Animal Ecology, Evolutionary Biology Centre, Uppsala University, Norbyva ¨gen 18D, S-752 36 Uppsala, Sweden
4Department of Biology, McGill University, Montreal, Quebec H3A 1B1, Canada
for sexual selection. Seahorses and pipefishes (family Syngnathidae) are extraordinary among fishes in their remarkable
adaptations for paternal care and frequent occurrences of sex-role reversals (i.e., female-female competition for mates),
offering exceptional opportunities to test predictions of sexual selection theory. During mating, the female transfers
eggs into or onto specialized egg-brooding structures that are located on either the male’s abdomen or its tail, where
they are osmoregulated, aerated, and nourished by specially adapted structures. All syngnathid males exhibit this form
of parental care but the brooding structures vary, ranging from the simple ventral gluing areas of some pipefishes to
the completely enclosed pouches found in seahorses. We present a molecular phylogeny that indicates that the di-
versification of pouch types is positively correlated with the major evolutionary radiation of the group, suggesting
that this extreme development and diversification of paternal care may have been an important evolutionary innovation
of the Syngnathidae. Based on recent studies that show that the complexity of brooding structures reflects the degree
of paternal investment in several syngnathid species, we predicted sex-role reversals to be more common among
species with more complex brooding structures. In contrast to this prediction, however, both parsimony- and likelihood-
based reconstructions of the evolution of sex-role reversal in pipefishes and seahorses suggest multiple shifts in sex
roles in the group, independent from the degree of brood pouch development. At the same time, our data demonstrate
that sex-role reversal is positively associated with polygamous mating patterns, whereas most nonreversed species
mate monogamously, suggesting that selection for polygamy or monogamy in pipefishes and seahorses may strongly
influence sex roles in the wild.
Modern theory predicts that relative parental investment of the sexes in their young is a key factor responsible
Comparative method, mating patterns, molecular phylogenetics, paternal care, sex-role reversal, sexual
Received February 25, 2002.Accepted December 19, 2002.
In the vast majority of animals, the male’s sole contribution
to his offspring is his sperm (Trivers 1972). As a result, even
when the observed sex ratio of males to females is equal, the
operational sex ratio is often biased toward males and males
almost universally compete more strongly for mates, while
females typically exert greater mate choice (Darwin 1871;
Emlen and Oring 1977). Although most egg-laying fishes
leave their eggs unprotected after spawning, sole male care
is the predominant pattern in those species that care for their
young (Blumer 1982). Paternal care is likely to increase off-
spring fitness, but may reduce the father’s ability to invest
in other offspring (i.e., a parental investment sensu Trivers
Sex roles are defined by mating competition (Emlen and
Oring 1977). Sex-role reversal occurs when femalesprimarily
compete for access to mates, as compared to conventional
sex roles with male-male mating competition (cf. mating
roles, Gwynne 1991). Traditionally, the concept of parental
investment has been used empirically to predict sex roles and
intensities of mating competition (Trivers 1972). More recent
studies have predicted mating competition by the sexual dif-
ference in potential reproductive rates, in which the potential
2Present address: Genetics and Evolution, Conservation Biology
Division, Northwest Fisheries Science Center, 2725 Montlake Bou-
levard East, Seattle, Washington 98112; E-mail: tony.wilson@
5Present address: Project Seahorse, Fisheries Centre, University
of British Columbia, 2204 Main Mall, Vancouver, British Columbia
V6T 1Z4, Canada.
reproductive rate is the population’s mean of offspring pro-
duction when not constrained by the availability of mates
(Clutton-Brock and Vincent 1991; Clutton-Brock and Parker
1992; Kvarnemo and Ahnesjo ¨ 1996, 2002; Parker and Sim-
mons 1996; Ahnesjo ¨ et al. 2001). If male parental investment
reduces their potential reproductive rate below that of fe-
males, the operational sex ratio (Emlen and Oring 1977) may
become female-biased, resulting in a reversal of sex roles,
with females competing more strongly for mates (Clutton-
Brock and Parker 1992; Parker and Simmons 1996; Kvar-
nemo and Ahnesjo ¨ 1996, 2002; Parker and Simmons 1996).
Recent theoretical work also indicates that parental invest-
ment is a primary determinant of sex roles (Kokko and Mon-
aghan 2001; Kokko and Johnstone 2002). Variation in both
degree of paternal care and the occurrence of sex-role re-
versals in pipefishes and seahorses offers opportunities to
explore hypotheses concerning the relationship between pa-
rental investment, sex roles, and sexual selection (Trivers
1972; Parker and Simmons 1996).
The family Syngnathidae (pipefishes and seahorses) is
characterized by remarkable adaptations for paternal care.
The female deposits eggs directly onto a specialized brooding
area or into a pouch located under the abdomen or the tail
of the male (Breder and Rosen 1966). This evolutionary in-
novation ensures males complete confidence of paternity
(Jones and Avise 1997; Jones et al. 1998, 1999), but at a
level of paternal investment that exceeds that of most other
vertebrates (Breder and Rosen 1966). Thereafter the embryos
are nourished, osmoregulated, and protected during a lengthy
SEXUAL SELECTION IN SYNGNATHID FISHES
period of male pregnancy (Berglund et al. 1986, 1989; Vin-
cent et al. 1992). The brooding structures vary in complexity
in five steps, from: (1) a simple unprotected ventral area for
gluing, (2) individual membranous egg compartments, (3)
protection of eggs in a pouch with pouch plates, (4) bilateral
pouch folds that grow together into a closed pouch, to (5)
the most complex and completely enclosed brooding pouch
of seahorses (Dawson 1985). There is a further significant
difference among species in that brooding may occur on the
tail (Urophori: A-type) or on the abdomen (Gastrophori: B-
type; Herald 1959). Among the Gastrophori, brooding struc-
tures vary in complexity from step 1 to step 3 (Dawson 1985).
Parental investment as defined by Trivers (1972) is ex-
tremely difficult to estimate, involving the assessment of ul-
timate fitness costs from expenditures in different currencies
(time, energy, etc.). It is important to recognize that parental
expenditures are not necessarily equivalent to parental in-
vestment, because expenditures such as parental guarding
may not necessarily carry a fitness cost. However, in many
cases, time and energy expenditures may be positively cor-
related and, in general, a large expenditure will often carry
larger costs and therefore represent a higher parental invest-
ment. In the limited number of syngnathids studied to date,
males of species with less complex brooding structures (e.g.,
Nerophis ophidion) spend less energy on their young than do
those brooding embryos in enclosed pouches with placen-
talike structures (e.g., Hippocampus zosterae, Syngnathus ty-
phle; Berglund et al. 1986; Masonjones 2001). If increasing
pouch complexity results in a general increase in male pa-
rental investment relative to females, we would expect to see
more frequent sex-role reversals in species with more com-
plex brood pouches. Intense mating competition among fe-
males (i.e., sex-role reversal) should also favor the evolution
of sexual dimorphism, where females are larger and more
colorful than males. True to predictions, although some pipe-
fishes retain conventional sex roles (e.g., Hippichthys peni-
cillus), several species are sex-role reversed (e.g., Nerophis
ophidion, Stigmatopora nigra, Syngnathus typhle), with fe-
males that are more vividly colored and striped than males
(Table 1). One notable exception to this pattern is the genus
Hippocampus, in which, although these species have the high-
est degree of pouch development, sex-role reversal has not
yet been documented (Vincent and Sadler 1995; Kvarnemo
et al. 2000; Masonjones and Lewis 2000).
