See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227298680

Evolution of habitat use by deep-sea mussels

Article  in  Marine Biology · February 2006

DOI: 10.1007/s00227-005-0115-1

CITATIONS READS

138 397

6 authors, including:

Joe Jones Yong-Jin Won

University of South Carolina Ewha Womans University
74 PUBLICATIONS   6,067 CITATIONS    103 PUBLICATIONS   2,374 CITATIONS   

SEE PROFILE SEE PROFILE

Peter J. Smith Robert C Vrijenhoek

Museum Victoria Monterey Bay Aquarium Research Institute
323 PUBLICATIONS   4,094 CITATIONS    340 PUBLICATIONS   25,819 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Molecular ecology of marine and aquatic organisms View project

Comparative Genomics of Marine Invertebrates and their Symbionts View project

All content following this page was uploaded by Yong-Jin Won on 26 May 2014.

The user has requested enhancement of the downloaded file.

Marine Biology (2006) 148: 841–851
DOI 10.1007/s00227-005-0115-1

R ES E AR C H A RT I C L E

W.J. Jones Æ Y-J. Won Æ P.A.Y. Maas Æ P.J. Smith

R.A. Lutz Æ R.C. Vrijenhoek

Evolution of habitat use by deep-sea mussels

Received: 4 May 2005 / Accepted: 30 August 2005 / Published online: 14 October 2005

Springer-Verlag 2005

Abstract Previous phylogenetic studies proposed that progressive evolution from shallow-to-deep habitats
symbiont-bearing mussels of the subfamily Bathymodi- appear to hold with a few instances of habitat reversals.
olinae (Bivalvia: Mytilidae) invaded progressively deeper
marine environments and evolved from lineages that
decomposed wood and bone to specialized lineages that
invaded cold-water hydrocarbon seeps and finally deep- Introduction
sea hydrothermal vents. To assess the validity of the
hypotheses, we examined two nuclear (18S and 28S Bathymodiolin mussels (Mytilidae: Bathymodiolinae)
rRNA) and two mitochondrial genes (COI and ND4) are among the dominant animals found at marine
from a broad array of bathymodiolin species that in- hydrothermal vents and cold-water seeps worldwide.
cluded several recently discovered species from shallow They rely mostly on chemoautotrophic endosymbionts
hydrothermal seamounts. Bayesian phylogenetic analy- for their nutrition but retain an ability to filter-feed,
sis and maximum-likelihood estimates of ancestral which together appear to contribute to their broad
character states revealed that vent species evolved mul- ecological success (Fisher et al. 1987). It was proposed
tiple times, and that reversals in vent and seep habitat that the deep-sea vent lineages originated recently from
use occurred within the sampled taxa. Previous older lineages that occupied cold-seeps and shallow-
hypotheses regarding evolution from wood/bone-to- water environments (Craddock et al. 1995a). Further-
seeps/vents are supported in that mid-ocean hydrother- more, it has been proposed that species exploiting sul-
mal vent species may represent a monophyletic group fide-rich decomposing wood and bone preceded the
with one noticeable reversal. Earlier hypotheses about origin of vent and seep species (Distel et al. 2000).
However, the generality of these patterns is not firmly
Communicated by R.T. Thompson, St. John’s
established, because they were derived from phyloge-
netic studies that examined a limited number of species
W.J. Jones (&) Æ Y-J. Won Æ R.C. Vrijenhoek mostly from the Atlantic Ocean. For example, support
Monterey Bay Aquarium Research Institute, for precedence of wood and bone specialists relies en-
Moss Landing, CA 95039, USA
E-mail: jones@mbari.org
tirely on phylogenetic placement of Benthomodiolus lig-
Tel.: +1-831-7751789 nicola as a basal lineage in the bathymodiolin clade. In
Fax: +1-831-7751620 addition, subsequent phylogenetic studies revealed that
some western Pacific mussel species can occur in vents
P.A.Y. Maas Æ R.A. Lutz and seeps (Miyazaki et al. 2004) and that vent-to-seep
Institute of Marine and Coastal Sciences, Rutgers University,
New Brunswick, NJ 08901, USA habitat reversals have occurred in some Atlantic mussel
lineages (Won et al. 2002).
P.J. Smith Nevertheless, broad-scale patterns of habitat evolu-
National Institute of Water and Atmospheric Research Ltd, tion (from wood/bone-to-seep/vent and from shallow-
Private Bag 14 901, Wellington, New Zealand
to-deep water) may be valid. New opportunities to test
Present address: Y-J. Won these hypotheses are possible, because the number of
Department of Life Sciences, Ewha Womans University, named species has doubled during the past decade (von
Seoul, Korea Cosel et al. 1994, 1997, 1999; Hashimoto and Okutani
Present address: P.A.Y. Maas
1994; Cosel and Olu 1998; Gustafson et al. 1998; von
School of Science, The College of New Jersey, Cosel 2002; Won et al. 2003b). Particularly relevant to
Ewing, NJ 08628-0718, USA the present analysis is the availability of new DNA se-
842

