Three new species and the molecular phylogeny of Antipathozoanthus from the Indo-Pacific Ocean (Anthozoa, Hexacorallia, Zoantharia)

Abstract In this study, three new species of macrocnemic zoantharians (Hexacorallia, Zoantharia) are described from localities in the Indo-Pacific Ocean including the Red Sea, the Maldives, Palau, and southern Japan: Antipathozoanthus obscurus sp. n., A. remengesaui sp. n., and A. cavernus sp. n. Although the genus Antipathozoanthus is currently restricted to species living on antipatharians, A. obscurus sp. n. is not associated with any living substrate and instead is found on coral reef carbonate substrate within narrow caves or cracks. The two new species that have association with antipatharians, A. remengesaui sp. n. and A. cavernus sp. n., can be distinguished by their relative coenenchyme development and the antipatharian species that each uses as substrate. Additionally, all new species described in this study have unique nuclear internal transcribed spacer region of ribosomal DNA (ITS-rDNA) sequences. Our results indicate that more phylogenetic studies focusing on increasing the numbers of species examined within each of the genera of Parazoanthidae are required in order to better understand the evolutionary history of substrate specificity within the family Parazoanthidae.


Introduction
Zoantharia Rafinesque, 1815 is the third most speciose order within the subclass Hexacorallia Haeckel, 1896. Zoantharians can be found in a wide variety of marine environments from intertidal zones to deep-sea cold seeps (e.g., Reimer et al. 2007b), and are characterized by having two rows of tentacles and the unique bilateral arrangements of the mesenteries, with most species forming clonal colonies without hard structures such as skeletons of the order Scleractinia. Zoantharia is currently divided into two suborders; Brachycnemina Haddon &Shackleton, 1891, andMacrocnemina Haddon &Shackleton, 1891, based on differences in the fifth pair of mesenteries from the dorsal directive. Zoantharians within suborder Macrocnemina are distributed worldwide, and are usually found in associations with other invertebrates. Within Macrocnemina, the largest family is Parazoanthidae Delage & Hérouard, 1901, which currently contains 13 genera . Most species of these genera live in association with other marine invertebrates, including antipatharians (Ocaña and Brito 2003;Sinniger et al. 2010), octocorals (Reimer et al. 2008;Bo et al. 2012;Sinniger et al. 2013), and sponges (Haddon and Shackleton 1891;Swain and Wulff 2007;Montenegro et al. 2015Montenegro et al. , 2016. Historically, establishing the taxonomic framework of Parazoanthidae was challenging due to relatively few diagnostic morphological characteristics (Sinniger et al. 2005;Montenegro et al. 2015), and the family was shown to be paraphyletic in initial molecular studies (Sinniger et al. 2005). Recently, however, studies based on molecular phylogeny combined with ecological data have greatly revised the taxonomy within the family Parazoanthidae (Sinniger et al. 2005Sinniger and Häussermann 2009;Montenegro et al. 2015Montenegro et al. , 2016. As a consequence of these studies, nine genera within Parazoanthidae have been described since 2008 and another genus, Bergia Duchassaing & Michelotti, 1860, has been resurrected. Key to this new taxonomic framework is the idea initially proposed by Sinniger et al. (2005Sinniger et al. ( , 2010) that different parazoanthid genera share long evolutionary histories with the associated marine invertebrates they use as substrates.
One of these recently erected genera is Antipathozoanthus Sinniger, Reimer & Pawlowski, 2010. As the generic name indicates, species in this genus utilize antipatharians (Hexacorallia, Antipatharia) as their obligate substrate. The genus currently includes two valid species; A. macaronesicus (Ocaña & Brito, 2003) from the eastern Atlantic and A. hickmani Reimer & Fujii, 2010 from the Galapagos Islands. Additionally, several potentially undescribed species have been reported from the Red Sea (Reimer et al. 2014b), the South China Sea , and Japan (Sinniger et al. 2010;Reimer et al. 2013Reimer et al. , 2014a. However, the species diversity of Antipathozoanthus spp. in the Indo-Pacific Ocean remains generally unknown. In this study, three new Antipathozoanthus species are formally described based on specimens collected from a number of regions in the Indo-Pacific Ocean, and the genus is redescribed based on these findings.

Materials and methods
Specimen collection. Antipathozoanthus specimens were collected between 2009 to 2016 from three localities in the Red Sea, three localities in the Maldives, five localities in Japan, and two localities in Palau (Fig. 1), with one comparative specimen of A. macaronesicus collected from Pico Island, Azores, Portugal. All specimens were collected by SCUBA. Specimen images were taken in situ for gross external morphological analyses. Collected specimens were preserved in 99.5% ethanol (Table 1).
