Fungia fungites (Linnaeus, 1758) (Scleractinia, Fungiidae) is a species complex that conceals large phenotypic variation and a previously unrecognized genus

Recent molecular phylogenetic analyses of scleractinian corals have resulted in the discovery of cryptic lineages. To understand species diversity in corals, these lineages need to be taxonomically defined. In the present study, we report the discovery of a distinct lineage obscured by the traditional morphological variation of Fungia fungites . This taxon exists as two distinct morphs: attached and unattached. Molecular phylogenetic analyses using mitochondrial COI and nuclear ITS markers as well as morphological comparisons were performed to clarify their phylogenetic relationships and taxonomic positions. Molecular data revealed that F. fungites consists of two genetically distinct clades (A and B). Clade A is sister to a lineage including Danafungia scruposa and Halomitra pileus , while clade B formed an independent lineage genetically distant from these three species. The two morphs were also found to be included in both clades, although the attached morph was predominantly found in clade A. Morphologically, both clades were statistically different in density of septal dentation, septal number, and septal teeth shape. These results indicate that F. fungites as presently recognized is actually a species complex including at least two species. After checking type specimens, we conclude that specimens in clade A represent true F. fungites with two morphs (unattached and attached) and that all of those in clade B represent an unknown species and genus comprising an unattached morph with only one exception. These findings suggest that more unrecognized taxa with hitherto unnoticed morphological differences can be present among scleractinian corals.


Introduction
Over the last two decades, molecular phylogenetic and subsequent morphological analyses have been applied to scleractinian corals (Cnidaria: Anthozoa) to infer phylogenetic relationships and to revise their taxonomy (Fukami et al., 2008;Budd et al., 2012;Huang et al., 2014a, b;Kitahara et al., 2016). For example, within the family Lobophylliidae Dai & Horng, 2009, molecular data showed that various genera were polyphyletic (Arrigoni et al., 2014a(Arrigoni et al., , b, 2015, conflicting with traditional morphology-based taxonomy. As a result of the search for morphological characters that reflect molecular phylogeny, several species and genera have been newly described taxonomically or resurrected (Arrigoni et al., 2015(Arrigoni et al., , 2016a(Arrigoni et al., , b, 2019Huang et al., 2016;Benzoni et al., 2018). In the family Fungiidae Dana, 1846, the taxonomy of 26 species were revised based primarily on molecular phylogenetic data (Gittenberger et al., 2011).
Hence, integrated analyses combining molecular and morphological data enable coral specialists to infer taxonomic positions more precisely and to find hidden species or cryptic lineages among corals. However, it is difficult to find specific morphological characteristics of hidden species or cryptic lineages in order to separate them from closely related species. One reason for this is due to colony formation, a trait typical of many corals that leads to large morphological variation among individual corallites (the cup-like skeletal structures of polyps) within a colony, and also between colonies. Such morphological variation can be caused by different environmental factors (Todd, 2008;Chen et al., 2011) or differences in genotypes (Carlon & Budd, 2002), eventually resulting in larger intraspecific variation. In order to solve this problem of detecting new morphological differences among closely related species, micromorphological analysis using scanning electron microscopy has been applied as aid in recent taxonomic revisions of corals (Gittenberger et al., 2011;Budd et al., 2012;Huang et al., 2014a, b;Arrigoni et al., 2014aArrigoni et al., , b, c, 2015Arrigoni et al., , 2016aArrigoni et al., , b, 2019. Fungia Lamarck, 1801, the type genus of the family Fungiidae, includes only one species, F. fungites, which is usually unattached (freeliving when full-grown) and common on shallow Indo-Pacific reefs. As with most other unattached fungiids (Hoeksema & Gittenberger, 2010;Hoeksema & Waheed, 2012;Hoeksema & Benzoni, 2013;Hoeksema 2014), larvae of this species settle on a solid substratum, remain attached by a stalk at the juvenile (anthocaulus) stage (Hoeksema, 1989), and become unattached in the adult stage (anthocyathus) after detaching a disc with a diameter of less than 50 mm from the stalk (Goffredo & Chadwick-Furman, 2003;Gilmour, 2004). However, a unique characteristic only in F. fungites, an attached morph (remaining attached with a disc of more than 50 mm in a diameter), has been reported in Thailand and Japan. In Thailand, Hoeksema & Yeemin (2011) reported that it remained attached with a disc up to 125 mm in diameter. In Japan, Nishihira & Veron (1995) also found the attached morph, but they considered it a different species, "Fungia sp. (Sessile)". In order to uncover whether the two morphs (the attached and unattached morphs) of F. fungites are separate species, and to determine whether F. fungites comprise cryptic lineages, we collected specimens of both morphs in Japan. We investigated their molecular phylogenetic positions, and studied micromorphological skeletal characters. While we found that the two morphs reflect intraspecific variation of F. fungites, we also discovered one likely new species, which is morphologically closely related to but genetically distant from the true F. fungites.