quences from several western Pacific species (Miyazaki In this and in previous studies, we found no evidence
et al. 2004) that were not integrated into prior studies. In for gender-biased mitochondria in bathymodiolin mus-
addition, discoveries of new western Pacific species that sels (Won et al. 2003a, b). Gonadal tissue is diffusely
live at very shallow cold-seeps and hydrothermal sea- distributed throughout the visceral mass and the pos-
mounts (Cosel and Marshall 2003; Smith et al. 2004) terior part of the mantle in bathymodiolins. Our DNA
provide opportunities to test hypotheses about habitat extracts were obtained from a frozen adductor muscle,
evolution. In this study, we examined DNA sequences so we were unable to assess the sex of individuals.
from two nuclear ribosomal genes (18S rDNA and 28S Nevertheless, we examined multiple individuals of all
rRNA) and two protein-coding mitochondrial genes species and found no evidence for the degree of intra-
(COI and ND4) to estimate a phylogeny of wood, bone, specific sequence divergence commonly associated with
vent, and seep bathymodiolins. We used enhanced sta- doubly uniparental inheritance (DUI) of mitochondria
tistical methods (i.e., Bayesian analysis) to construct a in other bivalves (Hoeh et al. 1996, 2002; Passamonti
robust bathymodiolin phylogeny, which in turn was and Scali 2001). In addition, we observed no hetero-
used to test hypotheses about patterns and directions of plasmy, an earmark of DUI (Hoeh et al. 1991), in the
evolution. present mitochondrial sequences. One case of hetero-
plasmy was previously observed with B. thermophilus
(Craddock et al. 1995b). A sample of nine females and
Materials and methods ten males from the same location revealed no association
between gender and mitochondrial haplotypes (P. Maas,
Specimens used to obtain original DNA sequences for unpublished data).
this study were collected from a range of depths and
habitats in the Atlantic, Pacific and Indian Oceans
(Table 1). GenBank accession numbers for these se- Phylogeny reconstruction
quences are listed with each collection locality. To ob-
tain these sequences, we extracted genomic DNA from Nucleotide composition of each gene was estimated
frozen samples of adductor muscle (50 mg) that was using MEGA 2.1 (Kumar et al. 1993). Bayesian relative
treated with the Qiagen Dneasy isolation kit, accord- rates test for each gene was performed with Bentho-
ing to manufacturer’s instructions (Qiagen Inc., Valen- modiolus lignicola as the outgroup using Cadence (ver.
cia , CA, USA). Polymerase chain reaction (PCR) 1.08 Wilcox et al. 2004) with the last 1,000 of the post
conditions for amplification of the gene regions were as burn-in Bayesian trees (see below).
follows: 30–100 ng of template DNA, 5 ll 10· buffer Saturation plots were constructed for COI and ND4
(supplied by manufacturer), 5 ll MgCl2 (2.5 lM), 2 ll using uncorrected ‘‘p’’ versus the gene-specific model as
of each primer (10 lM final conc.), 2.5 U of Taq poly- determined in Model Test 3.06 (Posada and Crandall
merase (Promega Inc., WI, USA), 5 ll of a 2 mM 1998). These gene-specific models were used to estimate
stock solution of dNTPs, and sterile H2O to a final- maximum-likelihood phylogenies in PAUP 4.0b10
volume of 50 ll. PCR was performed with a Cetus (Swofford 1998). Phylogenies were inferred from under
9700 DNA thermocycler (Perkin-Elmer Corp., CT, the maximum-parsimony and maximum-likelihood cri-
USA) with the appropriate primers and conditions as teria as implemented in PAUP 4.0b10 (Swofford 1998).
listed in Table 2. A heuristic approach with ten random additions of the
Polymerase chain reaction products were purified input taxa was used to search for the shortest parsi-
using Qiagen PCR purification KitTM (Qiagen Inc. mony trees. The robustness of parsimony trees was
Valencia, CA, USA) or Montage columns (Millipore, assessed by bootstrap analysis (Felsenstein 1985), using
Billerica, MA, USA). The purified template DNA was the heuristic search procedure in PAUP 4.0b10 with
sequenced using Big DyeTM Terminator cycle sequenc- 1,000 replicates and 10 random additions of the input
ing reaction kit (PE Biosystems, Foster City, CA, taxa. For all genes surveyed, the maximum-likelihood
USA) and ABI Prism 3100 DNA sequencers (Applied phylogeny was concordant with the Bayesian phylog-
Biosystems Inc., Foster, CA, USA). PCR products were eny and are not shown.
sequenced bidirectionally from each individual sample The partition homogeneity test (using the same
using the same forward and reverse primers as used in conditions as in the heuristic search) as implemented
PCR. DNA sequence alignments were initially con- in PAUP 4.0b10 was used as a proxy to determine the
structed using SequencherTM (Gene Codes Corp. Inc., overall concordance between gene trees. There was no
Ann Arbor, MI, USA). Secondary structure of rRNA significant difference between the overall tree distri-
(i.e., stems and loops) was inferred using the program bution of the two mitochondrial genes (P=0.57) or
GeneBee (Brodsky 1992). Amino acid translations of between 28S rRNA and the mitochondrial genes
COI and ND4 were performed in MEGA 2.1 (Kumar (P=0.65). For estimation of ancestral character
et al. 1993) using the invertebrate mitochondrial code states, we used the combined dataset of 28S, COI, and
(GenBank Code 5). Alignments for all four genes ana- ND4.
lyzed in this study are available in GenBank PopSet Bayesian phylogenetic trees were estimated for
(Accession Numbers in Table 1). individual genes using MrBayes version 3.0b4 (Huel-
Table 1 Specimen collection sites, species identifications, and GenBank accession numbers for bathymodiolin mussels

OTU Location/reference Ocean Latitude, Depth Habitat GenBank No.

basin longitude range (m)
18S 28S ND4 COI

Bathymodiolus heckerae BR Blake Ridge Atlantic 3230¢N; 7611¢W 2,155 Seep AY649830 AY781139 AY130245 AY649793
B. heckerae WFE W. Florida Escarp. Atlantic 2602¢N; 8455¢W 3,314 Seep AF221639 AY781138 AY130246 AY649794
B. azoricus Menez Gwen Atlantic 3717¢N; 3215¢W 866–2,330 Vent AY649822 AY781148 AF128534 AY649795
B. puteoserpentis Snakepit Atlantic 2322¢N; 4456¢W 3,023–3,510 Vent AF221640 AY781151 AF128533 AY649796
B. brooksi AC Alamiños Canyon Atlantic 2621¢N; 9429¢W 2,222 Seep AY649826 AY781136 AY130247 AY649797
B. brooksi WFE W. Florida Escarp. Atlantic 2602¢N; 8455¢W 3,314 Seep AY649825 AY781135 AY649805 AY649798
B. brevior MT Mariana Trough W. Pacific 1813¢N; 14442¢E 3,589 Vent AY649824 AY781150 AY649806 AY649799
B. brevior LBA Lau Basin W. Pacific 2313S; 17638¢W 1,750 Vent AY649827 AY781143 AY046277 AY275544
B. marisindicus Central Indian Ridge Indian 2353¢S; 6936¢E 3,289 Vent AY649818 AY781147 AY046279 AY275543
Ocean
B. thermophilus A 9N EPR E. Pacific 951¢N; 10418¢W 2,460–2,747 Vent AF221638 AY781141 AY649807 AF456285
B. thermophilus B 7S EPR E. Pacific 726¢S; 10447¢W 2,460–2,747 Vent AY649829 AY781142 AY649808 AF456303
B. aff. thermophilus 32S EPR E. Pacific 3152¢S; 11203¢W 2,331 Vent AY649823 AY781140 AY649809 AF456317
B. childressi Alamiños Canyon Atlantic 2621¢N; 9429¢W 540–2,222 Seep AF221641 AY781137 AY130248 AY649800
B. mauritanicus West Africa Atlantic 053¢N; 528¢W 1000–1267 Seep AY649828 AY781144 AY649810 AY649801
Gigantidas gladius Rumble III W. Pacific 3544¢S; 17830¢E 300–460 Vent AY649821 AY781149 AY649813 AY649802
B. tangaroa Cape Turnagain W. Pacific 4026¢S; 17858¢E 920–1,205 Seep AY649820 AY781134 AY649811 AY608439
Tamu fisheri Garden Banks Atlantic 2750¢N; 9210¢W 546–650 Seep AF221642 AY781132 AY649814 AY649803
Idas washingtonia Monterey Bay E. Pacific 377¢N; 12248¢W 960–1,910 Bone, AF221645 AY781146 AY649815 AY275546
wood
I. macdonaldi Garden Banks Atlantic 2750¢N; 9210¢W 650 Seep AF221647 AY781145 AY649816 AY649804
NZ-3 Macauley Cone W. Pacific 3013¢S; 17827¢W 200 Vent AY649819 AY781133 AY649812 AY608440
Benthomodiolus lignicola Chatham Rise W. Pacific 3441¢S; 17714¢W 826–1,174 Bone, AF221648 AY781131 AY649817 AY275545
wood
843
844

