High species diversity of the soft coral family Xeniidae (Octocorallia, Alcyonacea) in the temperate region of Japan revealed by morphological and molecular analyses

Abstract The soft coral family Xeniidae, commonly found in tropical and subtropical regions, consists of 20 genera and 162 species. To date, few studies on this family have been conducted in Japan, especially at higher latitudes. Although molecular phylogenetic analyses have recently been used to distinguish soft coral species, it is difficult to identify species and genera in this family due to the limited taxonomic indices and high morphological variation. In this study, we found a large Xeniidae community off the coast of Oshima Island (31°31.35'N, 131°24.27'E) at Miyazaki, Kyushu Island, located in the temperate region of Japan. The species composition and molecular phylogenetic relationships were investigated to uncover the species diversity of Xeniidae in this community. A total of 182 xeniid specimens were collected and identified to the species level, after which the samples were molecularly analyzed using a mitochondrial marker (ND2) and a nuclear marker (ITS) to infer the phylogenetic relationships. A total of 14 xeniid species were identified, including five undescribed species from five genera (Anthelia, Heteroxenia, Sympodium, Xenia, and Yamazatum). Miyazaki was identified as having the highest xeniid species diversity in Japan. The molecular phylogenetic trees inferred from each marker recovered very similar topologies: four genera (Anthelia, Heteroxenia, Sympodium, and Yamazatum) were monophyletic, whereas one (Xenia) was polyphyletic. Thus, except for Xenia, the morphological characteristics used for traditional taxonomy well reflected the phylogeny of the Xeniidae at the genus level. On the other hand, our results show that further taxonomic revisions of Xenia are needed.


Introduction
Tropical marine animals, including zooxanthellate alcyonacean corals (i.e., soft corals) abound in the southern part of the temperate region of Japan, due to the Kuroshio -a strong warm current running along the coast from the Ryukyu Archipelago to the mainland of Japan. However, studies looking into the zooxanthellate alcyonacean corals in Japan are limited, especially regarding the family Xeniidae Ehrenberg, 1828. Xeniidae comprises 20 genera and 162 species (Cordeiro et al. 2019), and is distributed mainly across the Red Sea and the Indian and Pacific Oceans. Additionally, a few species have been found in the south Atlantic (Kükenthal 1906) and Norwegian Sea (Koren and Danielssen 1883;Danielssen 1887;Grieg 1887;Jungersen 1892). Although 21 species from seven genera in this family have been recorded in Japan (Utinomi 1950(Utinomi , 1955(Utinomi , 1958Imahara 1996;Benayahu 2010), its current species diversity remains unknown due to the lack of recent surveys.
The present study describes a large community of xeniids found around Oshima Island (31°31.35'N, 131°24.27'E) at Miyazaki, Kyushu Island. As xeniids are uncommon in Japan, this is an unusual community. Due to this area's higher latitude, coral reef structures are usually not formed, but there are over 100 zooxanthellate scleractinian coral species (Nishihira and Veron 1995). This area was occupied previously by zooxanthellate scleractinian corals, which were damaged drastically during the 1980s by outbreaks of the coral-eating gastropod Drupella spp. and the crown-of-thorns seastar Acanthaster sp. (Takayama and Shirasaki 1990). Currently, the area is occupied mainly by xeniids, which are known as pioneer alcyonaceans in ecological succession in tropical coral reefs (Benayahu and Loya 1987). Therefore, the ecological context of Oshima Island may represent an initial stage of secondary succession, following the drastic reduction of zooxanthellate scleractinian corals in the coral community. Identifying xeniid species diversity in this area is important to understand how coral communities change over time at higher latitudes.
Species identification difficulties are common among the anthozoans due to their limited key taxonomic characteristics and high morphological variation and plasticity. Recently, molecular phylogenetic analyses have been used to overcome such limitations. In particular, molecular phylogenetic data have been used frequently in scleractinian corals to revise taxonomy, identify cryptic species and describe new species (e.g., Budd et al. 2012;Huang et al. 2014a, b). In the alcyonacean corals, molecular phylogenetic analyses have also been applied to several families (France and Hoover 2002;McFadden and Hutchinson 2004;McFadden et al. 2009 Alderslade, 2000, andSarcothelia Verrill, 1928 were paraphyletic with Xenia (COI, mtMutS, ND2, 28S rDNA). On the other hand, few molecular phylogenetic analyses have been performed in xeniids at the species level.
To date, the molecular data of xeniids indicate that mitochondrial ND2 marker is one of the best markers to infer the phylogenetic relationships among genera within many octocoral families (e.g., McFadden et al. 2006;McFadden et al. 2014), whereas nuclear ITS is a suitable marker to examine the octocorals' species-level relationships. In particular, the ITS marker has been used to investigate the relationships between closely related species in the soft coral genera such as Alcyonium Linnaeus, 1758 and Pseudopterogorgia Kükenthal, 1919(McFadden et al. 2001McFadden and Hutchinson 2004;Sánchez et al. 2007;Dorado and Sánchez 2009). The present study aimed to investigate the current species diversity of the family Xeniidae around Oshima Island, Japan, and to clarify this family's taxonomic issues at the species level, through molecular phylogenetic analyses using ND2 and ITS markers.

