SPECIATION AND DISPERSAL ALONG CONTINENTAL COASTLINES AND ISLAND ARCS IN THE INDO-WEST PACIFIC...

ORIGINAL ARTICLE

doi:10.1111/j.1558-5646.2011.01255.x

SPECIATION AND DISPERSAL ALONGCONTINENTAL COASTLINES AND ISLANDARCS IN THE INDO-WEST PACIFIC TURBINIDGASTROPOD GENUS LUNELLASuzanne Williams,1,2 Deepak Apte,3 Tomowo Ozawa,4 Fontje Kaligis,5 and Tomoyuki Nakano6

1Zoology Department, Natural History Museum, London SW7 5BD, United Kingdom2E-mail: [email protected]

3Bombay Natural History Society, Hornbill House, S. B. Singh Rd, Mumbai 400 001, India4Department of World Heritage, Cyber University, Nagoya Office, Ikegamicho-2-7-1-203, Chikusa-ku,

Nagoya 464-0029, Japan5Marine Science Study Program, Faculty of Fisheries and Marine Sciences, University of Sam Ratulangi,

Manado 95115 Indonesia6Department of Geology and Palaeontology, National Museum of Nature and Science, 3-23-1 Hyakunin-cho, Shinjuku-ku,

Tokyo 169-0073, Japan

Received August 12, 2010

Accepted February 1, 2011

Species trees were produced for the Indo-West Pacific (IWP) gastropod genus Lunella using MrBayes, BEAST, and ∗BEAST with

sequence data from four genes. Three fossil records were used to calibrate a molecular clock. Eight cryptic species were recognized

using statistical methods for species delimitation in combination with morphological differences. However, our results suggest

caution in interpreting ESUs defined solely by the general mixed Yule Coalescent model in genera like Lunella, with lower dispersal

abilities. Four almost entirely allopatric species groups were recovered that differ in ecology and distribution. Three groups occur

predominantly along continental coastlines and one occurs on island arrays. Sympatric species occur only in the torquata and

coronata groups along coastlines, whereas species in the cinerea group, distributed in two-dimensional island arrays, occur in

complete allopatry. Dispersal along island arcs has been important in the maintenance of species distributions and gene flow

among populations in the cinerea group. The emergence of new islands and their eventual subsidence over geological time has

had important consequences for the isolation of populations and the eventual rise of new species in Lunella.

KEY WORDS: Biogeography, Indo-West Pacific, speciation, species delimitation, tectonics.

Several interrelated exogeneous factors play pivotal roles in the

evolution of shallow-sea organisms. Four of the most impor-

tant factors are environment, geography, tectonics, and sea level.

These factors, in combination with biological differences, espe-

cially ecology and developmental mode, greatly affect patterns

of distribution and rates of speciation (Valentine and Jablonski

1982, 1983; Hellberg 1998; Collin 2003; Williams and Reid 2004;

Hoeksema 2007). In this study, we use a marine gastropod genus

to investigate the effects of these factors on species distributions

and patterns of speciation. In particular, we compare patterns

between species occurring in two different habitat types: one-

dimensional continental coastlines versus two-dimensional island

arrays.

The geography of a shallow-sea realm can be markedly af-

fected by tectonic events and changes to sea level. Continents

move laterally over time, coastlines change and islands subside

or emerge, resulting in the opening or closing passages and the

creation or loss of shallow-water habitat (Valentine and Jablonski

1 7 5 2C© 2011 The Author(s). Evolution C© 2011 The Society for the Study of Evolution.Evolution 65-6: 1752–1771

SPECIATION ON COASTLINES AND ISLAND ARCS IN LUNELLA

1982; Wilson and Rosen 1998; Hall 2001; Paulay and Meyer

2002; Kuhnt et al. 2004; Hall 2009). These changes to geography

result in changes to climate and environment, including salinity

(affected by freshwater runoff), sea-water temperature and trophic

resources (Valentine 1971; Hall 2009). Combined, these in turn

alter ocean current velocities and direction of flow (Kuhnt et al.

2004; Hall 2009), affecting gene flow among populations (Paulay

and Meyer 2002). Organisms react to all these changes by ex-

panding or reducing their distribution or by evolving in response

to changing local selection pressures (Valentine and Jablonski

1982), profoundly affecting the biogeography of a region.

The vetigastropod genus Lunella (family Turbinidae) is a

model group to test how species distributions and patterns of spe-

ciation differ between species distributed along continental coast-

lines and island arcs and to identify key drivers. The genus is found

only in the IWP, occurring from the western Indian Ocean, along

the east coast of Africa and the Arabian Peninsula, across to the

more continental islands of the West Pacific (e.g., New Zealand,

Fiji and American Samoa). It does not occur in the East Pacific or

Hawaii or on the oceanic island archipelagos of the Pacific (e.g.,

Marianas, Tuvalu or Kiribati). Individual species occur on rocky

shores in the intertidal zone either along continental coastlines

or on island arrays. Fertilization is external in all species that

have been examined (e.g., L. torquata, Joll 1980; Ward and Davis

2002). In L. smaragdus the young hatch 24 h after fertilization and

settlement probably occurs within four days (Grange 1976). Stud-

ies of reproductive behavior have shown similar development for

other turbinid species (e.g., Megastraea undosa, Guzman del Proo

et al. 2003; Turbo marmoratus, Yamaguchi 1993; Dwiono et al.

2001), and some turbinid species produce benthic egg masses with

crawl-away larvae (e.g., Turbo radiatus, Eisawy and Sorial 1974;

T. stenogyrus; Kono and Yamakawa 1999). As such, dispersal in

Lunella is likely to be limited between populations separated by

hundreds of kilometers. The fossil record, as for many rocky-

shore taxa, is poor, but some fossils do exist that can be used to

calibrate a molecular clock.

In this study, we use a near-complete, species-level phy-

logeny of the genus Lunella to investigate species distributions

and patterns of speciation. In particular, we investigate how these

differ between species distributed along continental coastlines

and island arcs. We compare patterns of distribution with a robust

phylogeny dated with fossil calibrations to infer mechanisms of

speciation. As cryptic species have already been found in Lunella

(Williams 2007), we use sequence-based methods to delimit evo-

lutionary lineages.

Materials and MethodsSAMPLING AND IDENTIFICATION

A total of 229 specimens of Lunella were sequenced, with sample

sizes ranging from seven to 95 per species recognized prior to

this study. We include four of the five species recognized in Alf

and Kreipl (2003), plus the recently identified L. jungi and L.

ogasawarana. We also include both Australian (L. undulata and L.

torquata) and New Zealand endemics (L. smaragdus) that have at

times been referred to the genus Turbo because molecular studies

show that they belong to Lunella (Williams and Ozawa 2006;

Williams 2007; this study). We were unable to obtain samples

from only one species; L. viridicallus from the Red Sea. As in our

previous papers, we continue to use the name L. coreensis for the

(usually) imperforate species found in Japan and Korea and use L.

granulata for the (usually) umbilicate species found in China and

Vietnam. The name L. moniliformis (used in Alf and Kreipl 2003)

is a synonym of L. granulata (Cernohorsky 1974). All specimens

are listed in Table S1.

Each species was sampled as widely as possible; as a result,

samples in this study cover the entire geographic range of the

genus although some areas of possible endemism were not sam-

pled (e.g., Red Sea, Madagascar, western Indonesia). Sequence

was obtained for COI from all but two samples, and 28S, 16S, and

12S sequences were obtained for at least three specimens from

each deeply divergent, reciprocally monophyletic clade identi-

fied in preliminary neighbor joining analysis of COI (not shown).

Previously published Lunella sequences are from Williams and

Ozawa (2006), Williams (2007) and Nakano et al. (2007).

In MrBayes analyses only, Lunella sequences were analyzed

together with previously published sequences of outgroup species

from the subfamily Turbininae (12 species in total from five differ-

ent genera; sequences from Williams and Ozawa 2006; Williams

2007).

DNA EXTRACTION, AMPLIFICATION, AND

SEQUENCING

The DNA extraction and amplification protocols described by

Williams and Ozawa (2006) and Williams et al. (2010) were

used to amplify portions of four genes: the nuclear 28S rRNA

gene (28S: 1496 bp) and three mitochondrial genes: cytochrome

oxidase subunit I (COI: 709 bp), 16S rRNA (16S: ∼610 bp),

and 12S rRNA (12S: ∼685 bp). Sequence reactions were per-

formed directly on purified PCR products using a BigDye Termi-

nator version 1.1 Cycle Sequencing Kit (Applied Biosystems, Life

Technologies Corporation, Carlsbad, CA) and run on an Applied

Biosystems 3730 DNA Analyser automated capillary sequencer.

Sequencing and PCR primers are listed in Table S2.

SEQUENCE ANALYSIS

Double stranded sequences were obtained for all gene fragments

and were edited using Sequencher (version 4.6 and version

4.8, Gene Codes Corporation, Ann Arbor, MI). All new se-

quences have been deposited in GenBank (accession numbers in

Table S1).

EVOLUTION JUNE 2011 1 7 5 3

SUZANNE WILLIAMS ET AL.

Alignments of both COI and 28S were unambiguous (both

including and excluding outgroups), requiring no insertions and

were done by eye in MacClade (version 4.08 OSX; Maddison

and Maddison 2003), checking amino acid translations in

COI. Alignment of mitochondrial ribosomal genes was more

complicated and these sequences were aligned using MAFFT

(Multiple Alignment using Fast Fourier Transform); a high speed

multiple sequence alignment program (version 6.0; Katoh et al.

2002; online http://align.bmr.kyushu-u.ac.jp/mafft/software/).

This programme has been shown to produce better alignments

than many more traditionally used programmes (e.g., Katoh et al.

2002). The Q-INS-i option was used, which is recommended

for a global alignment of highly diverged ncRNAs as secondary

structure of RNA is considered (Katoh and Toh 2008), and the

offset value was set at 0.1, as long gaps were not expected.

Poorly aligned sites in mitochondrial rRNA alignments were

identified using Gblocks Server (0.91b, Castresana 2000;

http://molevol.cmima.csic.es/castresana/Gblocks_server.html)

and removed from analyses. Parameters used in Gblocks allowed

for smaller final blocks and gap positions within the final blocks.

SPECIES DELIMITATION

Likely cryptic taxa were found in previous molecular studies

(Williams 2007) so we were concerned with delimiting species

correctly. We used sequence data to identify independently evolv-

ing lineages (or evolutionary significant units, ESUs; Moritz

1994) using the method of Pons et al. (2006), by estimating the

transition from coalescent to speciation branching patterns on an

ultrametric tree. This analysis exploits the predicted difference in

branching rates within and between species. Branching patterns

among species reflect the timing of speciation events (Yule 1924),

whereas branching rates within a species and among populations

reflect neutral coalescent processes (Kingman 1982; Monaghan

et al. 2009). This method identifies the point where a transition

between branching rates is most likely and the likelihood is com-

pared with a null model that all sequences are derived from a

single species (Pons et al. 2006; Monaghan et al. 2009).