Sampling and species identification
Two morphs (attached and unattached morphs) of F. fungites were collected by SCUBA diving on reefs at six islands of the Nansei Island group, southern Japan ( fig. 1). Depth of each specimen was also recorded. In Aka Is. and Iriomote Is., corals were collected with the permissions of the governor of Okinawa Prefecture (permission numbers 24-60, 31-53). After sampling, a fragment (<1 cm3) of each specimen was preserved in CHAOS solution (4M guanidine thiocyanate, 0.1% N-lauroyl sarcosine sodium, 10mM Tris-HCl pH 8, 0.1M 2-mercaptoethanol; Fukami et al., 2004) for DNA analysis, and the remnant samples were bleached for morphological analysis. In addition, we also collected specimens of Danafungia scruposa (Klunzinger, 1879) and Halomitra pileus (Linnaeus, 1758), which are known to be closely related to F. fungites (Gittenberger et al., 2011;Oku et al., 2017), and Lobactis scutaria (Lamarck, 1801) as outgroup. All specimens were identified at species level, Figure 1 Map of the sampling sites.

Molecular phylogenetic analysis
Total DNA from each specimen was extracted from tissue dissolved in CHAOS solution, using a conventional phenol/chloroform extraction method. The barcoding portion of the mitochondrial COI, and ITS of the nuclear ribosomal DNA (including partial 18S, ITS-1, 5.8S, ITS-2, and partial 28S) were amplified using polymerase chain reaction (PCR) with the primers COI mod F and R (Gittenberger et al., 2011) for COI, and primers 1S and 2SS (Wei et al., 2006) for ITS. PCR conditions described by Oku et al. (2017) were used in this study. The DNA sequences were determined by direct sequencing using ABI3730 sequencers (Applied Biosystems, Alameda, California, USA). All the DNA sequences obtained in the present study were submitted to DNA Data Bank of Japan, DDBJ (accession Nos. LC484501-LC484628). DNA sequences were aligned with Sequencher ver. 5.1 (Gene Codes, Ann Arbor, MI, USA). Phylogenetic trees were reconstructed using the neighbor-joining (NJ) and maximum-likelihood (ML) methods. For the NJ and ML, we assumed a model of nucleotide evolution obtained using MEGA ver. 7.0 (Kumar et al., 2016). The most appropriate models of nucleotide evolution were the Hasegawa-Kishino-Yano model for the COI marker, and Jukes-Cantor model with gamma distribution (G) for the ITS marker. MEGA was used to estimate the topologies for each marker and to conduct a bootstrap analysis (with 1000 replicates). For both COI and ITS trees, we used as outgroup L. scutaria, which is phylogenetically closest to our target species (Gittenberger et al., 2011;Oku et al., 2017). We also concatenated both markers and performed an analysis along with available DNA data to confirm the phylogenetic position of F. fungites within Fungiidae. These sequences were obtained from three previous studies (Fukami et al., 2008;Gittenberger et al., 2011;Oku et al., 2017), and accession numbers are included in supplementary fig. S1.