Table 2 PCR primers and temperature profiles

Gene Primer Sequence Temperature profile Reference

mtCOI HCO-2198 5¢-TAAACTTCAGGGTGACCAAAAAATCA-3¢ 94-4-[94-1-55-2-72-3.5]-72-10 (x35) (Folmer et al. 1994)

LCO-1490 5¢-GGTCAACAAATCATAAAGATATTGG-3¢ (Folmer et al. 1994)
mtCOI HCO-2148 5¢-CCYCTAGGRTCATAAAAAGA-3¢ 94-4-[94-1-55-2-72-3.5]-72-10 (x35) This study
LCO-1560 5¢-ATRCTDATTCGWATTGA-3¢ This study
mtNADH4 Arg BL 5¢-CAAGACCCTTGATTTCGGCTCA-3¢ 94-4-[94-40s-55-1-72-1]-72-10 (x37) (Bielawski and
Gold 1996)
NAP 2H 5¢-TGGAGCTTCTACGTGRGCTTT-3¢ (Arevalo et al. 1994)
18S rRNA 1F 5¢-TACCTGGTTGATCCTGCCAGTAG-3¢ 94-4-[94-40s-55-1-72-1]-72-7 (x37) (Giribet et al. 1996)
3R 5¢-AGGCTCCCTCTCCGGAATCGAAC-3¢ (Giribet et al. 1996)
3F 5¢-GTTCGATTCCGGAGAGGGA-3¢ 94-4-[94-40s-55-1-72-1]-72-7 (x37) (Giribet et al. 1996)
5R 5¢-CTTGGCAAATGCTTTCGC-3¢ (Giribet et al. 1996)
5F 5¢-GCGAAAGCATTTGCCAAGAA-3¢ 94-4-[94-40s-55-1-72-1]-72-7 (x37) (Giribet et al. 1996)
9R 5¢-GATCCTTCCGCAGG’TTCACCTAC-3¢ (Giribet et al. 1996)
28S rRNA LSUD1F 5¢-ACCCGCTGAATTTAAGCATA-3¢ 94-5-[94-1-55-1-72-2]-72-7 (x35) (Scholin and
Anderson 1994)
D3AR 5¢-ACGAACGATTTGCACGTCAG-3¢ (Scholin and
Anderson 1994)

senbeck and Ronquist 2001) using the codon model for (Bollback 2004) using the last 5,000 post burn-in trees
COI and ND4 and partitioning by structure (stem/loop) from the combined 28S, COI and ND4 dataset anal-
for 18S and 28S ribosomal RNAs. An additional series ysis.
of MrBayes analyses was performed on the combined
28S, COI, and ND4 dataset using the same parameters
as in the individual gene analysis. Parameters are Results
available upon request from the senior author. The
Monte Carlo Markov chain (MCMC) length was Nuclear 18S rRNA
1.1·106 generations with 6 markov chains, and we
sampled the chain every 100 generations to minimize The complete aligned 18S data set consisted of 1,805
autocorrelation. MCMC convergence was assessed by base pairs (bp). Altogether, 708 positions were desig-
visually inspecting the sample paths of model parame- nated as stems and 1,097 positions were designated as
ters (to determine an appropriate burn-in period) and loops (Table 3). Thirteen new 18S rRNA sequences were
by repeating the analysis at least three times with ran- obtained in the present study. Across both regions, we
dom initial parameter values (to assess the dependence found 240 variable sites, including 159 that were parsi-
of posterior distributions on initial conditions). Anal- mony informative. In addition to the taxa listed in Ta-
yses were run in parallel using two dual processor G5 ble 1, a number of additional taxa were analyzed to
Macintosh servers. determine an appropriate outgroup to root the 28S
Parameter estimates were graphically analyzed to rRNA and mitochondrial gene trees (GenBank accession
assess stability (Tracer ver. 1.0.1, Rambaut and Drum- numbers in parentheses): Adipicola arcuatilis
mond 2003). Log-likelihood values and associated (AF221644), Myrina pacifica (AF221646), Modiolus au-
parameters for sampled trees stabilized after approxi- riculatus (AF117735), M. americanus (AF229624), M.
mately 5–6·105 generations for both COI and ND4, modiolus (AF124210), Mytilus edulis (L33448), M. cali-
while it took approximately 4–5·105 generations for the fornianus (L33449), M. galloprovincialis (L33451), M.
18S and 28S rRNA data sets, respectively. Therefore, we trossulus (L24490), Musculus senhousei (AF124207),
conservatively used the last 5,000 sampled trees to esti- Musculus lateralis (AF229625), Hormomya domingensis
mate Bayesian posterior probabilities. If ‡95% of the (AF117736), Brachidontes variabilis (AJ389643), Geu-
sampled trees contained a given clade, we considered it kensia demissa (L33450), Lithophaga lithophaga
to be significantly supported by our data (sensu Wilcox (AF124208), L. nigra (AF124209), Atrina pectinata
et al. 2002). Divergence time among taxa was estimated (X90961), Chlamys islandica (L11232), Placopecten ma-
using a Bayesian approach (BEAST version 1.0.3, gellanicus (X53899).
Drummond and Rambaut 2003). The Bayesian analysis generated a topology that was
Maximum-likelihood (Mesquite version 1.04, broadly congruent with the maximum likelihood tree
Maddison and Maddison 2004) and Bayesian (SIM- obtained by Distel et al. (2000) (Fig. 1a). The wood-
MAP, Bollback 2004) estimates of ancestral character degrading specialist, Benthomodiolus lignicola, comprises
states (Huelsenbeck et al. 2003) for habitat and depth the sister-group to all other sampled deep-sea bathy-
were mapped on the combined 28S, COI and ND4 modiolins. Subsequent analyses only considered mem-
dataset. The rate and number of transformations be- bers of the Bathymodilinae with Be. lignicola used to
tween character states was estimated using SIMMAP root the tree.
 845

