A Phylogeny of the Family Poritidae (Cnidaria, Scleractinia) Based on Molecular and Morphological Analyses

The family Poritidae formerly included 6 genera: Alveopora, Goniopora, Machadoporites, Porites, Poritipora, and Stylaraea. Morphologically, the genera can be differentiated based on the number of tentacles, the number of septa and their arrangement, the length of the polyp column, and the diameter of the corallites. However, the phylogenetic relationships within and between the genera are unknown or contentious. On the one hand, Alveopora has been transferred to the Acroporidae recently because it was shown to be more closely related to this family than to the Poritidae by previous molecular studies. On the other hand, Goniopora is morphologically similar to 2 recently described genera, Machadoporites and Poritipora, particularly with regard to the number of septa (approximately 24), but they have not yet been investigated at the molecular level. In this study, we analyzed 93 samples from all 5 poritid genera and Alveopora using 2 genetic markers (the barcoding region of the mitochondrial COI and the ITS region of the nuclear rDNA) to investigate their phylogenetic relationships and to revise their taxonomy. The reconstructed molecular trees confirmed that Alveopora is genetically distant from all poritid genera but closely related to the family Acroporidae, whereas the other genera are genetically closely related. The molecular trees also revealed that Machadoporites and Poritipora were indistinguishable from Goniopora. However, Goniopora stutchburyi was genetically isolated from the other congeneric species and formed a sister group to Goniopora together with Porites and Stylaraea, thus suggesting that 24 septa could be an ancestral feature in the Poritidae. Based on these data, we move G. stutchburyi into a new genus, Bernardpora gen. nov., whereas Machadoporites and Poritipora are merged with Goniopora.


Introduction
The family Poritidae Gray, 1842 is distributed throughout the tropics [1] and includes over 140 nominal species [2], some of which are among the major coral reef framework builders. The family formerly included 6 extant genera (currently Alveopora is not included), 3 of which are species-rich [1]-Porites Link, 1807 (73 species), Goniopora de Blainville, 1830 (31 spp.), and Alveopora de Blainville, 1830 (18 spp.)-and 3 are monospecific-Stylaraea Milne Edwards & Haime, 1851, Poritipora Veron, 2000, and Machadoporites Nemésio, 2005. These last 3 are rare or restricted to peculiar environmental conditions or have a limited geographic distribution. Conversely, the 3 major genera are common throughout the Indo-Pacific, with Porites extending into the Atlantic, from tropical to temperate regions. The biological, ecological and biometeorological studies such as climate change using Porites are well known [3][4][5][6][7], and this genus has been studied extensively at the molecular level in comparison with other coral genera [8][9][10] although it is still very difficult to divide Porites into discrete species in some cases (e.g., P. lutea and P. lobata) [9] due to the tiny but highly variable morphological characters used in the classification. In contrast, there are very few ecological and molecular studies of other Poritidae [11,12], with the exception of our recent work showing the phylogenetic relationships of several species of Goniopora [13].
There has been some dispute over the classification of genera in the Poritidae since the revision by Dana [14]. In particular, the position of Alveopora has been contentious [15]. Vaughan and Wells [16] reinstated Alveopora in the Poritidae because of its similarity to Goniopora in ecology and polyp behavior, although in listing the skeletal characteristics of the genus, Wells [17] qualified most as ''except in Alveopora''. Alveopora may be confused with Goniopora in the field because the polyps of both genera have very long columns, a character not shared with other genera, and these are usually elongated, obscuring the skeleton. As Veron and Pichon [15] noted, Goniopora polyps have 24 tentacles whereas Alveopora polyps have 12 tentacles. Moreover, they state that ''no east Australian Alveopora shows any sign of having either of the patterns of septal fusion found in Porites or Goniopora.'' In support of this argument, recent molecular studies have indicated that Alveopora is genetically distant from Porites and Goniopora [18,19], although only a few specimens and species from each genus were analyzed in these studies. Based only on the evidence from these published genetic works, Alveopora was transferred from the Poritidae to the Acroporidae Verrill, 1902 by Dai and Horng [20], albeit without any formal diagnosis or discussion. Recently the characters of Alveopora have been discussed in relation to the Acroporidae [21,22].
Goniopora is easily distinguishable from other genera in the Poritidae, with 3 septal cycles and 24 septa (vs. 2 septal cycles and 12 septa for Porites, Alveopora, and Stylaraea) and larger corallites [15]. Machadoporites has 15-22 septa and hence fewer than Goniopora [23]. Poritipora skeletal morphology is quite similar to that of Goniopora due to the presence of 24 septa, but the former has smaller corallites and only 2 septal cycles [24]. Stylaraea is the only monospecific poritid genus to have been studied since its description. One ecological study [25] and a few morphological studies [26] have been reported for Stylaraea. Stylaraea punctata is possibly the smallest colonial zooxanthellate scleractinian coral and has been reported as a brooder [25]. This genus superficially resembles Porites, but is distinguished from it by reduced septation without the Porites pattern of fusion (e.g. septal triplet), and the absence of pali, although its phylogenetic position in the Poritidae has not been studied.
Recent molecular phylogenetic analyses of scleractinian corals [19,[27][28][29][30][31] revealed that molecular-based phylogenetic relationships sometimes conflict with traditional macromorphology-based taxonomy. This indicates that the current, common identifying characters within a family or genus of scleractinian corals do not always reflect their phylogenetic relationships. One such inconsistency is seen in the relationship between Atlantic and Indo-Pacific Faviidae Gregory, 1900 [27] (but see [32] for the taxonomic revision). Despite such conflicts, there are also many cases in which molecular data are consistent with traditional taxonomy. Moreover, some morphological characters previously not considered key characters have been found to be effective as diagnostics for the phylogenetic relationships in several cases [32][33][34]. Overall, most molecular studies conclude that molecular analysis is useful in discerning the relationships of species or genera and inferring the phylogeny. Forsman et al. [9] investigated the molecular-based phylogenetic relationships among species of Porites, using mainly the entire internal transcribed spacer region (ITS) of the nuclear ribosomal DNA. Their results showed that most species of Porites were clearly distinguishable genetically, highlighting the usefulness of ITS markers in inferring relationships in Porites at the species level.
In this study, we assess the relationship of all 5 genera in the Poritidae with Alveopora to revise the taxonomy, and infer the morphological changes in the evolutional lineage in this family, using both molecular and morphological analysis. Also to assess phylogenetic variation in the regional and species differences, the present study examines a large number of specimens collected with broad geographic ranges from mainly Japan water to the Indian Ocean, covering most of common species and some uncommon and rare species, together with the genetic data of Porites spp. from Forsman et al. [9].