The generated alignments of each marker were used to construct a concatenated alignment. All missing data, including gaps, were replaced with "N". All specimens of Antipathozoanthus included in the concatenated alignment included at least ITS-rDNA sequences. The concatenated alignment consisted of 1957 positions and 54 sequences. Phylogenetic analyses of the concatenated alignment were performed using maximum likelihood (ML) and Bayesian inference (BI), with gene partitions set for ML in RAxML v8 (Stamatakis 2014), and gene partitions for BI as indicated by jMod-elTest version 0.0.1 (Posada 2008) per each marker in MrBayes v3.2.2 (Huelsenbeck and Ronquist 2001) as shown below. Phylogeny reconstructions were performed for each marker using neighbor joining (NJ), ML and BI.
The NJ phylogeny reconstruction was performed using Geneious v8.1 (Kearse et al. 2012, http://www.geneious.com) with the Hasegawa-Kishino-Yano genetic dis-tance model (HKY) (Hasegawa et al. 1985) and 1000 replicates of bootstrapping. The best-fitting models for ML phylogeny reconstruction were performed by jModelTest under Akaike Information Criterion (AIC). The following models were suggested by jModelTest: TrN+I for the COI dataset; K80+G for the 16S-rDNA dataset; HKY+I+G for ITS-rDNA dataset. ML phylogenetic trees were constructed with PhyML (Guindon and Gascuel 2003) for each marker independently. PhyML was performed using an input tree generated by BIONJ with the models suggested by jModelTest, with 8 gamma-categories of substitution rates. Bootstrap replicates (1000) were conducted using the same parameters. The best fitting models for BI phylogeny reconstruction was performed by jModelTest under Bayesian Information Criterion (BIC). The following models were suggested by jModelTest: K80+G for the COI dataset; K80+G for the 16S-rDNA dataset; and HKY+I+G for the ITS-rDNA dataset. BI phylogenetic trees were constructed with the program MrBayes as a plug-in in Geneious with the models suggested by jModelTest. One cold and three heated Markov chain Monte Carlo (MCMC) chains with default temperature were run for 20,000,000 generations, subsampling frequency of 1000 and a burn in length of 3,000,000 (15%) for all alignments. Average Standard Deviation of Split Frequency (ASDOSF) values were <0.01 for all three Bayesian datasets.
Morphological analyses. Numbers of tentacles, polyp coloration, oral disk coloration, relative tentacle lengths, and polyp dimensions (oral disk diameter/polyp height) were examined using in situ images. Additionally, the relative development of the coenenchyme was examined using a dissecting microscope. Coenenchyme development was classified as 1) "highly developed coenenchyme" when polyps covered the antipatharian substrate completely, or 2) "poorly developed coenenchyme" when polyps did not completely cover the antipatharian substrate and the antipatharians were clearly visible. For internal morphological analyses, we observed mesentery arrangement and numbers, and location and shape of marginal muscle. Histological sections of 8 µm thickness were made and stained with hematoxylin and eosin after decalcification with Bouin's fluid for 24h.

Abbreviations
Remarks. Four of five formally described species grow mainly on antipatharians, but this character is not exclusive to all species in the genus as A. obscurus sp. n. is not associated with any host organism. Results of the current study showed that A. obscurus sp. n. is clearly placed within this genus according to COI and 16S-rDNA sequence analyses. Thus, these non-associated species/specimens are within the genus based on their phylogenetic position but do not fit the original definition of the genus by Sinniger et al. (2010). Diagnosis. External morphology: Open oral disks are approximately 5-10 mm in diameter, and polyps approximately 5-10 mm in height when open (Fig. 2). Polyps of a single colony are usually connected by a stolon forming a mesh-like network. Antipathozoanthus obscurus sp. n. has approximately 26-32 bright brown and/or orange tentacles that are as long as or longer than oral disk diameter. Polyps and coenenchyme have a heavily encrusted ectoderm including numerous various sand particles (usually 1 to 8 mm in size). Capitular ridges (= number of complete mesenteries) are slightly visible on tops (= capitulum) of closed polyps.
Internal morphology: Azooxanthellate. Fine sand particles and silica heavily encrusted into ectoderm and mesoglea. We could not obtain cross-sections or images to observe internal morphology such as mesenterial arrangement, marginal muscle or siphonoglyph due to heavy sand and silica encrustation.