Table 3 Base composition and nucleotide substitution patterns for Mitochondrial ND4
18S rRNA, 28S rRNA, ND4, and COI
Gene Nucleotide positions The complete aligned data set of ND4 consisted of
Loops Stems All 529 bp, which included 176 amino acids of the coding
region. Thirteen new ND4 sequences were obtained in
18S the present study. The inferred number of amino acid
%A 25.7 24.5 – 25.2 substitutions between taxa averaged 33.0 (range: 0–69).
%C 22.5 22.3 – 22.4
%G 27.0 27.3 – 27.1
Of the 529 nucleotide positions, 318 (60.1%) were vari-
%T 24.5 26.0 – 25.3 able and 254 (48.0%) were parsimony informative (Ta-
Variable sites 144 96 – 240 ble 3). About half of the substitutions occurred at third
Informative sites 95 64 – 159 positions (53.1%). Compositional biases across all three
TS/TV 1.95 0.96 – 1.11 codon positions were particularly exacerbated for thy-
28S mine and cytosine at third positions. Thymine was
%A 25.7 13.7 – 21.0 especially biased at second positions.
%C 29.4 37.0 – 32.4
%G 18.9 36.1 – 25.7
Parameters were estimated for a Bayesian analysis of
%T 26.0 13.1 – 21.0 ND4 (available upon request), which revealed a well-
Variable sites 77 33 – 110 resolved bathymodiolin phylogeny (Fig. 1c). The ND4
Informative sites 32 11 – 43 Bayesian phylogeny clearly places Bathymodiolus
TS/TV 2.4 6.8 – 3.45 brooksi within a clade containing B. thermophilis, B.
brevior, B. marisindicus, B. heckerae, B. azoricus and B.
First Second Third All puteoserpentis. ND4 also placed the undescribed New
ND4
Zealand species NZ-3 as a basal lineage in the Bathy-
%A 23.2 14.7 22.2 20.1 modiolinae. The Bayesian relative rates test identified no
%C 14.1 12.7 12.2 13.0 statistically significant heterogeneity in ND4 substitution
%G 29.9 22.1 21.6 24.6 rates among bathymodiolin taxa.\
%T 32.8 50.5 44.0 42.4
Variable sites 99 50 169 318
Informative sites 74 25 155 254 Mitochondrial COI
TS/TV 3.29 1.53 2.19 2.22
Ks – – – 1.17
Ka – – – 0.11 The complete aligned data set of COI consisted of
689 bp, which included 229 amino acids of the coding
COI
%A 26.4 14.3 26.6 22.4 region. Twelve new COI sequences were obtained in
%C 15.5 23.6 9.4 16.1 the present study. The inferred number of amino acid
%G 26.5 19.2 19.0 21.5 substitutions between taxa averaged 2.33 (range: 0–8).
%T 31.6 43.1 45.1 39.9 Altogether, 261 sites (37.9%) were variable mostly at
Variable sites 44 11 206 261
Informative sites 28 1 167 196
third positions (80%). Of the variable sites, 196
TS/TV 3.54 0.67 3.12 1.8 (28.4%) were phylogenetically informative: 28 at first
Ks – – – 1.12 positions; one at second positions; and 167 at third
Ka – – – 0.13 positions (Table 3). A relatively high ratio of transi-
Ka is the number of nonsynonymous substitutions per nonsynon-
tions to transversions was observed at first positions;
ymous site; Ks is the number of synonymous substitutions per however the ratio was lower at second positions.
synonymous site Nucleotide composition resembled that found in mito-
chondrial ND4, with excess thymine and deficiency of
cytosine across all positions. Pairwise comparisons
involving Benthomodiolus lignicola and bathymodiolin
Nuclear 28S rRNA taxa revealed possible saturation for third position
substitutions (data not shown). However, this problem
The complete aligned 28S rRNA data set consisted of was not as evident for first and second positions in the
938 bp. Altogether, 358 positions were designated as codons.
stems and 580 positions were designated as loops The undescribed New Zealand species, NZ-3, was
(Table 3). All 28S rRNA sequences reported in this supported as basal to all other sampled ingroup taxa
paper are new. Across both regions, we found 110 (Fig. 1d). The two recently described mussel species
variable sites, including 43 that were parsimony from New Zealand, Bathymodiolus tangaroa and Gig-
informative. The Bayesian analysis generated a topol- antidas gladius, were placed in the same clade as B.
ogy (Fig. 1b) that was substantially more resolved for childressi and B. mauritanicus. Inclusion of the partial
bathymodiolins than the 18S rRNA. Of particular note COI sequences (415 bp) from Miyazaki et al. (2004)
is the placement of Bathymodiolus tangaroa and Gig- strongly supported (1.00 BPP) the placement of B.
antidas gladius in the same clade as B. childressi and japonicus and B. platifrons in the same clade as B. chil-
B. mauritanicus. dressi (data not shown). The remaining well-supported
846

a 18S rRNA b 28S rRNA

B. heckerae BR
B. heckerae WFE
B. azoricus 1.00
99 B. heckerae BR
B. puteoserpentis
B. brooksi WFE B. heckerae WFE
1.00
B. brooksi AC 99 B. azoricus
B. brevior MT B. puteoserpentis
B. brevior LBA 1.00
99 B. brooksi WFE
B. marisindicus
B. thermophilus A B. brooksi AC

B. thermophilus B 1.00 0.98 B. brevior MT

-- --
B. aff. thermophilus B. brevior LBA
B. childressi
B. marisindicus
B. mauritanicus
G. gladius 0.96 B. thermophilus A
1.00
74 B. tangaroa --
B. thermophilus B
T.fisheri
1.00 B. aff. thermophilus
1.00 96 I. washingtonia
1.00 0.96
94 Adipicola arcuatilis* I. washingtonia
91 --
I. macdonaldi
I. macdonaldi
Myrina pacifica* 0.96
-- B. childressi
1.00 NZ-3 1.00
100 96 B. mauritanicus
Be. lignicola 0.98
1.00
--
100 B. tangaroa
Modiolinae
1.00 G. gladius
55 Mytilinae
1.00 Crenellinae T.fisheri
96
1.00 Lithophagainae
NZ-3
100
Family Pectinidae (outgroup)
Be. lignicola
0.01 0.01

