Estimation of divergence times in cnidarian evolution based on mitochondrial protein-coding genes and the fossil record

Molecular Phylogenetics and Evolution 62 (2012) 329–345

Contents lists available at SciVerse ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier .com/ locate /ympev

Estimation of divergence times in cnidarian evolution based on mitochondrialprotein-coding genes and the fossil record

Eunji Park a, Dae-Sik Hwang b, Jae-Seong Lee b, Jun-Im Song a, Tae-Kun Seo c,d, Yong-Jin Won a,⇑a Division of EcoScience, Ewha Womans University, 11-1 Daehyon-Dong, Sodaemun-Gu, Seoul 120-750, Republic of Koreab National Research Laboratory of Marine Molecular and Environmental Bioscience, Department of Chemistry, the Research Institute of Natural Sciences,College of Natural Sciences, Hanyang University, Seoul 133-791, Republic of Koreac Agricultural Bioinformatics Research Unit, Graduate School of Agricultural and Life Sciences, University of Tokyo, Tokyo 113-8657, Japand Division of Life Sciences, Korea Polar Research Institute, Songdo Technopark 7-50, Incheon 406-840, Republic of Korea

a r t i c l e i n f o

Article history:Received 28 February 2011Revised 1 October 2011Accepted 4 October 2011Available online 20 October 2011

Keywords:CnidariaHexacoralliaOctocoralliaMedusozoaDivergence timesMitochondrial genome

1055-7903/$ - see front matter � 2011 Elsevier Inc. Adoi:10.1016/j.ympev.2011.10.008

⇑ Corresponding author. Fax: +82 2 3277 4514.E-mail address: [email protected] (Y.-J. Won).

a b s t r a c t

The phylum Cnidaria is comprised of remarkably diverse and ecologically significant taxa, such as thereef-forming corals, and occupies a basal position in metazoan evolution. The origin of this phylumand the most recent common ancestors (MRCAs) of its modern classes remain mostly unknown, althoughscattered fossil evidence provides some insights on this topic. Here, we investigate the molecular diver-gence times of the major taxonomic groups of Cnidaria (27 Hexacorallia, 16 Octocorallia, and 5 Meduso-zoa) on the basis of mitochondrial DNA sequences of 13 protein-coding genes. For this analysis, thecomplete mitochondrial genomes of seven octocoral and two scyphozoan species were newly sequencedand combined with all available mitogenomic data from GenBank. Five reliable fossil dates were used tocalibrate the Bayesian estimates of divergence times. The molecular evidence suggests that cnidariansoriginated 741 million years ago (Ma) (95% credible region of 686–819), and the major taxa diversifiedprior to the Cambrian (543 Ma). The Octocorallia and Scleractinia may have originated from radiationsof survivors of the Permian–Triassic mass extinction, which matches their fossil record well.

� 2011 Elsevier Inc. All rights reserved.

1. Introduction

The phylum Cnidaria comprises over 9000 animal species, suchas corals, sea anemones, sea fans, sea pens, jellyfish, and hydroids,that have simple body forms and produce cnidae, the diagnosticfeature of the phylum (Daly et al., 2007; Fautin, 2009; Rogers,2009). Cnidarians are taxonomically subdivided into approxi-mately seven classes and 25 orders that exhibit diverse adaptationswith respect to their morphology, reproduction, and ecology (Dalyet al., 2007; Fautin, 2009; Rogers, 2009). Although they are distrib-uted worldwide in shallow to deep ocean environments, some lin-eages also occur in fresh water (Fabricius and Alderslade, 2001).Cnidarians originated early in the history of metazoan evolution,as indicated by fossil evidence (Ausich and Babcock, 1998; Cart-wright et al., 2007; Chen et al., 2002; Hagadorn and Waggoner,2000; Han et al., 2010) and molecular phylogenies (Dunn et al.,2008; Peterson et al., 2004, 2008).

Considering their early phylogenetic position, the followingquestions arise: when did the Cnidaria first appear, and when didthe major classes and subclasses diverge to produce the modernextant taxa? Although they are fascinating, these questions remain

ll rights reserved.

mostly unsolved. The hard calcareous skeletons of the Scleractinia,belonging to the Anthozoa, have left well-preserved fossils sincetheir first appearance in the Triassic (Stanley, 2003; Veron et al.,1996; Veron, 1995); however, the origins of the soft-bodied taxaare poorly understood. Ambiguity exists in assigning the limitedknown fossils to relatives of the modern taxa, as exemplified bythe soft corals of the Octocorallia (Ausich and Babcock, 2000;Bengtson, 1981).

During the past decade, a few molecular studies have attemptedto estimate divergence times for the deeper cnidarian nodes (Cart-wright and Collins, 2007; Medina et al., 2006; Peterson et al., 2004;Waggoner and Collins, 2004). Recent methodological advances inthe estimation of divergence times have made it possible to esti-mate the ages of phylogenetic nodes beyond calibration pointsbased on a limited number of fossil dates. Though the results mustbe carefully interpreted and guided by the fossil data, these ad-vances provide opportunities to assess the origins of major cnidar-ian lineages using the limited fossil records (Cartwright andCollins, 2007; Peterson et al., 2007).

Classical molecular clocks (Zuckerkandl and Pauling, 1962),assuming that the evolutionary rates are constant, have been usedto estimate divergence times for many taxa; for example, theyhave been employed to calculate deeper divergence times of meta-zoan phyla (Doolittle et al., 1996; Wray et al., 1996) and hard skel-

http://dx.doi.org/10.1016/j.ympev.2011.10.008

mailto:[email protected]


http://www.sciencedirect.com/science/journal/10557903

http://www.elsevier.com/locate/ympev

330 E. Park et al. / Molecular Phylogenetics and Evolution 62 (2012) 329–345

etal corals of the Scleractinia (Romano and Palumbi, 1996). How-ever, variability in mutation rates, generation lengths, effectivepopulation sizes, and functional constraints on gene products canproduce different evolutionary rates for various taxonomic groups.This variability has led to more sophisticated estimation proce-dures that relax the assumptions about rate homogeneity. The ‘‘lo-cal clock’’ approach (Yoder and Yang, 2000) assumes the existenceof multiple molecular clocks that are assigned to various taxo-nomic groups, within which evolutionary rates are assumed tobe constant. Nonparametric and ‘‘penalized likelihood’’ methods(Sanderson, 1997, 2002) assume different evolutionary ratesamong lineages to estimate lineage-specific rates and divergencetimes. Rate heterogeneity among lineages was further investigatedand modeled using a log-normal distribution under a Bayesianframework (Thorne et al., 1998). Nucleotide substitution rates inprotein-coding sequences can be partitioned for nonsynonymousand synonymous changes that can be modeled with bivariatelog-normal distributions (Seo et al., 2004). An advantage of themethod described by Seo et al. over that of Thorne et al. is thatchanges in selective pressure can be properly estimated. Bothmethods adopt fixed tree topologies and simple strategies for theincorporation of fossil data, but other Bayesian methods can jointlyestimate tree topologies and divergence times (Drummond andRambaut, 2007) and incorporate statistical models that deal withthe uncertainty of fossil dates (Yang and Rannala, 2006). To analyzecnidarian mitochondrial genomes, we adopted the methods of Seoet al. and Thorne et al. because they assume similar rate changemodels and particularly the former method provides an additionalfunction for estimating synonymous and nonsynonymous rates.

As single genes or combinations of genes, mitogenomes containuseful markers for phylogenetic analyses. Considerable effortswere previously directed towards producing highly resolved treesfor Anthozoa based on the mitochondrial cox1, msh1 and nad2genes in the Octocorallia (McFadden et al., 2006, 2009, 2010) andcox1 in the Scleractinia, which belongs to the Hexacorallia (Kitaha-ra et al., 2010). Though low substitution rates for these genes mayprovide insufficient information to distinguish some species andgenera of Octocorallia (McFadden et al., 2011), mitochondrialgenes have long been used for phylogenetic studies to capture glo-bal phylogenetic relationships among representative anthozoantaxa from families to suborders and orders (France and Hoover,2002; France et al., 1996; Hellberg, 2006; Shearer et al., 2002).

In general, mitochondrial genes of nonbilaterian animals, suchas the Medusozoa and Anthozoa, have been useful for clarifyinghigher phylogenetic relationships (Gissi et al., 2008; McFaddenet al., 2010). The combined information from several mitochondrialgenes increases confidence in estimates of higher evolutionaryrelationships and deep divergence times in anthozoan lineagesthat could not be achieved using fewer individual genes (Bruglerand France, 2007; Medina et al., 2006). While the small and largesubunit ribosomal RNA genes are often chosen for phylogeneticstudies of Medusozoa (Collins, 2002; Collins et al., 2006), a lackof mitogenomic data exists for this taxon. Before the present study,only one mitogenomic sequence for Scyphozoa and two for Hydro-zoa were available in GenBank.

Here, we explore published mitogenomic data that have accu-mulated for 39 species to date (27 Hexacorallia, 9 Octocoralliaand 3 Medusozoa) to estimate divergence times. We also se-quenced and characterized the complete mitogenomes from sevenoctocoral species and two jellyfish species, resulting in a total of 48cnidarians included in the present study. We investigated evolu-tionary rates and divergence times by applying sophisticatedBayesian methods to 13 protein-coding genes from all availablemitogenomic data. By assuming a log-normal rate change modeland separating nucleotide substitutions into synonymous andnonsynonymous partitions, we evaluated changes in selective

pressures. The obtained divergence time estimates are comparedwith those estimated from previous studies.

2. Materials and methods

2.1. Sample collection and DNA extraction

2.1.1. OctocoralsWe sequenced the complete mitogenomes of seven octocorals

(Dendronephthya castanea, Dendronephthya mollis, Dendronephthyaputteri, Dendronephthya suensoni, Scleronephthya gracillimum,Euplexaura crassa, and Echinogorgia complexa) from order Alcyona-cea. Among these species, three (D. castanea, D. suensoni, and S.gracillimum) were collected at depths between 10 and 20 m belowthe sea surface on the submerged rock cliff of Munseom Island,Korea in 2005. The seven octocorals represent two families (Nepht-heidae and Plexauridae) from the Alcyoniina–Holaxonia clade(McFadden et al., 2006) (Table 1). Several 2-cm long branches weredissected for each specimen and stored in 95% ethanol until use.Genomic DNA was isolated using a DNeasy Blood and Tissue Kit(Qiagen Inc., Valencia, USA) following the manufacturer’s protocol.We obtained DNA samples of the remaining four species (D. mollis,D. putteri, E. crassa, and E. complexa) from the Korean Coral Re-source Bank (Ewha Womans University, Korea). Voucher speci-mens of the seven octocorals were deposited at the NaturalHistory Museum in Ewha Womans University, Korea (Table 1 andSupplementary Table S1).

2.1.2. ScyphozoansThe two jellyfish species examined here, Aurelia aurita and

Chrysaora quinquecirrha, are distributed worldwide and have fre-quently been found in southern Korean coastal waters. Specimensof these two scyphozoans were collected from Korean coastalwaters near Incheon and Geoje-do, respectively. Individuals wereimmediately preserved in absolute ethanol for dehydration andstored at room temperature until use. Genomic DNA was isolatedfrom the stored tissues using Proteinase K treatment followed bychloroform extraction and isopropanol precipitation. The isolatedDNA was further purified using the DNeasy tissue kit (Qiagen,Valencia, CA). The voucher sample of A. aurita was deposited atthe Natural History Museum in Ewha Womans University, Korea(Table 1 and Supplementary Table S1).

2.2. PCR amplification and sequencing

2.2.1. OctocoralsTo amplify and read the complete mitochondrial genome se-

quences (Fig. 1), we used 31 pairs of primers that were previouslydesigned and successfully applied to for whole mitogenomesequencing of the soft coral species Dendronephthya gigantea (Parket al., 2010). PCR amplification reactions using the 31 primer pairswere performed with the following conditions: an initial denatur-ation step at 94 �C for 1 min followed by 35 cycles at 92 �C for 40 s,50 �C for 1 min, and 72 �C for 1 min, with a final extension step at72 �C for 7 min. The PCR products were purified with the LaboPassPCR purification kit (Cosmo Genetech Inc., Seoul, Korea) and se-quenced using an ABI3730XL instrument (Applied BiosystemsInc., Carlsbad, USA).

2.2.2. JellyfishesWe designed seven primer sets to amplify sequences from the

two scyphozoan jellyfishes (A. aurita and C. quinquecirrha) usingthe published mitochondrial genome of A. aurita (GenBank acces-sion no: NC_008446, Shao et al., 2006). PCR amplification reactionswere performed with the following conditions: 40 cycles at 98 �C

Table 1A list of the cnidarian mitochondrial genomes available in GenBank.

Class Subclass Order Suborder Family Species Total length(bp)

C + G(%)

Proteinno.

tRNAno.

