Testing alternative hypothesis of Neotrigonia (Bivalvia: Trigonioida) phylogenetic relationships using cytochrome c oxidase subunit I DNA sequences

12 Pages·1998·4.1 MB·English

by  Walter R Hoeh

MALACOLOGIA, 1998, 40(1-2): 267-278
TESTING ALTERNATIVE HYPOTHESES OF NEOTRIGONIA
(BIVALVIA:TRIGONIOIDA) PHYLOGENETIC RELATIONSHIPS USING
CYTOCHROME OXIDASE SUBUNIT DNA SEQUENCES
I
Walter R. Hoeh^ ^ Michael B. Black\ R. Gustafson\ Arthur E. Bogan^, RichardA. Lutz\ &
Robert 0. Vrijenhoek^
ABSTRACT
Uncertainties regardingthephylogenetic historywithin the Bivalvia have impeded attemptsto
understandevolutionwithinthegroup. Estimatingtheevolutionary relationshipssurroundingthe
Trigonioida has been especially problematic and has led to disparate hypotheses regarding (1)
the origin and subsequent diversification of unionoid bivalves and (2) autobranch gill character
state transitions. In orderto test alternative hypotheses of trigonioid phylogeny, 613 base pairs
of DNAsequenceofthecytochrome oxidase subunit I genewereanalyzedfrom 14speciesof
bivalves (ingroup) andthree speciesof non-bivalve mollusks (outgroup).All phylogeneticanaly-
ses, using either nucleotide or inferred amino acid sequences, produced trees that robustly
placed Neotrigonia (Trigonioida) as the sistertaxon to a monophyletic Unionoida. Furthermore,
the Autobranchia, Veneroida, and Mytiloida were supported as monophyletic groups in these
analyses,whereasthe Bivalviawas not.Thesephylogenetic relationshipssuggestthat(1) there
wasasingleinvasionoffreshwaterbyaunionoidbivalveancestor, (2)trigonioidratherthanven-
eroid bivalves gave rise to the Unionoida, (3) eitherthe eulamellibranchous orfilibranchous gill
condition has evolved multiple times within the Autobranchia, and (4) the molluscan bivalved
body plan may have evolved more frequentlythantraditional phylogenetic hypotheses suggest.
Key words: Bivalvia, Trigonioida, Unionoida, cytochrome oxidase I, mtDNA, phylogenetics,
convergence.
INTRODUCTION recently been proposed by Salvini-Plawen
& Steiner (1996; Fig. 1A) and Waller (1990;
The higher-level phylogenetic relationships Fig. IB). The former study is an exemplar in
within the Bivalvia are poorly understood at that explicit data matrices and phylogenetic
present (e.g., Allen, 1985). This statement is methodologies were presented. A major point
corroborated by the many disparate hypothe- ofdisagreementbetweenthetwophylogenies
ses of evolutionary relationships that have presented in Figure 1 is the placement of the
been proposed for the higher taxa within the Trigonioida, a relatively ancient bivalve taxon
Bivalvia (e.g., Purchon, 1990; Waller, 1990; that, althoughcurrentlydepauperate, wastax-
Starobogatov, 1992; Cope, 1996; B. Morton, onomically diverse during the Mesozoic (Cox
1996; Salvini-Plawen & Steiner, 1996). This et al., 1969; Allen, 1985). The Salvini-Plawen
plethora of phylogenetic hypotheses likely & Steiner hypothesis (Fig. 1A) indicates that
stems, in part, from arbitrary choices to ex- thetrigonioids are mostclosely related to pte-
clude certain types of characters and a gen- riomorph bivalves (represented herein by
eral lack of explicit and rigorous phylogenetic mytiloids), with the Veneroida closely related
analyses. Although major anatomical charac- to the Unionoida. Their proposed sistertaxon
ter suites distinguish two of the nominal sub- relationshipforthetrigonioid and pteriomorph
classes within the Bivalvia, that is, Proto- bivalves was supported by the shared pres-
branchia and Autobranchia, the ordinal-level ence of (1) byssate larvae and adults and (2)
relationships within each subclass are not abdominal sense organs in these taxa
stronglysupported by multipleshared-derived (Salvini-Plawen & Steiner, 1996). Alterna-
morphological features (e.g.. Waller, 1990). tively, the hypothesis of Waller (Fig. IB) indi-
Two contrasting phylogenies of the higher- catesthattheTrigonioida isthe sistertaxon to
level relationships within the Bivalvia have the Unionoida. However, no evidential sup-
'CenterforTheoreticalandAppliedGenetics,CookCollege, RutgersUniversity, NewBrunswick, NewJersey08903, USA:
[email protected]
^NorthCarolinaStateMuseumofNaturalSciences, P.O. Box29555, Raleigh, NorthCarolina,27626, USA.