We used the single-threshold, general mixed Yule-coalescent

(GMYC) model as implemented by SPLITS, code written by

T. Ezard, T. Fujisawa and T. Barraclough in R (version 2.10,

http://cran-project) to identify species from sequence variation in

COI, the “barcoding” gene. We then compared the validity of a

single gene approach with that of a multiple gene approach (based

on two loci) by comparing our results for COI with those based

on a tree produced by concatenation of all four genes.

Ultrametric trees for GMYC analyses were produced us-

ing BEAST (version 1.5.4.; Drummond and Rambaut 2007) us-

ing a relaxed lognormal clock, but without any fossil calibra-

tions and a fixed mean rate of substitutions set to one. We used

a constant coalescent prior, which is thought to be more con-

servative than a Yule prior for delimiting species (Monaghan

et al. 2009). Nucleotide substitution models used in prelimi-

nary analyses in BEAST were determined by MrModelTest us-

ing the hierarchical likelihood-ratio test (version 2.1, J. Nylan-

der, www.ebc.uu.se/systzoo/staff/nylander.html). Where multiple

models were suggested, the simplest was chosen. The best model

for each dataset was determined to be HKY + I + G for 16S and

COI, GTR + I + G for 12S and GTR + I for 28S. However, based

on preliminary analyses, nucleotide substitution models used in

GMYC/BEAST analyses were slightly different (HKY + I + G

for 16S and 12S, HKY + I for 28S and SDR06 for COI).

After preliminary runs, priors for various parameters were

changed, resulting in improved effective samples size (ESS) val-

ues. Each analysis ran for 50,000,000 generations with sample fre-

quency of 1000. The final trees were calculated based on 49,000

trees (after burnin of 1,001 generations) with maximum clade

credibility and median node heights. Length of burnin was deter-

mined by examination of traces in Tracer (version 1.5, avail-

able from http://beast.bio.ed.ac.uk/Tracer). Support for nodes

was determined using posterior probabilities (PP; calculated by

BEAST).

Average genetic distances for the “bar-coding” gene COI

were calculated using HKY + G + I (the model used in

phylogenetic analyses) and Kimura’s (1980) two-parameter

distances (K2P). The latter were calculated so that values

can be directly compared with other studies. Genetic dis-

tances and P ID (Liberal) among clades were calculated us-

ing a species delimitation plugin (B. Masters, V. Fan and H.

Ross; www.cebl.auckland.ac.nz/∼hros001/Software/SpDelim/)

for GeneiousPro (version 5.0.2; Drummond et al. 2009). P ID

(Liberal) is the mean probability of making a correct identifica-

tion of an unknown specimen of the focal species using BLAST

(best sequence alignment), DNA Barcoding (closest genetic dis-

tance) or placement on a tree, with the criterion that it falls sister

to or within a monophyletic species clade.

PHYLOGENY RECONSTRUCTION

A species-level phylogeny was produced using Bayesian infer-

ence as implemented in the programme BEAST. The analysis

was similar to that used for the GMYC analysis, but we used

a Yule prior, which is more appropriate for species-level phylo-

genies and assumes a constant speciation rate among lineages.

We used concatenated sequence from all four genes, and a single

individual to represent each species. Sequence variation was par-

titioned among genes and gene-specific nucleotide substitution

model parameters were used, with each gene allowed to evolve

at a different rate. We used similar nucleotide substitution mod-

els as were used in the 4gene BEAST/GMYC analysis (HKY +G + I for COI and 12S, HKY + I for 28S, HKY + G for 16S).

We used an uncorrelated relaxed, lognormal clock with three

1 7 5 4 EVOLUTION JUNE 2011


calibrations based on fossil evidence (see below). We ran the

analysis for 70,000,000 generations, sampling every 10,000 gen-

erations. The final species tree was a maximum clade credibility

tree with median node heights based on 6500 trees (after burnin

of 500 trees). In all analyses, sequences were assigned to species

based on results from species delimitation tests.

We also used the new Species Tree Ancestral Reconstruc-

tion (∗BEAST) method to confirm the topology obtained by the

BEAST analysis. It has been suggested that with some datasets

this method is superior to simple concatenation of multiple gene

partitions with respect to phylogenetic inference (Heled and

Drummond 2010). The ∗BEAST method coestimates gene trees

and a species tree and allows for the incorporation of multiple

exemplars of each species and the independent evolution of each

gene without fixing a single topology across genes (Heled and

Drummond 2010). Unlike “normal” BEAST analyses, it com-

bines priors for both speciation events and population genetics,

providing the opportunity to model for intraspecies polymorphism

and incomplete lineage sorting in phylogenetic estimation proce-

dures (Heled and Drummond 2010). In this analysis, all sequences

were included (as opposed to one per species in BEAST analyses),

with species defined a priori.

The ∗BEAST analysis ran for 900,000,000 generations with

a sample frequency of 10,000. The Yule tree prior was used for

species-level analyses and a constant coalescent model was used

for population-level analyses. Sequence variation was partitioned

among genes and gene-specific nucleotide substitution model pa-

rameters were used, with each gene allowed to evolve at a differ-

ent rate. We used the same nucleotide substitution models as were

used in the 4gene BEAST/GMYC analysis. Based on preliminary

analyses, we changed several of the priors, which resulted in im-

proved ESS values. A lognormal clock, but no fossil calibrations

were used in this analysis and mean rate was fixed to one. The

final ∗BEAST species tree was a maximum clade credibility tree

with median node heights based on 78,000 trees (after burnin of

12,000 trees). Length of burnin was determined by examination

of traces in Tracer. Support for nodes was determined using PP.

Although independent gene trees are obtained using∗BEAST, we did not use these trees as priors on population size

are such that there is a high likelihood of monophyly of species

as defined a priori (J. Heled, pers. comm. 2010). Therefore, in-

dividual gene trees were produced independently using Bayesian

inference as implemented in MrBayes (version 3.2, Huelsenbeck

and Ronquist 2001). Nucleotide substitution models used were

those suggested by MrModelTest, which were the same as those

for the ingroup alone, except for 16S (GTR + G + I). Analy-

ses were run for 7,500,000 generations with a sample frequency

of 1000. The first 2501 trees were discarded, so that 5000 trees

were accepted for each run after likelihood values had reached

a plateau. The datasets were analyzed in two independent runs,

and the final tree was computed from the combination of ac-

cepted trees from each run (a total of 10,000 trees). Conver-

gence between the two runs was tested by examining the poten-

tial scale reduction factors (PSRF) and standard deviation of split

frequencies.

FOSSIL CALIBRATIONS

Literature records state that the oldest fossil for Lunella is L.

miyarensis from the southern Ryukyu Islands, Japan, from the

Eocene Miyara group (MacNeil 1964). This group has been accu-

rately aged using foraminiferans and can be placed in the Priabo-

nian stage (37–34.2 Ma, Saito et al. 1984). The only specimen for

this species is the holotype, which is kept in the Smithsonian Insti-

tute (USNM 638646). Unfortunately, most of the fossil specimens

from this formation are poorly preserved, as the specimens occur

in consolidated hard limestone conglomerate and sandstone and it

is difficult to place this specimen accurately within the phylogeny.

We used it to constrain the cinerea and coronata groups, but not

species that were previously thought to belong in the genus Turbo.

The stem of the cinerea and coronata groups was constrained to

be at least 34.2 Ma (the youngest age of the fossil) (95% interval:

34.4–45 Ma; maximum age based on approximate mid-point of

Eocene, which allows both for the uncertainty of the fossil age,

and for the incompleteness of the fossil record) (mean in real

space: 2.5, log stdev: 1, offset 34.2).

We examined three specimens of L. kurodai (kindly donated

to the NHM by Susumu Tomida). This species is found in late

Early Miocene sediments of the Mizunami Group in central Japan

(approximately 16.5–17 Ma; S. Tomida, Chukyo Gakuin Univer-

sity, Nakatsugawa, Japan; pers. comm. 2010) and contemporane-

ous beds (Bihoku Group) in the Chugoku District, western Japan

(T. Ozawa, unpubl. data). Two of our specimens were collected

from the type locality (Kubohara sandstone of the Mizunami

Group in Kamigiri, Iwamura-cho, Ena City, Gifu Pref., Japan)

and one from Kujiri Bed, Mizunami Group in Inkyoyama, Kujiri,

Izumi-cho, Mizunami City, Gifu Prefecture, Japan (NHM PI TG

26505–07). Previous workers have suggested that L. kurodai is

most similar to L. coreensis (Itoigawa 1955). Comparisons of the

fossil specimens with our genetic samples suggest that shell shape

resembles that seen in several species within both the cinerea and

coronata groups. However, the pattern of spiral ribs made up of

small granules is found only in L. coreensis, L. granulata, L.

viridicallus and L. ogasawarana. Because these species represent

both subclades within the coronata group, we used L. kurodai to

date the age of the crown of the coronata clade. We used a log-

normal prior with 95% range from 16.5 Ma (the youngest age of

the fossil) to 34 Ma (which allows both for the uncertainty of the

fossil age, and for the incompleteness of the fossil record) (mean

in real space: 4.1, log stdev: 1, offset 16.15; 95% interval: 16.5 –

34 Ma).



Calibration of the cinerea clade was based on fossil spec-

imens of “Turbo versicolor” (=L. cinerea sensu latu) from

Junghun locations K (Cianjur, West Java) and R (Garut, Java) in

Indonesia (Leiden; RGM 11544). These locations have recently

been dated accurately by Willem Renema (Naturalis, Leiden, The

Netherlands), and his latest unpublished work suggests that both

correspond to the Middle Miocene. Of these two locations, loca-

tion R is older and likely to be approximately Langian (13–15 Ma)

(W. Renema pers. comm. 2010). A younger, complete specimen

from the oldest Pleistocene (also called “Turbo versiocolor”)

appears similar in size and shape to the central IWP form of

L. cinerea (Sonde, Central Java; Leiden; RGM 11545). However,

the Miocene samples are incomplete shells, showing some color

patterns, but lacking the columellar and lower part of the shell and

without the complete shell we are not confident to apply these fos-

sils to any particular lineage, although the species is definitely a

member of the cinerea group. We used the Miocene fossils to

calibrate the crown age of the cinerea group, with a lognormal

prior with 95% range from approximately 13 (the youngest age of

the oldest fossil) to 34 Ma (which allows both for the uncertainty

of the fossil age and incompleteness of the fossil record) (mean

in real space: 5, log stdev: 1, offset 12.6, 95% interval: 13 -34.1

Ma).