AccessionID

Anthozoa Hexacorallia Actiniaria – Edwardsiidae Nematostella sp. 16,389 39 13 2 NC_008164– Metridiidae Metridium senile 17,443 38 14 2 NC_000933

Antipatharia – Cladopathidae Chrysopathes formosa 18,398 39 13 2 NC_008411Corallimorpharia – Actinodiscidae Rhodactis sp. 21,376 38 13 2 NC_008159

– Discosomatidae Discosoma sp. (1) 20,908 38 13 2 NC_008071– Discosoma sp. (2) 20,912 38 13 2 NC_008072– Ricordeidae Ricordea florida 21,376 37 13 2 NC_008159

Scleractinia Astrocoeniina Acroporidae Acropora tenuis 18,338 37 13 2 NC_003522Anacropora matthai 17,888 38 13 2 NC_006898Montipora cactus 17,887 38 13 2 NC_006902

Pocilloporidae Madracis mirabilis 16,951 31 13 2 NC_011160Pocillopora eydouxi 17,422 30 14 2 NC_009798Pocillopora damicornis 17,415 30 14 2 NC_009797Seriatopora caliendrum 17,010 30 13 3 NC_010245Seriatopora hystrix 17,059 30 13 3 NC_010244Stylophora pistillata 17,177 29 13 2 NC_011162

Faviina Faviidae Colpophyllia natans 16,906 33 13 2 NC_008162Montastraea annularis 16,138 33 13 2 NC_007224Montastraea faveolata 16,138 33 13 2 NC_007226Montastraea franksi 16,137 33 13 2 NC_007225

Mussidae Mussa angulosa 17,245 33 13 2 NC_008163Rhizangiidae Astrangia sp. 14,853 31 13 2 NC_008161

Fungiina Agariciidae Agaricia humilis 18,735 40 13 2 NC_008160Pavona clavus 18,315 40 13 2 NC_008165

Poritidae Porites porites 18,648 36 13 2 NC_008166Siderastreidae Siderastrea radians 19,387 36 13 2 NC_008167

Zoanthidea – Parazoanthidae Savalia savaglia 20,764 48 14 1 NC_008827Anthozoa Octocorallia Alcyonacea Alcyoniina Alcyoniidae Sarcophyton glaucum 18,452 37 14 1 AF063191

AF064823Nephtheidae Dendronephthya

gigantea18,842 37 14 1 NC_013573

Nephtheidae Dendronephthyacastanea

18,907 37 14 1 GU047877a

Nephtheidae Dendronephthya mollis 18,844 37 14 1 HQ694725a

Nephtheidae Dendronephthya putteri 18,853 37 14 1 HQ694726a

Nephtheidae Dendronephthyasuensoni

18,885 37 14 1 GU047878a

Nephtheidae Scleronephthyagracillimum

18,950 37 14 1 GU047879a

Calcaxonia Isididae Acanella eburnea 18,616 37 14 1 NC_011016Isididae Keratoisidinae sp. 18,923 37 14 1 NC_010764

Holaxonia Gorgoniidae Pseudopterogorgiabipinnata

18,733 37 14 1 NC_008157

Plexauridae Euplexaura crassa 18,674 37 14 1 HQ694728a

Plexauridae Echinogorgia complexa 19,445 37 14 1 HQ694727a

Plexauridae Calicogorgia granulosa 20,246 38 16 1 GU047880Scleraxonia Briareidae Briareum asbestinum 18,632 37 14 1 NC_008073

Coralliidae Corallium konojoi 18,969 38 14 1 NC_015406Coralliidae Paracorallium japonicum 18,913 37 14 1 NC_015405

Hydrozoa Hydroida Anthomedusae Hydridae Hydra oligactis 16,314 23 13 2 NC_010214Hydridae Hydra magnipapillata 8194 22 6 2 NC_011220

7686 25 7 1 NC_011221Scyphozoa Semaeostomeae – Ulmaridae Aurelia aurita (1) 16,937 33 15 2 NC_008446

– Ulmaridae Aurelia aurita (2) 16,970 34 15 2 HQ694729a

– Pelagiidae Chrysaora quinquecirrha 16,774 31 15 2 HQ694730a

a Sequence information was collected in the present study.

E. Park et al. / Molecular Phylogenetics and Evolution 62 (2012) 329–345 331

for 20 s, 50 �C for 1 min, and 72 �C for 2 min. We used the step-outPCR method to obtain nucleotide sequences from two ends of themitogenome (cob, orf324 and partial orf969) (Wesley and Wesley,1997), as Medusozoa are known to have a linear mitogenomestructure (Bridge et al., 1992; Wesley and Wesley, 1997).

2.3. Gene annotation and sequence analysis

A total of 31 DNA fragments that were amplified and sequencedusing the 31 primer sets were aligned and assembled for the con-struction of a complete contig for the mitogenome of each investi-gated species on the basis of partially overlapping terminal

sequences between adjacent DNA segments using the AlignIR pro-gram (LiCoR, Lincoln, US). Subsequently, gene identification andannotation were carried out based on multiple sequence align-ments with the mitogenomes of six octocoral species that werepreviously published and annotated by other researchers (Gen-Bank accession nos.: DQ640646, DQ640649, EF622534, EF672731,FJ372991, and the combined sequence of AF064823 andAF063191 for Sarcophyton glaucum) (Table 1). ORF Finder (providedby NCBI) located many genes that were previously misannotated,so we corrected these cases and used the corrected data for thesubsequent analyses (Table 2). Additionally, because previousstudies on mitochondrial RNA genes of Octocorallia produced

cox1cox1

rnsrns

nad1nad1

cobcob

nad6nad6nad3nad3nad4L

nad4L

msh1msh1rnlrnl

nad2nad2

nad5nad5

nad4nad4

cox3cox3atp6atp6

atp8atp8

cox2cox2tRNAtRNA

cox1cox1

rnsrns

nad1nad1

cobcob

cox1cox1

rnsrns

nad1nad1

cobcob

cox1cox1

rns

nad1nad1

cobcob

cox3cox3atp6atp6

atp8atp8

cox2cox2

tRNAtRNA

nad6nad6nad3nad3nad4L

nad4L

msh1msh1

rnl

nad2 nad5

nad4nad4

nad6nad6

nad4Lnad4L

msh1msh1

rnlrnl

nad2nad2

nad5nad5 nad4nad4

cox3cox3

atp6atp6 atp8atp8

cox2cox2

tRNAtRNA

msh1msh1

rnlrnlnad2nad2

nad5nad5

nad4nad4

cox3cox3

atp6atp6

atp8atp8 cox2cox2

nad6nad6

nad3nad3nad4Lnad4L

P. japonicum

C. konojoi

K. spA. eburnea

Alcyoniidae Nephtheidae Gorgoniidae Plexauridae Briareidae

tRNAtRNA

A

D

CB

orf1

trn-

Met

cox3

atp6

atp8

trn-

Sec

cox2

cox1

orf2

cob

nad4

nad1

nad4

L

nad3

nad6

rns

nad5

nad2rnl

transcription

(a)

(b)

nad3nad3

Fig. 1. Mitogenomic organization of the octocorals and the two scyphozoan jellyfish species examined in this study. (a) Four different types (A, B, C, and D) of mitogenomes ofthe octocorals for which mitogenomes have been sequenced to date. The circular gene maps are illustrated with symbolic gene names according to the information presentedin Table 2 and represent gene arrangements on circular DNA. Type A includes 12 species from five families: Alcyoniidae (S. glaucum), Nephtheidae (D. gigantea, D. castanea, D.mollis, D. putteri, D. suensoni, and S. gracillimum), Gorgoniidae (P. bipinnata), Plexauridae (C. granulosa, E. crassa, and E. complexa), and Briareidae (B. asbestinum). The othertypes are as follows: type B is observed in Isididae (K. sp. and A. eburnea), type C in Coralliidae (P. japonicum), and type D in Coralliidae (C. konojoi). The thick circular linesaround the gene maps show contiguous stretches of genes that are inverted between two different mitogenome types. The circular lines with arrowheads inside the mapindicate the direction of transcription; dark and light gray represent the heavy and light strands of mitochondrial DNA, respectively. (b) Linear mitogenomes of Aurelia auritaand Chrysaora quinquecirrha sequenced in this study. The thick lines with arrowheads indicate the direction of transcription. The gene arrangement of these two species is thesame as that found for another individual sample of A. aurita published previously (GenBank accession no. NC_008446).


inconsistent annotations regarding the gene lengths in some spe-cies based on different sets of reference sequences for alignmentsand comparisons, we aligned our rns and rnl sequences with moredistantly related cnidarian species, such as A. aurita, Hydra oligactisand Pocillopora damicornis. Thus, we were able to finally obtain amore stable alignment of these genes with respect to their startingand ending points. We searched for possible tRNAs using the tRNAscan-SE program (Lowe and Eddy, 1997). Finally, Tandem RepeatFinder 4.0 (Benson, 1999) was used to find repeat motifs withinthe mitogenomes.

2.4. Taxon sampling

The mitogenomes of 27 hexacorals, 9 octocorals, 2 hydrozoansand 1 scyphozoan (for a total of 39) are currently accessible in Gen-Bank (Table 1). In addition, the seven octocoral mitogenomes (D.castanea, D. mollis, D. putteri, D. suensoni, S. gracillimum, E. crassaand E. complexsa) and two scyphozoan mitogenomes reported inthe present study were used for phylogenetic analyses. As out-groups, we included five representatives of different clades of dem-osponges (G0, G1, G2, G3, and G4) because demosponges havebeen repeatedly confirmed to be a basal outgroup to the Cnidaria

in molecular phylogenies based on a very large EST dataset (Dunnet al., 2008), 18S rRNA genes (Collins, 2002), nuclear coding genes(Peterson et al., 2004), and mitogenome sequences (Erpenbecket al., 2007; Kayal and Lavrov, 2008). The accession numbers ofthe outgroups are as follows: Oscarella carmela, NC_009090; Iger-nella notabilis, NC_010216; Aplysina fulva, NC_010203; Amphime-don compressa, NC_010201; and Agelas schmidti, NC_010213.

2.5. Phylogenetic analyses

For tree reconstruction, we carried out Bayesian phylogeneticanalyses. The tree was reconstructed using nucleotide sequenceswith MrBayes3.1.1 (Ronquist and Huelsenbeck, 2003). To constructa dataset, we included the sequences of 13 protein-coding genes(atp6 and 8, cob, cox1–3, and nad1–6 and 4L) from all of the 53operational taxonomic units noted above. Each protein set wasaligned in Mega 5.0 (Kumar et al., 2008), and ambiguous sites werethen eliminated with GBLOCK, allowing gap sites by indel variationwithin the final block (Talavera and Castresana, 2007). The result-ing sequences were concatenated, and a single amino acid align-ment was constructed (3539 a.a.). A nucleotide dataset wasproduced according to the standard amino acid alignment

Table 2Lengths of genes and IGRs of 16 octocoral species.