^Corresponding authorandcurrentaddress: Departmentof Biological Sciences, KentState University, Kent, Ohio44242,
USA
267
268 HOEH ETAL.
Veneroid 1
Bivalves ^
Unionoid
3
Bivalves
Trigonioid
Bivalves Q
3>2
Mytiloid
Bivalves
Protobranch
Bivalves
Non-bivalve
Mollusks
FIG. 1. Bivalve relationships, basedon morphological characteristics, accordingto Salvini-Plawen & Steiner
(1996;A) andWaller(1990; B). Both hypothesesindicateasingleoriginforthe Bivalvia(1) andfilibranchous
lamellibranch gills (2) butdifferintheirimplicationsforeulamellibranch gill evolution. HypothesisAindicates
a single origin foreulamellibranch gills (3) while indicates eithertwo origins (solid bars) ora single origin
followed by a reversal tothefilibranch condition in trigonioids (hatched bars).
port for this relationship was provided by ple, disparate hypotheses of trigonioid evolu-
Waller (1990). tionary affinities. According to the hypothesis
TheevolutionaryaffinitiesoftheTrigonioida of Salvini-Plawen & Steiner (1996; Fig. 1A),
have been a contentious subject for over a the placement of the trigonioids is consistent
hundred years (e.g., Steinmann, 1888; Neu- with a single origin of eulamellibranch gills
mayr, 1889; Cox, 1960; B. Morton, 1987; (tissue-linked gill filaments) from filibranch
Healy, 1989, 1996) and are central to an un- gills. In contrast, the phylogenetic placement
derstanding of bivalve mollusk character evo- of trigonioids in the hypothesis of Waller
lution, since representatives of the single ex- (1990; Fig. IB) suggests that there were ei-
tant genus (Neotrigonia) have a mixture of ther (1) two origins of eulamellibranch gills or
seemingly primitive (e.g., filibranchous gills (2) a single origin of eulamellibranch gills fol-
[possessing ciliary-linked gill filaments], lack lowed by a reversal to filibranch gills in the
of posterior mantle fusion, nacreous shells; trigonioids.
Cox, 1960;Allen, 1985; B. Morton, 1987) and The phylogenetic uncertainty within the Bi-
derived features (e.g., multi-vesicular sperm valvia impedes the rigorous testing of evolu-
acrosome; Healy, 1989). This mosaic of tionary hypotheses. Therefore, the establish-
seemingly primitive and derived character ment of a robust phylogenetic hypothesis for
states has contributed tothe erection of multi- highertaxawithin the Bivalviawill enablecriti-
NEOTRIGONIA PHYLOGENETIC RELATIONSHIPS 269
cal evaluations of (1) trigonioid evolutionary example, Skibinski et al., 1994: Zouros et al.,
relationships and (2) hypotheses of auto- 1994: Hoeh et al., 1996). DNA was also iso-
branch bivalve gill character state transitions. lated from representatives of three additional
Molecular systematic analyses have been molluscan classes (i.e., Gastropoda, Polypla-
useful in situations where morphological cophora, and Scaphopoda) for use in gener-
analyses were inconclusive (e.g., Avise, ating outgroup sequences. Subsequently, a
1994).Therefore, mitochondrial DNA(mtDNA) 710bp fragment of COI was PCR amplified
sequences, obtained from the cytochrome and cycle sequenced for each of the 12 taxa
oxidase subunit I (COI) gene, were used to as deschbed elsewhere (Folmer et al., 1994).
construct a phylogenetic hypothesis for the Both strands of the COI fragment were se-
major bivalve lineages represented in Figure quenced from each of two individuals from
1.COIwaschosenforthisanalysisbecauseof each terminal taxon to guard against PCR-
its slow rateofevolution relativetoother mito- based contamination artifacts. The resulting
chondrial protein coding genes, relative ease 12 COI sequences, plus the five veneroid bi-
of unambiguous sequence alignment (e.g., valve COI sequences from Baldwin et al.