LOCATING CHANGES IN DIVERSIFICATION RATE

We used the relative cladogenesis statistic (RC) implemented in

END-EPI (version 1.0, Rambaut et al. 1997), to determine whether

there were any significant increases in diversification rates across

the phylogeny. The RC statistic uses a constant-rates birth–death

model and calculates the probability that a particular lineage exist-

ing at time t will have k extant tips. This probability is compared to

the total number of extant tips and the statistic identifies branches

with higher than expected rates of cladogenesis.

ResultsSEQUENCE ANALYSIS

For datasets used in the BEAST, ∗BEAST and GMYC/BEAST

analyses (which included only ingroup samples) the COI dataset

of 658 bp included 240 variable base pairs, of which 236 were

phylogenetically informative. For 28S, of 1452 bp, 40 were vari-

able, of which 36 were phylogenetically informative. Alignments

including outgroups (for MrBayes analyses) of COI and 28S in-

cluded (respectively) 258 and 60 variable base pairs, of which 249

and 49 bp were informative.

After removal of ambiguous blocks from mitochondrial ribo-

somal genes, 543 bp of sequence from 16S remained to be used in

phylogenetic analyses of ingroup taxa (86% of 626 bp) of which

207 bp were variable and 201 bp phylogenetically informative. A

total of 625 bp of sequence from 12S (90% of 690 bp) was used in

analyses, of which 288 bp were variable and 283 bp phylogenet-

ically informative. For datasets including outgroups, analysis of

16S was based on 499 bp of sequence excluding ambiguous sites

(77% of 645 bp), of which 200 bp were variable and 185 bp phy-

logenetically informative. Analysis of 12S was based on 566 bp

of sequence excluding ambiguous sites (79% of 708 bp), of which

282 bp were variable and 263 bp phylogenetically informative.

SPECIES DELIMITATION

Ultrametric trees were obtained for COI and concatenated se-

quences from all four genes with all ESS values greater than

200. In the GMYC/BEAST COI tree all ESUs were supported by

PP > 90% except I, J, L∗, and M that received very low support

(PP < 70%) and P and P′ which received only moderate support

(PP = 85%) (Fig. 1). In the GMYC/BEAST four-gene tree all

ESUs were well supported, with PP = 100% (Fig. 2).

The GMYC species delimitation test showed that the like-

lihood of the null model (that all specimens belong to a single

species) was significantly poorer than the maximum likelihood

of the GMYC model both for COI (2179 vs. 2191; ratio: 24.9,

P = 1.6 × 10−5) and for the four-gene tree (421 vs. 442; ratio:

41.8, P = 4.3 × 10−9). The threshold time (T) from the branch

tips at which the speciation-coalescent transition occurred ranged

from 0.007 in COI to 0.005 substitutions per site, resulting in 21

clusters being recognized in both the COI tree and in the four-

gene tree, although these did not completely correspond to the

same entities. Using the equivalent of a 95% confidence interval,

the total number of entities identified ranged from 19 to 26 in

the COI tree and 20 to 22 in the four-gene tree based on model

substitutions at two log-likelihood units from the maximum (C.I.;

Monaghan et al. 2009). If the lower C.I. is used to define species

in the COI tree, only 19 clusters are recognized with ESUs L∗and M combined, as are ESUs I and J. In the four-gene tree using

the lower C.I. limit ESUs P and P′ are combined.

A comparison of ESUs defined by the two GMYC/BEAST

analyses revealed some differences. Two clades within L. coronata

were recognized as distinct ESUs in the COI tree (ESUs O and O′),but not in the four-gene tree. Likewise two different clades within

L. coronata were recognized as distinct in the four-gene tree, but

not in the COI tree (ESUs P and P′). The composition of two ESUs

(L and M) also differed slightly between the two analyses, with

samples from Vanuatu and one from New Caledonia occurring

along with samples from American Samoa in ESU L∗ in the COI

analysis but in ESU M (with the remainder of the New Caledonian

samples) in the four-gene analysis and the MrBayes analysis of

COI (Fig. 3).

Overall, we recognize 18 putative species in Lunella

(Tables 1, S1) of which 17 have been included in this study and

eight are new. ESUs that are conservatively treated as species

in this study are those that are recognized in both GMYC



analyses, when using the set of species suggested by their re-

spective lower confidence limits. A morphological examination

of specimens used in genetic analyses confirmed that these ESUs

can be distinguished from their sister species morphologically

either by shell characters and/or pattern of coloration on anten-

nae, head, and snout. A morphometric analysis of outline shape

also distinguishes some cryptic species (Williams et al. 2011).

Some clades occur in sympatry with sister taxa and/or have fixed

differences at 28S, both of which are consistent with clades be-

ing biological species (Table 1). In some cases, species shared

identical 28S genotypes (which is common in turbinids; e.g.,

Williams 2007) and morphological differences were small and

based on few individuals, so we consider our putative species

list as a working hypothesis, which will need to be confirmed by

more detailed studies. Estimated divergence times between sister

species were all greater than 1.8 Myr (Table 1). Genetic distances

between species/clades ranged from 5.4 to 46.6% (K2P = 4.5–

16.8; Table 1). P ID (liberal) for the 17 clades accepted as species

were all above 0.95 (0.96–1.00).

ESUs that are considered to represent either population struc-

ture or subspecies are L. cinerea from Japan and the central IWP

(ESUs I and J) and L. cinerea from Vanuatu, New Caledonia and

American Samoa (L and M). These clades were not supported

at the lower C.I. limits in the COI tree nor could they be dis-

tinguished on morphological grounds or by differences at 28S.

Four clades of L. coronata (O, O′, P and P′) were not combined

in any analysis, even at the lower C.I. cutoff (except P and P′),but the threshold time between the two pairs in the four-gene

GMYC tree is lower than that of some of the other pairs that we

combined. Therefore, we consider these four clades to represent

a single species despite some slight morphological differences of

the Gulf of Persia samples (ESU O′ tends to have a higher shell

spire) and a fixed (albeit polymorphic) difference at 28S between

the Indian and UAE clade (ESU O) and other L. coronata clades.

Average genetic distances for COI sequences among these clades

(considered to represent intraspecific variation) were also lower

than those observed among accepted species pairs, ranging from

2.7% to 4.8% (K2P = 2.3–4.0%; Table 1).

PHYLOGENETIC ANALYSES

Potential scale reduction factor in all MrBayes analyses were

1.00, showing that runs had converged (Gelman and Rubin

1992) and average standard deviation of split frequencies for

partitions with frequency >0.10 in at least one run approached

zero (28S: 0.000005, 12S: 0.001349, 16S: 0.000013, COI:

0.000117).

Ultrametric trees were obtained for BEAST analyses using

concatenated sequences from all four genes with all ESS values

greater than 200 (some more than an order of magnitude greater).

The total number of unique clades was 24 and the highest log

clade credibility was −0.37. It proved difficult to obtain ∗BEAST

analyses with acceptable ESS values even after very log run times,

so the analysis was run without calibrations (fixed mean rate).

An undated phylogeny was obtained with all ESS values greater

than 200 (some more than an order of magnitude greater). The

topology of the consensus species tree obtained from BEAST

and ∗BEAST analyses was identical, but support for nodes varied

slightly (Fig. 5)

All individual gene trees, except COI, recovered Lunella as

monophyletic (Figs. 3 and 4). The cinerea clade was recovered in

both BEAST and ∗BEAST trees and all three mitochondrial gene

trees and the coronata clade was recovered in both BEAST and∗BEAST trees and the 28S, 16S, and COI trees (Figs. 3 and 4). A

clade with L. torquata, L. undulata, and L. jungi was recovered in

all trees. The position of L smaragdus (endemic to New Zealand)

was unstable in the four independent gene trees, although in the

mitochondrial gene trees it always fell outside a clade containing

the cinerea and coronata groups. In both the BEAST tree and∗BEAST trees, it was sister to the cinerea and coronata clades.

The clades identified in the GMYC/BEAST analysis were also

recovered in individual gene trees, although some ESUs shared

the same 28S sequence. The phylogenetic relationships found in

this study are consistent with those in earlier studies with fewer

taxa (Williams 2007; Nakano et al. 2007).

LOCATING CHANGES IN DIVERSIFICATION RATE

No significant changes in rate were observed in the phylogeny

using the RC test so no further investigation was undertaken.

DiscussionSPECIES DELIMITATION

Overall, we recognize 18 putative species in Lunella, of which 17

are included in this study (listed in Table S1) and eight are new,

almost doubling the known diversity for this genus. Seven clades

Figure 1. Results of GMYC test on an ultrametric tree produced using BEAST for the cytochrome oxidase I sequences. (A) Tree with ESUs

and species names marked on the right-hand side. Specimen number and country of collection are given at nodes. ESUs considered to

represent a single species are bounded by gray boxes. The composition of ESU L∗ is similar, but not identical to ESU L in other trees.

Support for nodes are posterior probabilities, shown only for species-level relationships and PP > 50%. Branches in light gray indicate

population structure. The cinerea group is labeled with a solid circle, the torquata group with a star, and the smaragdus clade with a

triangle. (B). Lineage through time plot with gray line showing the threshold level suggested by GMYC. (C) Maximum likelihood plot

showing peak congruent with threshold limit in B.



Figure 2. Results of GMYC test on an ultrametric tree produced using BEAST for concatenated sequences from all four genes. (A) Tree

with ESUs and species names marked on the right-hand side. Specimen number and country of collection are given at nodes. ESUs

considered to represent a single species are bounded by gray boxes. Support for nodes are posterior probabilities, shown only for species

level relationships and PP > 50%. Branches in light gray indicate population structure. The cinerea group is labeled with a solid circle, the

coronata group with a square, the torquata group with a star, and the smaragdus clade with a triangle. (B) Lineage through time plot

with gray line showing the threshold level suggested by GMYC. (C) Maximum likelihood plot showing peak congruent with threshold

limit in B.



recognized by the GMYC/BEAST analyses correspond to cur-

rently recognized species (see Table 1), of which two have been

described only recently (L. jungi, in Lai 2006; L. ogasawarana,

in Nakano et al. 2007). However, two additional species, L.

cinerea and L. coronata, each correspond to a species complex.

The cinerea group contains nine clades, which we interpret here

as representing seven species. The West Indian Ocean coronata

group contains six clades (a maximum of five recognized in either

GMYC analysis), which we interpret here as representing three

species. The status of the only species not included in this study,

L. viridicallus, remains to be tested genetically. The newly recog-

nized species qualify as phylogenetic species, but in the cinerea

clade, where all sister species are allopatric it is not possible to

test whether they also qualify as biological species. Several names

exist in the literature that might be applied to these taxa, but a com-

prehensive revision of the genus is beyond the scope of this study

and will be published elsewhere (S. T. Williams, unpubl. data). It

is also possible that additional cryptic species may yet be found

in Lunella. Likely areas of endemism not investigated genetically

include the Red Sea (where L. viridicallus occurs; Alf and Kreipl

2003), Indonesia, and Madagascar.