Gene and Alcyoniidae Nephtheidae Isididae Gorgoniidae Plexauridae Briareidae Coralliidae

IGR S. glaucum D.gigantea

D.castanea

D.mollis

D.putteri

D.suensoni

S.gracillimum

A.eburnea

K. sp. P. bipinnata C.granulosa

E.crassa

E.complexa

B.asbestinum

C.konojoi

P.japonicum

cox1 1594 1597 1597 1597 1597 1597 1597 1597 1597 1597 1597 1597 1597 1597 1597 1597cox1–rnl �7 �7 �7 �7 �7 �7 �7 �7 �7 �7 �7 �7 �7 �7 �7 �7rns 1021a 1035a 1035 1035 1035 1035 1035 1022 1031 1036a 1019 1036 1036 717a, c 1043a 1043a

rnl–nad1 49 50 50 50 50 50 50 46 46 50 40 50 50 65 53 53nad1 972a 972 972 972 972 972 972 981 981 972 972 972 972 972 972 972nad1–cob 33 64 64 64 65 64 63 101 64 62 19 63 33 65 6 41cob 1161 1155a 1155 1155 1155 1155 1155 1161 1161 1161a 1152 1161 1155 1161a 1161 1194cob–nad6d 13 199 199 199 199 199 317 42 432 40 10 39 424 74 – –nad6 558 558 558 558 558 558 558 552 552 558 519 558 558 558 555 555nad6–nad3 43 43 43 43 43 43 43 16 17 55 886 41 43 54 29 35nad3 354 354 354 354 354 354 354 348 348 354 354 354 354 354 354 354nad3–nad4L 3 19 19 19 19 19 19 20 19 19 14 19 19 30 34 38nad4L 294 294 294 294 294 294 294 294 294 294 294 294 294 294 294 294nad4L–msh1d 14 13 13 13 13 13 13 – – 13 9 10 13 13 – 14msh1 2949 2940 2940 2940 2940 2940 2949 2964 2982 2958 2949 2949 2958 3066 2991 2991msh1–rnl 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0rnl 1928a 1958a 2023 1958 1956 2003 1953 1956a 1926a 1948a 1957 1927 1941 1987a 1957a 1971a

rnl–nad2 6 19 19 19 19 19 19 4 35 33 28 33 19 34 32 31nad2 1374 1365a 1365 1365 1365 1365 1365 1350a 1320 1374a 1374 1374 1359 1389a 1365a 1365a

nad2–nad5 �13 �13 �13 �13 �13 �13 �13 �12 �13 �11 �13 �13 �13 �13 �13 �13nad5 1818 1818 1818 1818 1818 1818 1818 1812 1812 1818 1818 1818 1818 1818 1818 1818nad5–nad4d 20 84 84 84 84 84 84 – – 97 92 90 514 49 24 24nad4 1449a 1449 1449 1449 1449 1449 1449 1449 1449 1449 1449 1449 1449 1449 1449 1449nad4–trnM 39 72 72 74 85 70 58 11 0 60 34 56 56 100 62 59trnM 71 71 71 71 71 71 71 71 71 71 71 71 71 71 71 71trnM–cox3 31 39 39 39 38 39 39 34 29 39 33 39 39 32 36 39cox3 786 786 786 786 786 786 786 786 786 786 786 786 786 786 786 786cox3–atp6 58 64 64 64 64 64 65 33 33 64 852 64 64 26 27 27atp6 708a 708 708 708 708 708 708 708a 708a 708 708 708 708 708 708 708atp6–atp8 18 24 24 24 24 24 24 26 21 24 19 24 23 24 20 20atp8 216a 216 216 216 216 216 216 216 216 216 216 216 216 216a 216 216atp8–cox2 20 22 22 22 22 22 22 4 4 22 125 22 22 28 17 20cox2 762 762 762 762 762 762 762 762 762 762 762 762 762 762 762 762cox2–cox1d 111 112 112 112 112 112 112 110 109 111 108 112 112 153 – –

Total genelength

18,015 18,038 18,103 18,038 18,036 18,083 18,042 18,029 17,996 18,062 17,997 18,032 18,034 17,905 18,099 18,146

Total pcgb 14,995 14,974 14,974 14,974 14,974 14,974 14,983 14,980 14,968 15,007 14,950 14,998 14,986 15,130 15,028 15,061Total IGR 438 804 804 806 817 802 908 587 927 671 2249 642 1411 727 870 767

Total length 18,453 18,842 18,907 18,844 18,853 18,885 18,950 18,616 18,923 18,733 20,246 18,674 19,445 18,632 18,969 18,913

a Newly corrected length according to screening by ORF Finder, as described in Section 2.3.b Abbreviation of protein-coding gene.c Previously, the length of B. asbestinum rns was deduced to 581 bp (Medina et al., 2006). In this study, however, the length was adjusted to 717 bp by alignment with other distantly related cnidarian species, including Aurelia

aurita, Hydra oligactis and Pocillopora damicornis. The difference was found at the 50 end of rns, which is consistent with a previous prediction that the excessively short length was likely due to incomplete alignment at both the 50

and 30 ends (Brugler and France, 2007). Nevertheless, our alignment approach for determining the boundary of B. asbestinum rns still results in some ambiguity due to its highly divergent sequence, and thus, 50 and 30 RACE analysismay be needed to confirm the boundary.

d IGR locations that differ among some octocoral groups. The species belonging to the families Isididae and Corallidae show inverted portions compared to a major type of mitogenome (Fig. 1). Because of the different genearrangements due to inversions, some IGRs were lost (–), and new IGRs were placed in different locations; A. eburnean: trnM–msh1 (47 bp) and nad4–nad4L (112 bp); K. sp.: trnM–msh1 (35 bp) and nad4–nad4L (103 bp); C. konojoi:cob–msh1 (80 bp), cox1–nad6 (405 bp), and cox2–nad4L (65 bp); P. japonicum: cox1–nad6 (209 bp) and cob–cox2 (96 bp).

E.Parket

al./Molecular

Phylogeneticsand

Evolution62

(2012)329–

345333


(10,617 bp). We used Mrmodeltest 2.3 to find the best nucleotidesubstitution model for Bayesian tree reconstruction (Nylander,2004). MrBayes analysis of the nucleotide sequences was per-formed with the following settings: partition; codon, rates, gam-ma; Nst, 6; number of generations, 5 million; number of parallelchains, 4; samplefreq, 5000; burn-in, 250. We used the last 750sampled trees to estimate Bayesian Posterior Probability (BPP). Fi-nally, the convergence of the Bayesian trees was checked usingindependent runs of MrBayes3.1.1 (Ronquist and Huelsenbeck,2003). After obtaining the Bayesian tree, the tree topology wasindependently compared to alternative tree topologies, includingthat of the once debated monophyly of the Scleractinia and theparaphyly of the Anthozoa to the Medusozoa using the Shimoda-ira–Hasegawa test (SH test) based on the maximum likelihoodmethod (Shimodaira and Hasegawa, 1999). The statistical confi-dence of our Bayesian tree compared to alternative topologieswas evaluated through a large number of bootstrap resamplings(n = 1000) with a one-tailed test using the RELL bootstrap methodimplemented in PAUP 4.0b10 (Kishino et al., 1990; Swofford,2002).

Fig. 2. A Bayesian tree of 48 cnidarian taxa based on the nucleotide sequences of 13 proright-hand margin delineate major taxonomic groups. Where the Bayesian Posterior Propresented on the corresponding node. Branches that were too short to be resolved inSupplementary Fig. S1. The paraphyly of the Anthozoa was strongly supported in the treeAll cnidarian artwork is courtesy of the Korean Coral Resource Bank, Ewha Womans Un

2.6. Estimation of divergence times and evolutionary rates

We estimated the divergence times and evolutionary rates ofcnidarian groups by applying two kinds of Bayesian methodsimplemented in Multidivtime (Thorne et al., 1998) and CodonRates(Seo and Kishino, 2008; Seo et al., 2004). In our analysis, weadopted more sophisticated approaches than a conventionalmolecular clock because rate heterogeneity between Anthozoaand Medusozoa has been noted (Dawson and Jacobs, 2001; Gov-indarajan et al., 2005). Furthermore, Fig. 2 shows the variabilityin the sum of branch lengths from the MRCA to each terminal node,which is a sign of rate heterogeneity among cnidarian lineages.

For the application of CodonRates and Multidivtime, we usedthe fixed tree topology shown in Fig. 2 and concatenated sequencesof 13 mitochondrial protein-coding genes. The prior ingroup roottime was assumed to follow a gamma distribution with a meanof 640 million years ago (Ma) and a standard deviation of 60 mil-lion years (MY). The mean and standard deviation of the gammadistribution were derived from the results of previous studies(see Table 1 of Rogers, 2009).

tein-coding genes (a total of 10,617 bp) of mitochondrial DNA. Vertical lines at thebability (BPP) equals 1.00, it is not shown, but where the BPP is less than 1.00, it isthe figure are depicted with black triangles, and detailed topologies are shown in. The Scleractinia were established as a monophyletic group in this tree (BPP = 1.00).iversity.

Table 3Fossil constraints.

Node number MRCA of clades Upper limit (Ma) Lower limit (Ma) References

#53 Pavona and Agaricia 33.1 32.9 Medina et al. (2006) and Veron (1995)#63 Acropora and Agaricia 55.1 54.9 Medina et al. (2006) and Veron (1995)#72 Astrangia and Montastraea 70.1 69.9 Medina et al. (2006) and Veron (1995)#52 Hexacorallia N/A 540 Han et al. (2010)#61 Medusozoa N/A 570 Chen et al. (2002)


Before estimating divergence times, we estimated branchlengths with codon and nucleotide models. By adopting Yang’s co-don model (Yang, 1998), we estimated the nonsynonymous andsynonymous branch lengths separately with CodonRates. Addi-tionally, by adopting the F84 model (Felsenstein, 1984) with fivecategories of gamma rate heterogeneity (Yang, 1994), we esti-mated the branch lengths of nucleotide substitutions with PAML(Yang, 2007). ML estimates from the F84 model were used as an in-put for Multidivtime.

At the ingroup root node, the prior mean and standard deviationof the evolutionary rate were set to 0.00156 and 0.00225 substitu-tions, respectively, per nucleotide per MY for Multidivtime. Weseparately estimated nonsynonymous and synonymous rates withCodonRates. This enabled us to investigate the change of selectivepressure, which provided an advantage over the estimation ofnucleotide substitution rates. For CodonRates, the median of theprior synonymous (nonsynonymous) rate at the ingroup root wasset to 0.00401 (0.000671) substitutions per codon per MY. Thestandard deviation of the prior synonymous (nonsynonymous) rateat the ingroup root was determined to be 0.00579 (0.000968) fromthe assumption of exponential distributions. These parameters forprior root rates were derived from the information of branchlengths and the mean of the prior root time. The prior mean andstandard deviation for the rate change parameter were 0.4 and0.4 for Multidivtime and 0.001 and 0.001 for CodonRates,respectively.

In our analysis, fossil information was available for five nodes(Table 3). Strict fossil calibration points were known for threenodes (nodes #53, #63, and #72 in Fig. 3) (Veron et al., 1996; Ver-on, 1995), which were incorporated into our analysis as 2-sidedfossil constraints with a very narrow interval of 0.2 MY. The othertwo fossil constraints (nodes #52 and #61) were incorporated asthe lower bounds of divergence times. The posterior distributionsof divergence times and evolutionary rates were estimated witha Markov chain Monte Carlo (MCMC) procedure in which the num-ber of burn-in generations, samples and intervals were 1,000,000,1000, and 1000, respectively, for both Multidivtime and Codon-Rates. We repeated the programs twice and confirmed the conver-gence of MCMC procedure.

3. Results

3.1. Octocoral mitogenomes

The seven new mitogenomes of octocorals sequenced in thisstudy showed a size range from 18,647 (E. crassa) to 19,445 bp(E. complexa) in length (Table 1) and included the same number(14) of protein-coding genes (atp6 and 8, cox1–3, cob, nad1–6 and4L and msh1) as those of the other nine octocoral species knownbefore the present study. As shown in Fig. 1, four types of genearrangements have been observed in Octocorallia and one type inMedusozoa to date. Among the 16 octocoral mitogenomes, 12(including all of our samples) showed type A gene arrangements(Fig. 1a), with 10 genes in the heavy strand (cox1, nad1, cob,nad6, nad4 nad4L, msh1, nad2, nad5 and nd4) and four in the light

strand (cox3, apt6, atp8 and cox2). When we quantitatively com-pared each gene and intergenic region (IGR) across the anthozoanmitogenomes, misannotated gene lengths were found for someoctocoral species (Table 2). Therefore, we corrected the misanno-tated gene lengths, as explained in Section 2.3, and then comparedthem again. As a result, we found that all of the corresponding pro-tein-coding genes across the octocorals exhibited similar lengthswith slight variations. Notably, some proteins (nad4, cox3, atp6,atp8 cox2 and cox1) showed the same lengths across species.

The seven new mitogenomes presented only one tRNA formethionine (trnM), and their sequences showed the same lengthof 71 bp, which holds for all of the other octocorals (Table 2). Allof the octocorals exhibited both the small and large subunits ofribosomal RNAs. The rns (small) sequence ranged from 1019 to1043 bp in most cases (Table 2). However, the rns sequence ofBriareum asbestinum was significantly shorter (717 bp) than thatof the other species, even though its boundary was adjusted fromthat given in the original report (581 bp) (Medina et al., 2006). IGRsaccounted for the greatest portion of the difference in genome size,with the two Plexauridae species (Calicogorgia granulosa and E.complexa) as the most extreme cases (Table 2). Nevertheless, eachconfamilial species showed similar lengths of IGRs (Nephtheidaeand Isididae), and in particular, the five congeneric species of Den-dronephthya presented identical IGR lengths at 12 of the 17 IGRs,indicating low genetic divergence among these species. All of thespecies of Dendronephthya retained approximately 18 units of the(GT)n microsatellite repeat between nad4 and trnM. In contrast,the number of this microsatellite repeat was variable betweenother species.

3.2. Scyphozoan mitogenomes

Two scyphozoan jellyfish species were found to present mitog-enomes that are arranged as a single linear DNA molecule (Fig. 1b).The total lengths of the A. aurita and C. quinquecirrha mitogenomeswere 16,971 and 16,775 bp, respectively (Table 4). The gene orderswere the same as that found in the previously known scyphozoanmitogenome of A. aurita, which was the only known sequence priorto this study (Shao et al., 2006) (Fig. 1b). The mitogenomes of A.aurita and C. quinquecirrha included 13 protein-coding genes(atp6 and 8, cox1–3, cob, and nad1–6 and 4L), two ribosomal RNAs(rns and rnl), and two tRNAs (tRNA-Sec and tRNA-Met) (Table 4).Both mitogenomes exhibited two unknown ORFs at the one endof the linear molecule, as reported for A. aurita (1) (Fig. 1b) (Shaoet al., 2006). A. aurita (2) presented unknown orfs with lengths of969 and 327 bp, while in C. quinquecirrha, these orfs were 543 and363 bp in length, respectively.