1.00
c ND4 0.99
100 B. heckerae BR d COI 1.00
B. heckerae BR
0.99 100
70 B. heckerae WFE --
1.00 B. heckerae WFE
1.00
96 96 B. azoricus
B. azoricus

B. puteoserpentis B. puteoserpentis
0.98
-- 1.00 0.96 1.00
B. brevior MT
100 B. brooksi WFE -- 100
B. brevior LBA*
B. brooksi AC
B. marisindicus*
1.00 B. brevior MT
1.00 1.00
95 B. thermophilus A*
100 B. brevior LBA* 1.00 100
98 B. thermophilus B*
B. marisindicus*
1.00
B. thermophilus A* B. aff. thermophilus*
99
1.00 0.96 1.00
100 B. thermophilus B* -- B. brooksi WFE
100

B. aff. thermophilus* B. brooksi AC

1.00 1.00
100 B. childressi B. childressi
1.00 0.98 100
100 B. mauritanicus 98 B. mauritanicus
1.00
0.98
95 B. tangaroa* B. tangaroa*
--
G. gladius G. gladius
T.fisheri
T.fisheri
1.00
I. washingtonia* I. washingtonia*
--
I. macdonaldi I. macdonaldi

NZ-3* NZ-3*

Be. lignicola* Be. lignicola*

0.1 0.1

Fig. 1 Bayesian trees of a 18S rRNA, b 28S rRNA, c ND4, and d are shown (lower number). Nodes with less than 0.95 BPP are
COI datasets. Previous published data are indicated with asterisks collapsed to polytomies. Maximum parsimony bootstrap values
(see Table 1). Scale bar indicates percent sequence divergence. Only less than 70% are indicated by ‘‘–’’. Additional outgroup taxa 18S
Bayesian posterior probabilities (BPP; upper number) greater than rRNA analyzed (see Materials and methods for GenBank numbers)
0.95 and maximum-parsimony bootstrap support greater than 70% are indicated by the subfamily clade names in Fig. 1a
 847

Fig. 2 Bayesian tree of 28S 1.00

rRNA, ND4, and COI 100 B. heckerae WFE
combined dataset. Scale bar
indicates percent sequence 1.00
divergence. Only Bayesian 100 B. heckerae BR
posterior probabilities (BPP; 0.95
upper number) greater than 0.95 -- B. azoricus
and maximum-parsimony 1.00
bootstrap support greater than
70% are shown (lower number).
-- B. puteoserpentis
Maximum parsimony bootstrap 0.99
values less than 70% are 82 B. brevior MT
indicated by ‘‘–’’. The four well- 1.00
supported clades (Nodes I-IV) 100
1.00 B. brevior LBA
are indicated in bold.
Synapomorphic amino acid --
changes in mtND4 are indicated B. marisindicus
by hash marks below which is
the relative position in the IV. 1.00
fragment sequenced and the 1.00 100 B. thermophilus A
amino acid change 98 1.00
100 B. thermophilus B
29
M->L B. aff. thermophilus
1.00
100 B. brooksi WFE
1.00
-- B. brooksi AC
1.00
100 B. childressi
1.00
I.
100 B. mauritanicus
1.00 III. 1.00
94 1.00 76
B. tangaroa
--
6 50 144
G. gladius
S->C L->F F->L 32
II.
P->S T.fisheri
1.00
--
I. washingtonia

73 166 I. macdonaldi
V->A F->L
NZ-3

Be. lignicola
0.1

relationships were found primarily at nodes that connect change in mtND4 (Fig. 2). First, the combined tree
OTUs of recognized species (i.e., B. thermophilus). clearly recognizes the placement of NZ-3 as basal to all
Support for many of the basal nodes was low. Bayesian other surveyed ingroup taxa (Fig. 2). Second, the genus
relative rates test of COI suggested homogenous Idas is supported as a distinct basal clade (Fig. 2). Third,
molecular rates among all taxa. the combined tree supports the existence of a distinct
group composed of Bathymodiolus childressi, B. mauri-
tanicus, B. tangaroa and Gigantidas gladius (Fig. 2).
Combined phylogenetic analysis Henceforth, we refer to this clade as the ‘‘childressi’’
group for the first named species in the group. Fourth,
Individually, the 28S, COI and ND4 sequences provided the tree supports a monophyletic group of species
varying degrees of phylogenetic resolution among including B. thermophilus, B. aff. thermophilus B.
members of the Bathymodiolinae; however a combined marisindicus, B. brevior, B. brooks, B. puteoserpentis, B.
phylogenetic analysis of the three genes identified four azoricus, and B. heckerae (Fig. 2). Henceforth, we refer
well-supported clades (Fig. 2). Additionally, each clade to this clade as the ‘‘thermophilus’’ group for the first
is supported by at least one synapomorphic amino acid named species in the group.
848

Fig. 3 Bayesian cladogram of a

COI, ND4, and 28S (after

B. aff. thermophilis

Gigantidas gladius
B. puteoserpentis
B. heckerae WFE

B. thermophilis A
B. thermophilis B
Fig. 2) with maximum-

B. brooksi WFE
B. heckerae BR

B. mauritanicus
B. marisindicus

I. washingtonia
likelihood estimates of ancestral

B. brevior LBA

B. brooksi AC
B. brevior MT

I. macdonaldi
character states for a depth and

B. childressi

Be. lignicola
B. tangaroa
B. azoricus
b habitat. Relative support for
ancestral character states are

T. fisheri
indicated by area of pie chart at

NZ-3
respective nodes. Values for the
three habitat states (v vent, s
seep, w/b wood/bone) are given
at nodes I and III. Asterisk
indicates a significant likelihood
that the particular character
state would have existed at a
particular node

shallow

deep

Node III
v 0.17* vent (v)
s 0.74*
w/b 0.09 seep (s)
Node I
v 0.36 wood / bone (w/b)
s 0.18
w/b 0.47