Microsatellite studies of genetic mating have revealed a
broad array of mating patterns in syngnathid fishes, ranging
from strict monogamy in Hippocampus subelongatus (Jones
et al. 1998) to various forms of multiple mating in many of
the pipefish species studied to date (Nerophis ophidion [Mc-
Coy et al. 2001]; Syngnathus floridae, S. scovelli, S. typhle
[Jones and Avise 2001]). Despite a methodological bias to-
wards the detection of multiple mating of male syngnathids
(McCoy et al. 2001), all mating patterns studied to date are
either polyandrous (multiple mating by females) or polyga-
mous (multiple mating by both males and females; Jones and
Avise 2001; McCoy et al. 2001), suggesting that access to
multiple mates may be common among female syngnathid
fishes. In species in which the potential reproductive rate of
females exceeds that of males, the operational sex ratio may
be biased towards females, intensifying female-female mat-
To reconstruct the evolution of male brooding structures
and study evolutionary patterns of sex-role reversal in the
Syngnathidae, we sequenced three mitochondrial genes from
a global sampling of syngnathid species, representing all ma-
jor male pouch types. If the degree of pouch complexity
accurately reflects relative paternal investment, theory (Triv-
ers 1972) predicts that sex-role reversal should be most prev-
alent in syngnathids with more complex brooding structures.
In addition to testing this hypothesis, we investigate the as-
sociation between mating patterns and sex roles in syngnathid
fishes (Vincent et al. 1992), explicitly incorporating phylo-
genetic data in a test of correlated trait evolution in the family
Syngnathidae. If mating patterns influence mating competi-
tion and thereby sexual selection (see above), we predict a
positive correlation between polygamy (where females fre-
quently show multiple matings) and sex-role reversals among
MATERIALS AND METHODS
Sample Collection, Polymerase Chain Reaction
Amplification and DNA Sequencing
We sampled a total of 48 individuals, including 44 spec-
imens from across the entire geographic range of the family
representing all major pouch types (Table 1). Morphological
work suggests that the Gasterosteiform and Syngnathiform
fishes are a monophyletic assemblage (Bowne 1984). With
this in mind, we included three outgroup species from the
sticklebacks (family Gasterosteidae; order Gasterosteifor-
mes): the threespine stickleback (Gasterosteus aculeatus;
New York), the blackspotted stickleback (Gasterosteus
wheatlandii; Rhode Island), and the ninespine stickleback
(Pungitius pungitius; Scotland, U.K.); and the Japanese tu-
besnout (Aulichthys japonicus; Kanagawa, Japan; family Au-
lorhynchidae; also within the order Gasterosteiformes; Bow-
ne 1984). Specimens were preserved in 70% ethanol and total
genomic DNA was extracted by proteinase K/SDS digestion
and purified by phenol-chloroform extraction and ethanol
precipitation (Kocher et al. 1989).
The polymerase chain reaction (PCR) was used to amplify
a 484-bp segment of the large subunit (16S) mitochondrial
ribosomal gene, a 352-bp segment of the small subunit (12S)
mitochondrial ribosomal gene, and the complete (1149 bp)
mitochondrial gene coding for cytochrome b: details of the
protocol and the primer sequences are published elsewhere
(Kocher et al. 1989; Palumbi et al. 1991; Pa ¨a ¨bo et al. 1991;
Taberlet et al. 1992; Wilson et al. 2001). DNA sequences
have been submitted to GenBank (accession numbers:
AF354940–AF355033, AF356040–AF356081, AF356539,
Sequence Alignment and Phylogenetic Reconstruction
The orthologous DNA sequences obtained were aligned,
using default settings, by Clustal W (Thompson et al. 1994)
and optimized by eye. Congruency of 12S, 16S, and cyto-
chrome b mitochondrial DNA datasets was tested with a like-
lihood-based congruency test (? ? 0.05; 10000 RELL boot-
ANTHONY B. WILSON ET AL.
TABLE 1. Syngnathid species, sampling locality (for study species), and pouch type with sex role (conventional: male-male competition for mates; reversed: female-female
competition for mates) and mating pattern from literature. Pouch type: A, tail brooder; B, abdominal brooder; 1–5, increasing pouch complexity (see text); e, everted pouch
folds; s, semi-inverted pouch folds; i, inverted pouch folds (only applicable to the A4 pouch type).
Hippocampus abdominalis Hippocampus barbouriHippocampus comes Hippocampus erectusHippocampus fuscus
A5 A5A5 A5A5
Conventional (Vincent 1994)
Monogamous (Perante et al. 2002)
Monogamous (Vincent 1994)
Hippocampus kudaHippocampus reidi Hippocampus subelongatusHippocampus whiteiHippocampus zosterae Hippocampus sp.
A5 A5 A5A5A5 A5
Conventional (Kvarnemo et al. 2000)Conventional (Vincent and Sadler 1995)Conventional (Masonjones and Lewis 2000)
Monogamous (Dauwe 1993; Nijhoff 1993)Monogamous (Jones et al. 1998) Monogamous (Vincent and Sadler 1995)Monogamous (Masonjones and Lewis 2000)
Syngnathus abasterSyngnathus acus Syngnathus floridaeSyngnathus fuscus Syngnathus leptorhynchus
A4iA4iA4i A4i A4i
Reversed (Fiedler 1954)Reversed (I. Ahnesjo ¨, pers. obs.)Reversed (Jones and Avise 2001)Reversed (Roelke and Sogard 1993)
Polygamous (Fiedler 1954) Polygamous (Vincent et al. 1995) Polygamous (Jones and Avise 2001)
Syngnathus louisianae Syngnathus rostellatusSyngnathus schlegeliSyngnathus scovelli Syngnathus taenionotusSyngnathus typhle
A4i A4i A4iA4i A4iA4i
Reversed (I. Ahnesjo ¨, pers. obs.)Reversed (Watanabe et al. 2000)Reversed (Jones and Avise 1997)
Reversed (Berglund et al. 1986, 1989)
Polygamous (Vincent et al. 1995)Polygamous (Watanabe and Watanabe 2001) Polygamous (Jones and Avise 1997)
Polygamous (Jones et al. 1999)
Hippichthys penicillusHyselognathus rostratus
Kuwait AustraliaAustralia Australia
Conventional (Watanabe et al. 1997)
Monogamous (Watanabe et al. 1997)
Pugnaso curtirostrisVanacampus phillipi Vanacampus poecilolaemus
Stigmatopora argus Stigmatopora nigra
Corythoichthys intestinalisCorythoichthys haemotopterus
AustraliaAustraliaAustralia Australia AustraliaGuam/Indonesia
A4sA4s A4sA4s A4sA4s A4s
Reversed (A. Kendrick, pers. comm.)1
Reversed (A. Kendrick, pers. comm.)1
Conventional (Gronell 1984) Reversed (Matsumoto and Yanagisawa 2001)
Monogamous (Gronell 1984) Monogamous (Matsumoto and Yanagisawa 2001)
Phycodorus equesPhyllopteryx taeniolatus
A2 A2 A2
Conventional (P. Groves, pers. comm.)2
Monogamous (P. Groves, pers. comm.)2
Dunckerocampus dactyliophorusDoryrhamphus excisus Doryrhamphus japonicusOosthethus brachyrus
B1 B1B2B3 B3B3
Reversed (I. Ahnesjo ¨, pers. obs.)Reversed (Berglund et al. 1986, 1989)
Conventional (A. Gronell, pers. comm.)3
Polygamous (Vincent et al. 1995)Polygamous (McCoy et al. 2001)
Monogamous (Kuiter 2001) Monogamous (Kuiter 2001) Monogamous (A. Gronell, pers. comm.)3
1Alan Kendrick, Fish Research Group, Murdoch University, Murdoch, Australia.