Collection
Most specimens (approximately 10 cm 3 in size) were collected from 17 sites in Japan (Fig. 1). Additionally some samples were collected from Malaysia, and western Indian Ocean (Table 1). All  Japanese sampling (AK, AM, AO, IK, IR, IS, KK, KS, MI, MO, OT, OU, SO, SR, SS, TN, TR; see Fig. 1) was performed in the frame of research projects by Japanese Society for Coral Taxonomy or by associate prof. H. Fukami at University of Miyazaki with sampling permission from each local government in Japan. Malaysia (PEN; see Fig. 1) sampling has taken place by local staffs in non-marine protected area, Songsong Island, under the permission of the research project by prof. Zulfigar Yasin and prof. Aileen Tan at Universiti Sains Malaysia. All western Indian Ocean sampling was also performed in the frame of research projects for which a sampling permission was delivered by local authorities and samples were shipped with CITES permits. AD, BA, BU, DJ, and MU are all sites in Yemen (Fig. 1). There, sampling has taken place in several missions and regular sampling permits were issued by Yemen Environmental Protection Agency (EPA) in Sana'a. Moreover, EPA staff supervised the activities in the field at all times. MY is Mayotte Island (Fig. 1). Sampling permits there were issued by the Direction de l'Agriculture et de la Foret de Mayotte, Service Environnement et Foret and by the Maritime Affairs Office. DJ are samples from Djibouti ( Fig. 1) taken during the Tara Oceans expedition and the sampling permits were delivered by the Aménagement du Territoire et de l'Environnement de Djibouti. Photos of each specimen were taken in the field (particularly for living polyps) and the depth and habitat were recorded. After collection, a small piece of each specimen was removed for use in DNA extraction (see below), and the remaining sample was bleached to investigate the skeletal morphology for species identification.