As has often been the case in studies to date there were more

GMYC clades than have been previously recognized as species by

traditional taxonomy (e.g., Papadopoulou et al. 2008; Monaghan

et al. 2009; Claremont et al. 2011). However, even with the benefit

of hindsight, we have not yet found strong evidence to support the

recognition of some clades as separate species and in this study,

we treat them as evidence of population structure. To delimit

species, we combined ESUs that were not supported in either tree

at the lower confidence limit and for which we could find no ad-

ditional evidence of specific status, and ESUs with splits younger

than those collapsed for other reasons. Detailed analysis of a gas-

tropod genus with higher dispersal abilities and therefore less

population structure (Stramonita, Claremont et al. 2011) suggests

that although the GMYC model identifies more ESUs than recog-

nized by traditional taxonomy, these ESUs nonetheless directly

correspond to biological species. However, our results suggest

that it may be necessary to use lower confidence limits to

distinguish between population structure and species in lower

Figure 3. Molecular phylogeny for Lunella with outgroups from

Turbininae based on Bayesian analysis of the single gene COI pro-

duced with MrBayes. Specimen number and country of collection

are given at nodes. ESUs are labeled on the right-hand side, fol-

lowing Figs. 1 and 2. Support values are posterior probabilities

(PP); branches with PP < 50% were collapsed. Not all PP values are

given for the sake of clarity. The cinerea group is labeled with a

solid circle, the coronata group with a square, the torquata group

with a star, and the smaragdus clade with a triangle. See Table S1

for details.



Ta

ble

1.

Sum

mar

yo

fES

Us

iden

tifi

edb

yG

MY

Can

alys

is(F

igs.

1an

d2)

and

dis

cuss

edin

the

text

;in

clu

din

gth

en

um

ber

of

spec

imen

sse

qu

ence

d(n

);ap

pro

xim

ate

geo

gra

ph

ical

ran

ge

(in

som

eca

ses

bas

edo

ng

enet

icse

qu

ence

so

nly

and

inth

eca

seo

fES

Us

Lan

dM

,b

ased

on

clad

esin

Fig

s.2–

4);

sist

ercl

ades

/sp

ecie

s(b

ased

on

rela

tio

nsh

ips

inth

eB

EAST

tree

inFi

g.

5);

div

erg

ence

tim

eb

etw

een

sist

ercl

ades

(Myr

;b

ased

on

Fig

.5)

;th

ep

rese

nce

of

mo

rph

olo

gic

ald

iffe

ren

ces

(eit

her

shel

lo

rso

ftti

ssu

e;Y-

yes,

N-

no

)b

etw

een

sist

er

clad

es/s

pec

ies

(bas

edp

rim

arily

on

seq

uen

ced

spec

imen

s);

wh

eth

ersi

ster

clad

es/s

pec

ies

occ

ur

insy

mp

atry

;th

eo

ccu

rren

ceo

ffi

xed

dif

fere

nce

sb

etw

een

sist

ercl

ades

/sp

ecie

sfo

r

seq

uen

cefr

om

28S

(P=

po

lym

orp

hic

fixe

dd

iffe

ren

cee.

g.,

Yvs

.T);

reci

pro

calm

on

op

hyl

yo

fcl

ades

/sp

ecie

sin

MrB

ayes

gen

etr

ees

(Fig

s.3

and

4);a

vera

ge

K2P

dis

tan

ceb

ased

on

CO

I

seq

uen

ced

ata

wit

hin

each

ESU

,an

db

etw

een

the

ESU

liste

dan

dth

ecl

ade

(lis

ted

inth

en

ext

colu

mn

)se

par

ated

by

the

smal

lest

gen

etic

dis

tan

ceu

sin

gC

OIs

equ

ence

on

ly;a

vera

ge

HK

Y+I

+Gd

ista

nce

bas

edo

nC

OIs

equ

ence

dat

ab

etw

een

ESU

and

the

clad

elis

ted

inth

en

ext

colu

mn

;an

dp

relim

inar

yco

ncl

usi

on

sab

ou

tES

Ust

atu

s—O

ne

spec

ies

=cl

ade

incl

ud

es

all

seq

uen

ces

ob

tain

edfo

ra

sin

gle

,d

iscr

ete

spec

ies;

—Po

pu

lati

on

stru

ctu

re=

clad

ein

clu

des

som

e,b

ut

no

tal

lse

qu

ence

so

bta

ined

for

asi

ng

lesp

ecie

s.N

od

ata

are

pro

vid

edfo

r

com

par

iso

nb

etw

een

ESU

sP

and

P′ o

rO

and

O′ b

ecau

seth

ese

clad

esw

ere

each

iden

tifi

edin

on

lyo

ne

of

the

GM

YC

/BEA

STtr

ees.

Bet

wee

nsi

ster

spec

ies/

clad

esR

ecip

roca

lmon

ophy

ly(M

rBay

es)

ESU

Spec

ies

nG

eogr

aphi

cal

rang

eSi

ster

clad

eD

ivtim

eM

yr

Mor

phdi

ffs

Sym

patr

icFi

xed

diff

s28

S

CO

I12

S16

S%

K2P

intr

aC

OI

%K

2Pin

ter

CO

I

% HK

Y+

I+G

inte

rC

OI

Clo

sest

spec

ies

(CO

Ige

netic

dist

ance

)K

2P/H

KY

+I+

G

Prel

imin

ary

conc

lusi

ons

AL

.jun

gi7

Taiw

anB

+C19

YN

YY

YY

0.1

14.0

31.2

L.u

ndul

ata

One

spec

ies

BL

.tor

quat

a10

Sout

hern

Aus

tral

iaC

8.2

YY

YY

YY

0.4

12.4

21.5

L.u

ndul

ata

One

spec

ies

CL

.und

ulat

a10

Sout

hern

Aus

tral

iaB

8.2

YY

YY

YY

0.7

12.4

21.5

L.t

orqu

ata

One

spec

ies

DL

.sm

arag

dus

12N

ewZ

eala

ndE

-T35

.1Y

NY

YY

Y0.

117

.546

.6L

.cin

erea

NW

Aus

tral

ia/L

.ci

nere

aFi

ji

One

spec

ies

EL

.cin

erea

NW

Aus

tral

ia9

Nor

thW

este

rnA

ustr

alia

F+G

3.2

YN

NY

YY

0.3

6.7

8.7

L.c

iner

eaSu

law

esi

One

spec

ies

FL

.cin

erea

Sula

wes

i4

Sula

wes

i,In

done

sia

G2.

3Y

NN

YY

Y0.

55.

67.

1L

.cin

erea

NE

Aus

tral

iaO

nesp

ecie

s

GL

.cin

erea

NE

Aus

tral

ia10

Nor

thE

aste

rnA

ustr

alia

F2.

3Y

NP

YY

Y0.

35.

67.

1L

.cin

erea

Sula

wes

iO

nesp

ecie

s

HL

.cin

erea

Indi

anO

cean

18N

orth

-eas

tern

Indi

anO

cean

E−G

6Y

NN

YY

Y0.

411

.317

.7L

.cin

erea

NW

Aus

tral

iaO

nesp

ecie

s

Ian

dJ

L.c

iner

eacI

WP

24C

entr

alIn

do-W

est

Paci

fic

E−H

7.8

YN

NY

YY

1.3

12.6

24.8

L.c

iner

eaN

WA

ustr

alia

One

spec

ies

-I

–L

.cin

erea

cIW

PA

7Ja

pan

J–

NN

NY

NY

0.1

2.3

2.7

L.c

iner

eacI

WP

B–

Popu

latio

nst

ruct

ure

-J

–L

.cin

erea

cIW

PB

17C

entr

alIn

do-W

est

Paci

fic

I–

NN

NY

NN

0.6

2.3

2.7

L.c

iner

eacI

WP

A–

Popu

latio

nst

ruct

ure

Co

nti

nu

ed



Ta

ble

1.

Co

nti

nu

ed

Bet

wee

nsi

ster

spec

ies/

clad

esR

ecip

roca

lmon

ophy

ly(M

rBay

es)

ESU

Spec

ies

nG

eogr

aphi

cal

rang

eSi

ster

clad

eD

ivtim

eM

yr

Mor

phdi

ffs

Sym

patr

icFi

xed

diff

s28

S

CO

I12

S16

S%

K2P

intr

aC

OI

%K

2Pin

ter

CO

I

% HK

Y+

I+G

inte

rC

OI

Clo

sest

spec

ies

(CO

Ige

netic

dist

ance

)K

2P/H

KY

+I+

G

Prel

imin

ary

conc

lusi

ons

KL

.cin

erea

Fiji

10Fi

jiL

M7.

6Y

NY

YY

Y0.

512

.423

.1L

.cin

erea

SWPa

cifi

cO

nesp

ecie

s

Lan

dM

L.c

iner

eaSW

Paci

fic

21So

uth-

Wes

tPac

ific

K7.

6Y

NY

YY

Y1.

812

.423

.1L

.cin

erea

Fiji

One

spec

ies

-L

–L

.cin

erea

SWPa

cifi

cA

5A

mer

ican

Sam

oaM

–N

NN

YY

N0.

24.

04.

8L

.cin

erea

SWPa

cifi

cB

–Po

pula

tion

stru

ctur

e

-M

–L

.cin

erea

SWPa

cifi

cB

16N

ewC

aled

onia

and

Van

uatu

L–

NN

NY

YN

0.4

4.0

4.8

L.c

iner

eaSW

Paci

fic

A–

Popu

latio

nst

ruct

ure

NL

.cor

onat

aop

hiol

item

orph

10U

nite

dA

rab

Em

irat

esO

−Q11

.5Y

YY

YY

Y<

0.1

16.8

43.8

L.c

oron

ata

coro

nate

One

spec

ies

Oan

dO

′

and

Pan

dP

′

L.c

oron

ata

coro

nate

mor

ph37

Wes

tInd

ian

Oce

anQ

5.2

YY

NY

YY

1.8

9.7

15.6

L.c

oron

ata

Om

anO

nesp

ecie

s

-O

and

O′

–L

.cor

onat

aco

rona

teA

and

B

24U

nite

dA

rab

Em

irat

esan

dIn

dia

Pan

dP

′–

NY

NN

NN

0.8

2.4

3.4

L.c

oron

ata

coro

nate

C–

Popu

latio

nst

ruct

ure

-P

and

P′

–L

.cor

onat

aco

rona

teC

13W

estI

ndia

nO

cean

Oan

dO

′–

NY

NN

NN

0.9

2.4

3.4

L.c

oron

ata

coro

nate

Aan

dB

–Po

pula

tion

stru

ctur

e

QL

.cor

onat

aO

man

6O

man

O−P

5.2

YY

NY

YY

0.4

9.3

15.6

L.c

oron

ata

coro

nate

One

spec

ies

RL

.cor

eens

is10

Sout

hE

astA

sia

S+T

4.6

YY

YY

YY

0.3

8.3

11.5

L.o

gasa

war

ana

One

spec

ies

SL

.oga

saw

aran

a10

Oga

saw

aran

aIs

land

s,Ja

pan

T1.