Species identification
For species identification, we first summarized the morphological characteristics for all species in the five genera we found in this study (Xenia, Heteroxenia, Sympodium, Yamazatum Benayahu, 2010, and Anthelia) from original descriptions and related references to define the criteria for each species (Suppl. materials 1-5: Tables S1-S5), and used the summary to identify specimens at the species level. Table 1 shows a list of all specimens collected in this study. All specimens are deposited at Miyazaki University, Fisheries Sciences (MUFS) for coral collections (-C). Regarding specimen identification, the following morphological characteristics were measured or counted under stereo microscope: colony height, length and width of stalk, presence of branches, length and width of polyp, length and width of tentacle, length and width of pinnule, number of rows of pinnules, number of pinnules in the aboral row, sclerites form and sclerites size. In addition, microstructure of sclerites was observed by scanning electron microscope (SEM) (HITACHI Tabletop Microscope TM1000) as this morphological trait has been used recently to separate xeniid species (Janes and Mary 2012).

DNA extraction, amplification, and sequencing
Tissue samples were kept in CHAOS solution for at least a week to dissolve proteins at room temperature. Total DNA was extracted from the CHAOS solution with tissue samples by conventional phenol/chloroform extraction method. We used the primers reported by McFadden et al. (2006) to amplify a fragment 5' end of the mitochondrial NADH-dehydrogenase subunit 2 gene (ND2) (16S647F: 5' -ACA CAG CTC GGT TTC TAT CTA CCA-3'; ND21418R: 5' -ACA TCG GGA GCC CAC ATA-3'). We also used two primers (1S: 5'-GGT ACC CTT TGT ACA CAC CGC CCG TCG CT-3'; 2SS: 5'-GCT TTG GGC GGC AGT CCC AAG CAA CCC GAC TC-3') (Wei et al. 2006) to amplify the internal transcribed spacer (ITS) of the nuclear riboso-mal RNA gene. All PCR reactions contained 1 μL of DNA solution, 1.6 μL of 2.5 mM dNTP Mixture, 2 μL of 10X Ex Taq buffer, 2 μL of each 10 mM primer, Ex taq (Ta-KaRa) 0.08 μL, and 11.32 μL of sterile distilled water. Amplifications of these markers were performed (GeneQ PCR Thermal Cycler) with the following thermal profile; 35 cycles of 90 sec at 94 °C, 60 sec at 58 °C, 60 sec at 72 °C. Amplified fragments were checked on 1% agarose gel electrophoresis. All the PCR products were subjected to digest excess primers and inactivation of dNTP using Exonuclease I (TaKaRa) and Shrimp Alkaline Phosphatase (TaKaRa). These DNA sequences were determined by ABI3000 using a research contract service (Ltd. FASMAC).

Sequence alignment and construction of phylogenetic trees
MEGA5 (Tamura et al. 2011) was used to manually align all the DNA sequences and to reconstruct phylogenetic trees. All indels were excluded from the analyses. Molecular phylogenetic trees were reconstructed using Neighbor-joining (NJ) method and maximum-likelihood (ML) method with model parameters (ND2: T92 + G, ITS: K2 + G) with 1000 bootstrap replicates. All the DNA sequences we obtained in this study were registered into DDBJ (accession nos. LC467016-LC467147).
In the present study, Xenia sp. 1, X. sp. 2, Yamazatum sp. 1, Sympodium sp. 1 and S. sp. 2 were identified as undescribed species for the following reasons: Xenia sp. 1 shared common morphological characteristics with the genus Xenia, such as the colony shape and the presence of oval sclerities, but presented also with unique needlelike sclerites with many small spines (Fig. 4J), which have never been re- ported in Xenia. Xenia sp. 2 was easily distinguishable from other Xenia species, as it presented many short branches extending from the top of colony, becoming hump-shaped (Fig. 2K). Yamazatum is a monotypic genus containing Y. iubatum Benayahu, 2010 and presenting two specific morphological characteristics: doubleheaded sclerites and a conspicuous crest on the sclerites' surface. Yamazatum sp. 1 presented a crest on the sclerites' surface ( Fig. 4I) but lacked double headed sclerites; in this species the sclerites were found only in the polyps, opposite to Y. iubatum, containing sclerites both in the surface and interior of the stalk layer and in the polyps. Furthermore, this species presents a branching stalk (Fig. 2I), opposite to Y. iubatum, which has a non-branching stalk. Sympodium sp. 1 and S. sp. 2 shared the common morphological characteristics of the genus Sympodium, such as a thin stolon-like sheet and no stalks in colony (Fig. 2G, H). However, both species found in this study presented unique sclerites, which differed from all eight known Sympodium species. Sympodium sp. 1 presented two types of sclerites; one a double headed sclerite, typical from Y. iubatum, located in the polyps (Fig. 4G), and an oval sclerite with protrusions like a mountain range, located on the coenenchyme (Fig. 4G). Sympodium sp. 2 presented disk-shaped sclerites throughout the whole colony, with smooth surfaces and no protrusions (Fig. 4H). Under a light microscope the sclerites of Sympodium sp. 1 were mostly colorless, whereas those of Sympodium sp. 2 were light brown (Fig. 3G, H).