Species identification
Species identification of Goniopora and Alveopora is difficult due to very limited skeletal characters and highly variable skeletal and polyp morphologies. The best solution to this common problem in the scleractinian corals is detailed analyses of the type material of each species [28,29]. In order to minimize the risk of misidentification in this study, we firstly made lists of characters for species identification for these two genera (Tables S1, and S2) using the original descriptions and related references [1,14,15,26,. These tables were used to identify our specimens to species, examining the skeletal morphology of each specimen using a VHX-1000 digital microscope (Keyence) or stereoscopic microscope. Traditionally, species identification of Goniopora and Alveopora is based on skeletal characters, but recently Veron and Pichon [15], Nishihira and Veron [59], and Veron [1] added polyp characters to support species identification. Therefore, we also considered polyp characters along with skeletal characters for species identification in this study. The skeletal specimens collected in Japan are retained at University of Miyazaki (MUFS) or Seto Marine Biological Laboratory (SMBL). Skeletal specimens collected in Malaysia are deposited at Universiti Sains Malaysia (MSL/USM), and skeletal specimens collected in western Indian Ocean are at University of Milano-Bicocca (UNIMIB).

Genetic analyses
A small sample (less than 1 cm 3 ) of each specimen was put in CHAOS solution to dissolve the tissues or fixed in 99% ethanol. Total DNA was extracted from CHAOS solution using the phenol/chloroform extraction method [61], and from the coral tissues preserved in ethanol using the DNeasy Blood & Tissue Kit (Qiagen). The barcoding region of the mitochondrial Cytochrome oxidase subunit I (COI) was amplified by the polymerase chain reaction (PCR) using the primers ZCO1 and ZCO1R [9]. The nuclear ribosomal ITS region (ITS) including the 39 end of the 18s rRNA, ITS-1, 5.8s, ITS-2, and the 59 end of the 28s rRNA was also amplified by PCR using the primers 1S and 2SS [62]. The PCR condition for these two markers was 94uC for 30 seconds followed by 30 or 35 cycles at 94uC for 30 seconds, 55uC or 60uC for 45 seconds, and 72uC for 90 seconds, with a final phase of 72uC for 5 minutes. For the mitochondrial region, PCR products were treated with Shrimp Alkaline Phosphatase (SAP) and Exonuclease I (ExoI) at 37uC for 40 minutes followed by 80uC for 20 minutes. The DNA sequences were then determined via a direct sequence method, using ABI3730 or ABI310 sequencer. PCR products of the nuclear marker were also directly sequenced, but when obtained sequences had more than double peaks in the chromatogram, they were sub-cloned into TA-vector (Promega) or TOPO10 (Invitrogen) and sequenced using ABI3730 or ABI310.
All DNA sequences obtained in this study were submitted to DDBJ (accession No. AB906942-AB907101, listed in Table 1).
A total of 15 COI and 26 ITS Goniopora sequences were taken from our previous study ( [13], see Table S3). Also, a number of sequences were downloaded from GenBank/DDBJ and included in the molecular analyses. For COI analysis, 30 sequences from 18 species of Porites ([9], Table S3), one sequence (AB441211 [19]) from Siderastrea siderea in the family Siderastreidae Vaughan an Wells, 1943, two sequences (AB441201-AB441202 [19]) from Galaxea fascicularis in the family Euphylliidae (Alloiteau, 1952), and two sequences (AB441216-AB441217 [19]) from Pavona spp. in the family Agariciidae Gray, 1847 (species name of AB441216 is registered as P. cactus in DDBJ, but it is revised P. decussata as the results of the reexamination of the skeleton). Also, in the family Acroporidae, four sequences (AF338425, AY451340-AY451342 [63,64]) from three species of Acropora, one sequence (AY903296 [65]) from Montipora cactus and one sequence (AY903295 [65]) from Anacropora matthai were used. For ITS, 67 sequences from 18 species of Porites from previous study ( [9], see Table S3) were used. Because there are many sequences of Porites for COI and ITS    registered in GenBank, the sequences of representative species from each of the different phylogenetic groups reported by Forsman et al. [9], were selected for this study. Finally, as Fukami et al. [19] and Kitahara et al. [31] showed that the family Dendrophylliidae Gray, 1847 is genetically closely related to the family Poritidae, specimens of Turbinaria peltata (family Dendrophylliidae) from Japan were also sampled for this study.