Brown, 1985: Simon etal., 1994; Russe etal., (1996), were readily aligned by eye using
1996), and demonstrated appropriateness for MacClade 3.05 (Maddison & Maddison,
this particularanalysis. 1992). Sixteen of the 17 OTUs produced se-
quences 613 bp in length, while that of
Geukensiawas 616 bp. The increased length
MATERIALSAND METHODS ofthe Geukensiasequencewasduetoan au-
tapomorphic, three nucleotide (single codon)
Organisms insertion event. These three contiguous nu-
cleotides, which are phylogenetically uninfor-
The molluscan taxa used in this study, with mative for the taxa considered herein, were
GenBank accession numbers and primary lit- deleted prior to all phylogenetic analyses. No
erature citation (where applicable), are as fol- additional hypothesized insertion or deletion
lows: (1) ingroup. Class Bivalvia, Subclass events were necessary to obtain the align-
Protobranchia, Order Solemyoida, Solemya ment utilized in the subsequent analyses.
velum (U56852), Order Nuculoida, Nucula The suitabilityofthe COI data set for phylo-
tenuis (U56851), Subclass Autobranchia, genetic analyses at the required hierarchical
Order Unionoida, Mutela rostrata (U56849), level was evaluated by plotting the substitu-
Amblemaplicata(U56841), Anodontacygnea tion pattern of transitions and transversions
( 56842), Margaritifera margarltifera for each codon position (e.g., Orti & Meyer,
(U56847), Order Trigonioida, Neotrigonia 1996). Furthermore, the degree of phyloge-
margarltacea (U56850), Order Mytiloida, netic signal within the COI data set was eval-
Modiolus modiolus (U56848), Geukensia de- uated using the g^ statistic of a random tree
missa (U56844), Order Veneroida (the five disthbution (from 100,000 random trees: e.g.,
veneroid COI sequences are from Baldwin et Hillis, 1991; Hillis & Huelsenbeck, 1992) as
al. [1996]), Corbicula fluminea (U47647), implemented in PAUP (Swofford, 1993). Phy-
Rangia cuneata (U47652), Mercenaria mer- logenetic analyses were carried out on the
cenaria (U47648), Mytilopsis leucophaeata COI nucleotide sequences using the maxi-
(U47649), Dreissena polymorpha (U47653), mum likelihood ([ML], DNAML in PHYLIP
(2) outgroup. Class Scaphopoda Dentalium 3.5c: Felsenstein, 1993), neighbor-joining
sp. (U56843), Class Gastropoda Lepetodrilus ([NJ], MEGA 1.02; Kumar et al., 1993), and
elevatus (U56846), Class Polyplacophora maximum parsimony ([MP], PAUP 3.1.1;
Katharinasp. (U56845). Swofford, 1993) algorithms. Katharina, the
only non-conchiferan mollusk taxon, was
Methods used to root the resulting topologies. Atransi-
tion/transversion ratio of 2.0 was utilized in
Total DNAwas isolated from somatic (man- the ML analyses and the gamma distance
tle) tissues of nine species of bivalves. Male (alpha = 0.5, using the Tamura-Nei model of
gonadaltissueswerespecificallyavoideddur- nucleotide sequence evolution) was used to
ing dissections to prevent comparisons of generate the pair-wise genetic distances for
non-orthologoussequences (duetotheactual the NJ analyses. This particular distance
orpotential presence ofdoubly uniparental in- takes into consideration among-site substitu-
heritance of mtDNA in some bivalve taxa: for tion rate variation (e.g., Yang, 1996). Further-