8Y

NY

YY

Y0.

14.

55.

4L

.gra

nula

taO

nesp

ecie

s

TL

.gra

nula

ta21

Sout

hE

astA

sia

S1.

8Y

NY

YY

Y0.

44.

55.

4L

.oga

saw

aran

aO

nesp

ecie

s



Fig

ure

4.

Mo

lecu

lar

ph

ylo

gen

yfo

rLu

nel

law

ith

ou

tgro

up

sfr

om

Turb

inin

aeb

ased

on

Bay

esia

nan

alys

iso

fin

dep

end

ent

gen

es28

SrR

NA

,12S

rRN

A,a

nd

16S

rRN

Ap

rod

uce

dw

ith

MrB

ayes

.Su

pp

ort

valu

esar

ep

ost

erio

rp

rob

abili

ties

(PP)

;bra

nch

esw

ith

PP<

50%

wer

eco

llap

sed

.No

tal

lPP

valu

esar

eg

iven

for

the

sake

of

clar

ity.

ESU

sd

iscu

ssed

inth

ete

xtar

e

lab

eled

on

the

rig

ht-

han

dsi

de.

The

cin

erea

gro

up

isla

bel

edw

ith

aso

lidci

rcle

,th

eco

ron

ata

gro

up

wit

ha

squ

are,

the

torq

uat

ag

rou

pw

ith

ast

ar,a

nd

the

smar

agd

us

clad

ew

ith

a

tria

ng

le.S

eeTa

ble

S1fo

rd

etai

ls.



Figure 5. Chronogram for Lunella based on Bayesian analysis of concatenated sequence from 28S rRNA, COI, 12S rRNA, and 16S rRNA

genes incorporating an uncorrelated relaxed, lognormal clock produced with BEAST. Branch lengths are proportional to time and scale in

millions of years (Myr) is given below. Tree is a maximum clade credibility tree with median node heights based on 6500 trees. Horizontal

gray bars correspond to 95% highest posterior density (HPD) interval for node heights (ages). The 95% HPD is shortest interval that

contains 95% of the sampled values. Support values are posterior probabilities (PP) from BEAST/∗BEAST. Species names and ESUs are

labeled on the right-hand side. The cinerea group is labeled with a solid circle, the coronata group with a square, the torquata group

with a star, and the smaragdus clade with a triangle.

dispersal groups like Lunella. In fact, three cinerea clades (ESUs

E, F, and G) show similar biogeographic distributions to divergent

clades in another rocky-shore gastropod Echinolittorina vidua

(Reid et al. 2006). Reid et al. (2006) decided that the E. vidua

clades belonged to a single, highly structured species but K2P

distances for COI sequences between the three cinerea clades

(5.6–6.7%) are higher than those observed among the correspond-

ing E. vidua clades (2.15–3.77%) supporting our decision to retain

these ESUs as species pending further studies.

Differences in the analysis of sequences from COI alone ver-

sus four genes suggest that some clades are identified for spurious

reasons. First, the GMYC analysis presupposes a well-resolved

tree, but in fact, several nodes in the COI tree defining clades

were not well supported (with PP < 50%). This is particularly

important when it affects branching order because in an ultramet-

ric sister branches have the same height, switching branches may

make populations appear to be more distinct than they really are

(compare relationships within the West Indian Ocean coronata

group in Figs. 1, 2 and 3). Second, GMYC analysis of sequences

from four genes (although only two loci) resulted in the loss of one

clade and changed the composition of another; suggesting that it

may always be useful to compare results from several datasets and

raises (again) the question of accuracy when barcoding a single

gene. Additional, independent nuclear loci may also be helpful, if

they evolve at a sufficiently fast rate to identify recent divergences.

Finally, the importance of sampling has been raised as a potential

problem with GMYC analyses (Lohse 2009, although see also

Papadopoulou et al. 2009), and we believe that the separation of

ESUs P and P′ in the four-gene study may have been the result

of more limited sampling in that tree. Despite these limitations



and used conservatively this method provides a nonsubjective

preliminary hypothesis for species delimitation, which can then

be used as a framework for further testing (Papadopoulou et al.

2009), helping to identify groups in need of greater taxonomic

work.

ENVIRONMENT

Various authors have noted that shallow-water marine species

show preferences for habitats that range from “continental,” along

continental coastlines or on high islands to “oceanic,” on low,

oceanic atolls (e.g., Reid 1986; Taylor 1997). Habitats along this

range differ in local nutrients, salinity, turbidity, and chlorophyll

among a vast range of other factors. It is not known exactly which

factors are important in determining distribution patterns, but it

may have to do with nutrients (Taylor 1997) and selection acting

on larvae (Williams and Reid 2004). Lunella occur on rocky-

shores along continental land margins or on high, continental

islands. Some species seem to prefer clearer water, but even these

species occur on high islands, not oceanic atolls (following defini-

tions in Springer 1982). For example, Lunella occurs in the Bonin

Islands, where the habitat is more continental and reefs are poorly

developed (Fukuda 1994), but not in the more oceanic Mariana

Islands.

Because all Lunella species share similar habitats and are

continental by preference, it seems unlikely that this habitat pref-

erence has had much impact on speciation in Lunella. However,

it may have been important in the evolution of deeper nodes. The

coronata clade is more common on continental coastlines whereas

its sister, the cinerea group, is more common on islands, suggest-

ing that ecology may have been important in the evolution of

these clades. Equally, distribution with respect to the littoral zone

may also have been important. Most of the cinerea and coronata

group occur in the high or mid intertidal zone, whereas L. smarag-

dus occurs in the mid to sub-littoral zone (Walsby 1977) as do L.

torquata and L. undulata (G. Vermeij, pers. comm. 2010). Similar

results have been found for other gastropods living in comparable

habitats (e.g., Nerita, Frey 2010, Echinolittorina, Williams and

Reid 2004; Reid et al. 2006; compare with Littoraria, Reid et al.

2010). Future studies involving ecological niche modeling may

be informative.

GEOGRAPHY: COASTLINES VERSUS ISLAND ARRAYS

Sympatric sister species occur only along coastlines in Lunella

in the coronata and torquata groups. In the torquata group, L.

torquata and L. undulata are sympatric across a large proportion

of their ranges along southern temperate Australia, although only

L. undulata is found in Tasmania and L. torquata occurs further

up the west coast of Australia. In the coronata group two cryptic

species apparently arose independently on the Arabian Peninsula

(ESUs N and Q). Both occur in sympatry over at least part of their

range with L. coronata coronate morph (ESUs O, O′, P and P′)although apparently not with each other. One is endemic to Oman

(L. coronata Oman ESU Q) and the other to the Gulf of Oman

coast of the United Arab Emirates and the Musandam Peninsula,

an exclave of Oman (G. Feulner, pers. comm. 2011) (L. coronata

ophiolite morph ESU N). Both of these localities harbor high lev-

els of endemicity with many other small range endemic species

(e.g., gastropods Trochita dhofarensis and Haliotis mariae, kelp

Ecklonia, Taylor and Smythe 1985; Turbo jonathani, Dekker et al.

1992; parrotfish, Randall and Hoover 1995; Eoacmaea omanen-

sis [as Patelloida], Kirkendale and Meyer 2004; cuttlefish Sepia

pharaonis, Anderson et al. 2010). The Oman endemics occur in a

region of upwelling and may have evolved as a result of the colder

thermal regime (Taylor and Smythe 1985; Randall and Hoover

1995; Kemp 1998; Kirkendale and Meyer 2004) or increased nu-

trients. Alternately, ocean currents may act as a physical barrier

to larval dispersal.

Species tend to range long distances along linear coastlines,

even if their dispersal ability is poor (Valentine and Jablonski

1982, 1983). Range end-points are not randomly distributed, as

might be expected if they were affected by dispersal ability, but

rather cluster at biogeographic boundaries, often corresponding

to a shift in thermal regime or the confluence of water masses

(Valentine and Jablonski 1983). As predicted, the most widely

distributed species is L. coronata coronate morph, which occurs

along the coast of east Africa and the Arabian Peninsula into

India. Other coronata and torquata species also range long dis-

tances with species boundaries most likely set by temperature.

Range limits along the coasts of east Africa, southern Australia,

and southeast Asia also mirror those seen in many other taxa.

The range end point of L. coronata coronate morph in India cor-

responds with an area of low salinity and high turbidity due to

freshwater run-off from river deltas. Species in these groups with

small ranges are L. jungi and L. ogasawarana, which occur only

on islands, and the two Arabian endemics.

Populations within L. coronata coronate morph are not pan-

mictic. Genetic structure within this species suggests that there

are semi-permeable boundaries to gene flow along the Arabian

Peninsula, although not along the east coast of Africa. Population

samples of the L. coronata coronate morph from the Persian Gulf

(ESU O′) have higher spired shells than the African clade (ESU

P and P′) and there is a fixed, albeit polymorphic, difference in

a nuclear gene in Indian and Gulf of Oman populations (ESU

O). All three distinct populations (ESUS, O, O′, and P and P′)co-occur in UAE (in the Gulf of Oman). Likely partial barriers

to gene flow for L. coronata coronate morph are the Persian Gulf

and long stretches of unsuitable habitat between the UAE and

India where long sandy beaches occur, and few rocky headlands.

However, similar patterns have also been observed in other

organisms that do not share the same habitat preferences. For



example, the cuttlefish Sepia pharonis species complex includes

several divergent clades, with clades from the Persian Gulf and

Gulf of Oman highly divergent from both central (India and Thai-

land) and western Indian Ocean clades (Red Sea and Gulf of

Aden) (Anderson et al. 2010).

Population structure is also observed in L. torquata, which

was thought by early malacologists to be two species: one found in

western Australia and the other in east Australia. These two forms

are distinct, but intermediates occur in central south Australia.

There are two clades in the COI tree; one corresponding to samples

from western Australia and the other with samples from both

New South Wales (southeast Australia) plus western Australia.