Molecular phylogenetic analysis
Electropherograms and DNA sequences were checked and edited using Sequencher (Gene Code Co.) and SeeView 4.3.0 [66]. DNA sequences were aligned with MAFFT 7 [67] using the L-INS-i option. Then, all sites with indels and several sites with alignment ambiguities were excluded manually from the subsequent analyses. Two aligned DNA datasets (COI and ITS) used in this study are shown in supplementary information (Datasets S1, S2). Pairwise genetic distances were calculated as p-distance using MEGA 4.0.2 [68]. Phylogenetic trees were reconstructed by neighbor-joining (NJ) and maximum likelihood (ML). For NJ, PAUP* 4.0b10 [69] was used to infer the topologies for both COI and the ITS markers using Kimura 2-parameter model [70] and to conduct bootstrap analysis (1000 replicates). For ML, we assumed a model of nucleotide evolution obtained by using the Akaike Information Criterion (AIC) as implemented in MrModeltest 2.2 [71]. The most appropriate models of nucleotide evolution were TrN with invariant (I) and gamma (G) parameters (TrN+I+G) for the COI marker, and TrNef +I+G for ITS marker. PAUP* was used to reconstruct a best ML tree using a heuristic search and the tree-bisection-reconnection branch swapping method. GARLI (Genetic Algorithm for Rapid Likelihood Inference) 0.951 [72] was preferred to PAUP* for the bootstrap estimation as the former is less time consuming. Using GARLI, optimal ML topologies were searched with default setting using the models selected by MrModeltest (TrN+I+G for COI, TrNef +I+G for ITS) and bootstrap analyses (500 replicates) were conducted for each marker. MrBayes 3.2.2 [73] was also used to conduct Bayesian analyses under the same models. Four parallel chains of 1-4610 6 generations were run for each marker. Trees were sampled every 100 generations, and the initial 25% of the total trees as burn-in were discarded. The remaining trees were pooled to produce a 50% majority rule consensus tree. The average standard deviation of split frequencies after 4610 6 generations was 0.002069 for COI, and ones after 2.4610 6 generations was 0.009967 for ITS. All topologies obtained in these analyses were quite similar, so that only the ML tree inferred using PAUP* is used in this study.
In addition, we combined COI and ITS data and analyzed them with same methods as each marker using the GTR+I+G model for the nucleotide substitution (the average standard deviation of split frequencies after 1.0610 6 generations was 0.009909).

Nomenclature Acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix ''http://zoobank.org/''. The LSID for this publication is: urn:lsid:zoobank.org:pub: 6975D790-3A4F-466A-ABFA-D922E6675B4B. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Species identification
Twenty samples of Alveopora and 58 samples of Goniopora were analyzed in this study (Table 1). Although a few species have species-specific polyps, such as Goniopora albiconus, polyp characters vary greatly in the field. For example, terete tentacles, a typical polyp character of G. tenuidens, are also seen in G. burgosi.
All 7 specimens of Stylaraea punctata were found in very shallow water (1 m) on a sandy beach in Aakajima Island, Okinawa, Japan (Fig. 1). Notably, all of them were attached to dead coral skeletons of the genus Acropora. Their size is very small (less 1 cm) and they have only 5 or 6 corallites. Tentacle and septal numbers were both 12 in all of them ( Fig. 2A-D).
Poritipora paliformis Veron, 2000 has 24 septa with typically 2 septal cycles (long and short), 6 pali and no columella reported in the literature [24]. Two samples we collected in Taketomi Island, Japan (first record in the Pacific Ocean) had no elongating polyps in the field and had a cellular appearance (Fig. 3A), which is a feature of P. paliformis, as shown in Veron [1,24]. The skeletal morphologies are also consistent in the literature, although the second cycle is not well developed in some corallites; however, many had 24 septa with 2 cycles (Fig. 3B). Therefore, we identified these 2 samples as P. paliformis. This species was described in Veron [1] without designating type material, and then it was redescribed [24] designating the holotype. However, the hototype of this species is not valid following ICZN [74], and the specimens listed in Veron [1] are regarded as part of the syntype series. Therefore, Dash means no data. Asterisk shows accession number referred from Kitano et al. [13]. Note that more than one ITS sequences were obtained by sub-cloning from a single specimen in several samples while ITS from other samples were determined by direct sequencing of PCR products. the holotype of this species listed in Veron [24] is to be considered a lectotype. Four specimens collected in the Gulf of Aden (Fig. 3EF), which is near the type locality of Machadoporites tantillus (Claereboudt & Al-Amri, 2004), were identified as M. tantillus because they are consistent with the original description of this species [23].
Diagnostic morphological characters among genera in the family Poritidae are summarized in Table 2.