Habitat and distribution. Antipathozoanthus obscurus sp. n. is found in low-light environments such as within crevasses of reef slopes and reef floors, and coral reef caves. Specimens were found from 3 to 15 m. This species has been found from the Red Sea and Okinawa.
Differential diagnosis. Antipathozoanthus obscurus sp. n. is easily distinguished from all other Antipathozoanthus species, including the two other new species in this study, which all have associations with antipatharians. A. obscurus sp. n. is not associated with antipatharians and instead is found on coral reef carbonate substrate within caves or cracks. Additionally, the cnidome of A. obscurus sp. n. is different from all other known Antipathozoanthus species, including the other new species in this study, as there are no medium holotrichs in any tissue of A. obscurus sp. n., and instead only large holotrichs are found in all tissues. Although A. obscurus sp. n. is not associated with antipatharians, phylogenetic data indicate that A. obscurus sp. n. is very closely related to other Antipathozoanthus species associated with antipatharians, with identical COI and 16S-rDNA sequences to those of A. macronesicus (EU591618).
Remarks. The samples of Antipathozoanthus obscurus sp. n. in the present study contain two morphotypes; one with bright brown tentacles that are longer than the oral disk (MISE-TF54); and the other morphotype with orange tentacles that are only as long as the oral disk (MISE-BISE1, MISE-BISE3, MISE-JDR190, MISE-JDR191, MISE-JDR192, MISE-JDR279, MISE-KU1, MISE-TF78, MISE-TF148). However, the sequences of all specimens formed a monophyletic clade and therefore we have described A. obscurus sp. n. in this study as containing two morphotypes. Genetic vari- Table 2. Cnidae types and sizes observed in three new Antipathozoanthus species. Frequency: relative abundance of cnidae type in decreasing order; numerous, common, occasional, rare (n = number of cnidae). Occasional ation in all three genetic markers in the samples of A. obscurus sp. n. was observed, and the possibility remains that A. obscurus sp. n. may contain cryptic species. Thus, we have excluded specimen MISE-BISE3 from the type series, although it was tentatively identified as A. obscurus sp. n. Further specimens and fine-scale genetic analyses are required to better understand if there is any cryptic diversity within this species.
Etymology. Antipathozoanthus obscurus sp. n. is named from the Latin "obscura" meaning "dark", as this species can be found in dark environments.
Habitat and distribution. Antipathozoanthus remengesaui sp. n. has been found on the sides and/or floors of cave entrance, and always on Antipathes. Specimens were collected from depths of 9 to 40 m. This species is known from Palau, Kagoshima in Japan, the Maldives, and the Red Sea.
Differential diagnosis. In the Pacific, Antipathozoanthus remengesaui sp. n. can be distinguished from A. hickmani by the development of the coenenchyme and in part by polyp size; the larger polyps (4-12 mm in diameter and 4-15 mm in height) of A. hickmani are connected by a well-developed coenenchyme on Antipathes galapagensis, while the slightly smaller polyps (4-8 mm in diameter and 3-8 mm in height in situ) of A. remengesaui sp. n. are either connected by a poorly developed coenenchyme or may even be solitary on Antipathes. Additionally, the cnidomes of these species are different; A. hickmani does not have spirocysts in the column, while A. remengesaui sp. n. has spirocysts in the column.

Remarks.
The Antipathozoanthus remengesaui sp. n. specimens found in Kagoshima, Japan have different morphological features compared to the specimens found in all other regions. Specimens collected from Kagoshima have relatively large polyps (6-8 mm in diameter, and approximately 5-8 mm in height in situ) compared to specimens from other regions. The coloration of oral disks is also different between Kagoshima and other regions; A. remengesaui sp. n. from Kagoshima has a bright brown oral disk, while those from other regions have pink oral disks. However, sequences of these specimens collected from all regions formed a monophyletic clade for all genetic markers including ITS-rDNA. In terms of substrate organisms, A. remengesaui sp. n. collected from all regions in this study was associated with black corals of the genus Antipathes. Here, we have described this group as a single species, A. remengesaui sp. n., based on phylogeny and substrate specificity, although we have excluded some specimens for which we could not amplify ITS-rDNA successfully from the type series.
Etymology. Antipathozoanthus remengesaui sp. n. is named after Tommy Esang Remengesau, Jr., the current president of the Republic of Palau, who has greatly contributed to marine research and conservation in Palau.