Molecular phylogenetic analyses
From the collected 14 species (78 samples), we obtained 673-707 bases of ND2 and 910-1039 bases of ITS. Molecular phylogenetic trees using the NJ and ML methods showed very similar topologies. Therefore, in this study, only ML trees for each marker are shown (Figs 5, 6). These trees showed that the family Xeniidae was monophyletic in the Alcyonacea, and that the xeniid species were separated into seven clades. Clade I included Xenia plicata and X. sp. 1. Although the ND2 tree showed an absence of genetic differences between these two species (Fig. 5), the ITS tree showed that they were clearly separated from each other (Fig. 6). Clade II included X. kuekenthali and X. novaecaledoniae, and clade III included only one species, Yamazatum. sp. 1. The ND2 tree showed that clade III formed a sister group with clades I and II with Xenia spp., whereas the ITS tree showed that clade III formed a sister group with only clade I. Clade IV contained all three Heteroxenia species (H. cf. elisabethae, H. medioensis, H. minuta). Clade V contained a single species X. sp. 2. Clades VI and VII contained Sympodium spp. and Anthelia spp., respectively. Thus, four genera (Anthelia, Heteroxenia, Sympodium, and Yamazatum) were monophyletic (clades III, IV, VI, VII) whereas Xenia was polyphyletic (clades I, II, V) because clades III and IV with Heteroxenia and Yamazatum were included within clades of Xenia.

Comparison between phylogenetic relationships and morphological characteristics
In the present study, except Xenia, all genera were monophyletic (clades III, IV, VI, VII). Therefore, the synapomorphy reflecting each of the four clades is consistent with the key morphological characteristics for each genus. On the other hand, only Xenia was polyphyletic (clades I, II, and V). Therefore, to determine the synapomorphy for each clade, the morphological characteristics of the species in these three clades were compared. In clade I, including X. plicata and X. sp. 1, the synapomorphy is a colony form 25-40 mm in height and without secondary branches. Clade II, including X. novaecaledoniae and X. kuekenthali, presented a colony form similar to clade I (typical and no secondary branches), but shorter (10-20 mm in height). It is noteworthy that, although the family Xeniidae is taxonomically defined as presenting oval sclerites, both species in clade II lacked sclerites. Clade V, with just X. sp. 2, was characterized by a unique colony form, comprising a stalk measureing about 10 mm high and 20 mm in diameter, and many short branches extending from the top, becoming hump-shaped. This type of colony form has not been reported previously in the genus Xenia.

Ecological succession in temperate coral communities
Alcyonacean corals (soft corals) have been known as pioneers in coral reefs (Benayahu and Loya 1987;Fabricius 1995), as well as negative indicators of the early developmental processes of the zooxanthellate scleractinian corals (Maida et al. 1995(Maida et al. , 2001. Thus, alcyonacean corals play an important role for ecological succession in coral reefs. Around Oshima Island, zooxanthellate scleractinian corals were dominant until the 1980s, probably representing the late stage of ecological succession in the coral community. Subsequently, these corals were damaged by Drupella spp. and Acanthaster sp. (Takayama and Shirasaki 1990). Currently, many zooxanthellate alcyonacean corals inhabit the top of dead coral skeletons, which may represent the initial stage of the secondary ecological succession in this coral community. In fact, Endean (1976) reported that Alcyonacea attached onto dead coral skeletons after feeding damage by Acanthaster sp. One of the most dominant alcyonacean corals in Oshima Island is Xeniidae, which may be related to its faster growth, rapid colony migration and asexual reproduction (Benayahu and Loya 1985). Although no species diversity data pertaining to hard and soft corals are currently available from the time when hard corals were dominant, the fact that the three-dimensional structures constructed by the zooxanthellate scleractinian corals are gone, suggests that the biota in Oshima Island might have been dramatically different than the present one. Therefore, it would be worthwhile to continuously investigate the change of biota in this area, to understand the process of ecological succession of the benthic and coral community at this higher latitudinal region.