A break has been observed in genetic studies of many temperate

marine taxa between east and west Australia, with the boundary

at the southeast corner of Australia (e.g., Waters et al. 2005; Ayre

et al. 2009; Colgan and Schreiter 2011). This break is thought

to be due in part to a historical barrier to dispersal in the form

of a land bridge between mainland Australia and Tasmania in

the Pleistocene (Dawson 2005). Modern day dispersal may also

be limited by the confluence of two ocean currents and a stretch

of habitat unsuitable for many rocky-shore taxa (Billingham and

Ayre 1996; Ayre et al. 2009). In some species, there are two clear

clades (e.g., Nerita “atramentosa”; Waters et al. 2005), but in L.

torquata there appears to have been gene flow from NSW into

west Australia subsequent to a period of isolation (see also Waters

et al. 2007). Conversely there is no evidence at all of a break across

the same region in L. undulata (see also Ayre et al. 2009 for other

similar examples).

Unlike the torquata and coronata groups, which are dis-

tributed along one-dimensional coastlines, the cinerea group is

distributed on two-dimensional array of islands and continental

shelf segments. Species ranges in this habitat are predicted to

depend upon dispersal ability and habitat preference. In keeping

with this prediction, all species in the cinerea group have small

ranges, consistent with their likely low dispersal ability. Within

the group, the two species found along the coast of Australia have

longer ranges and L. cinerea SW Pacific is likely to have a range

including Solomon Islands (no genetic data), New Caledonia,

Vanuatu, and American Samoa.

Although exact ranges are not known for the species making

up the cinerea group that occur in the central IWP and southwest

Pacific, to date none have been found in sympatry. In fact, all

seven species are allopatric, which is consistent with allopatric

speciation. In island arrays, particularly archipelagos of oceanic

island atolls, allopatric speciation may be common in species

with lower dispersal abilities where dispersal between archipela-

gos may occur stochastically and infrequently (e.g., Paulay and

Meyer 2002; Meyer 2003; Kirkendale and Meyer 2004; Williams

and Reid 2004; Meyer et al. 2005). Species on the periphery with

smaller ranges may also be subject to a higher rate of specia-

tion and extinction (Jablonski and Roy 2002; Jablonski and Hunt

2006).

Broad continental shelf segments (included in two-

dimensional habitats) may also have played an important role in

maintaining regional diversity. During the Quaternary, sea-level

fluctuations were frequent and of great amplitude, with sea level

dropping possibly by as much as 135 m (Peltier 2002; Kuhnt

et al. 2004). The effect of lowered sea level was profound in the

central IWP, where it exposed large expanses of shallow conti-

nental shelf and markedly changed the geography of the region. It

has been suggested that many marine organisms went locally ex-

tinct on the Sunda Shelf (Hoeksema 2007; Bellwood et al. 2011),

northwest Australia (Kuhnt et al. 2004) and from islands in the

west Pacific as a result of habitat loss during periods of lowered

sea level (Paulay 1990). However, the existence of species that

predate Pleistocene glaciations and that are (currently) endemic

to nearshore northwest and central northern Australia (e.g., L.

cinerea NW Australia) suggests that not all species were extir-

pated. Either species survived in pockets of refugia or only partic-

ular habitats were affected. Indeed Paulay (1990) showed that on

island atolls qualitative, not quantitative loss of habitat resulted in

extinction.

TECTONIC PROCESSES

Oceanic–oceanic plate convergence results in subduction of one

plate under the other and in the process a deep trench may be

formed along the plate margin and volcanic arcs are created,

often forming emergent island chains (Hall 2001). These arcs

provide stepping-stones for dispersal (Springer 1982; Hall 2001;

Paulay and Meyer 2002). Dispersal via island arcs is likely to be

of greatest importance in regions that are tectonically active like

Southeast Asia, which has been active throughout the Cenozoic

(Hall 2001, 2002, 2009). Island arcs are likely to appear more

often and be more transient in regions that are tectonically active,

affecting gene flow among populations, as they become either

better connected or more isolated. Because tectonic features can

disappear over very short periods of time (Hall 2001), the loss of

stepping-stones will result in fragmented populations that may fol-

low independent evolutionary trajectories leading to genetically

structured populations and eventually speciation. Alternatively,

emergence of new island chains may reconnect fractured popu-

lations before they have had a chance to speciate or allow the

dispersal of new species back into ancestral habitats.

One possible example of speciation as a result of the loss

of stepping-stones is found in Japan. Lunella granulata, which

occurs on mainland Asia, Taiwan, and in the Okinawa Prefecture

of Japan is sister to L. ogasawarana, an endemic of the Bonin

Islands (known in Japan as the Ogasawara Group; Fig. 6B) some

1000 km south of Tokyo. This sister relationship between the

Bonin Islands and Okinawa is similar to that observed for other



Figure 6. Distribution maps for (A) cinerea group and (B) coronata group showing distributions of cryptic taxa based on genetic samples.

Gray shading indicates approximate range of each species complex. Open circles show the location of museum samples examined. Museum

records have not been identified to species. Colored icons indicate ESUs (according to keys in figures). Note that cladograms indicate

BEAST/∗BEAST tree topology, but branch length has no meaning.



endemic species (Fukuda 1994, 1995). In a study of a low-

dispersal patellogastropod, Nakano et al. (2009) explained a simi-

lar pattern by suggesting that an ancestral limpet species dispersed

to the Bonin Islands from mainland Japan on the Kuroshio Cur-

rent, possibly with the help of stepping-stones in the form of

islands between Izu and Bonin Islands. They suggested that these

islands became emergent as a result of falling sea levels during

periods of glaciation and/or by tectonic activity, because there

are several sea mounts along the Shitito-Iwojima oceanic ridge.

The loss of these “stepping-stone” islands by further tectonic pro-

cesses or rising sea levels would have resulted in the isolation

and subsequent divergence of the populations on the Ogasawara

Islands (Nakano et al. 2009). Paleogeographic reconstructions

confirm that the Izu-Bonin-Mariana arc was likely to have been

locally and intermittently emergent at sites of volcanic activity

(Hall 2001).

The estimated age of the divergence between L. granulata

and L. ogasawarana is approximately 1.8 Ma (95% HPD: 1.0–

2.7 Ma; Fig. 5). The Bonin Islands formed during the Palaeo-

gene, but emerged above sea level in the Pleistocene (Kaizuka

1977; Imaizumi and Tamura 1984). Molecular clock estimates

for endemic plants (Ito and Ono 1990) and other shallow marine

gastropods (limpets, Nakano et al. 2009) are also consistent with

Plio–Pleistocene origins.

One of the most important tectonic events in the IWP during

the Oligo-Miocene was the collision of the Australia and New

Guinea plate with the southeast extremity of the Eurasian plate

and the Philippines-Halmahera-New Guinea arc system approxi-

mately 25 Ma (Hall 1998). Molecular and fossil evidence suggests

that this was followed by a period of rapid diversification of zoox-

anthellate corals and associated shallow-water faunas between 20

Ma and 25 Ma (Kohn 1990; Wilson and Rosen 1998; Barber and

Bellwood 2005; Read et al. 2006; Alfaro et al. 2007; Alfaro et al.

2007, 2009; Williams 2007; Williams and Duda 2008; Renema

et al. 2008; Frey and Vermeij 2008; Bellwood et al. 2011). How-

ever, although clades of this age were identified in Lunella (e.g.,

cinerea + coronata groups, 25 Ma; 95% HPD: 20.2–29.7 Ma), no

statistical evidence of increased cladogenesis was observed in this

group, or in a previous study of mangrove snails (Littoraria, Reid

et al. 2010). Moreover, although the rocky-shore genus Echinolit-

torina showed a substantial increase in the rate of cladogenesis,

the increase was statistically nonsignificant (P = 0.07; Williams

and Duda 2008). This suggests that fauna living high in the in-

tertidal zone on rocky shores or mangroves benefited less from

the increase in habitat and the increase in diversity of niches and

trophic resources than reef-associated organisms.

ACKNOWLEDGMENTSWe are grateful to D. Reid, A. Kohn, G. Vermeij, G. Paulay, G. Feulner, M.Hellberg, J. Taylor, B. Marshall, and R. Hall for reading the manuscript

and for thoughtful comments. We also thank G. Feulner, H. Dekker,K. Kreipl, A. Alf, C. Meyer, B. Marshall, P. Lozouet, F. Wesselingh, andG. Paulay for discussions about turbinids; D. Reid, C. Meyer, G. Paulay,F. Naggs, and A. Kohn for interesting discussions about biogeographyof southeast Asia; T. Fujisawa, J. Heled, P. Foster, M. Claremont, and C.Drummond for discussion about analyses; and P. Marko, A. Kohn, andD. Eernisse for an invitation to speak at an AMS symposium that helpedto shape this article. We thank S. Tomida for donating fossil specimensand are grateful to W. Renema for allowing us to use unpublished fossildates. We also thank G. Shimpi and Y. Shouche of National Center ofCell Studies and N. Thakur for sequencing samples from India and J.Llewellyn-Hughes and C. Griffin for operating automated sequencers atthe NHM. We are grateful to the following people for providing samples,access to loan material (of both recent and fossil material) or help withfieldwork: T. Al Abdessalaam (Environment Agency, Abu Dhabi), M.Bemis (UF, Gainesville), P. Bouchet (MNHN, Paris), F. Boneka (USR,Manado), B. Buge (MNHN, Paris), J.-J. Cassan (Department of EconomicDevelopment and Environment, Noumea), B. Chan (AS, Taiwan), M.-H.Chen (KMU, Taiwan), A. Fields (JCU, Townsville), G. Feulner (Dubai,UAE), M. Florence (SI, Washington DC), M. Frey (UC, Davis), J. George(NHM, London), G. Giribet (MCZ, Harvard), E. Glover (NHM, London),T. Haga (University of Tokyo), G. Harasewych (SI, Washington DC), D.Herbert (Natal Museum, Natal), V. Heros (MNHN, Paris), R. Hornby(Nautica Environmental Associates, Abu Dhabi), P. Kendrick (CALM,Dampier), K. Kreipl (Meeresmuseum Oehringen, Germany), J. Liddelow(CALM, Dampier), I. Loch (AM, Sydney), L. Kirkendale (UVic, GreaterVictoria), H. Kinyo (Japan), P. Kuklinski (PAS, Sopot), K. Lam (UHK,Hong Kong), P. Maestrati (MNHN, Paris), M. Malaquias (NHM,London), B. Marshall (Te Papa Tongarewa, Wellington), R. Marwoto(LIPI, Indonesia), C. Meyer (SI, Washington DC), A. Miller (AM, Syd-ney), R. Moolenbeek (ZMA, Amsterdam), C. Muller (CALM, Dampier),T. Nangammbi (Natal Museum, Natal), B. Olivera (University of Utah),M. Pandolfi (Service de l’environnement marin, Noumea), G. Paulay(UF, Gainesville), N. Pilcher (Marine Research Foundation, Sabah), R.Pouwer (Naturalis, Leiden), R. Przeslawski (GA, Canberra), N. Puil-landre (MNHN, Paris), D. Reid (NHM, London), B. Richer de Forges(IRD, Noumea), M. Ruf (Marine Research Foundation, Sabah), A. SaadAl-Cibahy (Environment Agency, Abu Dhabi), N. Saguil (Philippines),T. Shimira (Japan), S. Slack-Smith (WAM, Perth), J. Slapcinsky (UF,Gainesville) T. Tajima (Japan), J. Taylor (NHM, London), K. Tilbrook(AUT, Auckland), J. Todd (NHM, London), G. Walker (NHM, London),M. Wallace (South Africa), T. Waller (SI, Washington DC), F. Wells (Dept.Fish., Perth), F. Wesseling (Naturalis, Leiden), C. Whisson (WAM, Perth),R. Willan (Museum and Art Gallery of the Northern Territory, Darwin), G.Williams (UHK, Hong Kong), P. Williams (Perth, Australia), I. Zagorskis(AIMS, Townsville). This study was supported in part by a grant fromthe Natural Environment Research Council to STW (NE/C507453/1),JSPS to TN (207024). Fieldwork by STW was partially funded by theSystematics Research Fund of the Linnean Society of London and theSystematics Association.