Common name. Momoiro-mame-tsuno-sunaginchaku (new Japanese name). Diagnosis. External morphology: Polyps in situ are approximately 4-15 mm in diameter when oral disk is expanded, and approximately 3-10 mm in height (Figure 2). Colonial zoantharian with polyps connected by highly developed coenenchyme on Myripathes. Antipathozoanthus cavernus sp. n. has approximately 32-40 translucent tentacles of approximately 1 to 5 mm in length. Tentacle lengths are either as long as or slightly shorter than expanded oral disk diameter. Polyps have orange oral disk with orange or light orange ring around oral disk. When polyps are closed, capitular ridges are present and observed clearly, numbering approximately 16-20. The capitulum is orange or light orange in color. Polyps encrusted with visible sand particles (1-8 mm) in their coenenchyme and ectodermal tissue. Polyps usually much more encrusted than coenenchyme. Colonies attached on axis from proximal extremity to base of Myripathes.
Habitat and distribution. Antipathozoanthus cavernus sp. n. is found on the sides and/or floor of cave entrances, and on steep slopes, and always on Myripathes. Specimens were collected from depths of 19 to 39 m.
Differential diagnosis. Antipathozoanthus cavernus sp. n. occurs in similar environments as A. remengesaui sp. n., but these species can be distinguished by their coenenchyme development and by the generic identity of the antipatharian host. A. remengesaui sp. n. is associated with genus Antipathes (family Antipathidae) covered by a poorly developed coenenchyme, while A. cavernus sp. n. is associated with genus Myripathes (family Myripathidae) covered by a highly developed coenenchyme. A. cavernus sp. n. can be distinguished from A. hickmani by a different coloration and by its antipatharian association; A. cavernus sp. n. does not have red or cream colored polyps as seen in A. hickmani. Additionally, A. hickmani is associated with Antipathes galapagensis, while A. cavernus sp. n. is associated with genus Myripathes. A. macaronesicus is easily distinguishable from A. cavernus sp. n. by their polyp coloration (orange and light orange versus pinkish and yellowish, and their antipatharian host (genus Antipathes versus genus Myripathes). Finally, all species above have unique ITS-rDNA sequences.
Etymology. Antipathozoanthus cavernus sp. n. is named from the Latin "caverna" meaning "cave", as this species is found in caves.

Phylogenetic analyses
Concatenated alignment. All Antipathozoanthus species together formed a large monophyletic clade within the Parazoanthidae with complete support (ML = 100%, BI = 1) in the concatenated (COI+16S-rDNA+ITS-rDNA) alignment phylogeny (Fig. 5). Within the Antipathozoanthus clade, the various Antipathozoanthus species were divided into two subclades, an 'associated' subclade consisting of species associated with antipatharians, and a 'non-associated' subclade consisting only of A. obscurus sp. n. found directly on non-biotic substrates.  A. hickmani and A. cavernus sp. n.). A. macaronesicus formed a subclade with very strong support (ML = 97%, BI = 1). COI. All Antipathozoanthus species formed a large monophyletic clade within the Paraozoanthidae with a very strong support (NJ = 99%, ML = 99%, BI = 1) in the COI phylogeny (Suppl. material 2). Within the clade, Antipathozoanthus species were divided into two subclades (associated subclade + non-associated subclade). The topology within the large monophyletic associated subclade was very similar to that as seen in the 16Sr-DNA phylogeny. Both the associated subclade (NJ =77%, ML = 66%) and the non-associated subclade had moderate support (NJ = 86%, ML = 85%, BI = 0.99). Sequences of Antipathozoanthus species within each of the subclades showed no differences in sequences. The difference in sequences between the associated subclade and the non-associated subclade was 3 bp (0.69%).
16S-rDNA. All Antipathozoanthus species formed a large monophyletic clade within the Parazoanthidae with generally high support (NJ = 99%; ML = 85%; BI = 1) in the 16Sr-DNA phylogeny (Suppl. material 3). Within this large clade, Antipathozoanthus species were divided into two subclades; an associated subclade (A. macaronesicus, A. hickmani, A. remengesaui sp. n., and A. cavernus sp. n.); and the other subclade not associated with antipatharians (A. obscurus. sp. n.; 'non-associated subclade'). The associated subclade formed only in NJ phylogenetic tree with moderate support in the 16S-rDNA tree (NJ = 78%), while the non-associated subclade had strong support in each phylogeny (NJ = 97%; ML = 95%; BI = 0.96). Sequences of Antipathozoanthus species within the associated subclade were identical with the exception of A. hickmani (EU333757), which differed by one base substitution, while within the non-associated subclade there were a few small sequence differences (maximum difference 3 bp). Differences of sequences between the associated and the non-associated subclades were 4-6 bp (0.67 to 1.01%).