Character evolution in deep-sea mussels pattern recognized by Craddock and coworkers appears
to be supported. Nevertheless, some notable exceptions
The combined phylogeny provides an evolutionary exist. One member of this ‘‘thermophilus’’ clade,
framework against which we can test various hypotheses B. azoricus, has also invaded a shallower portion of the
about character evolution in bathymodiolin mussels. Mid-Atlantic Ridge (type locality Menez Gwen, 866 m).
We treated habitat depth and type in each species as Similarly, evolutionary transitions between habitat
discrete character states and examined their traces in the types appear to be complex. Hydrothermal vent mussels
phylogeny. To test Craddock et al. (1995a) hypothesis do not comprise a derived monophyletic clade. Gigantidas
that these mussels diversified from shallow- to deep-wa- gladius and undescribed species NZ-3 were found on
ter habitats, we divided the most shallow known depth of shallow hydrothermal seamounts and they stem from
each species into those that are greater or lesser than basal lineages. Nevertheless, mussels found at deep-sea
1,000 m (Fig. 3a) Clearly, the taxa stemming from basal hydrothermal vents, the ‘‘thermophilus’’ clade (node IV,
nodes (NZ-3, Idas and most of the ‘‘childressi’’ clade) Fig. 2) appears to be monophyletic. Thus, hydrothermal
occur in shallow sites. The ‘‘thermophilus’’ clade defined habitats have been invaded multiple times, but deep-sea
by node IV all tend to live at deep sites, so the general hydrothermal habitats may have been invaded once.
 849

Habitat reversals are evident within the ‘‘thermophilus’’ OTUs (Fig. 1) with respect to a previous analysis by
clade. Two species, Bathymodiolus brooksi and B. hecke- Distel et al. (2000). The present phylogeny strongly
rae, have independently invaded deep-water cold seeps. supports the placement of Benthomodiolus lignicola as a
Distel et al. (2000) argued that mussel species basal lineage within the monophyletic subfamily Bay-
exploiting sulfide-rich decomposing wood and bone thymodiolinae. Unfortunately, the phylogeny produced
preceded the origin of vent and seep species (Distel et al. by this highly conserved gene does not resolve rela-
2000). Benthomodiolus lignicola, a species found on tionships among bathymodiolin genera. To obtain better
wood and bones, was used to root the present bathy- resolution at this level, we generated a combined phy-
modiolin trees, so it obviously cannot be used to test this logenetic tree based on sequences from a nuclear gene,
hypothesis. The only other species known to occupy 28S rDNA, and two mitochondrial protein-coding
wood and bone habitats, Idas washingtonia, also stems genes, COI and ND4. Restricting our analysis to
from a basal branch in the combined tree, but its rela- bathymodiolins and rooting with Be. lignicola allowed
tive, I. macdonaldi, lives in cold seeps. Another basal us to circumvent a problem that occurred if the analysis
lineage, undescribed species NZ-3, was found on a was extended to the family Mytilidae as a whole. The
shallow hydrothermal seamount. The ‘‘childressi’’ group two protein-coding genes exhibited significant saturation
also is relatively basal in this tree and it mostly com- of synonymous nucleotide substitutions across the
prises cold seep species. Thus, given our current family.
knowledge of deep-sea mussels, it is impossible to assess The combined gene tree rooted with Benthomodiolus
the habitat state of ancestral nodes in this tree—vent, lignicola revealed four well-supported clades. Unde-
seep, wood and bone are all probable character states scribed species NZ-3 from a shallow hydrothermal se-
(Fig. 3b) amount off New Zealand branched basally to all other
bathymodiolin lineages. Node II links the Idas species.
Node III links five species placed in three genera:
Evolutionary age of Bathymodilinae ‘‘Bathymodiolus’’ childressi, B. mauritanicus, B. tangaroa
and Gigantidas gladius and Tamu fisheri. Gustafson et al.
Although prone to saturation across the Mytilidae, COI (1998, p. 89) questioned placement of ‘‘B.’’ childressi, the
sequences could be used to estimate divergence time for first species described from this clade, in the genus
the Bathymodiolinae. We used previously published Bathymodiolus:
estimates of divergence rates for mitochondrial COI in ‘‘ ‘‘Bathymodiolus’’ (their quotes) childressi possesses
deep-sea mussels (Won et al. 2003b). Using the 1%/MY a combination of morphological characters not seen
rate based on divergence between Bathymodiolus ther- in any previously described deep-sea mytilid genus;
mophilus and B. aff. thermophilus across the Easter mi- however, genetic distance measures (at that time) ...do
croplate region of the East Pacific Rise (Won et al. not clearly separate this species from other members
2003b), we estimated the divergence of bathymodiolin of the genus Bathymodiolus. So as to avoid erecting a
mussels from the lineage leading to Benthomodiolus lig- new mono-specific genus, this species is provisionally
nicola would have occurred about 30.2 MY ago (range: placed in Bathymodiolus. ...‘‘Bathymodiolus’’ chil-
23.7–37.1 MY). Second, we substituted a more conser- dressi differs from all other species referred to
vative rate of 0.5%/MYfor COI divergence among Bathymodiolus in having multiple separation of the
invertebrate taxa (Knowlton 2000) that is consistent posterior byssal retractors (similar to Modiolus), a
with the slower rates also seen in some vent–endemic single posterior byssal retractor scar, and a rectum
annelids (Chevaldonné et al. 2002). Accordingly, diver- that enters the ventricle posterior to the level of the
gence of other bathymodiolin mussels from the lineage auricular ostia.’’
leading to Benthomodiolus lignicola would have occurred
about 60.8 MY (range 48.2–74.3). Despite the wide Separate analyses of published mitochondrial COI se-
range of estimates obtained with these Bayesian cali- quences (Miyazaki et al. 2004) also allow us to confi-
brations (23.7–74.3 MY), molecular evidence clearly dently place B. platifrons and B. japonicus in the
places the origin of bathymodiolins in the Cenozoic or ‘‘childressi’’ clade. Placement of Tamu in this clade is less
late Mesozoic. certain, however. Inclusion of Tamu is supported by the
COI and combined trees, but not by the 28S tree. More
studies involving additional gene markers are needed to
Discussion resolve relationships between the ‘‘childressi’’ clade and
Tamu. Nevertheless, the present analyses clearly reveal
The present analysis of DNA sequences from four gene that Bathymodiolus constitutes a paraphyletic taxon that
segments provides the first comprehensive phylogenetic requires further investigation with detailed morpholog-
analysis of bathymodiolin mussels from vents, seeps and ical and molecular investigations. Until these matters are
other reducing environments. To assess the position of resolved, we recommend that the provisional use of
the Bathymodiolinae within the family Mytilidae, we ‘‘Bathymodiolus’’ (in quotes) when applied to members
examined 18S ribosomal DNA sequences from 13 new of the childressi clade, as follows: ‘‘B.’’ platifrons,
850