2Paul Groves, Hilarys, Australia.
3Ann Gronell, CSIRO, Hobart, Australia.
SEXUAL SELECTION IN SYNGNATHID FISHES
strap replicates; Waddell et al. 2000), using maximum like-
lihood (ML) topologies generated from individual gene anal-
yses as well as the overall ML tree (likelihood parameters
estimated by Modeltest Ver. 3.06; Posada and Crandall
1998). Free parameters (including branch lengths) were re-
optimized for each dataset on all topologies tested.
Neighbor-joining (NJ) distance and maximum parsimony
(MP) analyses were performed with PAUP Ver. 4b10 (Swof-
ford 2000), with indels coded as missing data. Maximum-
parsimony trees were constructed with full heuristic searches
(500 bootstrap replicates) using random addition (10 repli-
cates), the tree-bisection-reconnection (TBR) branch-swap-
ping algorithm and the MULPARS option. For parsimony
analyses, a transversion/transition weighting of two was used.
Neighbor-joining analyses (500 bootstrap replicates) applied
a HKY model of substitution (Hasegawa et al. 1985), with
transition/transversion ratio (2.22), gamma shape parameter
(0.57), proportion of invariable sites (0.41), and nucleotide
frequencies (A, 0.3092; C, 0.2413; G, 0.1519; T, 0.3031)
estimated from the dataset using Modeltest (Posada and Cran-
Bayesian methods of phylogenetic inference, incorporating
posterior probabilities, were calculated using MrBayes Ver.
2.01 (Huelsenbeck and Ronquist 2001), with parameters es-
timated from the dataset under the HKY ? I ? ? model. A
(MCMCMC) was run for 100,000 generations starting from
a random tree (sampling every 25 generations) using four
heated chains (temp ? 0.2). The MCMCMC reached apparent
stationarity at 20,000 generations and trees sampled during
this burn-in of the model were not used in further analyses.
Bayesian analyses were replicated in four runs with identical
The overall ML tree was estimated following a heuristic
search on starting tree topologies generated by distance, par-
simony and Bayesian analyses. Initial branch lengths were
estimated using Rogers-Swofford approximation and opti-
mized following Newton-Raphson optimization (100 branch-
length smoothing passes, ? ? 0.000006). Branch swapping
followed the TBR algorithm. The optimal ML tree was iden-
tical in each of the three replicates.
To investigate whether rates of sequence evolution in the
Syngnathidae deviate from the neutral expectation of clock-
like behavior, the likelihood of the ML tree was recalculated
with the constraint of a molecular clock (Rambaut parame-
terization for clock optimization). A comparison of the mo-
lecular clock tree with the unconstrained topology was tested
with a likelihood-ratio test (LRT). Finally, to determine over-
all support for the ML tree, well-supported internal groupings
were shuffled and the deviation of these alternativetopologies
from the ML tree was tested using a likelihood-based Shi-
modaira-Hasegawa (SH) test (Shimodaira and Hasegawa
Carlo Markov Chain
Reconstruction of Character State Evolution
Prior to the reconstruction of ancestral character states,
duplicated taxa were pruned from the dataset, parameters
were re-estimated using Modeltest (Posada and Crandall
1998) and the ML topology was recalculated as outlined
To investigate directional evolution of brooding structures
in the Syngnathidae, Multi-state Ver. 0.6 (Pagel 1999a) was
used to estimate ancestral character states at each node in-
dependently for Urophori and Gastrophori lineages based on
brood-pouch morphology for extant taxa (Table 1). A two-
parameter forward-backward model of character evolution
incorporating independent forward and backward rates of
character transformation was initially tested against an un-
constrained model of character change. Ancestral character
states were estimated by optimizing the overall ML on the
phylogeny with individual nodes fixed independently for all
possible character states. Preference for one ancestral state
over another was deemed significant if the difference in their
log-likelihoods was greater than two (Pagel 1999a).
A gradual model of evolution predicts that evolutionary
change should accrue slowly and uniformly over the phy-
logenetic history of a group of organisms. Deviations from
this model suggest that character change has been punctuated,
possibly indicating lineage-specific selection pressures (Pa-
gel 1999b). To quantify the rate of brood pouch evolution,
the likelihood of an unconstrained model incorporating a
scaling parameter was compared to a model where the scaling
parameter was fixed at 0 (punctuated rate of trait change) for
both the Urophori and Gastrophori using Multi-state (Pagel
1999a). The significance of this comparison was tested using
the likelihood ratio statistic: ?2[loge(H0/H1)] ? ?2, df ? 1.
Available empirical data on both sex roles and mating pat-
terns for each species of syngnathid included in our phylo-
genetic analyses (Table 1) were mapped onto our molecular
phylogeny using the accelerated transformation(ACCTRAN)
feature in MacClade Ver. 3.08a (Maddison and Maddison
1992) in order to determine the most parsimonious character
states for each of taxa for which data were unavailable. To
test for monophyly of sex roles in the Syngnathidae, the ML
tree was recalculated, constraining sex-role reversed, non-
role reversed, and outgroup species (Table 1) to be mono-
phyletic groups. This topology was compared with that of
the unconstrained ML tree using a SH test (Shimodaira and
A likelihood-ratio test incorporating branch lengths im-
plemented by DISCRETE Ver. 4.0 (Pagel 1994a) tested the
correlation between brood pouch development and sex-role
reversal in the family Syngnathidae independently for the
Urophori and Gastrophori. Six- and four-state pouch vari-
ables were collapsed into a series of binary traits for this
comparison as suggested by Pagel (1994b). Sex roles were
characterized as conventional (i.e., predominantly male-male
competition for mates) or reversed (i.e., predominant female-
female competition for mates). Mating pattern was classified
as either monogamous (socially and genetically) or polyga-
mous (including both polygamy, in which both males and
females mate multiple times for a single brood or clutch; and
polyandry, in which only females mate multiple times for a
single breeding event.). A comparison of the log-likelihood
of an independent model of trait evolution (four parameters)
to that of a dependent model of trait evolution (eight param-
eters) tested the correlation between brood pouch complexity,
sex-role reversal, and mating patterns within the Syngnath-
ANTHONY B. WILSON ET AL.