LITERATURE CITEDAlf, A., and K. Kreipl. 2003. The family Turbinidae, subfamily Turbininae,

genus Turbo. Conchbooks, Hackenheim, Germany.Alfaro, M. E., F. Santini, and C. D. Brock. 2007. Do reefs drive diversification

in marine teleosts? Examples from the pufferfishes and their allies (OrderTetraodontiformes). Evolution 61:2104–2126.

Alfaro, M. E., C. D. Brock, B. L. Banbury, and P. C. Wainwright. 2009.Does evolutionary innovation in pharyngeal jaws lead to rapid lineagediversification in labrid fishes? BMC Evol. Biol. 9:255.



Anderson, F. E., R. Engelke, K. Jarrett, T. Valinassab, K. S. Mohamed, P. K.Asokan, P. U. Zacharia, P. Nootmorn, C. Chotiyaputta, and M. Dunning.2011. Phylogeny of the Sepia pharaonic species complex (Cephalopoda:Sepiida) based on analyses of mitochondrial and nuclear DNA sequencedata. J. Mollus. Stud. 77:65–75.

Ayre, D. J., T. E. Minchinton, and C. Perrin. 2009. Does life history predictpast and current connectivity for rocky intertidal invertbrates across amarine biogeographic barrier? Mol. Ecol. 18:1887–1903.

Barber, P. H., and D. R. Bellwood. 2005. Biodiversity hotspots: evolutionaryorigins of biodiversity in wrasses (Halichoeres: Labridae) in the Indo-Pacific and new world tropics. Mol. Phylogenet. Evol. 35:235–253.

Bellwood, D. R., W. Renema, and B. R. Rosen. 2011. Biodiversity hotspots,evolution and coral reef biogeography: a review. In D. Gower, K. G.Johnson, B. R. Rosen, J. Richardson, L. Ruber and S. T. Williams, eds.Biotic evolution and environmental change in Southeast Asia. LinneanSociety, London.

Billingham, M. R., and D. J. Ayre. 1996. Genetic subdivision in the subti-dal, clonal sea anemone Anthothoe albocincta. Marine Biol. 125:153–163.

Castresana, J. 2000. Selection of conserved blocks from multiple align-ments for their use in phylogenetic analysis. Mol. Biol. Evol. 17:540–552.

Cernohorsky, W. O. 1974. Type specimens of Mollusca in the UniversityZoological Museum, Copenhagen. Rec. Auckland Inst. Mus. 11:143–192.

Claremont, M., S. T. Williams, T. G. Barraclough, and D. G. Reid. 2011.The geographic scale of speciation in a marine snail with high potentialdispersal. J. Biogeogr. In press.

Colgan, D. J., and S. Schreiter. 2011. Extrinsic and intrinsic influences on thephylogeography of the Austrocochlea constricta species group. J. Exp.Mar. Biol. Ecol. 397:44–51.

Collin, R. 2003. Phylogenetic relationships among Calyptraeid gastropodsand their implications for the biogeography of marine speciation. Syst.Biol. 52:618–640.

Dawson, M. N. 2005. Incipient speciation of Catostylus mosaicus (Scypho-zoa, Rhizostomeae, Catostylidae), comparative phylogeography and bio-geography in south-east Australia. J. Biogeogr. 32:515–533.

Dekker, H., R. G. Moolenbeek, and P. S. Dance. 1992. Turbo jonathani, anew turbinid species from the southern coast of Oman (Gastropoda:Turbinidae). J. Conchol. London 34:225–229.

Drummond, A. J., and A. Rambaut. 2007. BEAST: bayesian evolutionaryanalysis by sampling trees. BMC Evol. Biol. 7:214.

Drummond, A. J., B. Ashton, M. Cheung, J. Heled, M. Kearse, R. Moir,S. Stones-Havas, T. Thierer, and A. Wilson. 2009. Geneious v 4.7,Available from http://www.geneious.com.

Dwiono, S. A. P., Pradina, and P. C. Makatipu. 2001. Spawning and seedproduction of the green snail (Turbo marmoratus L.) in Indonesia. Sec-retariat of the Pacific Community Trochus Information Bulletin 7:9–13.

Eisawy, A. M., and A. E. Sorial. 1974. The egg masses and developmentof Turbo radiatus Gmelin. Bull. Inst. Oceanography Fisheries, Egypt4:221–235.

Frey, M. A. 2010. The relative importance of geography and ecology in speciesdiversifcation: evidence from a tropical marine intertidal snail (Nerita).J. Biogeogr. 37:1515–1528.

Frey, M. A., and G. J. Vermeij. 2008. Molecular phylogenies and historicalbiogeography of a circumtropical group of gastropods (Genus: Nerita):implications for regional diversity patterns in the marine tropics. Mol.Phylogenet. Evol. 48:1067–1086.

Fukuda, H. 1994. Marine Gastropoda (Mollusca) of the Ogasawara Is-lands (Bonin) Islands. Part 2: Neogastropoda, Heterobranchia and fossilspecies, with faunal accounts. Ogasawara Research 20:1–126.

Fukuda, H. 1995. Marine Gastropoda (Mollusca) of the Ogasawara Islands(Bonin) Islands. Part 3: Additional records. Ogasawara Research 21:1–142.

Gelman, A., and D. B. Rubin. 1992. Inference from iterative simulation usingmultiple sequences (with discussion). Stat. Sci. 7:457–511.

Grange, K. R. 1976. Larval development in Lunella smaragda (Gastropoda:Turbinidae). N. Z. J. Mar. Freshw. Res. 10:517–525.

Guzman del Proo, S. A., T. Reynoso-Granados, P. Monsalvo-Spencer, and E.Serviere-Zaragoza. 2003. Larval and early development of the wavy tur-ban snail, Megastraea undosa (Wood, 1828) (Gastropoda: Turbinidae).The Veliger 46:320–324.

Hall, R. 1998. The plate tectonics of Cenozoic SE Asia and the distributionof land and sea. Pp. 99–131 in R. Hall and J. D. Holloway, eds. Bio-geography and geological evolution of SE Asia. Backhuys Publishers,Leiden.

———. 2001. Cenozoic reconstructions of SE Asia and the SW Pacific:changing patterns of land and sea. Pp. 35–56 in I. Metcalfe, J. M.B. Smith, M. Morwood, and I. D. Davidson, eds. Faunal and floralmigrations and evolution in SE Asia–Australasia. A. A. Balkema (Swets& Zeitlinger Publishers), Lisse.

———. 2002. Cenozoic geological and plate tectonic evolution of SE Asia andthe SW Pacific: computer based reconstructions, models and animations.J. Asian Earth Sci. 20:353–431.

———. 2009. Southeast Asia’s changing paleogegraphy. Blumea 54:148–161.

Heled, J., and A. J. Drummond. 2010. Bayesian inference of species treesfrom multilocus data. Mol. Biol. Evol. 27:570–580.

Hellberg, M. E. 1998. Sympatric sea shells along the sea’s shore: the geogra-phy of speciation in the marine gastropod Tegula. Evolution 52:1311–1324.

Hoeksema, B. W. 2007. Delineation of the Indo-Malayan centre of maximummarine biodiversity: the coral triangle. Pp. 117–178 in W. Renema,ed. Biogeography, time and place: distributions, barriers and islands.Springer, Dordrecht.

Huelsenbeck, J. P., and F. Ronquist. 2001. MRBAYES: bayesian inference ofphylogenetic trees. Bioinformatics 17:754–755.

Imaizumi, T., and T. Tamura. 1984. Geomorphology of the Chichijima andHahajima Islands. Bull. Ogasawara Res. 8:3–11.

Ito, M., and M. Ono. 1990. Allozyme diversity and the evolution of Crepidi-

astrum (Compositae) on the Bonin (Ogasawara) Islands. Bot. Magaz.Tokyo 103:449–459.

Itoigawa, J. 1955. Molluscan fauna of the Mizunami group in the IwamuraBasin. Memoirs of the College of Science, University of Kyoto, SeriesB 22:127–143 + pls V-VI.

Jablonski, D., and G. Hunt. 2006. Larval ecology, geographic range, andspecies survivorship in Cretaceous mollusks: organismic versus species-level explanations. Am. Nat. 168:556–564.

Jablonski, D., and K. Roy. 2002. Geographical range and speciationin fossil and living molluscs. Proc. R. Soc. London 270:401–406.

Joll, L. M. 1980. Reproductive biology of two species of Turbinidae (Mollusca:Gastropoda). Aust. J. Mar. Freshw. Res. 31:319–336.

Kaizuka, S. 1977. Geology and geomorphology of the Bonin Islands. Bull.Ogasawara Res. 1:29–34.

Katoh, K., and H. Toh. 2008. Recent developments in the MAFFT multiplesequence alignment program. BMC Bioinform. 9:212.

Katoh, K., K. Misawa, K. Kuma, and T. Miyata. 2002. MAFFT: a novel methodfor rapid multiple sequence alignment based on fast Fourier transform.Nucleic Acids Res. 30:3059–3066.

Kemp, J. M. 1998. Zoogeography of the coral reef fishes of the SocotraArchipelago. J. Biogeogr. 25:919–933.



Kimura, M. 1980. A simple method for estimating evolutionary rate ofbase substitutions through comparative studies of nucleotide sequences.J. Mol. Evol. 16:111–120.

Kingman, J. F. C. 1982. On the genealogy of large populations. J. Appl.Probability 19A:27–43.

Kirkendale, L., and C. Meyer. 2004. Phylogeography of the Patelloidea pro-

funda group (Gastropoda: Lottidae): diversification in a dispersal-drivenmarine system. Mol. Ecol. 13:2749–2762.

Kohn, A. J. 1990. Tempo and mode of evolution in Conidae. Malacologia32:55–67.