Phylogeny and taxonomy of the Xeniidae
Heteroxenia and Yamazatum were monophyletic, although Xenia were closely related to both genera (Figs 5, 6). Although Heteroxenia presents dimorphic polyps composed of autozooids (normal polyps) and siphonozooids (i.e., no tentacles in polyps, but functional for inhalation and discharge of seawater), siphonozooids only develop when the colony is sexually mature (Gohar 1940;Fabricius and Alderslade 2001). Thus, Heteroxenia and Xenia can only be superficially distinguished during the breeding season, since during the non-breeding season Heteroxenia contains one type of polyp only (autozooids). The present study shows that Xenia and Heteroxenia can be clearly separated in the molecular trees, although some colonies of Heteroxenia were found not to form siphonozooids. These colonies were morphologically identified as Heteroxenia, based on the colony size and shape, the autozooids, the pinnules and the sclerites, despite the occurrence of dimorphic polyps. Although the presence or absence of siphonozooids, an important morphological characteristic for Alcyonacea's generic classification, was confirmed for Xenia and Heteroxenia, molecular phylogenetic analyses of all the 11 species of Heteroxenia are necessary to properly define the taxonomic position of this genus.
In the present study, the phylogenetic position of Yamazatum sp. 1 was ambiguous as this species formed a sister group with clade I in the ND2 tree (Fig. 5), and with both clades I and II in the ITS tree (Fig. 6). Currently, several xeniid genera, including Yamazatum are taxonomically classified based only on sclerite surface microstructure (Bayerxenia Alderslade, 2001;Ingotia Alderslade, 2001;Ixion Alderslade, 2001;Orangaslia Alderslade, 2001;Ovabunda;Fasciclia Janes, 2008;Conglomeratusclera Benayahu et al., 2018;Caementabunda Benayahu et al., 2018;and Yamazatum). Although most of these genera have never been analyzed molecularly, a recent molecular phylogenetic analysis revealed that Ovabunda belonged to the same clade as Xenia (Haverkort-Yeh et al. 2013;McFadden et al. 2014), which, in the present study, is also in the clade of Yamazatum. Therefore, detailed comparisons between molecular data and the sclerite microstructure will be needed for future xeniid taxonomic classification.
Xenia was polyphyletic, particularly due to X. sp. 2 (Figs 5, 6). Clade V with X. sp. 2 was closer to clade IV with Heteroxenia than other Xenia clades (clades I and II). Xenia sp. 2 exhibited slight but substantial differences from its congeners in terms of colony morphology, as their colony shapes lacked branching, exhibiting dome-shaped protrusions (Fig. 7). Considering that Heteroxenia presents specific characteristics that distinguish it from Xenia, such as dimorphic polyps, the species X. sp. 2 may be assigned to a new genus, although this requires further investigations into the morphological characteristics of other genera not observed in present study.
Two undescribed species, S. sp. 1 and S. sp. 2, were found in Sympodinium, and presented different sclerites and microstructure types (Fig. 4) from their congeners. Currently, this genus has only eight species, S. abyssorum Danielssen, 1887, S. caeruleum (Ehrenberg, 1834, S. fuliginosum Ehrenberg, 1834, S. hyalinum Grieg, 1887, S. norvegicum Koren & Danielssen, 1883, S. punctatum May, 1898, S. splendens Thomson & Henderson, 1906 (Cohn, 1908). Their type localities are the Red Sea for S. caeruleum and S. fuliginosum, Norwegian Sea for S. abyssorum, S. hyalinum and S. norvegicum, Indian Ocean S. punctatum and S. splendens, and Madagascar for S. tamatavense. Except for S. caeruleum, all species have never been recorded in the Pacific region, probably due to the lack of research into this genus. Therefore, more species are likely to be found in the Pacific region in the future.
Studies on the species composition and biodiversity of alcyonacean corals have drawn considerably less attention than those on scleractinian corals, since alcyonacean corals do not form the same three-dimensional structures with their hard skeletons as scleractinian corals, and, therefore, provide less habitat for other animals. However, coral communities have been reported to shift from scleractinian corals to alcyonacean corals in the future, if ocean acidification persists (Inoue et al. 2013). Thus, further ecological and taxonomic studies of alcyonacean corals are needed. Although the current taxonomic classification of alcyonacean corals is still underdeveloped, this may be improved by further molecular analyses and accurate species identification will improve this situation.