270 HOEH ETAL.
more, MP and NJ (again using gamma dis- were invariant, while 315 (102from 1stcodon
tances, alpina=0.5) analyseswereconducted positions; 59 from 2nd; 154 from 3rd) were
on the inferred COI amino acid sequences phylogenetically informative using the parsi-
(using the Drosophila mtDNA genetic code). mony criterion.Analysis ofthetree length dis-
Multiple random terminal taxon addition order thbution of 100,000 randomly generated
runs, combined with global branch rearrange- trees, using all 17 taxa, suggests that there is
ment options, were employed to generate a significant amount of hierarchical structure
topologies from ML and MP analyses. These within the transformed COI data s«et (g^ =
options increasedtheprobabilityoffindingthe -0.794; with 384 variable sites, p 0.01;
actual besttopology undereach ofthetwoop- Hillis & Huelsenbeck, 1992). This analysis
timalitycriteria (e.g., Hendy et al., 1988: Mad- was repeated using only the Corbicula. Den-
dison, 1991). The robustness of the resulting tallum. Katharina. Lepetodhlus. Modiolus,
topologies was evaluated by bootstrap analy- Neothgonia. and Solemya COI sequences
ses (1,000 replicates for MP and NJ, 100 (seven taxa) in order to minimize the number
replicates for ML). of closely related taxa present in the hierar-
The best COI-based topology derived from chical structure analysis. Significant structure
the DNAML analysis was compared with the was still present in this truncated COI data
phylogenetic hypotheses presented in Figure matrix (g^ = -0.527; with 295 variable sites, p
1 using the Kishino-Hasegawa test (paired z < 0.01), which suggests that the hierarchical
test; Kishino & Hasegawa, 1989) as imple- structure present in the original data matrix
mented in DNAML.Tothisend,thetopological was not solely due to the presence of closely
constraints option in PAUP was used to gen- related taxa. The findings from the plots of
erate82usertrees(alltrees<fivestepslonger substitution pattern and g, statistics are con-
than the shortesttrees found by PAUP) repre- sistent with the hypothesis that significant
senting the two tree topologies (forthe partic- phylogenetic signal exists in the transformed
ulartaxaevaluatedherein)in Figure 1. Eachof COI nucleotide data matrix and validate the
these user trees was then compared to the latter's use in this particularphylogeneticcon-
bestDNAMLtreebytheKishino-Hasegawaal- text (e.g., Swofford et al., 1996). Since amino
gorithm.Thistestevaluatedthesignificanceof acid substitution rates are lower than the un-
any potential incongruence between the mor- derlying nucleotide substitution rates (e.g., Li
phology- (Figure 1) and COI-basedtopologies & Graur, 1991), itfollows thatthe inferred COI
(Swofford et al., 1996). In addition, character amino acid sequences are not saturated and,
optimization, using MacClade (Maddison & therefore, also appropriate for this level of
Maddison, 1992), wascarried outon the COI- phylogenetic analysis.
based topologies to investigate their implica- The best tree produced by ML analysis of
tions for morphological character evolution the transformed COI nucleotide mathx is pre-
within the Bivalvia. sented in Figure 3 along with bootstrap per-
centages for the NJ (above branches, 1,000
replicates), MP (below branches, 1,000 repli-
RESULTS cates), and ML (in parentheses, 100 repli-
cates) analyses. Only bootstrap percentages
Scatter plots of the relationship between greaterthan 50% are shown. This topology is
the number of transitional and transversional largely congruent with the best trees pro-
substitutions at each of the three codon posi- duced by MP (three equally parsimonious
tions and the proportion of nucleotide differ- trees) and NJ analyses (trees not shown).
ences (all positions) for the COI sequences However, the topological relationships of the
revealed that only transitional substitutions at non-autobranch taxa on the best MP trees
the third codon position had reached satura- were identical to those portrayed in the trees
tion (Fig. 2). Because saturated categories of derived from analyses of the inferred CO!
substitution can contribute to erroneous esti- amino acid sequences (see below).