Kono, N., and H. Yamakawa. 1999. Spawning and early development ofthe small turban shell, Marmarostoma stenogyrus (Vetigastropoda:Turbinidae). Venus 58:217–221.

Kuhnt, W., A. Holbourn, R. Hall, Zuvela, and R. Kase. 2004. Neogene Historyof the Indonesian Throughflow. Pp. 299–320 in P. Clift, P. Wang, W.Kuhnt, and D. E. Hayes, eds. Continent-Ocean Interactions within EastAsian Marginal Seas. Geophysical Monograph Series 149. AmericanGeophysical Union, Washington, DC.

Lai, K.-Y. 2006. A new species of Turbo from Taiwan (Gastropoda:Turbinidae). Bull. Malacol. Taiwan 30:37–42.

Lohse, K. 2009. Can mtDNA barcodes be used to delimit species? A responseto Pons et al. (2006). Syst. Biol. 58:439–442.

MacNeil, F. S. 1964. Eocene megafossils from Ishigaki-shima, Ryukyu-retto.Geological Survey of Professional Paper 399-B.

Maddison, D. R., and W. P. Maddison. 2003. MacClade. Version 4.06 OSX.Sinauer Associates, Sunderland, MA.

Meyer, C. P. 2003. Molecular systematics of cowries (Gastropoda: Cypraei-dae) and diversification patterns in the tropics. Biol. J. Linn. Soc. 79:401–459.

Meyer, C. P., J. B. Geller, and G. Paulay. 2005. Fine-scale endemism oncoral reefs: archipelagic differentiation in turbinid gastropods. Evolution59:113–125.

Monaghan, M. T., R. Wild, M. Elliot, T. Fujisawa, M. Balke, D. J. G. InWard, D. C. Lees, R. Ranaivosolo, P. Eggleton, T. G. Barraclough, andA. P. Vogler. 2009. Accelerated species inventory on Madagascar usingcoalescent-based models of species delimitation. Syst. Biol. 58:298–311.

Moritz, C. 1994. Defining “evolutionary significant units” for conservation.Trends Ecol. Evol. 9:373–375.

Nakano, T., K. Takahashi, and T. Ozawa. 2007. Description of an endangerednew species of Lunella (Gastropoda: Turbinidae) from the OgasawaraIslands. Venus 66:1–10.

Nakano, T., I. Yazaki, M. Kurokawa, K. Yamaguchi, and K. Kuwasawa. 2009.The origin of the endemic patellogastropod limpets of the OgasawaraIslands in the northwestern Pacific. J. Molluscan Stud. 75:87–90.

Papadopoulou, A., J. Bergsten, T. Fujisawa, M. T. Monaghan, T. G.Barraclough, and A. P. Vogler. 2008. Speciation and DNA barcodes:testing the effects of dispersal on the formation of discrete sequenceclusters. Philos. Trans. R. Soc. Lond. B 363:2987–2996.

Papadopoulou, A., M. T. Monaghan, T. G. Barraclough, and A. P. Vogler.2009. Sampling error does not invalidate the Yule-Coalescent Model forspecies delimitation. A response to Lohse (2009). Syst. Biol. 58:442–444.

Paulay, G. 1990. Effects of late Cenozoic sea-level fluctuations on the bivalvefaunas of tropical oceanic islands. Paleobiology 16:415–434.

Paulay, G., and C. Meyer. 2002. Diversificaton in the tropical Pacific: com-parisons between marine and terrestrial systems and the importance offounder speciation. Integr. Comp. Biol. 42:922–934.

Peltier, W. R. 2002. Comments on the paper of Yokoyama et al. (2000), entitled“Timing of the Last Glacial Maximum from observed sea level minima”.Quat. Sci. Rev. 21:409–414.

Pons, J., T. G. Barraclough, J. Gomez-Zurita, A. Cardoso, D. P. Duran, S.Hazell, S. Kamoun, W. D. Sumlin, and A. P. Vogler. 2006. Sequence-based species delimitation for the DNA taxonomy of undescribed in-sects. Syst. Biol. 55:595–609.

Rambaut, A., P. H. Harvey, and S. Nee. 1997. End-Epi: an application forreconstructing phylogenetic and population processes from molecularsequences. Comput. Appl. Biosci. 13:303–306.

Randall, J. E., and J. P. Hoover. 1995. Scarus zufar, a new species of parrotfishfrom southern Oman, with comments on endemism of the area. Copeia3:683–688.

Read, C. I., D. R. Bellwood, and L. van Herwerden. 2006. Ancient ori-gins of Indo-Pacific coral reef fish biodiversity: a case study ofthe leopard wrasses (Labridae: Macropharyngodon). Mol. Phylogenet.Evol. 38.

Reid, D. G. 1986. The littorinid molluscs of mangrove forests in the Indo-Pacific region: the genus Littoraria. British Museum (Natural History),London.

Reid, D. G., K. Lal, J. Mackenzie-Dodds, F. Kaligis, D. T. J. Littlewood, andS. T. Williams. 2006. Comparative phylogeography and speciesboundaries in Echinolittorina snails in the central Indo-West Pacific.J. Biogeogr. 33:990–1006.

Reid, D. G., P. Dyal, and S. T. Williams. 2010. Global diversification ofmangrove fauna: a molecular phylogeny of Littoraria (Gastropoda: Lit-torinidae). Mol. Phylogenet. Evol. 55:185–201.

Renema, W., D. R. Bellwood, J. C. Braga, K. Bromfield, H. Hall, K. G.Johnson, P. Lunt, C. P. Meyer, I. B. McMonagle, R. J. Morley, et al.2008. Hopping hotspots: global shifts in marine biodiversity. Science321:654–657.

Saito, T., H. Okado, and K. Kaiho. 1984. Biostratigraphy and internationalcorrelation of the Paleogene system in Japan. Publication DepartmentEarth Science, Faculty of Science, Yamagata Univ., Japan.

Springer, V. G. 1982. Pacific plate biogeography with special reference toshore fishes. Smithson. Contrib. Zool. 367:1–182.

Taylor, J. D. 1997. Diversity and structure of tropical Indo-Pacific benthiccommunities: relation to regimes of nutrient input. Pp. 178–200 in R.F. G. Ormond, J. D. Gage, and M. V. Angel, eds. Marine biodiversity:patterns and processes. Cambridge Univ. Press, London.

Taylor, J. D., and K. R. Smythe. 1985. A new species of Trochita (Gastropoda:Calyptraeidae) from Oman; a relict distribution and association withupwelling areas. J. Conchol. 32:39–48.

Valentine, J. W. 1971. Plate tectonics and shallow marine diversity and en-demism, an actualistic model. Syst. Biol. 20:253–264.

Valentine, J. W., and D. Jablonski. 1982. Major determinants of the bio-geographic pattern of the shallow-sea fauna. Bulletin de la SocieteGeologique de France 24:893–899.

———. 1983. Speciation in the shallow sea: general patterns and biogeo-graphic controls. Pp. 201–226 in R. W. Sims, J. H. Price, and P. E. S.Whalley, eds. Evolution, time and space: the emergence of the biosphere.Systematics Association Special Volume 23. Academic Press, London.

Walsby, J. R. 1977. Population variations in the grazing turbinid Lunella

smaragda (Mollusca: Gastropoda). N. Z. J. Mar. Freshwater Res.11:211–238.

Ward, D. W., and A. R. Davis. 2002. Reproduction of the turban shell Turbo

torquatus Gmelin 1791 (Mollusca: Gastropoda), in New South Wales,Australia. Mar. Freshw. Res. 53:85–91.

Waters, J. M., T. M. King, P. M. O’Loughlin, and H. G. Spencer. 2005. Phylo-geographical disjunction in abundant high-dispersal littoral gastropods.Mol. Ecol. 14:2789–2802.

Waters, J. M., G. A. McCulloch, and J. A. Eason. 2007. Marine biogeo-graphical structure in two highly dispersive gastropods: implications fortrans-Tasman dispersal. J. Biogeogr. 34:678–687.



Williams, S. T. 2007. Origins and diversification of Indo-West Pacific marinefauna: evolutionary history and biogeography of turban shells (Gas-tropoda, Turbinidae). Biol. J. Linn. Soc. 92:573–592.

Williams, S. T., and D. G. Reid. 2004. Speciation and diversity on tropicalrocky shores: a global phylogeny of snails of the genus Echinolittorina.Evolution 58:2227–2251.

Williams, S. T., and T. Ozawa. 2006. Molecular phylogeny suggests polyphylyof both the turban shells (family Turbinidae) and the superfamily Tro-choidea (Mollusca: Vetigastropoda). Mol. Phylogenet. Evol. 39:33–51.

Williams, S. T., and T. F. J. Duda. 2008. Did tectonic activity stimulate Oligo-Miocene speciation in the Indo-West Pacific? Evolution 62:1618–1634.

Williams, S. T., K. M. Donald, H. G. Spencer, and T. Nakano. 2010. Molecu-lar systematics of the marine gastropod families Trochidae and Callios-tomatidae (Mollusca: Superfamily Trochoidea). Mol. Phylogenet. Evol.54:783–809.

Williams, S. T., A. Hall and P. Kuklinski. 2011. Utility of elliptic Fourierdescriptors of shell shape for distinguishing cryptic taxa in the gastropodgenus Lunella (Turbinidae). Am. Malacol. Bull. In press.

Wilson, M. E. J., and B. R. Rosen. 1998. Implications of paucity of corals inthe Paleogene of SE Asia: plate tectonics or centre of origin? Pp. 165–195 in R. Hall, and J. D. Holloway, eds. Biogeography and geologicalevolution of SE Asia. Backhuys Publishers, Leiden.

Yamaguchi M. 1993. Green snail. Pp. 497–511 in A. Wright and L. Hill, eds.Nearshore marine resources of the South Pacific. Institute for PacificStudies, Suva, Fiji.

Yule, G. U. 1924. A mathematical theory of evolution based on the conclusionsof Dr J. C. Willis, F.R.S. Philos. Trans. R. Soc. Lond. B Biol. Sci. 213:21–87.

Associate Editor: M. Hellberg

Supporting InformationThe following supporting information is available for this article:

Table S1. List of taxa used in study. Lunella species in order of clades named in Figure 5, followed by outgroups.

Table S2. Forward (F) and reverse (R) PCR primers (also used in sequencing), and forward (FS) and reverse (RS) internal

sequencing primers.

Supporting Information may be found in the online version of this article.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting information supplied by the

authors. Any queries (other than missing material) should be directed to the corresponding author for the article.


SPECIATION AND DISPERSAL ALONG CONTINENTAL COASTLINES AND ISLAND ARCS IN THE INDO-WEST PACIFIC...

Documents

Transcript of SPECIATION AND DISPERSAL ALONG CONTINENTAL COASTLINES AND ISLAND ARCS IN THE INDO-WEST PACIFIC...