3.3. Molecular phylogeny and divergence times of cnidarians

3.3.1. PhylogenyThe Bayesian tree based on the nucleotide sequences of 13

mitochondrial protein-coding genes resolved most of the tradi-tional taxonomic groups of Cnidaria well. The nodes of ingroupCnidaria were highly supported by BPP value of 1.0, except fortwo nodes within the octocorallian clade (Fig. 2). The tree indicated

#93

#66

#53

#72

050100150200250300350400450500550600650700750

Rob

ust c

lade

Cor

allim

orph

aria

Com

plex

cla

de

800850

Seriatopora hystrixSeriatopora caliendrumStylophora pistillataPocillopora eydouxiPocillopora damicornisMadracis mirabilisAstrangia sp. Mussa angulosaColpophyllia natansMontastraea franksiMontastraea faveolataMontastraea annularisPorites poritesSiderastrea radiansPavona clavusAgaricia humilisAcropora tenuisAnacropora matthaiMontipora cactusRicordea floridaRhodactis sp.Discosoma sp. (1)Discosoma sp. (2)Chrysopathes formosaNematostella sp.Metridium senileSavalia savagliaAurelia aurita (2)Aurelia aurita (1)Chrysaora quinquecirrhaHydra oligactisHydra magnipapillataBriareum asbestinumKeratoisidinae sp.Acanella eburneaCorallium konojoiParacorallium japonicumSarcophyton glaucumPseudopterogorgia bipinnataEuplexaura crassaCalicogorgia granulosaEchinogorgia complexaScleronephthya gracillimumDendronephthya putteriDendronephthya suensoniDendronephthya gigianteaDendronephthya castaneaDendronephthya mollis

Hexacorallia

33 Ma

70 Ma

570 Ma

55 Ma

540 Ma

#73

#91

#61

#79

#67

#77

#63

#92

#93

570

540

#

#92

Ma

Ma

#61

#779

#91

#

1

#67

#77

7

#66

#7270 Ma

55 Ma

#73

#63

33

a

#533 Ma

*0.74

*0.96

#76

Octocorallia

Medusozoa

#52

#89

#69

#90

Paleozoic ciozoneCciozoseMciozoretorpoeNirTyrC PerCarDevSilOrdCamEdi Neo

PalCreJur

Fig. 3. Bayesian estimates of divergence times for cnidarians. Divergence times are depicted within the evolutionary relationship among the sampled taxa that wasreconstructed by Bayesian phylogenetic analysis (Fig. 2) in such a way that each node and branch corresponds to each specific past time point and duration on a horizontallyscaled geological time scale in units of millions of years ago (Ma) (bottom axis). The tree topology is the same as the Bayesian tree (Fig. 2), except that outgroups(demosponges) are excluded, and branch lengths are rescaled to the absolute geological time. The Bayesian posterior probabilities of the Bayesian tree are also presented herewith an asterisk (�) only if the values are less than 1.00. In other cases, we did not show these values for the sake of simplicity. The green horizontal lines placed on each noderepresent the 95% credibility region of divergence time. Five fossil calibration points used in this analysis are indicated by red circles with their corresponding ages. For thesake of brevity, we marked only some nodes with their identification numbers (#) as designated for the CodonRates and Multidivtime analyses. Terminal branches withdifferent colors represent different taxonomic groups of Octocorallia at the suborder level: purple for Scleraxonia, orange for Calcaxonia, light blue for Alcyoniina, and red forHolaxonia. Geological time scale abbreviations: Cry (Cryogenian), Edi (Ediacaran), Cam (Cambrian), Ord (Ordovician), Sil (Silurian), Dev (Devonian), Car (Carboniferous), Per(Permian), Tri (Triassic), Jur (Jurassic), Cre (Cretaceous), Pal (Paleogene), and Neo (Neogene).


that the subclass Octocorallia is closer to Medusozoa than to thesister anthozoan subclass Hexacorallia, which differs from the tra-ditional taxonomy (Fig. 2). Furthermore, this unexpected phyloge-netic relationship was highly supported (BPP = 1.00). It wasevident that all octocorals are highly diverged from the Hexacoral-lia, with a long branch being observed between these groups.Within the Hexacorallia, skeletal corals belonging to the Scleracti-

nia were supported as a natural group showing a sister relationshipwith the non-skeletal corals, the Corallimorpharia.

The phylogenetic relationships found for octocorals in the pres-ent study are identical to those reported previously based on anal-ysis of just two mitochondrial genes (McFadden et al., 2006).Therefore, it should be pointed out that the results reported heresupport those published previously and do not offer any novel in-

Table 4The lengths and IGRs of genes of the scyphozoan jellyfishes, A. aurita (2) and C.quinquecirrha.

Gene A. aurita (2) C. quinquecirrha

pcga IGR pcga IGR

371 182orf1 969 19 543 507orf2 327 0 363 0rnl 1818 0 1794 0cox1 1581 48 1572 72cox2 771 5 747 7trn-Sec 70 5 70 5atp8 204 �1 201 �1atp6 705 �1 708 �1cox3 786 11 786 4trn-Met 71 1 71 5nad2 1320 0 1317 0nad5 1818 0 1818 0rns 957 0 987 0nad6 573 �47 549 �17nad3 360 �10 360 �10nad4L 303 �1 303 �1nad1 972 0 972 �1nad4 1443 3 1443 4cob 1140 380 1152 264

Total 16,971 16,775

a Abbreviation of protein-coding gene.


sights into the global phylogeny of the Octocorallia, except thatlonger DNA sequences of multiple mitochondrial genes could leadto enhanced phylogenetic resolution among very closely relatedcongeneric species. Additionally, it is of note that congeneric spe-cies of Montastraea of the Scleractinia and Dendronephthya of theOctocorallia showed very short branch lengths (Fig. 2). Neverthe-less, the order of branches within each group was reliably recon-structed with high BPPs, indicating an increased number ofinformative sites for multiple genes (Supplementary Fig. S1). Thecongruent phylogenetic pattern of the Octocorallia shown in boththe present tree and the previous one is similar in the following re-spects. The subordinal Alcyoniina group was not supported as amonophyletic group. Instead, its branches are mixed with otherHolaxonia groups, constituting a Alcyoniina–Holaxonia clade(McFadden et al., 2006). The node between the scleraxonian B.asbestinum and the other octocorals in the tree was found to occu-py the most basal position of the octocorallian clade. However, thethree scleraxonian species did not constitute a monophyletic cladeas a group.

The new mitogenome sequence of the moon jellyfish A. auritacollected from the Yellow Sea near Korea showed high divergencefrom the other previously known A. aurita, which is greater thanthat of the general species level shown by the other cnidarians(Fig. 2). This result suggests that the circumglobal inhabitant A.aurita consists of at least two cryptic species, which is consistent

Table 5Shimodaira–Hasegawa test.

Tree no. Tree topology

1 Demosponge ((Scleractiniaa, b, Corallimorpharia2 Demosponge (((Scleractiniaa Corallimorpharia)3 Demosponge (((Scleractiniaa, b, Corallimorphar4 Demosponge (((Scleractiniaa, b, Corallimorphar

To investigate the reliability of the phylogenetic relationships among the majogenes addressed in the present study, Shimodaira–Hasegawa test was conductewas on the monophyly of the Scleractinia and the paraphyly of the Anthozoa i* Significant p-values are shown in trees 3 and 4 (0.001 and 0.009, respectivelytree and are not reliable. Notably, tree 2 presents an insignificant p-value (0.22

a Scleractinia indicates the complex clade.b Scleractinia indicates the robust clade.

with previous studies (Dawson and Jacobs, 2001; Schroth et al.,2002).

To evaluate the resultant phylogenetic relationship among themajor groups, we compared alternative tree topologies using theShimodaira–Hasegawa test (SH test, Table 5). The SH test rejectedmonophyly of the Anthozoa with significant p-values (Table 5).However, the problematic paraphyly of the Scleractinia previouslyreported by Medina et al. (2006) was not rejected (p = 0.223), indi-cating that there are not sufficient synapomorphic nucleotide sitesin the 13 protein-coding genes for Corallimorpharia or Scleractinia.Note the short branch length from the MRCA of the Scleractinia andCorallimorpharia to the MRCA of robust and complex clades(Fig. 2).

3.3.2. Bayesian estimation of divergence times and evolutionary ratesThe divergence time of each tree node was estimated with

CodonRates and Multidivtime, and these values are presented asthe posterior means accompanying 95% credible regions (CRs) (Ta-ble 6). In addition, we obtained and compared independent molec-ular estimates for the divergence times of some cnidarians fromprevious studies, when the estimates corresponded to any of thenodes in the present tree shown in Fig. 2 (Table 6). The CodonRatesresults are also represented with the posterior median and 95% CRat each node of the Bayesian tree (Fig. 3). However, for comparisonwith the Multidivtime results, the posterior means given byCodonRates are presented in Table 6. Our results indicated thatthe origin of cnidarians (node #93) and the earlier MRCAs of theHexacorallia (#52) and Medusozoa (#61), and between theMedusozoa and Octocorallia (#92) might have occurred duringthe Neoproterozoic era (95% CRs: 544–819), prior to the Cambrian(543 Ma). None of the 95% CRs of the estimates overlapped withthe beginning of the major diversification of life that took placeduring the Cambrian Explosion, approximately 543 Ma. Accordingto the cnidarian samples addressed in the present study, the mod-ern anthozoans of the Octocorallia (#91) and Scleractinia (#67) areindicated to have descended from MRCAs in the late Paleozoic andearly Mesozoic, respectively. These MRCAs appear to either havesurvived or descended from the survivors of the Permian–Triassic(P/T) mass extinction event that occurred 251 Ma. Notably, theOctocorallia exhibited a long internal branch from the Precambrianancestor (696 Ma, CR: 648–767) to the MRCA of the modern crowngroup (292 Ma, CR: 173–421), leaving an approximately 400 MYgap between nodes (Fig. 3). The 95% CRs of the Scleractinia pre-sented fairly narrow intervals. For example, the MRCA of the Scler-actinia (#67) was estimated to have appeared approximately243 Ma (95% CR: 203–286), and this date very closely coincidedwith the fossil date of approximately 234–241 Ma. However, thedivergence time of each node of the Octocorallia showed broaderrange of CRs compared to the Hexacorallia, except for the internalnodes of the genus Dendronephthya (Fig. 3). The MRCA of all of theoctocoral taxa (#91: 292 Ma, CR: 173–421) was located in the

p-Value

), (Medusozoa, Octocorallia)), Scleractiniab), (Medusozoa, Octocorallia)) 0.223ia), Medusozoa), Octocorallia) 0.001*

ia), Octocorallia), Medusozoa) 0.009*

r cnidarian taxonomic groups based on the mitochondrial protein-codingd with some major alternative tree topologies. Our focus in these analysesn the Bayesian tree (Fig. 2).), which indicate that these trees are considerably different from the best3). Tree 2 showed the same results as found by Medina et al. (2006).

Table 6Divergence times of major bifurcation events during cnidarian evolution. For each node, the posterior mean and 95% credible region (CR), which were estimated by CodonRatesand Multidivtime, are shown in the third and fourth columns, respectively. Estimates from other references are presented in the fifth column. The unit of time is millions of years(MY).