mates of evolutionary history (e.g., Swofford The Kishino-Hasegawa test results (Table
et al., 1996), all first and second position sub- 1) indicate that the morphology-based topol-
stitutions together with only transversions at ogy represented in Figure 1Awas significantly
the third codon position were included in the worse (p < 0.05) than the best topology from
COI nucleotide data matrix used for phyloge- MLanalyses ofthetransformed COI data ma-
netic analyses. Of the 613 nucleotide posi- trix (Figure 3). However, the morphology-
tions in the transformed COI data matrix, 229 based topology represented in Figure 1 was
NEOTRIGONIA PHYLOGENETIC RELATIONSHIPS 271
272 HOEH ETAL.
-Dreissena
100
100
1(100) -Mytilopsis
(64)
Veneroid
-Rangia
(57) Bivalves
19090 -Mercenaria
(98)
66 -Corbicula
82
(51) -Modiolus
_25_ ~] Mytiloid
75 J Bivalves
(83) Geukensia
71 Amblema
62
(55)
88 -Margaritifera Unionoid
Bivalves
97 Anodonta
-
83
(67)
.m. Mutela
97
(100)
Neotrigonia
75
55
NEOTRIGONIA PHYLOGENETIC RELATIONSHIPS 273
-Dreissena
100
100
-Mytilopsis
SL.
Veneroid
-Mercenaria Bivalves
_ZS_
.5a_
98 -Rangia
-Corbicula ""
55
Modiolus
~] Mytiloid
_99f3l_ J Bivalves Autobranch
-Geukensia
Bivalves
~
Margaritifera
51
Anodonta
-84_ Unionoid
Bivalves
83 Amblema
56
53
_2Z_ Mutela
98
Trigonioid-I
Neotrigonia Bivalve
53
Protobranch
Nucula
Bivalve
Dentalium (TuskShell)
Protobranch
Solemya
Bivalve
_Z2_
55
Lepetodrilus (Snail)
Katharina (Chiton)
FIG. 4. Besttreetopology produced by neighbor-joining analysisofthe inferred CO! aminoacid sequences.
Numerals are bootstrap percentages for NJ (above branches, 1,000 replicates) and MP (below branches,
1.000 replicates) analyses. Only bootstrapvaluesgreaterthan 50%are shown.
sels (Unionoida) than to the filibranchous placement of Neotrigonia as sister taxon to
mytiloids (mean level of bootstrap support = the Unionoida.
98%). This level of bootstrap support sug- Another noteworthy aspect of all of the
gests a robust resolution of the evolutionary above phylogenetic analyses was the ab-
relationships of Neotrigonia among the taxa sence of support for bivalve mollusk mono-
utilized herein (Hillis & Bull, 1993). phyly. In the phylogenetic hypothesis repre-
This particular phylogenetic placement of sented in Figure 3, the protobranch bivalves,
Neotrigonia was also manifest in the best Nucula and Solemya. are portrayed as a
trees (not shov\/n) constructed from analyses clade,withthegastropod, Lepetodrilus. asthe
ofthe COI nucleotidesequenceswhen (1) the sister taxon to that clade. The phylogenetic
native (untransformed) COI sequences were hypothesis represented by Figure 4 portrays
utilized (ML, MP and NJ analyses), (2) all Nuculaasthesistertaxon oftheAutobranchia
codon positions were coded fortransversions while Solemya is the sister taxon to Lepeto-
only (ML, MP, and NJ analyses), and (3) only drilus. While our analyses were somewhat
first and second codon positions were used limited due to the relatively small number of
(ML, MP, and NJ analyses). These results nucleotides and taxa analyzed, the fact that
taken together are consonant with the hy- none of the best trees or bootstrap trees gen-
pothesis that there is a strong phylogenetic erated from theseanalysesgavesupportfora
signal in the COI sequences supporting the monophyletic Bivalviasuggeststhatthe impli-