COI phylogeny
We obtained 69 COI sequences from all 5 genera in the Poritidae with Alveopora, 3 sequences from Turbinaria peltata and Astreopora spp., and one from Montipora venosa (Table 1). A total of 473 positions were used (120 polymorphic sites with 109 informative sites) and no indels were observed. A phylogenetic tree was reconstructed using these data, including sequences from GenBank/DDBJ (see Methods). Siderastrea siderea was used as an outgroup, based on the phylogenetic position of the Scleractinia shown by Fukami et al. [19]. The COI phylogenetic tree showed that all 18 species of Porites are monophyletic. Moreover, the 13 Goniopora species we examined in this tree are also monophyletic, with the notable exception of G. stutchburyi (Fig. 4). This species and Stylaraea punctata are sister taxa (Fig. 2), and together they form a sister group to Porites. Machadoporites tantillus and Poritipora paliformis are nested within Goniopora (except G. stutchburyi). On the one hand, within the Goniopora phylogeny, M. tantillus forms a clade with the western Indian Ocean species G. somaliensis and G. cf. somaliensis. On the other hand, P. paliformis forms a clade with G. minor and G. columna, although G. somaliensis and G. minor are polyphyletic. Notably, all G. somaliensis from Japan are genetically distant from the G. somaliensis specimens from the western Indian Ocean. Uncorrected genetic pdistances between G. stutchburyi and Porites and between G. stutchburyi and the remaining Goniopora spp. were very similar (approximately 0.02).
All samples of Alveopora are genetically distant from all other poritids (p-distance 0.08-0.10), but closely related to the family Acroporidae (0.06). The phylogenetic position of Alveopora is unclear due to low bootstrap values, but it forms a sister group with Astreopora spp. In addition, sequences from T. peltata (family Dendrophylliidae) form a sister clade of all poritids except Alveopora and are positioned between Alveopora and the other poritids.

ITS phylogeny
We obtained a total of 84 sequences of ITS from all 5 genera in the Poritidae (Table 1). In this study, we excluded Alveopora from ITS analysis because ITS regions were highly variable between Alveopora and other genera and they were hardly aligned.
The phylogenetic relationships among Porites, Goniopora, Stylaraea, Poritipora, and Machadoporites were inferred using ITS (Fig. 5). The 68 Porites sequences from Forsman et al. [9] and 26 Goniopora sequences from Kitano et al. [13] were also added for this analysis (see Table S3). A total of 347 positions were used (108 polymorphic sites with 89 informative sites). This ITS tree also showed similar topology to the COI tree as described above. In particular, Stylaraea punctata and G. stutchburyi are sister taxa. Poritipora paliformis formed a clade with G. minor and G. columna. One specimen of G. minor in the Poritipora clade is from the western Indian Ocean and others are from Japan. Machadoporites tantillus formed a clade with G. somaliensis and another 3 species (G. cf. somaliensis, G. sp.1, and G. sp.2), all of which were collected from the western Indian Ocean. Other western Indian Ocean specimens (G. albiconus, G. ciliatus) and Malacca Strait specimens (G. albiconus, G. pendulus) were included in a major clade of Goniopora spp. Meanwhile, species relationships of Goniopora were less resolved because Porites and Goniopora have many indels in their rDNA sequences and phylogenetic information sites were largely excluded.
The tree using combined data of COI and ITS showed a similar topology to the one for ITS (Fig. S1).