SEXUAL SELECTION IN SYNGNATHID FISHES
and Berlocher 1987) and Bayesian phylogenetic analyses (Huelsenbeck and Ronquist 2001) based on the combined dataset of 1603 bp
with branch lengths as estimated from maximum-likelihood analysis. Numbers on branches represent bootstrap values from distance and
parsimony analyses and consensus support for Bayesian analysis. Archived samples (S accession numbers listed after species names)
and data on specimen collections are housed at the Evolutionary Biology Centre (Uppsala Univ.). For descriptions of pouch types, see
text and Table 1. Doryrhamphus and Dunckerocampus are recognized here as separate genera as suggested by Kuiter (1998). Photographs
reprinted with permission from Kuiter (2001), top to bottom: Hippocampus bleekeri (Rudie Kuiter); Syngnathus acus (Charles Hood);
Phyllopteryx taeniolatus, Stigmatopora nigra (Rudie Kuiter); Corythoichthys spp. (Rudie Kuiter); Doryrhamphus excisus (Phil Woodhead);
Gasterosteus aculateus (Rudie Kuiter); Aulichthys japonicus (Tomonori Hirata).
Consensus phylogenetic tree constructed from neighbor-joining distance (Saitou and Nei 1987), maximum-parsimony (Swofford
idae. Significance values for phylogenetic tests for correlation
(nine tests) were adjusted following sequential Bonferroni
correction (Rice 1989).
Molecular Phylogeny of Syngnathid Fishes
Mitochondrial sequences of 12S rDNA, 16S rDNA, and
cytochrome b were collected, collated and aligned for 48, 45,
and 40 specimens, respectively, resulting in a total sequence
length of up to 1985 bp per specimen. Analyses of cyto-
chrome b sequence data revealed third codon saturation of
transitions for Kimura-2-parameter distances greater than
0.20 (data not shown). A likelihood-based congruency test
did not reject the congruency of the dataset (P ? 0.352 for
overall ML topology). Subsequent analyses were based on
up to 1603 bp of sequence data for each individual.
Concatenated DNA sequences analyzed with NJ distance,
MP and Bayesian phylogenetic analyses resulted in identical
topologies for major groups of syngnathid fishes (Fig. 1). All
three analyses support monophyly of the Urophori (tail
brooding) and Gastrophori (abdominal brooding) lineages of
syngnathid fishes. While phylogenetic resolution was low at
the base of the Urophori, monophyly of Solegnathus and Phyl-
lopteryx (single egg membrane compartments type A2), Stig-
matopora and Corythoichthys (semi-inverted pouch type
A4s), Syngnathus (inverted pouch type A4i), and Hippocam-
pus (completely enclosed pouch type A5) were all supported
by our multigene analyses (Fig. 1).
A ML model incorporating lineage-specific rates of se-
quence divergence fit the sequence data significantly better
than a model constrained to fit a molecular clock (free model
?ln L ? ?15448.35; constrained model ?ln L ? ?15585.63;
?2? 137.28; df ? 46; P ? 0.001). Subsequent analyses did
not enforce a molecular clock constraint.
A Shimodaira-Hasegawa test supported the monophyly of
the Gastrophori and Urophori (P ? 0.05 for all topologies
tested), but polytomic relationships among well-supported
internal groupings within the Urophori could not be rejected
as significantly different from the ML topology (Table 2).
Although variation in pouch type complexity is evident in a
group of syngnathid pipefishes endemic to Australia (A4e
and A4s pouch types in Urocampus/Vanacampus clade; Fig.
1), specific brooding structures are generally restricted to the
monophyletic groups outlined above (see Fig. 1).
Reconstruction of Brooding Structure Evolution
The evolution of brooding structures within the Gastro-
phori and Urophori was reconstructed using both MP and
ML methods. Parsimony reconstruction suggests six changes
in brooding structures within the Urophori and three changes
within the Gastrophori (Fig. 2). ML reconstructions using a
single forward-backward transition rate did not significantly
differ from those with independent transition parameters
(Urophori: free model ?ln L ? ?19.2988; constrained model
?ln L ? ?24.0225; ?2? 9.4474; df ? 28; P ? 1.000;
Gastrophori: free model ?ln L ? ?5.7829; constrained mod-
el ?ln L ? ?7.5201; ?2? 3.4744; df ? 10; P ? 0.968).
Subsequent analyses were conducted using the two-parameter
forward-backward transition model.
Ancestral character states, incorporating branch length in-
formation, were estimated using Multi-state (Pagel 1999a).
Maximum-likelihood analyses could not reject a directional
mode of evolution of brooding structures in either the Uro-
phori or Gastrophori (Fig. 2). Although a punctuated model
of brood structure evolution could not be rejected for the
Gastrophori (?2? 2.41, P ? 0.120), brood structureevolution
within the Urophori approximates a gradual model of evo-
lution (?2? 7.97, P ? 0.005; scaling parameter ? 3.49).
Phylogenetic Reconstruction of Sex Role and Mating
To test the predicted positive association between male
parental care (as measured by brood structure complexity)
and sex-role reversal, we used parsimony and ML approach-
es. Mapping sex roles on the pruned ML tree suggests at least
four changes in sex roles within the Syngnathidae (Fig. 3a).
The log-likelihood estimate of the topology estimated with-
out the constraint of sex-role monophyly is significantly bet-
ter than that estimated with sex-role-reversed and nonrole-
reversed species constrained as monophyletic lineages (Table
2; SH-test ln-likelihoods: unconstrained ?14667.84 vs. con-
strained ?14867.35; ? ? 199.51; significant at P ? 0.001).
A likelihood-ratio test (LRT) failed to support a relation-
ship between sex roles and male pouch development within
either the abdominal or tail-brooding lineages (Table 3; P ?
0.05 for all comparisons after Bonferroni correction). Al-
though our phylogeny indicates that male brooding structures
have been highly conserved over the evolutionary history of
the group (Fig. 1), sex roles appear to have reversed multiple
times, independent of pouch type variation within the family
Parsimony reconstruction of mating pattern evolution on
the ML tree indicates that four changes in patterns of mating
have occurred within the Syngnathidae (Fig. 3b). We tested
for a correlation between sex-role reversal and mating pat-
terns in syngnathid fishes. We performed a likelihood-ratio
test, mapping ecological data on sex roles and mating patterns
ANTHONY B. WILSON ET AL.
TABLE 2. Shimodaira-Hasegawa test of alternate topologies (Shimodaira and Hasegawa 1999). Maximum-likelihood tree, log-likelihood
differences, and P-values for alternative topologies tested. Lmax, maximum likelihood; L?, likelihood of tree ?. Significantly different
topology at *P ? 0.05, **P ? 0.001. Groupings tested (pouch type; for legend see Table 1): 1, Hippocampus spp. (A5); 2, Syngnathus
spp. (A4i); 3, Australian endemics (A4e, A4s); 4, Solegnathus/Phyllopteryx (A2); 5, Corythoichthys (A4s); 6, Stigmatopora (A4s); 7,
Oostethus/Doryrhamphus/Dunckerocampus (B2, B3); 8, Nerophis/Entelurus (B1); 9, outgroup.