Node number MRCA CodonRates (95% CR) Multidivtime (95% CR) Other studies (range of estimation)

#93 Ingroup of cnidarians 741 (686, 819) 842 (796, 895) 875 (700, 1200)a

711.7 (389, 1035)b

595 (561, 626)c

#61 Medusozoa 607 (571, 670) 753 (684, 814) 790 (640, 940)a

#92 Medusozoa and Octocorallia 696 (648, 767) 795 (743, 853) –#91 Octocorallia 292 (173, 421) 499 (320, 647) –#52 Hexacorallia 601 (544, 684) 549 (540, 574) 517.5b

#77 Scleractinia and Corallimorpharia 276 (235, 323) 250 (200, 301) 264 (240, 280)d

#67 Robust and complex clades 243 (203, 286) 204 (158, 259) >300e

#76 Corallimorpharia 206 (158, 253) 187 (130, 246) 121 (110–132)d,f

#73 Robust clade 105 (96, 113) 96 (87, 106) –#66 Complex clade 90 (80, 103) 82 (74, 93) –#90 Octocorallia without B. asbestinum 241 (140, 356) 445 (284, 599) –#89 Alcyoniina–Holaxonia 147 (77, 242) 350 (210, 494) –#69 Scleraxonia and Calcaxonia 199 (114, 309) 410 (253, 560) –

a Waggoner and Collins (2004): 18S rRNA gene, 74 taxa (Anthozoa, Scyphozoa, Cubozoa, and Hydrozoa), QDate method.b Cartwright and Collins (2007): 18S and 28S rRNA genes, 159 metazoan taxa (Porifera, Ctenophora, Anthozoa, Staurozoa, Scyphozoa, Cubozoa, Hydrozoa, Placozoa,

Acoelomorpha, Protostomia, and Deuterostomia), penalized likelihood method (r8s).c Peterson et al. (2004): 7 nuclear protein-coding genes, 23 metazoan taxa (only two cnidarian species: one anthozoan and one hydrozoan), penalized likelihood method

(r8s).d Medina et al. (2006): 13 mitochondrial protein-coding genes, 20 taxa (Octocorallia and Hexacorallia), penalized likelihood method (r8s).e Romano and Palumbi (1996): mitochondrial 16S rRNA gene, 34 taxa (Scleractinia), average genetic distance (29.4%).f The tree node is different from our node, #76, in that it is the MRCA of the present four corallimorpharian taxa and another corallimorpharian species, Corynactis sp.,

which is basal to the former ingroup. Because there are no mitochondrial protein-coding genes available from the sample of Corynactis sp. used in Medina et al. (2006), weexcluded it in our study.


Permian period, and the second most basal node (#90: 241 Ma, CR:140–356) fell within the upper Triassic period.

The separately estimated amounts of nonsynonymous and syn-onymous substitutions calculated by CodonRates are shown as thebranch lengths in Fig. 4a and b, respectively. The branch lengthswere measured in terms of the number of substitutions per codonin both nonsynonymous and synonymous cases. By combining thelikelihoods of two phylogenies (Fig. 4a and b) and fossil informa-tion from five nodes, we obtained the posterior distributions ofthe nonsynonymous and synonymous rates at each node, for whichthe medians are shown in Fig. 4c and d, respectively. Furthermore,using the posterior distributions of the nonsynonymous and syn-onymous rates, we obtained the posterior distribution of omega,which represents the strength of selective pressure applied to theprotein-coding genes (Fig. 4e). The size of the gray circles is pro-portional to the logarithm of the rates or omega. The amount ofnucleotide substitution estimated by PAML is shown in Fig. 4f.While branch lengths were measured on a ‘per codon basis’ inFig. 4a and b, they were measured on a ‘per nucleotide basis’ inFig. 4f.

For prior root time, we considered a gamma distribution with amean and standard deviation of 640 Ma and 60 MY, respectively.While the 95% confidence interval (CI) of this distribution wasroughly from 520 to 760 Ma, the 95% CRs of the posterior roottimes given by CodonRates and Multidivtime were from 686 to819 Ma and from 796 to 895 Ma, respectively. Both CRs were olderthan the CI of the prior root time, and the shift of the CRs towardthe past was due to two lower-bounded fossil constraints at nodes#52 and #61 (Fig. 3; Table 3). The CR at the root node indicated byMultidivtime was older than that obtained using CodonRates.Additionally, we observed relatively large differences betweenthe two CRs estimated with CodonRates and Multidivtime at nodes#61, #69, #89, #90, #91, and #92 (Table 6). It is worth noting thatthe 95% CRs given by CodonRates and Multidivtime did not overlapat node #61.

Fig. 4f shows the branch lengths of the phylogeny that wereestimated using the F84 model and gamma rate heterogeneity.The pattern of branch lengths shown in Fig. 4f was similar to that

associated with nonsynonymous evolution as presented in Fig. 4a.That is, the Hydra group was the most rapidly evolving, and theevolution of the Octocorallia was slower than that of a robust cladeand faster than that of a complex clade. Additionally, the Scypho-zoa were observed to have evolved as rapidly as the robust clade.The pattern of synonymous branch lengths (Fig. 4b) was slightlydifferent than that of the nonsynonymous branch lengths(Fig. 4a), in which the evolution of the Hydra was as fast as thatof the robust clade, and the Scyphozoa evolved more slowly thanthe robust clade. In Fig. 4a and b, the nonsynonymous and synon-ymous branch lengths leading to the robust clade within the Scler-actinia were relatively long. This led to the high rate estimates atthe position of the MRCA of the robust clade (Fig. 4c and d). Be-cause both the nonsynonymous and synonymous rates were high,the omega estimate at the MRCA of the robust clade was not ashigh and was similar to those of other groups (Fig. 4e). Withinthe medusozoan group, the nonsynonymous branch lengths ofthe Hydra clade were longer than those of the robust clade, andthe nonsynonymous branch lengths of the Scyphozoa were as longas those of the robust clade (Fig. 4a). However, the synonymousbranch lengths of the Hydra clade were as long as those of the ro-bust clade, whereas the synonymous branch lengths of the Scypho-zoa were shorter than those of the robust clade (Fig. 4b). This led tothe increased omega in the ancestral lineages of the Scyphozoaclade (Fig. 4e). Using the posterior medians of the rates at eachnode shown in Fig. 4c and d, we investigated the rate changes overtime. Fig. 5a and b shows how nonsynonymous and synonymousrates have changed during cnidarian evolution. The distributionsof the evolutionary rates estimated from different classes werewidely distributed and they largely overlapped.

4. Discussion

4.1. Comparison of cnidarian mitogenomes

Among all of the cnidarian mitogenomes sequenced to date, thelargest genome size has been found for Rhodactis sp. (21,376 bp,

(a) (b)

(d)(c)

Fig. 4. Branch length and rate estimates. Nonsynonymous (a) and synonymous (b) branch lengths were estimated with a codon model and measured in units ofnonsynonymous and synonymous substitutions per codon. The posterior medians of nonsynonymous (c) and synonymous (d) rates and omega (e) are represented on eachnode. The size of circles is proportional to the logarithm of the rate (c, d) or to the logarithm of omega (e). Branch lengths were estimated with the F84 model (f) and measuredin units of nucleotide substitutions per nucleotide site.


(e) (f)

Fig. 4 (continued)


Corallimorpharia), while the smallest is that of Montastraea franksi(16,137 bp, Scleractinia) (Table 1). These two species belong to thesubclass Hexacorallia, which includes very diverse taxonomicgroups, from stony corals (Scleractinia), black corals (Antipatharia),and sea anemones (Actiniaria) to Zoanthidea (Zoanthids). Theoctocorals likewise include very diverse groups of marine species,ranging from blue corals and organ pipe corals to diverse sea pens,soft corals, and stoloniferous corals as well as a variety of sea fanswith varied colony morphologies and even free-living solitary pol-yps (genus Taiaroa). Therefore, there is a greater potential for var-iability in the mitogenomes of octocorals than the currently knownrange shown in Table 1. The relatively large mitogenome size of thesubclass Octocorallia compared to other anthozoans mainly resultsfrom the additional msh1 gene present (�3 kb), which is a homologof bacterial MutS and is regarded as a synapomorphic character ofthis taxonomic group (Beagley et al., 1995; Pont-Kingdon et al.,1998). The puzzling origin of this gene in the mitochondrial gen-ome of the Octocorallia has recently been attributed to a horizontalgene transfer from a non-eukaryotic source (Bilewitch and Degnan,2011). Variation in the length of IGRs is another important factorinvolved in the size differences of the octocorallian mitogenomes.None of the homologous regions was found to have exactly thesame IGR length across all of the octocorals. As it is known thatthe two species from Corallidae exhibit a different organizationthan observed for the other octocoral mitogenomes (Uda et al.,2011), four different types of gene arrangements have been re-ported in the Octocorallia to date (Fig. 1a). In the family Isididae,two species (Keratoisidinae sp. and Acanella eburnea) exhibit aninversion of five genes from nad4 to msh1 (mitogenome type B,Fig. 1a) (Brugler and France, 2008; Van der Ham et al., 2009).

Although octocoral mitogenomes have generally been found tobe devoid of repetitive sequences, direct variant repeats (DVRs)were recently discovered in the mitogenome of C. granulosa (Parket al., 2011), and the Japanese octocorals Paracorallium japonicumand Corallium konojoi have been found to exhibit inverted repeatsof 44- and 47-bp short sequences (Uda et al., 2011).

Scyphozoans have two additional ORFs, including a putativeDNA-dependent polymerase, orf969 (Shao et al., 2006). A smallnumber of tRNAs is a common characteristic of cnidarian mitoge-nomes. The Cnidaria have been found to present only one to threetRNAs. Two species of Pocilloporidae (Seriatopora caliendrum andSeriatopora hystrix) exhibit a duplication of trnW, resulting in onetrnM and two trnWs. Zoanthideans have only trnM, identical tooctocorals (Tables 1 and 2). The Octocorallia are known to haveonly one tRNA for methionine, and thus, other tRNAs are assumedto be imported from the cytoplasm (Beagley et al., 1995). Noexception to this pattern was detected in the present study. Medu-sozoans exhibit two to three tRNAs (Table 4).

The members of the Medusozoa are extremely different fromthe Anthozoa in several respects. Medusozoa, including the Scy-phozoa, Cubozoa, and Hydrozoa, present a linear structure in theirmitochondrial genome, which has been regarded as a unique char-acteristic of this group (Bridge et al., 1992). Furthermore, themitogenome of Hydra magnipapillata exists as two separate chro-mosomes (Voigt et al., 2008). The medusozoan mitogenomes showa tendency to be compact (Kayal and Lavrov, 2008; Shao et al.,2006; Voigt et al., 2008), with average sizes of 16 kb being ob-served, as shown in Table 4. Our mitogenome data from an individ-ual of A. aurita collected in Korean seawater revealed the existenceof cryptic species; the two specimens of A. aurita collected in Kand-

(a)

(b) Log rate

Time (Ma)

Robust cladeComplex cladeCorallimorphariaMedusozoaOctocoralliaOthers

0 100 200 300 400 500 600 700 800

0

-2

-4

-6

-8

-10

-12

-14

Log rate0

-2

-4

-6

-8

-10

-12

-14

Time (Ma)0 100 200 300 400 500 600 700 800

Fig. 5. Trend of nonsynonymous and synonymous rate changes. The posteriormedians of nonsynonymous (a) and synonymous (b) rates from Fig. 4c and d areplotted along the axis of time.


alaksha Bay of the White Sea (Table 1, A. aurita (1)) and the YellowSea of Korea (Table 1, A. aurita (2)) showed a significant differencein their DNA sequences, which reflects their distant geographiclocations. A previous phylogeographic study of A. aurita in worldoceans indicated that seven sibling species of the named speciesexist, and their genealogies corresponded highly to their geo-graphic regions (Dawson and Jacobs, 2001).

4.2. Phylogeny of cnidarians

From a molecular phylogenetic perspective, the phylum Cni-daria is subdivided into two monophyletic clades, which has beensupported by 16S and 18S ribosomal RNA data, mitogenome struc-tures, morphological features and differences in life cycle charac-teristics (Bridge et al., 1992, 1995; Collins, 2002; Kim et al.,1999). These previous studies showed that the class Anthozoa,including the subclasses Hexacorallia and Octocorallia, constitutesa monophyletic clade, and the remaining three classes (Hydrozoa,Scyphozoa, and Cubozoa) are clustered into the medusozoan clade(Daly et al., 2007). However, our Bayesian tree of cnidarians basedon the nucleotide sequences of 13 protein-coding genes repro-duced a paraphyletic relationship of the Anthozoa with theMedusozoa, showing a sister relationship between the Octocoralliaand Medusozoa with strong support. This paraphyly has also beenindicated in previous studies (Lavrov et al., 2008; Shao et al., 2006)

that employed mitogenomic data. Considering the early divergencetimes of cnidarians in the metazoan evolution and the relativelyshort intervals between bifurcating events associated with themain taxonomic groups of Cnidaria at the either subclass or classlevel, it is likely that there may be a substantial degree of variancebetween different molecular markers in their efficacy. Phylogeniesreconstructed using the nuclear 18S and 28S genes consistentlyshow that the branch lengths from the cnidarian root to the MRCAsof the Anthozoa and Medusozoa are substantially shorter com-pared to those from the MRCAs to the present-day taxa (Cartwrightand Collins, 2007; Collins, 2002; Collins et al., 2006). Thus, the dis-crepancy between trees with respect to the phylogenetic relation-ship between the Anthozoa and Medusozoa could be accounted forby such a potential variance between genes. Additionally, theremay be different sources of bias in molecular data: for example,long-branch attraction (Bergsten, 2005; Brinkmann et al., 2005;Philippe et al., 2005), heterogeneous nucleotide substitution ratesamong different lineages of cnidarians (Baurain et al., 2007),incomplete taxon sampling (Omland et al., 1999; Rannala et al.,1998; Rosenberg and Kumar, 2001; Zwickl and Hillis, 2002), com-positional heterogeneity (Foster, 2004; Foster and Hickey, 1999;Jermiin et al., 2004; Lockhart et al., 1994; Nesnidal et al., 2010),generation time effects or a significant change in the evolutionarymode in the history of the mitogenomes (Thomas et al., 2010), aswell as combinations of these factors. We paid particular attentionto potential errors due to compositional heterogeneity in mitoge-nomes among Cnidaria. Indeed, the G + C contents found for thetwo hydrozoans (average of 23.2%) were dramatically lower thanthose of all of the anthozoans (average of 36.0%), while those ofthe three scyphozoans (average of 32.7%) were moderately lower(Table 1). We speculate that the apparent compositional heteroge-neity of the Hydra might have resulted in the excessively longbranch associated with this group and have affected the systematicerror of the unrealistic paraphyly between the Anthozoa andMedusozoa (Fig. 2). This indicates the need for special cautionregarding the use of mitochondrial data for phylogenetic analyses,including in Hydrozoa and other higher taxonomic groups. Theo-retically and empirically, compositional heterogeneity might belessened by dropping idiosyncratic portions of DNA sequencesand/or a few taxa and choosing appropriate nucleotide substitutionmodels (Foster, 2004; Foster and Hickey, 1999; Jermiin et al., 2004;Lockhart et al., 1994; Nesnidal et al., 2010).