274 HOEH ETAL.
cations of these results be seriously consid- suggests that the protobranch bivalve genera
ered. Nuculaand Solemyaare more closely related
to the snail, Lepetodrilus, than to the other bi-
valve taxa in the analysis. Thus, the shared
DISCUSSION presence of bipectinate gills and hypobran-
chial glands (J. E. Morton, 1988), esophageal
Morton (1987) argued, based partially on and stomach similarities (Salvini-Plawen,
the anatomical discontinuities between trigo- 1988), ultrafiltration site similarities (Andrews,
nioids and unionoids, that Neotrigonia was 1988), oxygen transport molecule similarities
a transitional taxon, phylogenetically inter- (Mangum et al., 1987), and flattened pedal
mediate between the presumed ancestral areas in both protobranch bivalves and primi-
protobranch bivalves and the more derived tive gastropods may be due to shared com-
pteriomorph bivalves. This hypothesis is con- mon ancestry rather than to the retention of
sistent with that of Salvini-Plawen & Steiner ancestral characterstates.
(1996; Fig. 1A). However, the results of the The phylogenetic relationships of Neotrigo-
COI sequence analyses (Figs. 3, 4) strongly nia. Nucula. and Solemya. as deduced from
suggest that the extant representative of the the COI analyses presented herein (e.g..
Trigonioida, that is, the genus Neotrigonia, is Figs. 3, 4), suggest that a significant amount
the sister taxon to unionoid bivalves, as sug- of convergent anatomical and conchological
gested in the hypothesis of Waller (1990; Fig. evolution hastaken placewithin the Mollusca.
IB). This phylogenetic propinquity is sup- There is a great deal of precedent for this
ported by similarities in shell structure (Taylor statement (e.g., Allen, 1985; Purchon, 1990;
etal., 1969, 1973;Tevesz & Carter, 1980), gill Davis, 1994; Salvini-Plawen & Steiner, 1996).
speculation (Taylor et al., 1969, 1973), sperm An important evolutionary implication that
morphology (Popham, 1979; Healy, 1989), stems from the phylogenetic placement of
and gill cilia patterns (Atkins, 1937; Tevesz, Neotrigonia as sister taxon to a freshwater
1975). Furthermore, the indicated monophyly mussel clade is that it corroborates previous
of the Unionoida is consistent with the hy- hypotheses of convergence (e.g.. Waller,
pothesis of a single invasion of freshwater by 1990) in the evolution of autobranch bivalve
the ancestral unionoid. This finding corrobo- gill structure, i.e., either the filibranchous or
rates the hypothesis of a dramatic evolution- the eulamellibranchous gill condition evolved
arytransition in larval morphology, that is, be- at least twice in the evolutionary history of
tween glochidium and haustorium/lasidium these bivalve taxa (Fig. 5A). The latter possi-
morphology, during unionoid phylogenesis. bility is favored by Waller (1990). The evolu-
Evaluating the directionality of this character tion of eulamellibranchous gill organization
state transition will require further, broad- (which facilitated larval brooding) may have
scale phylogenetic analyses. been a necessary antecedent to the success-
Another interesting result is the placement ful colonization and subsequent marked evo-
of Solemya (in Figs. 3, 4) and Nucula (in Fig. lutionary diversification in freshwater habitats
3), both protobranch bivalves, amongthenon- by unionoid bivalves.