Discussion
Distribution of Antipathozoanthus species in the Indo-Pacific Ocean. Antipathozoanthus hickmani is found in only the Galapagos with A. cf. hickmani reported from the coast of Ecuador (Bo et al. 2012), suggesting an East Pacific distribution. On the other hand, A. remengesaui sp. n. was found in the Red Sea, the Maldives, Palau, and mainland Japan and Okinawa, Japan, while A. cavernus sp. n. was found in the Maldives, Palau, and mainland Japan, and A. obscurus sp. n. was found in the Red Sea and Okinawa, Japan. Additionally, unidentified Antipathozoanthus species have been previously reported from the central Indo-Pacific Ocean (Reimer et al. 2014a), the South China Sea , and mainland Japan (Reimer et al. 2013). These results indicate that the three new Antipathozoanthus species described herein are likely widely distributed across the Indo-Pacific Ocean, and also that Antipathozoanthus species diversity is higher than has been previously known.
Evolution of macrocnemic zoantharians in caves. Antipathozoanthus obscurus sp. n. without host was found in similar environments as the 'associated' Antipathozoanthus species, but this species does not associate with antipatharians and is instead directly attached to coral reef carbonate. Ocaña and Brito (2003) explained the relationship between Antipathozoanthus and antipatharians as a case of facultative parasitism, although this association still requires further research. It has been revealed that some macrocnemic species gain an advantage in plankton feeding by utilizing substrate organisms that filter feed in environments where plankton organisms are scarce (e.g., Hydrozoanthus species on oligotrophic coral reefs; Di Camillo et al. 2010), and this could be one reason that most Antipathozoanthus spp. utilize antipatharians as substrate. However, moderate currents conducive to plankton feeding may occur in coral reef caves by inflow of tidal currents or terrestrial runoff (Iliffe and Kornicker 2009), and it may be unnecessary to have an association with antipatharians for obtaining sufficient plankton in such environments. Additionally, in marine caves, there are fewer predators of zoantharians, such as fishes (e.g., Bussotti et al. 2002), and perhaps fewer competitors for substrate space. Such environments may promote the speciation of 'non-associated' zoantharian species as seen here with A. obscurus sp. n.
All new species in the present study are azooxanthellate, and this trait is common within macrocnemic zoantharians to the exception of some species such as Bergia cutressi (West, 1979) and Nanozoanthus harenaceus Fujii & Reimer, 2013. Irei et al. (2015 suggested that cave-dwelling Palythoa species within Brachycnemina lost their zooxanthellae to adapt to environments in caves and cracks. On the other hand, macrocnemic cave-dwelling species may originally have lacked zooxanthellae rather than undergone a loss of zooxanthellae. However, more investigations are needed to evaluate the species diversity of zoantharians in caves to more comprehensively understand the evolution of these zoantharian species.
Substrate specificity within Antipathozoanthus. Within the family Parazoanthidae, different generic lineages likely have long evolutionary histories associated with their substrate organisms, based on the fact that many parazoanthid genera form monophyletic clades in accord to their substrates (Sinniger et al. 2005(Sinniger et al. , 2010Montenegro et al. 2015). In this study, we found two different subclades within Antipathozoanthus (Fig. 5, Suppl. Materials 2-4) that corresponded to substrate differences. The genetic distances between the associated subclade (A. hickmani, A. macaronesicus, A. remengesaui sp. n., A. cavernus sp. n.) and the non-associated subclade of A. obscurus sp. n. were 0.60% (COI) to 1.01% (16S-rDNA). Additionally, we observed characteristic insertions and deletions in the 16S-rDNA between the associated and non-associated subclades. Although the two clades formed in accordance to their substrate (antipatharians, coral reef carbonate), we consider the genetic distances between the two clades as intra-generic based on previous comparisons of genetic distances (Sinniger et al. 2010). While many taxonomic and molecular studies focusing on the family Parazoanthidae have been conducted using various genetic markers (e.g., Sinniger et al. 2005Sinniger et al. , 2010Reimer et al. 2008;Montenegro et al. 2015), little research has been conducted focusing on the phylogenetic relations within different parazoanthid genera, except for studies examining Bergia, Parazoanthus, and Umimayanthus, which are all associated with sponges (Sinniger et al. 2005;Montenegro et al. 2015Montenegro et al. , 2016Carreiro-Silva et al. 2017). There is a need for more phylogenetic studies focusing on increasing the numbers of species examined within each of the genera of Parazoanthidae in order to better understand the evolutionary history of substrate specificity and other traits within the family Parazoanthidae (Swain et al. 2015(Swain et al. , 2016.