Although there is a problematic paraphyly in the mitochondrialtree, it is unlikely that the molecular dates at the deep nodes #93,#92, #61, and #52 would significantly diverge from the present95% CRs of them (Fig. 3). Because the intervals from the root node(#93) to the others (#92, #52, and 61) are equivalent to the CRsand overlapped within a certain geological time duration, a changeof branch order, such as monophyly of the Anthozoa from the pres-ent paraphyly, would not lead to a significant difference in thedates other than slight changes. Therefore, we adhered to theBayesian tree for Bayesian analyses of divergence times, carefullyinterpreting the resultant estimates under their CRs (Fig. 3).

4.3. Bayesian estimation of divergence times and evolutionary rates

We note that the relatively large differences between theCodonRates and Multidivtime time estimates at nodes #61, #69,#89, #90, #91, and #92 (Table 6) are due to insufficient fossil infor-mation and taxon sampling. As described in Sections 2.6 and 3two-sided fossil constraints (nodes #53, #63 and #72 in Fig. 3)are located in robust and complex clades. We consider that mostinformation for rate changes comes from the nodes neighboringthese two-sided constraints. The information for rate changes fromthe robust and complex clades is combined with two lower-bounded fossil constraints (nodes #52 and #61) for time


estimation of the rest of cnidarian clades. There is a large time gapbetween the three two-sided constraints and the two lower-bounded constraints, and appropriate fossil information is not cur-rently available during this time period. Furthermore, there hasbeen a relatively small number (five) of taxa within the Medusozoasince the time of their distant MRCA (>570 Ma), and there is no fos-sil information that can directly be employed for calibration of theOctocorallia. Under these circumstances, time estimates for thecnidarian clades to the exclusion of the robust and complex cladestend to be more dependent on estimated branch lengths, which arestrongly affected by the adopted sequence evolution models.

The pattern of branch lengths shown in Fig. 4f is not identical tothat in Fig. 2 because the former were estimated with ML methodsand the latter with Bayesian methods, and the adopted sequenceevolution models are different. CodonRates utilized synonymousas well as nonsynonymous information, which appears to producetime estimates at some nodes that are inconsistent with those ofMultidivtime. We note that more intensive taxon sampling wouldseparate long branches leading to the Hydra and Scyphozoa intofragments of short branches and improve the accuracy of estima-tion with additional fossil information.

Previous studies based on mitochondrial genes of anthozoansshowed a very low divergence among mitochondrial genes in boththe Hexacorallia and Octocorallia (France and Hoover, 2002;McFadden et al., 2006; Shearer et al., 2002). In addition, it has beennoted that anthozoan mitochondrial DNA has evolved more slowlythan that of the Medusozoa (Govindarajan et al., 2005). However,this notion is based on a limited number of cases, and furthermore,most evidence of slow substitution rates in anthozoans came fromthe complex clade of Scleractinia, which apparently shows shortbranch lengths in the Bayesian tree (Fig. 2). A few representativestudies of the complex clade showed that the substitution rateassociated with mitochondrial genes ranges from approximately0.00019 to 0.0018 substitutions per nucleotide site per MY, whichis based on the synonymous (0.00056) and nonsynonymous(0.00019) rates of the cytochrome oxidase subunit I gene (COI) oftwo stony coral species (Hellberg, 2006) and the range of 0.001–0.0018 found for the cytochrome b gene (cytb) of Acropora species(Van Oppen et al., 1999). Note that these studies have differenttime depths for the calibration points in their analyses. For com-parison with these values, if we convert the log rate of the complexclade (Fig. 4c and d) into a similar scale rate (substitutions per co-don per MY), the following ranges are estimated between maxi-mum and minimum nodes for the rates: nonsynonymous,0.00014–0.00205 and synonymous, 0.00039–0.00788. The largeranges indicate that there is measurable variation in the substitu-tion rate, even in the same clade, and careful interpretation andanalysis is needed for comparison of wider taxonomic groups. Nev-ertheless, it should be noted that the present maximum rate ofsynonymous site substitutions in the complex clade is approxi-mately twice as slow than as that (0.0195 substitutions per nucle-otide site per MY) of the third codon of the COI gene of hydrozoanspecies (Obelia geniculata) (Govindarajan et al., 2005). However,analysis of a sufficient number of taxa from the Medusozoa willbe necessary to achieve an accurate evaluation regarding this issue.

Our results regarding substitution rates of mitochondrial DNAsare based on multiple genes, large taxonomic samplings from sev-eral classes of Cnidaria, and explicit Bayesian models of molecularevolution that allow for more realistic rate changes over time. Oneof the many advantages of our Bayesian analysis is that we can di-rectly measure the evolutionary rates at each node, which allowsus to evaluate rate changes over time, as shown in Fig. 5a and b.Because we suffer from a paucity of taxon samples in some groups,it is still premature to infer any conclusive pattern based on com-parison of rate changes over time, although octocorallian mito-chondrial genes appear to evolve more slowly than those of the

other groups. Therefore, we need to be careful in discussing the ra-pid evolution of the Medusozoa and the slow evolution of Antho-zoa. Although pairwise distances among medusozoan species aregreater than those among Anthozoa (Fig. 4a, b, and f), the ratesof some anthozoan nodes are higher than those of some meduso-zoan nodes and vice versa (Fig. 5a and b). Therefore, it appears tobe undesirable to summarize and compare the evolutionary ratesof certain taxonomic groups with single values. We consider it tobe highly recommended to examine the rates at each node individ-ually and take into consideration individual information collec-tively for comparison of different groups’ evolutionary rates. TheBayesian estimation procedures adopted here are very useful toolsfor this purpose, and we believe that our analysis will provide aclue for further studies of cnidarian evolution.

4.4. Divergence times and fossil records of cnidarians

The present divergence time estimates provide new insightsinto the deep evolutionary history of the Cnidaria. The Precambrianorigin of the Cnidaria and their subsequently diverged lineages isconsistent with recently discovered fossil evidence for cnidarians,including fossils of sea anemones (Han et al., 2010), jellyfishes(Cartwright et al., 2007; Hagadorn and Waggoner, 2000), hydrasand many other cnidarians (Chen et al., 2002). Han et al. (2010) re-cently found microfossils from the southern province of China withmorphological characteristics that fall within the range of the mod-ern Hexacorallia, excluding the Ceriantharia, and these microfossilshave been dated to approximately 540 Ma. This means a group ofhexacorallians already existed in the beginning of the Cambrian.Based on this new finding, we hypothetically considered an ageof minimum 540 Ma as a constraint for the MRCA of the hexacor-allian taxa (#52). As a result, the age of this node was estimatedas 601 Ma (CR: 544–684), which largely overlaps with the Ediaca-ran Period (542–630 Ma). Similarly, we added another hypotheti-cal calibration point to the MRCA of the Medusozoa (#61)according to a reliable fossil of Ediacaran hydra dated to approxi-mately 570 Ma (Chen et al., 2002). Interestingly, Cartwright et al.(2007) reported the discovery of exceptionally preserved fossilsof jellyfishes from the middle Cambrian (approximately 505 Ma),which also circumstantially supported our constraint of the old cal-ibration age to that of node #61. If we take into account that soft-bodied cnidarians do not often leave convincing fossils, the oldages of the deep nodes are quite acceptable estimates. A few com-parable previous studies also proposed old ages for some majorcnidarian lineages (#93, #61, and #52) (Table 6); however, thedata and methods employed in these studies were different fromthose used in the present study. It is likely that the old ages ofthe basal nodes of cnidarians may account for the recently discov-ered fossils in the Neoproterozoic strata that have been suspectedto be cnidarians but are difficult to directly connect to the moderncnidarian taxa.

Such old ages of the major cnidarian lineages should thereforebe considered in the reevaluation of the Precambrian and Cam-brian fossil record, particularly in relation to fossils that are asyet unconfirmed or debated due to their ambiguous morphologicalcharacteristics compared to extant species. In the case of the Octo-corallia, early fossil records are very sparse (Fabricius and Alders-lade, 2001). Nevertheless, a few fossils have been proposed torepresent Paleozoic octocorals including one from the lower Ordo-vician (471–478 Ma) (Cope, 2005) and one from the lower Silurian(425–435 Ma) (Bengtson, 1981). The Ordovician fossil is the earli-est known gorgonian coral belonging to the order Alcyonacea, andthe Silurian fossil had been the earliest known representative ofthe order Alcyonacea before the study conducted by Cope (2005).These fossils match well with the long interval of the octocoralliantree branch extending back from the Permian to the Neoproterozo-


ic (Fig. 3). Furthermore, the existence of the long branch in thePaleozoic might help us to interpret the questionable octocoral fos-sil found in the Burgess Shale of the middle Cambrian (530 Ma)that has been debated regarding its affinity to either octocoralsor echinoderms (Ausich and Babcock, 1998, 2000).

In addition to the early evolution of the root of cnidarians, ourresults provide insight into the diversification of the modern anth-ozoans since the Permian–Triassic boundary (Fig. 3; Table 6). In thecase of the Hexacorallia, their rich fossil record documents suchdramatic events. For instance, the dominant Paleozoic hexacorals,such as Tabulata and Rugosa corals, went completely extinct dur-ing the Permian, and the modern Scleractinia suddenly appearedin the middle Triassic, approximately 234–241 Ma (Stanley,2003; Veron, 1995). Their sudden emergence may not be a surpriseif we take into account the Permian–Triassic mass extinction thatswept approximately 90–95% of marine species to extinction. Thismajor mass extinction may well have had an impact on the evolu-tion of the Hexacorallia (Veron, 1995). Our molecular estimates ofdivergence times for Scleractinia agree quite well with the fossilrecord (Veron et al., 1996; Veron, 1995). Although the three fossilcalibration points came from internal nodes of the Scleractinia,the estimates for the other Scleractinia nodes, even at nodes olderthan the fossil dates, matched the fossil record very accurately (Ta-ble 6). Medina et al. (2006) first applied the three fossil dates tomitogenomic data from hexacorallians, which were also used inthis study, to estimate divergence dates. They reached an estimateequivalent to that produced in the present study of between 240and 288 Ma for the origination of the Scleractinia. However, theirestimate of the MRCA of the Corallimorpharia differs from oursdue to the monophyly of the Scleractinia in our Bayesian tree(Fig. 2; Table 6).

Given the present taxon samples of the Octocorallia, which lacktwo orders (Pennatulacea and Helioporacea) of those considered inBayer’s three order system (Bayer, 1981; Daly et al., 2007; Fabri-cius and Alderslade, 2001), our inference regarding the evolution-ary pattern of the octocorals is confined to the order Alcyonaceain the strict sense. Although our taxon samples are biased towardthe order Alcyonacea, they may represent early bifurcating eventsin the evolution of the extant octocorals because our samples in-volve the three most divergent clades and one of the basal species,B. asbestinum, in McFadden et al.’s phylogeny (2006). However, itshould be noted that McFadden et al.’s phylogeny is derived fromthe same underlying data as used in the present study, i.e., twoprotein-coding mitochondrial genes, so this is not truly an inde-pendent comparison. Additionally, our octocorallian clades are de-fined by molecular data but not morphological groups; e.g., nopennatulaceans (sea pens) or helioporaceans (blue corals) haveyet been considered. Therefore, if more divergent morphospeciesare added to the current dataset, it is likely that the current posi-tion of the MRCA of modern octocorals could be shifted back some-what further, shortening the gap between the molecular date andthe earliest known fossil record. Following the Permian–Triassic(P/T) mass extinction event approximately 251 Ma, it becomes evi-dent that alcyonacean octocorals began to diversify, with the originof their bifurcating events being detected in the Cretaceous period(Fig. 3), although these events proportionally reflect the large tax-on sampling from the Alcyoniina–Holaxonia assemblage. The fre-quent branching pattern observed coincides with the fossilrecord during the Cretaceous (Schlagintweit and Gawlick, 2009).Because the fossil record of octocorals in the upper Cretaceous isrelatively well preserved, a great deal of detailed information onchanges in specific taxonomic groups can be retrieved. Accordingto the paleobiology of the octocorals, the number of alcyonaceangenera began to increase in the upper Cretaceous (approximately100 Ma), and this age has been pinpointed in the middle of theAlcyoniina–Holaxonia clade (Fig. 3).