9 (ML pruned tree)
10 (Sex-role reversal monophyly)
(((1,2), (3, 4)), (5, 6)), (7, 8)), 9
((((Hippichthys, 1, 2), (3, 4)), (5, 6)), (7, 8)), 9
(((1, 2), (3, 4), (5, 6)), (7, 8)), 9
((((1, 2, (3, 4)), (5, 6)), (7, 8)), 9
(((1, 2, (3, 4), (5, 6)), (7, 8)), 9
((((1, 2), (7, 8)), (5, 6)), (3, 4)), 9
((((1, 2), (7, 8)), (3, 4)), (5, 6)), 9
((1, 2), (3, 4), (5, 6), (7, 8)), 9
(((1, 2), (3, 4)), (5, 6)), (7, 8)), 9
((conventional, sex-role reversed), outgroup)0.000**
collected from the literature and reconstructed from the par-
simony analysis onto our ML phylogeny (Fig. 3b). Results
were significant (LRT: ?2? 19.3234, df ? 4, P ? 0.0007),
demonstrating a strong positive association betweensex-roles
and mating patterns, in which monogamous species have con-
ventional patterns of mating and polygamous species are gen-
erally sex-role reversed.
Our molecular data demonstrate an extensive morpholog-
ical divergence in lineages with diverse and highly special-
ized adaptations for paternal care. In syngnathids, the evo-
lution of increased pouch development has occurred in par-
allel in both abdominal and tail-brooding lineages, leading
to the remarkable diversity of male brooding structures ob-
served today. At the same time, multiple evolution of sex-
role reversal within the family Syngnathidae suggests that
pouch type variation may not directly reflect relative parental
investment, or that the relationship between pouch type and
the sexual difference in potential reproductive rates (and
thereby the operational sex ratio) is more dynamic rather than
being a direct consequence of their brooding structure. In
contrast, the strong association between sex roles and mating
patterns in syngnathid species highlights the relationship be-
tween polygamous mating, in which multiple mating may be
more frequent among females than males, and sex-role re-
Mitochondrial Phylogeny Supports Parallel Evolution of
Major Pouch Lineages
In accordance with the assumed evolutionary significance
of male brooding structures, phylogenetic analyses indicate
that distinct pouch types each generally represent monophy-
letic lineages of species. Our total molecular evidence dataset
indicates that most major pouch types arose once early in the
evolution of syngnathid fishes, and punctuated evolution of
brooding structures in abdominal brooding pipefishes sug-
gests that morphological radiation may have occurred in a
relatively short burst of morphological change in this group
(Fig. 1). Our molecular data also support Herald’s (1959)
classification of these fishes into tail- and abdominal brood-
ers, demonstrating the independent radiations of morpholog-
ical structures in the Urophori (A-type: tail pouch) and Gas-
trophori (B-type: abdominal pouch). Our phylogeny suggests
that the primary split between these two lineages occurred
at the same time or shortly before the major morphological
radiation of male brooding structures and subsequent radi-
ation of species (Wilson et al. 2001). The early diversification
of the ancestral syngnathid into tail and abdominal brooders
is consistent with results from a karyotypic study, which
highlights a possible total-genome duplication in the abdom-
inal-brooding lineage (Vitturi et al. 1998). Brooding struc-
tures within these two lineages independently increased in
complexity, culminating in the completely enclosed brood
pouches located on the tail of seahorses (Herald 1959; type
A5) and the well-defined abdominal pouch of Oostethus bra-
chyrus (type B3), the most complex abdominal pouch type.
Repeated Shifts in Sex Roles
Detailed behavioral and ecological studies of a subset of
syngnathid species have revealed substantial variation in pat-
terns of sex roles in the family (Vincent et al. 1992). The
lack of association between sex roles and pouch type vari-
ation demonstrated in our study suggests that the relationship
between parental care and sex roles in these fishes may not
be as straightforward as predicted. Our conclusions are based
on the assumption that pouch type variation accurately re-
flects relative parental investment. Although it is likely that
paternal energy expenditures are positively correlated with
increasing pouch complexity (Berglund et al. 1986; Mason-
jones 2001), the assessment of true parental investment (sen-
su Trivers 1972) may be more complex, involving the ulti-
mate fitness costs of both energy and time expenditures in
both sexes. Independent of our assumptions, however, it is
clear that it is not possible to map sex roles on the consensus
phylogeny of syngnathids without allowing multiple ap-
pearances of role reversal (Fig. 3a; see results of SH test
The lack of correlation between sex roles and brooding
structures in our study may be partly explained by results
from several recent studies, which demonstrate that sex roles
and intensity of competition for mates can be predicted by
the differences in potential reproductive rates between the
sexes, which are influenced by environmental factors such as
SEXUAL SELECTION IN SYNGNATHID FISHES
FIG 2. Most parsimonious reconstruction of the evolution of brooding structures in the (a) Urophori and (b) Gastrophori mapped on the
pruned maximum-likelihood phylogenetic trees for each group using MacClade Ver. 3.08a (Maddison and Maddison 1992). Maximum-
likelihood ancestral character states were also estimated at each node using Multi-state Ver. 0.6 (Pagel 1999a). Possible character states
at nodes in which more than a single ancestral state is equally likely are indicated in parentheses.
ANTHONY B. WILSON ET AL.
(Fig. 1) using MacClade Ver. 3.08a (Maddison and Maddison 1992) with empirical data coded as boxes where available. Character states
Most parsimonious reconstruction of (a) sex-role reversal and (b) mating pattern mapped on the consensus phylogenetic tree
SEXUAL SELECTION IN SYNGNATHID FISHES
for Doryrhamphus excisus, Dunckerocampus dactyliophorus, and Phyllopteryx taeniolatus derived from closely related taxa. Character
states for outgroups from references in McLennan et al. (1988). The most parsimonious reconstruction of these characters on this phylogeny
involves multiple independent origins of sex roles (four steps) and mating patterns (three steps) within the Syngnathidae. Sex roles are
uncorrelated with male brood pouch development (LRT: P ? 0.05 for all comparisons; Table 3). In contrast, there is a strong correlation
between sex roles and mating pattern in the Syngnathidae (LRT: P ? 0.001).
temperature and food availability (Kvarnemo and Ahnesjo ¨
1996, 2002; Ahnesjo ¨ et al. 2001). In sand gobies (Pomatos-
chistus minutus; Kvarnemo 1994), temperature influences
sexual differences in the potential reproductive rate, causing
variation in the intensity of mating competition. Mating com-
petition in the sex-role reversed pipefish Syngnathus typhle
is also strongly influenced by fluctuations in temperature(Ah-
nesjo ¨ 1995). During a warm breeding season, males have
shorter pregnancies and are available more frequently for
mating than under colder ambient water temperatures (Vin-
cent et al. 1994). Even though females always compete for
access to mates in Syngnathus typhle, they will compete even
more intensely when temperatures are reduced, due to a more
female-biased operational sex ratio caused by the extended
period of male pregnancy (Vincent et al. 1994; Ahnesjo ¨
Sex roles may also shift depending on food availability as
demonstrated in the bush cricket (Gwynne and Simmons
1990; Simmons 1992). While the energetic intake and re-
productive investment of female bush crickets on a limited
diet was significantly lower than that for those allowed un-
limited access to food, no significant difference in either en-
ergetic intake or reproductive investment could be detected
in males raised under same the two food regimes (Gwynne
and Simmons 1990; Simmons 1992). As a result, male pa-
rental investment in bush crickets became even more im-
portant when food resources were scarce and female com-
petition for mates was intensified under these conditions
(Gwynne and Simmons 1990; Simmons 1992).