Phylogenetic relationships of Goniopora, Stylaraea, and Porites
Our molecular data showed that Goniopora can be divided into 2 distinct groups (Figs. 4, 5). One contains most Goniopora species together with Machadoporites and Poritipora, whilst the other is monospecific for G. stutchburyi. Moreover, G. stutchburyi is monophyletic with Stylaraea punctata and these 2 species form a sister group to all Porites. Stylaraea punctata has very unique morphological characters among the zooxanthellate scleractinian corals, such as the smallest (less 1 cm) mature colony size with only 5 or 6 corallites [25 and our observation of the release of zooxanthellate planulae from samples AK92 (maximum diameter 7.15 mm) and AK93 (4.00 mm)], and 12 septa ( = 12 tentacles) without a specific septal pattern such as in Porites or Goniopora. Moreover, the calices of this species are devoid of pali. Thus, these features allow to distinguish Stylaraea from the other genera in this family. Although Alveopora can bear resemblance to Stylaraea, the two genera can be distinguished based on the columella (not present or very weakly developed in Alveopora and star-like columella in Stylaraea) and the elongation of polyps (long polyps for Alveopora and no elongation for Stylaraea). Goniopora stutchburyi also has its own specific characters, such as coralla that appear to have a smooth surface, dense septa and a columella composed of a star-shaped central rod (Figs. 2G, H). Thus, G. stutchburyi is easily distinguishable from all other congeneric species. Although the smooth appearance of the skeleton of this species and small corallite size may lead to misidentification as Porites, G. stutchburyi can be differentiated from Porites on the basis of its septal number (24 vs. 12). Thus, morphologically, the only character shared by S. punctata and G. stutchburyi is a columella composed of a star-shaped central rod (Figs. 2D, H), which is absent in other Goniopora species but common in Porites. This type of columella could be a synapomorphy in the lineage including Porites, Stylaraea, and G. stutchburyi. Considering that molecular and morphological data agree in distinguishing G. stutchburyi from the other Goniopora species and S. punctata, we establish Bernardpora Kitano & Fukami gen. nov. (urn:lsid:zoobank.org:act:9C2FE523-A491-45AE-BC22-B528CA68C040, also see below for details) and move G. stutchburyi into it. Comparison of diagnostic morphological characters between genera is summarized in Table 2.
Bernard [26] proposed that the septal formula of Porites derives from that of Goniopora by reduction of the third septal cycle, referring to the typical septal pattern of Goniopora as the gonioporoid pattern. Veron and Pichon [15] showed that G. stutchburyi typically has this septal pattern (Fig. 2G). In this study, we proved that G. stutchburyi is a basal species of Porites, and our results strongly support Bernard's hypothesis that Porites is derived from Goniopora-like morphologies. This conclusion is also supported by the fossil record: Goniopora extends back to the Cretaceous, but Porites only to the Eocene [16]. Thus, the presence of 24 septa would seem to be an ancestral feature in the Poritidae. The fact that taxa in the family Dendrophylliidae, the closest related outgroup of the Poritidae, have more than 24 septa would appear to support this. According to Veron and Pichon [15] and Veron [1], Goniopora burgosi and G. somaliensis are morphologically similar to G. stutchburyi in corallite size and height, respectively. Furthermore, G. burgosi appears to form only 12 septa as a third cycle of septa is reduced or absent in whole colonies, as is typical of Porites (Goniopora usually forms 24 septa). However, in our genetic analysis, G. burgosi and G. somaliensis were included in the major Goniopora clade, and their phylogenetic position was distant from Porites even though they have fewer than 24 septa. The typical gonioporoid pattern was also observed in some colonies of G. lobata and G. djiboutiensis, but they were not closely related to G. stutchburyi. These results suggest that each of these morphological characters alone (corallite size, depth, and septal formula) would not be sufficient to explain the species relationships in this genus.