‘‘B.’’ japonicus, ‘‘B.’’ childressi, ‘‘B.’’ mauritanicus and significantly, the present results are consistent with the
‘‘B.’’ tangaroa. Depending on the placement of Tamu hypothesis that shallow-water hydrothermal seamounts,
inside or outside this group, either Tamu or Gigantidas cold seeps, and hydrocarbon deposits may have served
might have priority as the genus name for this group. as refugia against large-scale extinction events in deep
For now, we suggest that the name Bathymodiolus be oceanic basins. We urge comprehensive phylogenetic
restricted to the monophyletic group of taxa that stems studies of other animal taxa found on seamounts, vents
from the well–supported clade defined by node IV. and seeps to test the hypothesis that such shallow-water
A general pattern of progressive evolution from habitats may have provided refugia during deep-sea
shallow-to-deep water is supported by the combined anoxic/dysoxic events and opportunities for re-radiated
tree, though reversals occur. Eight of nine basal lineages into deep-sea environments.
occur in relatively shallow environments (<1,000 m). In
contrast, only one Bathymodiolus lineage (node IV) oc- Acknowledgments We gratefully appreciate the efforts of the pilots
curs shallower than 1,000 m. Bathymodiolus azoricus is of the deep-sea submersibles Alvin, Johnson Sea Link, Shinkai 6500
and Tiburon during our oceanographic expeditions over the past
found at 866 m at Menez Gwen on the Mid-Atlantic 15 years. We thank J. Childress, C. Fisher, I. MacDonald for
Ridge, though it also is found at deeper sites (1,710– providing mussel specimens from the Gulf of Mexico, and C.R.
3,350 m). Idas washingtonia, a basal lineage, has a Young for help with the Bayesian analysis. We also thank P.
remarkable depth range (960–1,919 m) (Baco and Smith Braccio for help using XGrid and parallel versions of MrBayes.
C.L. Van Dover provided comments on data interpretation. P.J.S.
2003). We recently found this species on whale bones at was supported by the New Zealand Foundation for Research Sci-
2,890 m in Monterey Bay, California (R.C. Vrijenhoek, ence and Technology (contract # COIX0028). This study was
personal observation). It is difficult to trace habitat funded by the US National Science Foundation (OCE8917311,
depth for these mussels because many of the species have OCE9212771, OCE9302205, OCE9529819, OCE9633131,
OCE9910799, ESI0087679, OCE0327353 and OCE0241613) and
been found at only one or a few localities. Continued the David & Lucile Packard Foundation via the Monterey Bay
exploration of shallower vents, seamounts, wood falls Aquarium Research Institute.
and whale bones may significantly alter our limited view
of bathymetric limits for many of these taxa.
Furthermore, we suspect that the apparent pattern of Reference
evolution from shallow-to-deep habitats may represent a
recent re-radiation of mytilids into deep-sea environ- Arevalo E, Davis SK, Sites JW Jr (1994) Mitochondrial DNA se-
ments. Jacobs and Lindberg (1998) argued that deep-sea quence divergence and phylogenetic relationships among eight
vent communities experienced global extinction events chromosome races of the Sceloperus grammicus complex
(Phrynosomatidae) in Central Mexico. Syst Biol 43:387–418
associated with oceanic anoxic/dysoxic events at the end Baco AR, Smith CR (2003) High species richness in deep-sea
of the Mesozoic (Cenomanian–Turonian boundary, chemoautotrophic whale skeleton communities. Mar Ecol Prog
93 MY ago) and early Cenozoic (end Palaeocene, Ser 260:109–114
55 MY ago). Though their argument is broadly consis- Bielawski JP, Gold JR (1996) Unequal synonymous substitution
rates within and between two protein-coding mitochondrial
tent with the molecular clock evidence for late Mesozoic genes. Mol Biol Evol 13:880–992
and early to mid Cenozoic radiations of the dominant Bollback JP (2004) SIMMAP version 1.0: a program for stochastic
vent and seep invertebrate taxa, fossil evidence suggests mapping of molecular and morphological characters. Copen-
that some vent and seep taxa have persisted since the hagen, Denmark. http://brahms.ucsd.edu/simmap.html
early and mid Mesozoic (Little and Vrijenhoek 2003). Brodsky LI, Vasisiev AV, Kalaidzidis YL, Osipov YS, Tatuzov
RL, Feranchuk SI (1992) GeneBee: the program package for
The global and complete nature of hypothesized biopolymer structure analysis. Dimacs 6:127–139
extinction events is debatable; nevertheless, periodic Chevaldonné P, Jollivet D, Desbruyères D, Lutz RA, Vrijenhoek RC
extinctions followed by re-radiation of crown-taxa from (2002) Sister-species of eastern Pacific hydrothermal-vent worms
stem-ancestors provides a plausible explanation for the (Ampharetidae, Alvinelidae, Vestimentifera) provide new mito-
chondrial clock calibration. Cahiers Biol Mar 43:367–370
discrepancy between fossil and molecular evidence for von Cosel R (2002) A new species of bathymodioline mussel
the ages of vent and seep taxa (Little and Vrijenhoek (Mollusca, Bivalvia, Mytilidae) from Mauritania (West Africa),
2003). Basal bathymodiolins, such as Benthomodiolus with comments on the genus Bathymodiolus Kenk & Wilson,
lignicola and undescribed species NZ-3, might be relics 1985. Zoosystema 24:259–271
von Cosel R, Comtet T, Krylova E (1997) Two new species of
of lineages that survived deep-sea extinction events. Bathymodiolus from hydrothermal vents on the Mid-Atlantic
However, it is unknown whether they represent lineages Ridge. Cahiers Biol Mar 38:145–146
that continuously occupied shallow habitats, or if they von Cosel R, Comtet T, Krylova E (1999) Bathymodiolus (Bivalvia:
were derived from deep-sea lineages that secondarily Mytilidae) from hydrothermal vents on the Azores triple junc-
invaded shallow habitats, like Bathymodiolus azoricus, tion and the Logatchev hydrothermal field, Mid-Atlantic Ridge.
The Veliger 42:218–248
part of a deep-sea vent lineage that has invaded a shal- von Cosel R, Marshall BA (2003) Two new species of large mussels
low vent near the Azores. (Bivalvia: Mytilidae) from active submarine volcanoes and a
Additional species of bathymodiolin mussels will cold seep off the eastern North Island of New Zealand, with
undoubtedly be discovered as more of the world’s description of a new genus. The Nautilus 117:31–46
von Cosel R, Métivier B, Hashimoto J (1994) Three new species of
oceans are explored. We anticipate that the present gene Bathymodiolus (Bivalvia: Mytilidae) from hydrothermal vents
trees will provide a robust foundation for adding new in the Lau Basin and the Fiji Basin, Western Pacific, and the
branches and testing evolutionary hypotheses. More Snake Pit Area, Mid-Atlantic Ridge. The Veliger 37:374–392
 851