Given that sex roles can shift and the intensity of mating
competition can vary in relation to environmental factorsboth
within and between species, one might expect variation in
these characters independent of phylogenetic relationships.
However, in the Syngnathidae, although a strict correlation
between sex-role reversal and pouch development is rejected,
there is, at the same time, an intriguing consistency in the
similarity of sex roles and pouch type at higher taxonomic
levels (e.g., within the genera Syngnathus and Hippocampus;
Table 1; Figs. 1, 3a).
Correlated Evolution of Sex Roles and Mating Patterns
Although male pouch development does not predict the
presence of sex-role reversal in the Syngnathidae, analysis
of our data highlights a strong correlation between sex-role
reversal and mating patterns in syngnathid species, as sug-
gested by Vincent et al. (1992). Syngnathids exhibiting sex-
role reversal are generally polygamous, whereas thosespecies
with nonreversed patterns of mating competition are monog-
amous. Among seahorses (Hippocampus spp.), the monog-
amous mating pattern is well documented both socially (Vin-
cent and Sadler 1995; Kvarnemo et al. 2000; Masonjones and
Lewis 2000; Perante et al. 2002) and genetically (Jones and
Avise 2001) and social monogamy has been confirmed among
Corythoichthys spp. pipefishes (Gronell 1984; Matsumoto and
Yanagisawa 2001). In both of these genera, the female trans-
fers a full clutch to her mate, either into the male’s brooding
pouch (seahorses) or as a sheet of eggs onto the brooding
structure of Corythoichthys spp.
Polygamy occurs in the genus Syngnathus, where both
males and females mate multiple times during a single breed-
ing period. As a result, males brood eggs from severalfemales
concurrently (Berglund et al. 1989; Vincent et al. 1995; Jones
and Avise 2001). In contrast, in species with more simple
brooding structures, males receive their entire brood from a
female during a single mating (e.g., Nerophis and Entelurus)
but females are capable of mating with additional males well
before the previous male has finished brooding the clutch
(i.e., polyandry; Berglund et al. 1989; Vincent et al. 1995).
The pattern of monogamous species showing conventional
sex roles has, so far, only a single exception. The pipefish
Corythoichthys haematopterus clearly mates monogamously,
but in a wild population where the operational sex ratio was
consistently skewed toward an excess of females, behavioral
observations indicated female-female competition (Matsu-
moto and Yanagisawa 2001).
The association between sex roles and mating patterns is
intriguing and its causes, constraints, and consequences war-
rant further attention. One may, however, speculate on eco-
logical differences associated with the pattern. Several of the
polygamous and sex-role reversed species live in eelgrass
habitats and tend to occur at relatively high densities, have
no home ranges, and are comparatively mobile (e.g., Syng-
nathus typhle, S. rostellatus, S. scovelli and Nerophis ophi-
dion; Vincent et al. 1995; Jones and Avise 1997). In contrast,
monogamous species with a conventional mating pattern tend
to occur at lower densities, have reduced mobility, and fixed
home ranges (i.e., Corythoichthys spp., Hippocampus spp.;
Gronell 1984; Vincent and Sadler 1995; Kvarnemo et al.
2000; Perante et al. 2002). These differences in population
densities may influence opportunities for remating. A similar
ecological association has been found in polyandrous birds,
where species with male-only care are generally found at
lower densities compared to species with female-only care
Although a skewed operational sex ratio or differences in
quality of potential mates may create competition for mates
in monogamous species, considerable mating competition in
polygamous animals generally results in more extensive sex-
ual dimorphisms as compared to monogamous species (An-
dersson 1994). When a polygamous mating pattern results in
more frequent opportunities for female multiple mating (i.e.,
polyandry), intensified mating competition among females
may lead to a reversal of conventional sex roles. This close
correlation between sex-role reversal and polyandry has also
ANTHONY B. WILSON ET AL.
TABLE 3. Comparison of independent versus dependent (correlated evolution) maximum-likelihood models (Pagel 1994a) of the evolution
of brooding structures and sex roles within the Urophori and Gastrophori. Brooding structures were sequentially dichotomized to facilitate
testing with DISCRETE Ver. 4.0 (Pagel 1994a). Significance of these comparisons was tested with a likelihood ratio test (df ? 4). For
pouch type legend, see Table 1.
Character 0 Character 1IndependentDependent
no pouch (outgroup)
no pouch (outgroup)
been identified in several species of the avian orders Grui-
formes and Charadriiformes (Emlen and Oring 1977), high-
lighting the potential significance of mating patterns on sex
roles in natural populations.
In reconstructing complex ecological traits, there is a pos-
sibility of introducing systematic error due either to biased
taxonomic sampling and/or lack of data for taxa included in
the study. Although our molecular phylogenetic study in-
cludes the widest taxonomic sampling of syngnathid species
presented to date, and syngnathid mating behaviors are ar-
guably some of the best studied, ongoing ecological studies
on syngnathid fishes continue to be critical to further explore
the findings presented here. The low level of statistical power
in such reconstructions also decreases the possibility to detect
significant correlations between ecological variables, adding
additional confidence to the significant correlation observed
between sex-role reversal and polygamy, but suggesting that
the identification of more frequent transitions of ecological
characters on the syngnathid phylogeny could potentially lead
to significant correlations between pouch development and
sex-role reversal. Additional behavioral study of syngnathid
fishes and a more highly resolved molecular phylogeny of
the family Syngnathidae will clearly help to address these
We thank M. S. Abdullah, M. A. Bell, W. Chan, C. Dayton,
P. Franzoi, J. de Greef, A. Jordan, A. Kendrick, the Klubbans
Biological Station, H. Lessios, F. Leung, C. Linaker, H. Ma-
sonjones, R. E. Matheson, G. Orti, D. Reznick, D. R. Rob-
ertson, R. Ruiz-Carus, R. L. Teixeira, and Y. Yanagisawa for
assistance with collections. Thanks to J. Garcı ´a-Moreno, M.
Pagel, R. Ota, and P. J. Waddell for constructive discussions.
Special thanks to T. Clutton-Brock, D. Hosken, L. Kvarnemo,
M. Ah-King, S. A. Lourie, and especially B. Crespi for help-
ful feedback on manuscript drafts. We are grateful to J. M.