Phylogenetic position of Alveopora
None of the 8 species of Alveopora we analyzed is closely related to Goniopora, despite the 2 genera having similar elongating polyps and tentacles. This suggests that the typically elongated polyps in these genera have appeared independently as result of convergent evolution. Alveopora is also genetically distant from Porites. Both genera have 12 tentacles (12 septa), but our genetic data indicate that this character is not synapomorphic, which is also supported morphologically by the difference in septal plan in both genera (spin-like for Alveopora and fused for Porites). Bernard [26] asserted that Alveopora did not belong to the Poritidae, and Veron and Pichon [15] also cast doubt on this classification. Along with our genetic data, the morphological differences between Alveopora and other genera in the Poritidae, which have previously been noted [15], support the hypothesis that Alveopora does not belong in the Poritidae. Wallace [21] also noted that Alveopora has the Acroporidae characteristic of synapticulothecate skeleton but does not have coenosteum. As reported by previous studies, the phylogenetic position of Alveopora is close to the Acroporidae [18,19,31,62]. The ITS of Alveopora is also very divergent, a characteristic shared by acroporids, but not observed in other families [62]. However, the phylogenetic position of Alveopora is still unclear because different topologies (forming a sister group with Astreopora or outside of Astreopora) were also obtained in COI (Fig. 4) and rDNA [62] trees. Our preliminary analyses of rDNA also showed that Alveopora was positioned outside of the Astreopora (data not shown). A complete evaluation of the phylogenetic position of Alveopora is outside the scope of this study. To perform such an evaluation, morphological comparison among all genera of Acroporidae would be necessary (Wallace et al., in preparation).
Phylogenetic relationships of the monospecific genera Stylaraea, Machadoporites, and Poritipora Poritipora and Machadoporites are found within the Goniopora lineage in all molecular phylogenetic trees. This is supported by morphology. Machadoporites differs from Goniopora by having fewer septa (fewer than 24) and smaller calices (,1.7 mm). However, some Goniopora species can have superficially similar characters. For example, G. burgosi has typically 12-15 septa, as shown in the original description ( [51], Fig. 3G). A similar pattern is also observed in G. pendulus (Fig. 3H). Moreover, the G. minor calices were described as 1.5-2 mm in size in the original description [50]. Thus, characters such as ''fewer than 24 septa,'' and ''small size calices'' are not enough to separate Machadoporites from Goniopora. In addition, M. tantillus forms a clade with G. somaliensis and other Goniopora species from the western Indian Ocean.
Similar to Goniopora, Poritipora has 24 septa, but the 2 genera can be distinguished by the difference in the number of septal cycles: 2 in P. paliformis and 3 in Goniopora. However, for several Goniopora species, primary and secondary cycles of septa are equal or subequal, such as in the case of G. minor (Fig. 3D). Therefore, the character ''two cycles of septa'' is not enough to separate Poritipora from Goniopora. In addition, P. paliformis forms a clade with G. minor and G. columna.
On the one hand, Machadoporites and Poritipora are considered junior synonyms of Goniopora and their taxonomy is hence revised hereafter.
On the other hand, the type material of P. paliformis (Fig. 3C) and our samples (Fig. 3B) look similar to the type material of G. minor (Fig. 3D) shown in Crossland [50]. Goniopora minor has a similar size of corallites, 12 equally sized septa for the primary and secondary cycles, small or absent septa in tertiaries, and 4-6 pali. The development of the columella was described as ''large,'' but it is composed only of joined septa, which is the same pattern as that of Poritipora. Considering that most G. minor examined in this study (one colony of G. minor was genetically separated; Figs. 4 and 5) formed a clade with P. paliformis with little genetic difference, P. paliformis may be a morphological variant of G. minor.

Regional differences
Our specimens were collected mostly from Japanese waters, but some of them including Machadoporites were collected from the western Indian Ocean and the Malacca Strait. Although several species from these regions, such as G. albiconus and G. ciliatus, were included in the clade with specimens collected from Japanese waters, 4 specimens, including G. somaliensis from the western Indian Ocean, formed their own clade, whereas G. somaliensis from Japanese waters was distant from the western Indian Ocean clade and included in a major clade of Goniopora spp. This suggests that morphological convergence may have occurred between the western Pacific and the western Indian Ocean populations of these species. Recently, Arrigoni et al. [75] showed that numerous cases of intraspecific divergence between Indian Ocean and Pacific Ocean populations were present in the families Merulinidae and Lobophylliidae. Moreover, Keshavmurthy et al. [76] reported that a widely distributed species, Stylophora pistillata, comprises 4 divergent clades corresponding to different regions, such as the western Pacific and the Red Sea, suggesting that their clades are  Table 1, Table S3). Grey in color for Alveopora, green for Porites, purple for Stylaraea, blue for 'Poritipora', and orange for 'Machadoporites'. Goniopora is shown by bars in black. Bernardpora is shown by bar in red. doi:10.1371/journal.pone.0098406.g004 divergent at the species level. A similar observation on divergence was made with regard to the octocoral family Melithaeidae [77], indicating that species did not cluster according to their present morphological classification but instead clustered according to a biogeographical pattern such as the Indo-Pacific, Red Sea and Indian Ocean. Thus far, many studies of scleractinian corals have focused on higher level taxonomy and are based on material sampled from one or a few nearby Indo-Pacific regions. However, species-level analyses among regions are the next necessary step in the ongoing revolution in scleractinian taxonomy.