bivalve outgroup taxa. This observation is not The implications ofthe hypothesized phylo-
anartifactoftheparticularrootingschemeem- genetic relationships of Nuculaand Solemya,
ployed in Figures 3 and 4. It is not possible to as inferred from analyses of COI sequences,
root either of these topologies such that all of are more profound. It is suggested thatthe bi-
the bivalve taxa represented therein form a valved phenotype has evolved at least three
monophyletic group. In all of the phylogenetic times during the evolution ofthe Mollusca; (1)
analysesoftheCOI sequences, Solemyawas in the ancestorofthe Juliidae, a relatively de-
either(1)thesistertaxonto Lepetodrilusor(2) rived family within the opisthobranch gas-
in a clade with Nucula and Lepetodrilus. tropods (not represented in the analyses
Therefore, these analyses provide some sup- herein), (2) in the ancestor of the Autobran-
port for the hypothesis that the currently-rec- chia, and (3) in the ancestor of the Proto-
ognized molluscan taxon Bivalvia is a poly- branchia (assuming a monophyletic Proto-
phyletic assemblage. The non-monophyletic branchia, Fig.5B). Ifthegenera Tuarangiaand
status of the Bivalvia was supported by a re- Pseudomyonaarefoundtobebivalved mono-
cent phylogenetic analysis of 18S rDNA se- placophorans (Runnegar, 1983) rather than
quences(Adamkewiczetal., 1997;fig.2). Fur- autobranch bivalves (Mackinnon, 1982; Berg-
thermore, the topology in our Figure 3 Madsen, 1987),andiftheSolemyoidaandNu-
NEOTRIGONIA PHYLOGENETIC RELATIONSHIPS 275
#S^
¿•^* />^^
.N<» ^ ^.<>^
'^ jxo .<^ .d* x)"^ ej^" o"^ í^ á^ ®
^^
v^^ <J
lamellibranch gills
absent
filibranch
eulamelh'branch
equivocal
^
eVç»? ^^ <^Ä <c0^^ íJOr^^$i».4^>^>J5v"^J;^^.<^^<^^/\tí^^' ^?^> ^4<ö<
ris
bivalveshell
absent
^I HI
present
FIG. 5. Morphologicalcharacteroptimization using MacCladeonthebesttopologyproducedfrom maximum
likelihood analysis of the transformed CO! nucleotide matrix using Katharina as the outgroup. A. The most
parsimonious estimate of autobranch bivalve gill character state transitions for the taxa included in these
analyses isthateitherthefilibranchousoreulamellibranchouscharacterstateevolved at leasttwice. B. The
most parsimonious estimate of character state transitions forthe taxa included in these analyses suggests
thatthe bivalved phenotype hasevolved at leasttwice. Justificationforthe useof Katharinaastheoutgroup
forthemolluscantaxaincluded inthisanalysisisprovided by numerousstudies (e.g., Salvini-Plawen. 1980,
1985, 1988, 1990; Wingstrand, 1985; Eernisse, etal., 1992; Salvini-Plawen & Steiner, 1996).
276 HOEH ETAL.
culoida had independent origins (Purchon, tural Experiment Station Contribution #D-
1978; Salvini-Plawen & Steiner, 1996), the bi- 32104-3-98andwassupported bystatefunds
valvedconditionwould have hadtoevolvedat and NSFgrantsOCE9633131 OCE9302205,
,
leastfive times within the Mollusca. This strik- and OCE8917311 (to R. V. and R.A. L.).
ing assessment is nonetheless not totally un-
expected when evaluated in the context of (1)
the great evolutionary mutability of body plan LITERATURE CITED
exemplifiedbythephylum Mollusca(e.g.,J. E.
Morton & Yonge, 1964; Allen, 1985; Willmer, ADAMKEWICZ, S. L., M. G. HARASEWYCH, J.
1990) and (2) the multiple origins of bivalved BLAKE, D. SAUDEK&C. J.BULT 1997,Amole-
external shells in fourphyla (Thomas, 1988). cular phylogeny ofthe bivalve mollusks. Molecu-
Under the hypothesis of a polyphyletic Bi- larBiologyandEvolution. 14: 619-629.
valvia, the degree of morphological conver- ALLEN, J.A., 1985, The Recent Bivalvia:theirform
gence on the bivalved body plan varies con- and evolution. Pp. 337-403, in E. R. trueman &
M. R. CLARKE, eds.. TheMollusca, Vol. 10, Evolu-
siderably within the extant mollusks. In the tion. Academic Press, Inc., Orlando, Florida,
case of the opisthobranch gastropod genus U.S.A.