Morrissey, P. J. Perl, and P. A. Ritchie for technical assis-
tance and to S. A. Lourie for assistance with identification
of Hippocampus material. Cover photos were provided cour-
tesy of R. Kuiter. This work was funded by a Canadian Na-
tional Sciences Engineering and Research Council PGSB
scholarship and NRC Research Associateship to ABW, grants
from the Swedish Natural Science Research Council to IA,
the Ernest Cook Research Fellowship from Somerville Col-
lege, Oxford, and the Darwin Senior Research Fellowship to
ACJV, and research grants from the Deutsche Forschungs-
gemeinschaft, the U.S. National Science Foundation and the
Fond der Chemischen Industrie to AM.
Ahnesjo ¨, I. 1995. Temperature affects male and female potential
reproductive rates differently in the sex-role reversed pipefish,
Syngnathus typhle. Behav. Ecol. 6:229–233.
Ahnesjo ¨, I., C. Kvarnemo, and S. Meriliata. 2001. Using potential
reproductive rates to predict mating competition among indi-
viduals qualified to mate. Behav. Ecol. 12:397–401.
Andersson, M. 1994. Sexual selection. Princeton Univ. Press,
Berglund, A., G. Rosenqvist, and I. Svensson. 1986. Reversed sex
roles and parental energy investment in zygotes of two pipefish
(Syngnathidae) species. Mar. Ecol. Prog. Ser. 29:209–215.
———. 1989. Reproductive success of females limited by males
in two pipefish species. Am. Nat. 133:506–516.
Blumer, L. S. 1982. A bibliography and categorization of bony
fishes exhibiting parental care. Zool. J. Linn. Soc. 76:1–22.
Bowne, P. S. 1984. Systematics and morphology of the Gasteros-
teiformes. Pp. 28–60 in M. A. Bell and S. A. Foster, eds. The
evolutionary biology of the threespine stickleback. Oxford Univ.
Press, Oxford, U.K.
Breder, C. M., and D. E. Rosen. 1966. Modes of reproduction in
fishes. Natural History Press, New York.
Clutton-Brock, T. H., and G. A. Parker. 1992. Potential reproductive
rates and the operation of sexual selection. Q. Rev. Biol. 67:
Clutton-Brock, T. H., and A. C. Vincent. 1991. Sexual selection
and the potential reproductive rates of males and females. Nature
Darwin, C. 1871. The descent of man, and selection in relation to
sex. Murray, London.
Dauwe, B. 1993. Ecology of the seahorse Hippocampus reidi on
the coral reefs of Bonaire (N.A.): habitat use, reproduction and
interspecific interactions. M. Sc. thesis. Rijksuniversiteit, Gron-
ingen, The Netherlands.
Dawson, C. E. 1985. Indo-pacific pipefishes (Red Sea to the Amer-
icas). Gulf Coast Research Laboratory, Ocean Springs, MS.
Emlen, S. T., and L. W. Oring. 1977. Ecology, sexual selection,
and the evolution of mating systems. Science 197:215–223.
Fiedler, K. 1954. Vergleichende verhaltensstudien an seenadeln,
schlangennadeln und seepferdchen (Syngnathidae). Z. Tierp-
Gronell, A. M. 1984. Courtship, spawning, and social organization
SEXUAL SELECTION IN SYNGNATHID FISHES
of the pipefish, Corythoichthys intestinalis (Pisces: Syngnathi-
dae) with notes on two congeneric species. Z. Tierpsychol. 65:
Gwynne, D. T. 1991. Sexual competition among females—what
causes courtship role reversal? Trends Ecol. Evol. 6:118–121.
Gwynne, D. T., and L. W. Simmons. 1990. Experimental reversal
of courtship roles in an insect. Nature 346:172–174.
Hasegawa, M., H. Kishino, and T. Yano. 1985. Dating of the human-
ape splitting by a molecular clock of mitochondrial DNA. J.
Mol. Evol. 22:160–174.
Herald, E. S. 1959. From pipefish to seahorse—a study of phylo-
genetic relationships. Proc. Calif. Acad. Sci. 29:465–473.
Huelsenbeck, J. P., and F. Ronquist. 2001. MrBayes: Bayesian in-
ference of phylogenetic trees. Bioinformatics 17:754–755.
Jones, A. G., and J. C. Avise. 1997. Microsatellite analysis of ma-
ternity and the mating system in the Gulf pipefish Syngnathus
scovelli, a species with male pregnancy and sex-role reversal.
Mol. Ecol. 6:203–213.
———. 2001. Mating systems and sexual selection in male-preg-
nant pipefishes and seahorses: insights from microsatellite-based
studies of maternity. J. Hered. 92:150–158.
Jones, A. G., C. Kvarnemo, G. I. Moore, L. W. Simmons, and J.
C. Avise. 1998. Microsatellite evidence for monogamy and sex-
biased recombination in the Western Australian seahorse Hip-
pocampus angustus. Mol. Ecol. 7:1497–1505.
Jones, A. G., G. Rosenqvist, A. Berglund, and J. C. Avise. 1999.
The genetic mating system of a sex-role-reversed pipefish (Syng-
nathus typhle): a molecular inquiry. Behav. Ecol. Sociobiol. 46:
Kocher, T. D., W. K. Thomas, A. Meyer, S. V. Edwards, S. Pa ¨a ¨bo,
F. X. Villablanca, and A. C. Wilson. 1989. Dynamics of mito-
chondrial DNA evolution in animals: amplification and sequenc-
ing with conserved primers. Proc. Natl. Acad. Sci. USA 86:
Kokko, H., and R. A. Johnstone. 2002. Why is mutual mate choice
not the norm? Operational sex ratios, sex roles and the evolution
of sexually dimorphic and monomorphic signalling. Philos.
Trans. R. Soc. Lond. B 357:319–330.
Kokko, H., and P. Monaghan. 2001. Predicting the direction of
sexual selection. Ecol. Lett. 4:159–165.
Kuiter, R. H. 1998. Pipefishes of the syngnathid genus Dunckero-
campus (Syngnathiformes: Syngnathidae), with a description of
a new species from the Indian Ocean. Aqua: J. Ichthylol. Aquat.
Kuiter, R. H. 2001. Syngnathiformes pictorial guide. Zoonetics,
Kvarnemo, C. 1994. Temperature differentially affects male and
female reproductive rates in the sand goby: consequences for
operational sex ratio. Proc. R. Soc. Lond. B 256:151–156.
Kvarnemo, C., and I. Ahnesjo ¨. 1996. The dynamics of operational
sex ratios and competition for mates. Trends Ecol. Evol. 11:
———. 2002. Operational sex ratios and mating competition. Pp.
366–382 in I. Hardy, ed. Sex ratio handbook. Cambridge Univ.
Press, Cambridge, U.K.
Kvarnemo, C., G. I. Moore, A. G. Jones, W. S. Nelson, and J. C.
Avise. 2000. Monogamous pair bonds and mate switching in the
Western Australian seahorse Hippocampus subelongatus. J. Evol.
Maddison, W. P., and D. R. Maddison. 1992. MacClade Ver. 3.