Taxonomic account
Below we propose the description of the new genus Bernardpora gen. nov. and the revised diagnosis of Goniopora, based on the original descriptions and subsequent information resulting from this study. See Table 1 for the museum abbreviations.
Family Poritidae Gray, 1847 Type genus: Porites Link, 1807 Diagnosis [1,16,60]: Massive, laminar or ramose colonies; corallites vary in size but usually small and mostly compacted closely without coenosteum, with one or two synapticular rings. Walls and septa are porous. Septa usually 12 to 24. Septa formed by 3 to 8 nearly vertical trabeculae, and innermost trabeculae of certain septa differentiated as pali.
Remarks: There are four extant genera in Family Poritidae, Porites, Goniopora, Stylaraea and Bernardpora gen. nov. All are zooxanthellate corals. Porites is the only genus distributed throughout the tropics. Others are Indo-Pacific. Based on our results we confirm that the genus Alveopora does not belong to the same lineage as the family Poritidae. Although a full evaluation of the position of Alveopora is not completed yet, it is certain that Alveopora is closely related to other genera in the family Acroporidae ( [20], this study).
Genus Mural trabeculae always present. Columella trabeculae usually present with star-shaped granules. The wall is really simple, but the incipient synapticulae, seen starting from the sides of septal granules, may become complete and form an inner synapticular wall.
Generic Revised diagnosis [1,16,26,this study]: Massive, columnar or ramose, rarely encrusting colonies. Corallites are circular or polygonal. Calice diameter 1-10 mm. Septa 24 in two or three cycles, or between 24 and 12 in two or three cycles, composed of 4 to 8 trabeculae. Pali and columella may develop. Columellae are composed of anastomosed septal dentations or arranged synapticula and fused inner ends of septa. Wall structure is synapticulothecal. Polyps usually elongate during the day (note that G. paliformis does not elongate polyps during the day).
Remarks: Poritipora and Machadoporites are considered as junior synonyms of Goniopora. Distribution: Indo-Pacific [1]. Species number: 33 [1,15, Diagnosis: Stylaraea is a monospecific genus with only known species, S. punctata. Therefore, the characters of this genus are those of S. punctata. Colonies are tiny (usually less 10 mm in size) and from ''cushion-shaped crusts'' [46]. Calices are concavate and around 1 mm diameters. Septal number is 12 (''2 cycles of 6 each'' [15]) without specific septal pattern. Septa are composed of rows of star-shaped granules. Primary septa may reach to collumellae. Columella is composed of a star-shaped central rod such as Porites or Bernaldopora. Wall structure is synapticulothecal.
Remarks: Distribution: Indo-Pacific [1]. Diagnosis: Bernardpora is a monospecific genus with only known species, B. stutchburyi. Therefore, the characters of this genus are those of B. stutchburyi. Encrusting or sub-massive colonies. Figure 5. Molecular phylogenetic relationships of genera of the Poritidae except of Alveopora based on ITS sequences. Letter (a, b, c, d) after sample code indicates that different alleles were obtained from a single coral sample by cloning. Numbers on/below main branches show bootstrap values (.50%) in ML and NJ analyses, and Bayesian posterior probability (.0.8). Stars show specimens collected from western Indian Ocean, and triangles show ones collected from Malacca Strait. Sample codes or accession numbers are shown after species names (see Table 1). Green in color for Porites, purple for Stylaraea, blue for 'Poritipora', orange for 'Machadoporites', red for Bernardpora, and black for Goniopora. doi:10.1371/journal.pone.0098406.g005 Corallites round to polygonal and very shallow with smooth appearance. Calices are around 2 mm diameters. Septal number is mostly 24 with clear gonioporoid pattern shown in Bernard [26]. Septa are dense with up to seven multiple-spine-shaped septal teeth. Septal teeth closest to collumellae are indistinguishable from pali. Primary and secondary septa nearly reach to collumellae. Columella is composed of a star-shaped central rod such as Stylaraea and Porites. Wall structure is synapticulothecal. Polyps elongate but very short during the day.
Remarks: Distribution: Indo-Pacific [1]. Species number: 1 Etymology: The generic name is in honor of the coral scientist Henry M. Bernard.  Table 1, Table S3). Grey in color for Alveopora, green for Porites, purple for Stylaraea, blue for 'Poritipora', and orange for 'Machadoporites'. Goniopora is shown by bars in black. Bernardpora is shown by bar in red.

(TIF)
Dataset S1 Nexus data file of the COI sequence alignments used for the analyses.

(NXS)
Dataset S2 Nexus data file of the ITS sequence alignments used for the analyses.