In summary, our divergence time estimates highlight the factthat the extant scleractinians and octocorals have common fea-tures related to radiating evolution during the late Mesozoic peri-od, which might suggest a common historical constraint imposedon these two extremely divergent taxonomic groups. Our findingssurprisingly coincide with the rise and diversification of other mar-ine organisms in the Mesozoic era (Vermeij, 1977). Multiple fac-tors, such as continental breakup, expansion of the Tethys Seaand shallow habitats followed by fragmentation and isolation,changes in ocean temperature and chemistry, changes in predationpressure, the symbiotic associations between corals and their algalpartners, and the combined effects of these factors, might have dri-ven the evolution of marine metazoans over geological time(Beauvais, 1984; Erwin, 1993; Ezaki, 1998; Fischer, 1981; Flügeland Senowbari-Daryan, 2001; Isozaki, 1997; Railsback andAnderson, 1987; Stanley, 2003; Wignall and Twitchett, 1996).

5. Conclusions

The present estimation of divergence times for the phylum Cni-daria and its modern classes based on mitochondrial protein-cod-ing genes and fossil data provides new insights and a hypotheticalfoundation for future studies on the origin and earlier evolution ofcnidarians. Our application of Bayesian methods, which allow formore realistic rate changes over time and explicitly consider theheterogeneity of substitution rates of mitochondrial DNAs, anduse of five reliable fossil dates enabled us to estimate the deepdivergence times of the major cnidarian taxa with some degreeof confidence. Although the estimates of divergence time accom-pany variable credible regions (CRs) depending on the extent oftaxon sampling and the presence or absence of calibration pointsfor the fossil record in a certain taxonomic group, two pronouncedpatterns related to the long evolutionary history of cnidarianscould be inferred: the origin of the Cnidaria (736 Ma, CR: 686–819) and the subsequent divergence of the Hexacorallia (601 Ma,CR: 544–684), Octocorallia and Medusozoa (696 Ma, CR: 648–767) and of the Scyphozoa and Hydrozoa (607 Ma, CR: 571–670)might have occurred prior to the Cambrian (543 Ma). In contrast,the modern alcyonacean octocorals (292 Ma, CR: 173–421) andhard corals (Scleractinia of Hexacorallia, 243 Ma, CR: 203–286)may have originated from the MRCAs appearing far later, followingthe earlier divergence times of the cnidarian classes. The latter im-plies that only a few evolutionary lineages of Paleozoic coralsmight have survived the Permian–Triassic mass extinction andthen radiated, adapting to the dramatically changed Mesozoicocean environments. The fossil record of the octocorallian andhexacorallian corals supports this interpretation and matchesmolecular estimates of divergence times well. However, we wereunable to estimate the divergence times of internal nodes of themodern medusozoans and the other cnidarian taxa in this studydue to insufficient taxon samples.

Acknowledgments

We thank Robert C. Vrijenhoek, Steven H.D. Haddock, and twoanonymous reviewers for their constructive criticism, which hasimproved this paper. We also thank the members of the KoreanCoral Resource Bank at Ewha Womans University for generouslyproviding the specimens of octocoral species examined in the pres-ent study. The present work was supported by the grant for ‘‘TheGenetic Evaluation of Important Biological Resources’’ (No. 074-1800-1844-304) funded by the National Institute of Biological Re-sources (NIBR), Korean Government, to Y.-J. Won. T.-K. Seo wassupported by Grant-in-Aid for Scientific Research(C-23570108) ofJapan.


Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ympev.2011.10.008.

References

Ausich, W.I., Babcock, L.E., 1998. The phylogenetic position of Echmatocrinusbrachiatus, a probable octocoral from the Burgess Shale. Palaeontology 41, 193–202.

Ausich, W.I., Babcock, L.E., 2000. Echmatocrinus, a Burgess Shale animalreconsidered. Lethaia 33, 92–94.

Baurain, D., Brinkmann, H., Philippe, H., 2007. Lack of resolution in the animalphylogeny: closely spaced cladogeneses or undetected systematic errors? Mol.Biol. Evol. 24, 6–9.

Bayer, F.M., 1981. Key to the genera of Octocorallia exclusive of Pennatulacea(Coelenterata: Anthozoa), with diagnosis of new taxa. Proc. Biol. Soc. Wash. 94,902–947.

Beagley, C., Macfarlane, J., Pont-Kingdon, G., Okimoto, R., Okada, N., Wolstenholme,D., 1995. Mitochondrial genomes of Anthozoa (Cnidaria). Prog. Cell Res. 5, 149–153.

Beauvais, L., 1984. Evolution and diversification of Jurassic Scleractinia. Palaeontogr.Am. 54, 219–224.

Bengtson, S., 1981. Atractosella, a Silurian Alcyonacean octocoral. J. Paleontol. 55,281–294.

Benson, G., 1999. Tandem repeats finder: a program to analyze DNA sequences.Nucl. Acids Res. 27, 573–580.

Bergsten, J., 2005. A review of long branch attraction. Cladistics 21, 163–193.Bilewitch, J., Degnan, S., 2011. A unique horizontal gene transfer event has provided

the octocoral mitochondrial genome with an active mismatch repair gene thathas potential for an unusual self-contained function. BMC Evol. Biol. 11, 228.

Bridge, D., Cunningham, C., DeSalle, R., Buss, L., 1995. Class-level relationships in thephylum Cnidaria: molecular and morphological evidence. Mol. Biol. Evol. 12,679–689.

Bridge, D., Cunningham, C., Schierwater, B., DeSalle, R., Buss, L., 1992. Class-levelrelationships in the phylum Cnidaria: evidence from mitochondrial genomestructure. Proc. Natl. Acad. Sci. USA 89, 8750–8753.

Brinkmann, H., Van Der Giezen, M., Zhou, Y., De Raucourt, G., Philippe, H., 2005. Anempirical assessment of long-branch attraction artefacts in deep eukaryoticphylogenomics. Syst. Biol. 54, 743–757.

Brugler, M., France, S., 2007. The complete mitochondrial genome of the black coralChrysopathes formosa (Cnidaria: Anthozoa: Antipatharia) supports classificationof antipatharians within the subclass Hexacorallia. Mol. Phylogenet. Evol. 42,776–788.

Brugler, M., France, S., 2008. The mitochondrial genome of a deep-sea bamboo coral(Cnidaria, Anthozoa, Octocorallia, Isididae): genome structure and putativeorigins of replication are not conserved among octocorals. J. Mol. Evol. 67, 125–136.

Cartwright, P., Collins, A., 2007. Fossils and phylogenies: integrating multiple linesof evidence to investigate the origin of early major metazoan lineages. Integr.Comp. Biol. 47, 744–751.

Cartwright, P., Halgedahl, S.L., Hendricks, J.R., Jarrard, R.D., Marques, A.C., Collins,A.G., Lieberman, B.S., 2007. Exceptionally preserved jellyfishes from the MiddleCambrian. PLoS One 2, e1121.

Chen, J.Y., Oliveri, P., Gao, F., Dornbos, S.Q., Li, C.W., Bottjer, D.J., Davidson, E.H.,2002. Precambrian animal life: probable developmental and adult cnidarianforms from Southwest China. Dev. Biol. 248, 182–196.

Collins, A.G., 2002. Phylogeny of Medusozoa and the evolution of cnidarian lifecycles. J. Evol. Biol. 15, 418–432.

Collins, A.G., Schuchert, P., Marques, A.C., Jankowski, T., Medina, M., Schierwater, B.,2006. Medusozoan phylogeny and character evolution clarified by new largeand small subunit rDNA data and an assessment of the utility of phylogeneticmixture models. Syst. Biol. 55, 97–115.

Cope, J.C.W., 2005. Octocorallian and hydroid fossils from the Lower Ordovician ofWales. Palaeontology 48, 433–445.

Daly, M., Brugler, M.R., Cartwright, P., Collins, A.G., Dawson, M.N., Fautin, D.G.,France, S.C., Opresko, D., Rodriguez, E., Romano, S., 2007. The phylum Cnidaria: areview of phylogenetic patterns and diversity 300 years after Linnaeus. Zootaxa1668, 127–182.

Dawson, M.N., Jacobs, D.K., 2001. Molecular evidence for cryptic species of Aureliaaurita (Cnidaria, Scyphozoa). Biol. Bull. 200, 92–96.

Doolittle, R.F., Feng, D.F., Tsang, S., Cho, G., Little, E., 1996. Determining divergencetimes of the major kingdoms of living organisms with a protein clock. Science271, 470–477.

Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis bysampling trees. BMC Evol. Biol. 7, e214.

Dunn, C.W., Hejnol, A., Matus, D.Q., Pang, K., Browne, W.E., Smith, S.A., Seaver, E.,Rouse, G.W., Obst, M., Edgecombe, G.D., 2008. Broad phylogenomic samplingimproves resolution of the animal tree of life. Nature 452, 745–749.

Erpenbeck, D., Voigt, O., Adamski, M., Adamska, M., Hooper, J., Worheide, G.,Degnan, B., 2007. Mitochondrial diversity of early-branching metazoa isrevealed by the complete mt genome of a haplosclerid demosponge. Mol.Biol. Evol. 24, 19–22.

Erwin, D.H., 1993. The Great Paleozoic Crisis: Life and Death in the Permian.Columbia Univ. Press, New York.

Ezaki, Y., 1998. Paleozoic Scleractinia: progenitors or extinct experiments?Paleobiology 24, 227–234.

Fabricius, K., Alderslade, P., 2001. Soft Corals and Sea Fans: A Comprehensive Guideto the Tropical Shallow Water Genera of the Central-West Pacific, the IndianOcean and the Red Sea. Australian Institute of Marine Science, Townsville.

Fautin, D., 2009. Structural diversity, systematics, and evolution of cnidae. Toxicon54, 1054–1064.

Felsenstein, J., 1984. Distance methods for inferring phylogenies: a justification.Evolution 38, 16–24.

Fischer, A.G., 1981. Climatic oscillations in the biosphere. In: Nitecki, M.H. (Ed.),Biotic Crises in Geological and Evolutionary Time. Academic Press, New York,pp. 103–131.

Flügel, E., Senowbari-Daryan, B., 2001. Triassic reefs of the Tethys. In: Stanley, G.D.,Jr. (Ed.), The History and Sedimentology of Ancient Reef Systems. KluwerAcademic/Plenum Publishers, New York, pp. 217–249.

Foster, P.G., 2004. Modeling compositional heterogeneity. Syst. Biol. 53, 485–495.Foster, P.G., Hickey, D.A., 1999. Compositional bias may affect both DNA-based and

protein-based phylogenetic reconstructions. J. Mol. Evol. 48, 284–290.France, S., Hoover, L., 2002. DNA sequences of the mitochondrial COI gene have low

levels of divergence among deep-sea octocorals (Cnidaria: Anthozoa).Hydrobiologia 471, 149–155.

France, S.C., Rosel, P.E., Agenbroad, J.E., Mullineaux, L.S., Kocher, T.D., 1996. DNAsequence variation of mitochondrial large-subunit rRNA provides support for atwo-subclass organization of the Anthozoa (Cnidaria). Mol. Mar. Biol.Biotechnol. 5, 15–28.

Gissi, C., Iannelli, F., Pesole, G., 2008. Evolution of the mitochondrial genome ofMetazoa as exemplified by comparison of congeneric species. Heredity 101,301–320.

Govindarajan, A., Halanych, K., Cunningham, C., 2005. Mitochondrial evolution andphylogeography in the hydrozoan Obelia geniculata (Cnidaria). Mar. Biol. 146,213–222.

Hagadorn, J.W., Waggoner, B., 2000. Ediacaran fossils from the southwestern GreatBasin, United States. J. Paleontol. 74, 349–359.

Han, J., Kubota, S., Uchida, H., Stanley Jr., G.D., Yao, X., Shu, D., Li, Y., Yasui, K., 2010.Tiny sea anemone from the Lower Cambrian of China. PLoS One 5, e13276.