Analysis of phylogeny and character evolution. Sinauer Asso-
ciates, Sunderland, MA.
Masonjones, H. D. 2001. The effect of social context and repro-
ductive status on the metabolic rates of dwarf seahorses (Hip-
pocampus zosterae). Comp. Biochem. Physiol. 129A::541–555.
Masonjones, H. D., and S. M. Lewis. 2000. Differences in potential
reproductive rates of male and female seahorses related to court-
ship roles. Anim. Behav. 59:11–20.
Matsumoto, K., and Y. Yanagisawa. 2001. Monogamy and sex role
reversal in the pipefish Corythoichthys haematopterus. Anim.
McCoy, E. E., A. G. Jones, and J. C. Avise. 2001. The genetic
mating system and tests for cuckoldry in a pipefish species in
which males fertilize eggs and brood offspring externally. Mol.
McLennan, D. A., D. R. Brooks, and J. D. McPhail. 1988. The
benefits of communications between comparative ethology and
phylogenetic systematics: a case study using gasterosteid fishes.
Can. J. Zool. 66:2177–2190.
Nijhoff, M. 1993. Reproductive ecology of the seahorse Hippocam-
pus reidi on a Bonaire coral reef. M. Sc. thesis. Rijksuniversiteit,
Groningen, The Netherlands.
Owens, I. P. F. 2002. Male-only care and classical polyandry in
birds: phylogeny, ecology and sex differences in remating op-
portunities. Philos. Trans. R. Soc. Lond. B 357:283–293.
Pa ¨a ¨bo, S., W. K. Thomas, K. M. Whitfield, Y. Kumazawa, and A.
C. Wilson. 1991. Rearrangements of mitochondrial transfer RNA
genes in marsupials. J. Mol. Evol. 33:426–430.
Pagel, M. 1994a. Detecting correlated evolution on phylogenies: a
general method for the comparative analysis of discrete char-
acters. Proc. R. Soc. Lond. B 255:37–45.
———. 1994b. Discrete. University of Reading, Reading, PA.
———. 1999a. The maximum likelihood approach to reconstructing
ancestral character states of discrete characters on phylogenies.
Syst. Biol. 48:612–622.
———. 1999b. Inferring the historical patterns of biological evo-
lution. Nature 401:877–884.
Palumbi, S. R., A. P. Martin, S. L. Romano, W. O. D. McMillan,
L. Stice, and G. Grabowski. 1991. The simple fool’s guide to
PCR. University of Hawaii, Honolulu, HI.
Parker, G. A., and L. W. Simmons. 1996. Parental investment and
the control of sexual selection: Predicting the direction of sexual
competition. Proc. R. Soc. Lond. B 263:315–321.
Perante, N. C., M. G. Pajaro, J. J. Meeuwig, and A. C. J. Vincent.
2002. Biology of a seahorse species, Hippocampus comes, in the
central Philippines. J. Fish. Biol. 60:821–837.
Posada, D., and K. A. Crandall. 1998. Modeltest: testing the model
of DNA substitution. Bioinformatics 14:817–818.
Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution
Roelke, D. L., and S. M. Sogard. 1993. Gender-based differences
in habitat selection and activity level in the northern pipefish
(Syngnathus fuscus). Copeia 1993:528–532.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new
method for reconstructing phylogenetic trees. Mol. Biol. Evol.
Shimodaira, H., and M. Hasegawa. 1999. Multiple comparisons of
log-likelihoods with applications to phylogenetic inference. Mol.
Biol. Evol. 16:1114–1116.
Simmons, L. W. 1992. Quantification of role reversal in relative
parental investment in a bush cricket. Nature 358:61–63.
Swofford, D. 2000. PAUP*. Phylogenetic analysis using parsimony
(*and other methods). Ver. 4b10. Sinauer Associates, Sunder-
Swofford, D., and S. H. Berlocher. 1987. Inferring evolutionary
trees from gene frequency data under the principle of maximum
parsimony. Syst. Zool. 36:293–325.
Taberlet, P. A., A. Meyer, and J. Bouvet. 1992. Unusual mito-
chondrial DNA polymorphism in two local populations of blue
tit (Parus caeruleus). Mol. Ecol. 1:27–36.
Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. Clustal
W: improving the sensitivity of progressive multiple sequence
alignment through sequence weighting, position-specific gap
penalties and weight matrix choice. Nucleic Acids Res. 22:
Trivers, R. L. 1972. Parental investment and sexual selection. Pp.
136–179 in B. Campbell, ed. Sexual selection and the descent
of man. Heinemann, London.
Vincent, A. 1994. Seahorses exhibit conventional sex roles in mat-
ing competition, despite male pregnancy. Behaviour 128:
Vincent, A., and R. M. Sadler. 1995. Faithful pair bonds in wild
seahorses, Hippocampus whitei. Anim. Behav. 50:1557–1569.
Vincent, A., I. Ahnesjo ¨, and A. Berglund. 1994. Operational sex
ratios and behavioural sex differences in a pipefish population.
Behav. Ecol. Sociobiol. 34:435–442.
1386 Download full-text
ANTHONY B. WILSON ET AL.
Vincent, A., I. Ahnesjo ¨, A. Berglund, and G. Rosenqvist. 1992.
Pipefishes and seahorses: Are they all sex role reversed? Trends
Ecol. Evol. 7:237–241.
Vincent, A., A. Berglund, and I. Ahnesjo ¨. 1995. Reproductive ecol-
ogy of five pipefish species in one eelgrass meadow. Environ.
Biol. Fish. 44:347–361.
Vitturi, R., A. Libertini, M. Campolmi, F. Calderazzo, and A. Maz-
zola. 1998. Conventional karyotype, nucleolar organizer regions
and genome size in five Mediterranean species of Syngnathidae
(Pisces, Syngnathiformes). J. Fish. Biol. 52:677–687.
Waddell, P. J., H. Kishino, and R. Ota. 2000. Rapid evaluation of
the phylogenetic congruence of sequence data using likelihood
ratio tests. Mol. Biol. Evol. 17:1988–1992.
Watanabe, S., and Y. Watanabe. 2001. Brooding season, sex ratio
and brood pouch development in the seaweed pipefish, Syng-
nathus schlegeli, in Otsuchi Bay, Japan. Ichthyol. Res. 48:
Watanabe, S., Y. Watanabe, and M. Okiyama. 1997. Monogamous
mating and conventional sex roles in Hippichthys penicillus
(Syngnathidae) under laboratory conditions. Ichthyol. Res. 44:
Watanabe, S., M. Hara, and K. Watanabe. 2000. Male internal fer-
tilization and introsperm-like sperm of the seaweed pipefish
(Syngnathus schlegeli). Zool. Sci. (Tokyo) 17:759–767.
Wilson, A. B., A. Vincent, I. Ahnesjo ¨, and A. Meyer. 2001. Male
pregnancy in seahorses and pipefishes (family Syngnathidae):
rapid diversification of paternal brood pouch morphology in-
ferred from a molecular phylogeny. J. Hered. 92:159–166.
Corresponding Editor: B. Crespi