von Cosel R, Olu K (1998) Gigantism in Mytilidae. A new Knowlton N (2000) Molecular genetic analyses of species bound-
Bathymodiolus from cold seep areas on the Barbados Accre- aries in the sea. Hydrobiologia 420:73–90
tionary Prism. Anim Biol 321:655–663 Kumar S, Tamura K, Nei M (1993) MEGA: molecular evolu-
Craddock C, Hoeh WR, Gustafson RG, Lutz RA, Hashimoto J, tionary genetics analysis. Pennsylvania State University, Uni-
Vrijenhoek RC (1995a) Evolutionary relationships among deep- versity Park
sea mytilids (Bivalvia: Mytilidae) from hydrothermal vents and Little CTS, Vrijenhoek RC (2003) Are hydrothermal vent animals
cold-water methane/sulfide seeps. Mar Biol 121:477–485 living fossils? Trends Ecol Evol 18:582–588
Craddock C, Hoeh WR, Lutz RA, Vrijenhoek RC (1995b) Maddison DR, Maddison WP (2004) Mesquite, version 1.04 (build
Extensive gene flow in the deep-sea hydrothermal vent mytilid g21). Tucson, AZ. http://mesquiteproject.org
Bathymodiolus thermophilus. Mar Biol 124:137–146 Miyazaki J-I, Shintaku M, Kyuno A, Fujiwara Y, Hashimoto J,
Distel DL, Baco AR, Chuang E, Morrill W, Cavanaugh C, Smith Iwasaki H (2004) Phylogenetic relationships of deep-sea mussels
CR (2000) Do mussels take wooden steps to deep-sea vents? of the genus Bathymodiolus (Bivalvia: Mytilidae). Mar Biol
Nature 403:725–726 144:527–535
Drummond A, Rambaut A (2003) BEAST version 1.0.3: Bayesian Passamonti M, Scali V (2001) Gender-associated mitochondrial
evolutionary analysis sampling trees. Oxford, UK. http:// DNA heteroplasmy in the venerid clam Tapes philippinarum
evolve.zoo.ox.ac.uk/beast/ (Mollusca Bivalvia). Curr Genet 39:117–124
Felsenstein J (1985) Confidence limits in phylogenies: an approach Posada D, Crandall KA (1998) MODELTEST: testing the model
using the bootstrap. Evolution 39:783–791 of DNA substitution. Bioinformatics 14:817–818
Fisher CR, Childress JJ, Oremland RS, Bidigare RR (1987) The Rambaut A, Drummond A (2003) Tracer: a program for analysing
importance of methane and thiosulfate in the metabolism of the results from Bayesian MCMC programs such as BEAST &
bacterial symbionts of two deep-sea mussels. Mar Biol 96:59–71 MrBayes. Oxford, UK. http://evolve.zoo.ox.ac.uk/soft-
Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R (1994) DNA ware.html?id=tracer
primers for amplification of mitochondrial cytochrome C oxi- Scholin CA, Anderson DM (1994) Identification of group- and
dase subunit I from metazoan invertebrates. Mol Mar Biol strain-specific genetic markers from globally distributed Alex-
Biotechnol 3:294–299 andrium (Dinophyceae). I. RFLP analysis of SSU rRNA genes.
Giribet G, Carranza S, Baguna J, Riutort M, Ribera C (1996) First J Phycol 30:744–754
molecular evidence for the existence of a Tardigrada + Ar- Smith PJ, McVeagh SM, Won YJ, Vrijenhoek RC (2004) Genetic
thropoda clade. Mol Biolo Evol 13:76–84 heterogeneity among New Zealand species of hydrothermal vent
Gustafson RG, Turner RD, Lutz RA, Vrijenhoek RC (1998) A mussels (Mytilidae: Bathymodiolus). Mar Biol 144: 537–545
new genus and five species of mussels (Bivalvia, Mytilidae) from Swofford DL (1998) PAUP* Phylogenetic analysis using parsi-
deep-sea sulfide/hydrocarbon seeps in the Gulf of Mexico. mony (*and other methods). Sinauer, Sunderland
Malacologia 40:63–113 Wilcox TP, Garcı́a de Leon FJ, Hendrickson DA, Hillis DM (2004)
Hashimoto J, Okutani T (1994) Four new mytilid mussels associ- Convergence among cave catfishes: long-branch attraction and
ated with deep-sea chemosynthetic communities around Japan. a Bayesian relative rates test. Mol Phylogenet Evol 31:1101–
Venus 53:61–83 1113
Hoeh WR, Blakley KH, Brown WM (1991) Heteroplasmy suggests Wilcox TP, Zwickl DJ, Heath TA, Hillis DM (2002) Phylogenetic
limited biparental inheritance of Mytilus mitochondrial DNA. relationships of the dwarf boas and a comparison of Bayesian
Science 251:1488–1490 and bootstrap measures of phylogenetic support. Mol Phylo-
Hoeh WR, Stewart DT, Sutherland BW, Zouros E (1996) Multiple genet Evol 25:361–371
origins of gender-associated mitochondrial DNA lineages in Won Y, Hallam SJ, O’Mullan GD, Vrijenhoek RC (2003a) Cyto-
bivalves (Mollusca: Bivalvia). Evolution 50:2276–2286 nuclear disequilibrium in a hybrid zone involving deep-sea
Hoeh W, Stewart D, SI G (2002) High fidelity of mitochondrial hydrothermal vent mussels of the genus Bathymodiolus. Mol
genome transmission under the doubly uniparental mode of Ecol 12:3185–3190
inheritance in freshwater mussels (Bivalvia: Unionoidea). Evo- Won Y, Young CR, Lutz RA, Vrijenhoek RC (2003b) Dispersal
lution 56:2252–2261 barriers and isolation among deep-sea mussel populations
Huelsenbeck JP, Nielsen R, Bollback JP (2003) Stochastic mapping (Mytilidae: Bathymodiolus) from eastern Pacific hydrothermal
of morphological characters. Syst Biol 52:131–158 vents. Mol Ecol 12:169–184
Huelsenbeck JP, Ronquist F (2001) MRBAYES: Bayesian infer- Won Y-J, Maas PAY, Dover CLV, Vrijenhoek RC (2002) Habitat
ence of phylogenetic trees. Bioinformatics 17:754–755 reversal in vent and seep mussels: seep species, Bathymodiolus
Jacobs DK, Lindberg DR (1998) Oxygen and evolutionary patterns heckerae, derived from vent ancestors. Cahiers Biol Mar
in the sea: onshore/offshore trends and recent recruitment of 34:387–390
deep-sea faunas. Proc Natl Acad Sci USA 95:9396–9401

View publication stats