Hellberg, M., 2006. No variation and low synonymous substitution rates in coralmtDNA despite high nuclear variation. BMC Evol. Biol. 6, 24.

Isozaki, Y., 1997. Permo-Triassic boundary superanoxia and stratified superocean:records form lost deep sea. Science 276, 235–238.

Jermiin, L.S., Ho, S.Y.W., Ababneh, F., Robinson, J., Larkum, A.W.D., 2004. The biasingeffect of compositional heterogeneity on phylogenetic estimates may beunderestimated. Syst. Biol. 53, 638–643.

Kayal, E., Lavrov, D., 2008. The mitochondrial genome of Hydra oligactis (Cnidaria,Hydrozoa) sheds new light on animal mtDNA evolution and cnidarianphylogeny. Gene 410, 177–186.

Kim, J., Kim, W., Cunningham, C.W., 1999. A new perspective on lower metazoanrelationships from 18S rDNA sequences. Mol. Biol. Evol. 16, 423–427.

Kishino, H., Miyata, T., Hasegawa, M., 1990. Maximum likelihood inference ofprotein phylogeny and the origin of chloroplasts. J. Mol. Evol. 31, 151–160.

Kitahara, M., Cairns, S., Stolarski, J., Blair, D., Miller, D., 2010. A comprehensivephylogenetic analysis of the Scleractinia (Cnidaria, Anthozoa) based onmitochondrial CO1 sequence data. PLoS One 5, e11490.

Kumar, S., Nei, M., Dudley, J., Tamura, K., 2008. MEGA: a biologist-centric softwarefor evolutionary analysis of DNA and protein sequences. Brief Bioinform. 9,299–306.

Lavrov, D., Wang, X., Kelly, M., 2008. Reconstructing ordinal relationships in theDemospongiae using mitochondrial genomic data. Mol. Phylogenet. Evol. 49,111–124.

Lockhart, P.J., Steel, M.A., Hendy, M.D., Penny, D., 1994. Recovering evolutionarytrees under a more realistic model of sequence evolution. Mol. Biol. Evol. 11,605–612.

Lowe, T., Eddy, S., 1997. tRNAscan-SE: A program for improved detection of transferRNA genes in genomic sequence. Nucl. Acids Res. 25, 955–964.

McFadden, C., France, S., Sanchez, J., Alderslade, P., 2006. A molecular phylogeneticanalysis of the Octocorallia (Cnidaria: Anthozoa) based on mitochondrialprotein-coding sequences. Mol. Phylogenet. Evol. 41, 513–527.

McFadden, C., Sanchez, J., France, S., 2010. Molecular phylogenetic insights into theevolution of Octocorallia: a review. Integr. Comp. Biol. 50, 389–410.

McFadden, C., van Ofwegen, L., Beckman, E., Benayahu, Y., Alderslade, P., 2009.Molecular systematics of the speciose Indo-Pacific soft coral genus, Sinularia(Anthozoa: Octocorallia). Invertebr. Biol. 128, 303–323.

McFadden, C.S., Benayahu, Y., Pante, E., Thoma, J.N., Andrew Nevarez, P., France, S.C.,2011. Limitations of mitochondrial gene barcoding in Octocorallia. Mol. Ecol.Res. 11, 19–31.

Medina, M., Collins, A., Takaoka, T., Kuehl, J., Boore, J., 2006. Naked corals: skeletonloss in Scleractinia. Proc. Natl. Acad. Sci. USA 103, 9096–9100.

Nesnidal, M.P., Helmkampf, M., Bruchhaus, I., Hausdorf, B., 2010. Compositionalheterogeneity and phylogenomic inference of metazoan relationships. Mol. Biol.Evol. 27, 2095–2104.

Nylander, J., 2004. MrModeltest v2. Program Distributed by the Author.Evolutionary Biology Centre, Uppsala University 2.

Omland, K., Lanyon, S., Fritz, S., 1999. A molecular phylogeny of the New Worldorioles (Icterus): the importance of dense taxon sampling. Mol. Phylogenet.Evol. 12, 224–239.



Park, E., Kim, B., Won, Y.-J., 2010. The complete mitochondrial genome ofDendronephthya gigantea (Anthozoa: Octocorallia: Nephtheidae). Kor. J. Syst.Zool. 26, 197–201.

Park, E., Song, J.I., Won, Y.J., 2011. The complete mitochondrial genome ofCalicogorgia granulosa (Anthozoa: Octocorallia): Potential gene novelty inunidentified ORFs formed by repeat expansion and segmental duplication.Gene 486, 81–87.

Peterson, K.J., Cotton, J.A., Gehling, J.G., Pisani, D., 2008. The Ediacaran emergence ofbilaterians: congruence between the genetic and the geological fossil records.Philos. Trans. Roy. Soc. Lond. B: Biol. Sci. 363, 1435–1443.

Peterson, K.J., Lyons, J.B., Nowak, K.S., Takacs, C.M., Wargo, M.J., McPeek, M.A., 2004.Estimating metazoan divergence times with a molecular clock. Proc. Natl. Acad.Sci. USA 101, 6536–6541.

Peterson, K.J., Summons, R.E., Donoghue, P.C.J., 2007. Molecular palaeobiology.Palaeontology 50, 775–809.

Philippe, H., Zhou, Y., Brinkmann, H., Rodrigue, N., Delsuc, F., 2005. Heterotachy andlong-branch attraction in phylogenetics. BMC Evol. Biol. 5, 50.

Pont-Kingdon, G., Okada, N., Macfarlane, J., Beagley, C., Watkins-Sims, C., Cavalier-Smith, T., Clark-Walker, G., Wolstenholme, D., 1998. Mitochondrial DNA of thecoral Sarcophyton glaucum contains a gene for a homologue of bacterial MutS: apossible case of gene transfer from the nucleus to the mitochondrion. J. Mol.Evol. 46, 419–431.

Railsback, L.B., Anderson, T.F., 1987. Control of Triassic seawater chemistry andtemperature on the evolution of post-Paleozoic aragonite-secreting faunas.Geology 15, 1002–1005.

Rannala, B., Huelsenbeck, J., Yang, Z., Nielsen, R., 1998. Taxon sampling and theaccuracy of large phylogenies. Syst. Biol. 47, 702–710.

Rogers, A.D., 2009. Cnidarians (Cnidaria). In: Hedges, S.B., Kumar, S. (Eds.), TheTimetree of Life. Oxford University Press, USA, pp. 233–238.

Romano, S.L., Palumbi, S.R., 1996. Evolution of scleractinian corals inferred frommolecular systematics. Science 271, 640–642.

Ronquist, F., Huelsenbeck, J., 2003. MrBayes 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574.

Rosenberg, M., Kumar, S., 2001. Incomplete taxon sampling is not a problem forphylogenetic inference. Proc. Natl. Acad. Sci. USA 98, 10751–10756.

Sanderson, M.J., 1997. A nonparametric approach to estimating divergence times inthe absence of rate constancy. Mol. Biol. Evol. 14, 1218–1231.

Sanderson, M.J., 2002. Estimating absolute rates of molecular evolution anddivergence times: a penalized likelihood approach. Mol. Biol. Evol. 19, 101–109.

Schlagintweit, F., Gawlick, H.J., 2009. The incertae sedis Carpathoporella Dragastan,1995, from the Lower Cretaceous of Albania: skeletal elements (sclerites,internodes/branches, holdfasts) of colonial octocorals. Facies 55, 553–573.

Schroth, W., Jarms, G., Streit, B., Schierwater, B., 2002. Speciation andphylogeography in the cosmopolitan marine moon jelly, Aurelia sp. BMC Evol.Biol. 2, 1.

Seo, T.K., Kishino, H., 2008. Synonymous substitutions substantially improveevolutionary inference from highly diverged proteins. Syst. Biol. 57, 367–377.

Seo, T.K., Kishino, H., Thorne, J.L., 2004. Estimating absolute rates of synonymousand nonsynonymous nucleotide substitution in order to characterize naturalselection and date species divergences. Mol. Biol. Evol. 21, 1201–1213.

Shao, Z., Graf, S., Chaga, O., Lavrov, D., 2006. Mitochondrial genome of the moon jellyAurelia aurita (Cnidaria, Scyphozoa): a linear DNA molecule encoding a putativeDNA-dependent DNA polymerase. Gene 381, 92–101.

Shearer, T., Van Oppen, M., Romano, S., Wörheide, G., 2002. Slow mitochondrialDNA sequence evolution in the Anthozoa (Cnidaria). Mol. Ecol. 11, 2475–2487.

Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods withapplications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116.

Stanley, G.D., 2003. The evolution of modern corals and their early history. Earth-Sci. Rev. 60, 195–225.

Swofford, D., 2002. PAUP: Phylogenetic Analysis using Parsimony, Version 4.0 b10.Sinauer Associates, Sunderland, MA.

Talavera, G., Castresana, J., 2007. Improvement of phylogenies after removingdivergent and ambiguously aligned blocks from protein sequence alignments.Syst. Biol. 56, 564–577.

Thomas, J., Welch, J., Lanfear, R., Bromham, L., 2010. A generation time effect onthe rate of molecular evolution in invertebrates. Mol. Biol. Evol. 27, 1173–1180.

Thorne, J.L., Kishino, H., Painter, I.S., 1998. Estimating the rate of evolution of therate of molecular evolution. Mol. Biol. Evol. 15, 1647–1657.

Uda, K., Komeda, Y., Koyama, H., Koga, K., Fujita, T., Iwasaki, N., Suzuki, T., 2011.Complete mitochondrial genomes of two Japanese precious corals,Paracorallium japonicum and Corallium konojoi (Cnidaria, Octocorallia,Coralliidae): notable differences in gene arrangement. Gene 476, 27–37.

Van der Ham, J., Brugler, M., France, S., 2009. Exploring the utility of an indel-rich,mitochondrial intergenic region as a molecular barcode for bamboo corals(Octocorallia: Isididae). Mar. Genom. 2, 183–192.

Van Oppen, M.J.H., Willis, B.L., Miller, D.J., 1999. Atypically low rate of cytochrome bevolution in the scleractinian coral genus Acropora. Proc. Roy. Soc. Lond. B: Biol.Sci. 266, 179–183.

Vermeij, G.J., 1977. The Mesozoic marine revolution: evidence from snails,predators and grazers. Paleobiology 3, 245–258.

Veron, J., Odorico, D., Chen, C., Miller, D., 1996. Reassessing evolutionaryrelationships of scleractinian corals. Coral Reefs 15, 1–9.

Veron, J.E.N., 1995. Corals in Space and Time: The Biogeography and Evolution ofthe Scleractinia. Cornell Univ. Press, Ithaca.

Voigt, O., Erpenbeck, D., Worheide, G., 2008. A fragmented metazoan organellargenome: the two mitochondrial chromosomes of Hydra magnipapillata. BMCGenom. 9, 350.

Waggoner, B., Collins, A.G., 2004. Reductio ad absurdum: testing the evolutionaryrelationships of Ediacaran and Paleozoic problematic fossils using moleculardivergence dates. J. Paleontol. 78, 51–61.

Wesley, U.V., Wesley, C.S., 1997. Rapid Directional Wallc With in DNA Clones byStep-Out PCR. Methods Mol. Biol. Humana Press, Totawa, pp. 279–285.

Wignall, P.B., Twitchett, R.J., 1996. Oceanic anoxia and the end Permian massextinction. Science 272, 1155.

Wray, G.A., Levinton, J.S., Shapiro, L.H., 1996. Molecular evidence for deepPrecambrian divergences among metazoan phyla. Science 274, 568–573.

Yang, Z., 1994. Maximum likelihood phylogenetic estimation from DNA sequenceswith variable rates over sites: approximate methods. J. Mol. Evol. 39, 306–314.

Yang, Z., 1998. Likelihood ratio tests for detecting positive selection and applicationto primate lysozyme evolution. Mol. Biol. Evol. 15, 568–573.

Yang, Z., 2007. PAML 4: phylogenetic analysis by maximum likelihood. Mol. Biol.Evol. 24, 1586–1591.

Yang, Z., Rannala, B., 2006. Bayesian estimation of species divergence times under amolecular clock using multiple fossil calibrations with soft bounds. Mol. Biol.Evol. 23, 212–226.

Yoder, A.D., Yang, Z., 2000. Estimation of primate speciation dates using localmolecular clocks. Mol. Biol. Evol. 17, 1081–1090.

Zuckerkandl, E., Pauling, L.B., 1962. Molecular disease, evolution, and geneticheterogeneity. In: Kasha, M., Pullman, B. (Eds.), Horizons in Biochemistry.Academic Press, New York, pp. 189–225.

Zwickl, D., Hillis, D., 2002. Increased taxon sampling greatly reduces phylogeneticerror. Syst. Biol. 51, 588–598.

Documents

Estimation of divergence times in cnidarian evolution based on mitochondrial protein-coding genes and the fossil record