Julia, the shell has converged on the mono- ANDREWS, E. ., 1988, Excretorysystemsofmol-
myarian (= single adductor muscle) bivalve luscs. Pp. 381-448, in E. R. TRUEMAN & M. R.
condition while the soft anatomical character- CLARKE, eds., The Mollusca. Vol. 11, Form and
istics are clearly those of a gastropod (Kay, Function.Acaóemlc Press, Inc., San Diego, Cali-
1968). However, the degree of convergence fornia, U.S.A.
in body plans between autobranch and proto- ATKINS, D., 1937, On the ciliary mechanisms and
bbortahncshheblilvaalnvdesaisnamtuocmihcaglreacthearraacntderinsvtoaltvees.s iicnnrgtoesdrcreoevpliiaccteaislonSoschniipetsnhceoef.gill7las9m.:elQ3lu3ia9brr-ta3en7rcl3hy.s.JoPuarrntalII:ofsoMrit--
The phylogenetic hypothesisdisplayed in Fig- AVISE, J. C. 1994, Molecularmarkers, naturalhis-
ure 3 suggeststhat in thedistinctevolutionary tory and evolution. Chapman & Hall, New York,
histories of the autobranch and protobranch NewYork. U.S.A. 511 pp.
mollusks, in addition to the convergent bi- BALDWIN, B. S.. M. BLACK, O. SANJUR, R.
vofalhveeaddshaelnld, trhaedurleaw,e(r2e)ocroignivnesrgofenltab(i1al) lpoaslspes,s G19U9S6T.AAFSdOiaNg,nosR.tiAc.LmoUlTeZcu&laRr. marVkReIrJEfoNrHOzeEbKr,a
(3) origins of the dimyarian adductor muscle mussels (Dreissenapolymorpha) and potentially
condition, and (4) origins of fibrous ligament. cuola-rocMcaurrirniengBiboivlaolgvyesa:ndmiBtiooctheocnhdnroilaolgyC.OI5.:9M-o1l4e.c-
Thus, the evolutionary mutability of body BERG-MADSEN, V., 1987, Tuarangia hom Born-
planswithinthe Mollusca may begreaterthan holm (Denmark) and similaritiesin Baltoscandian
that suggested bythe traditional classification and Australasian late Middle Cambrian faunas.
schemes. Phylogenetic evaluations of addi- Alcheringa. 11: 245-259.
tional molecular and morphological data sets BROWN, W. M., 1985, The mitochondrial genome
are neededtotestthe hypothesized polyphyly of animals. Pp. 95-130, in R. j. macintyre, ed.,
ofthe Bivalviaandfurtherdecipherpatternsof Molecular evolutionary genetics. Plenum, New
molluscan body plan evolution. York. NewYork, U.S.A.
COPE,J. W., 1996,TheearlyevolutionoftheBi-
valvia. Pp. 361-370, inJ. TAYLOR, ed.. Originand
ACKNOWLEDGMENTS evolutionary radiation of the Mollusca. Oxford
University Press. London, England.
Wethank B. Baldwin for helpwith sequenc- COtXh,e LB.ivRa.l,vi1a9.60P,rTohceoeudgihntgssonofthteheclMasasliafciocaltoigoincaofl
ing the veneroids used in this analysis. S. L. SocietyofLondon. 34: 60-88.
Baldauf, D.J. Berg, S. I. Guttman, S. L. Neale, COX, L. R., R NUTTALL & E. R. TRUEMAN,
M. O'Connell, P. O'Reilly, D. T Stewart, B. W. 1969, General features of the Bivalvia. Pp.
Sutherland, and T. R. Waller provided helpful 2-129, in R. MOORE, ed.. Treatise on inverte-
comments on earlier versions of this manu- bratepaleontology PartN,Vol. 1.TheGeological
script. Neothgonia specimens were provided Society of America, Inc. and The University of
by S. Boyd, and Mutela specimens were pro- Kansas, Lawrence, Kansas, U.S.A.
DAVIS, G. M., 1994, Molecular genetics and taxo-
vided byG. Soliman. R. Bielerprovided assis-
tance with literature acquisition. We also wish 2n:o3m-i2c3d.iscrimination. The Nautilus. Supplement,
to thanktwo anonymous reviewers and G. M. EERNISSE, D. J., J. S. ALBERT & F E. ANDER-
Davis forcomments on earlierversions ofthe SON, 1992,AnnelidaandArthropodaare notsis-
manuscript. This study is NewJerseyAgricul- ter taxa: a phylogenetic analysis of spiralian