Morphological analysis
To investigate the morphological differences of two morphs in F. fungites, we first classed them into two growth stages -immature (juvenile) and mature (full-grown) -because corals in the immature stage usually exhibit atypical morphology (Baird and Babcock, 2000;Babcock et al., 2003). We defined the immature stage as having a diameter of less than 50 mm, because F. fungites typically detaches itself from the substrate when reaching approximately this size (Goffredo & Chadwick-Furman, 2003;Gilmour, 2004). The mature stage for both morphs was defined as having a diameter of 50 mm or more. We examined corallum diameter, number of septa, and density of septal dentation and costal spines for all specimens (fig. 2) using a digital microscope (VHX-1000, Keyence). In addition to these morphological skeleton exanimations, we examined the micromorphological characters of septal teeth and septal side with scanning electron microscopy (SEM) using TM-1000 (Hitachi High-Technologies Corp., Tokyo, Japan). To avoid measurement bias for density of septal dentation (teeth), we randomly selected five septa from all septa reaching around the mouth and counted the number of septal teeth within 1 cm of the middle part of each selected septum ( fig. 2b). Similarly, to assess the density of costal spines, we randomly selected five out of all costae and counted the number of costal spines within 1 cm of the middle part of each selected costa ( fig. 2b). For these characteristics, the mean values of specimens were calculated from five replicates. The Kruskal-Wallis test was used to test whether density of septal dentation and costal spines were significantly different between three groups (two morphs and immature specimens). Non-parametric pairwise analyses were done using the Steel-Dwass test. Finally, significant differences between two samples for morphological characteristics were tested using the Mann-Whitney U test. Statistical tests were performed using R ver. 3.5.3 (R Core Team, 2019).

Molecular analysis
For COI analysis, DNA sequences of immature specimens (12 specimens), and two morphs (11 for attached, 31 for unattached) of F. fungites were obtained, in addition to those of D. scruposa (five specimens) and H. pileus (4) ( For ITS analysis, DNA sequences of 54 specimens (12 specimens for immature, 11 for attached, 31 for unattached) of F. fungites were obtained, in addition to those of five specimens of D. scruposa and two specimens of H. pileus (table 1). We obtained 937 positions, including 40 polymorphic sites with 24 parsimony-informative sites, and all indels were deleted from the analysis. An ITS phylogenetic tree showed a topology similar to that of the COI tree, in which F. fungites was divided into genetically distant clades ( fig. 4). Overall, bootstrap values of the ITS tree were lower than those of the COI tree. Danafungia scruposa and H. pileus were phylogenetically positioned between two clades A and B, forming sister clades with clade A, as in the COI tree.
For concatenated COI-ITS analysis, we obtained 1,087 positions, including 178 polymorphic sites with 105 parsimony-informative sites, and all indels were deleted from the analysis. In this tree, F. fungites was also divided into two distant clades. The DNA sequence of one sample of F. fungites used in Gittenberger et al. (2011) was in clade B (supplementary fig. S1).

Morphological comparison
We focused on the comparison between clades, and compared morphological data of specimens between clades A and B, because the two morphs were included in both clades. Morphological data of the two morphs and immature specimens from each clade are summarized in table 2. Three morphological differences were observed in the specimens (including immature and two morphs) between clades A and B. The first was density of septal dentations, which appeared to be the most useful characteristic for distinguishing between clades. It was significantly different (Mann-Whitney U test: U = 12.5, N = 54, P < 0.0001) between all specimens of clades A (8-22 teeth per cm) and those of clade B (12-33 teeth per cm) ( fig. 5, table 2), whereas density The second was the number of septa in relation to corallum diameter. The number of septa increased according to increasing corallum size in both clades ( fig. 6), and was significantly higher in clade A (3.17-5.31) than clade B (2.85-4.54) (table 2, Mann-Whitney U test: U = 119, N = 54, P = 0.0002). The third was the shape of septal teeth. In clade A, these were regularly or irregularly angular in immature specimens and the attached morph, and regularly or irregularly lobate and angular in the unattached morph ( fig. 7). In contrast, in clade B, there was fine septal dentation in immature specimens and the attached morph, and angular septal teeth in the unattached morph ( fig. 8). Because of these differences between clades, septal teeth look coarser in clade A than in clade B.
To clarify the morphological differences in growth stages between and within clades, we performed pairwise comparisons for density of septal dentation, which was a major morphological difference between two clades, among the three groups (immature specimens, attached morphs, and unattached morphs). For the attached morph, the density of septal dentation in clade B (22-26 teeth per cm) was higher than those in clade A (9-21 teeth per cm), although we did not test statistically for the attached morph in clade B because there was only one sample.  (table 3). For micromorphology, we could not find clear differences in the septal teeth and septal sides between clades A and B ( fig. 9) because the morphology was too variable even within each clade.

Species complex
We discovered a statistical difference in the density of septal dentation of specimens between the two clades of F. fungites regardless    of morphs and growth stages. Moreover, they were also different in the number of septa per corallum and shape of septal teeth ( fig. 6). These results revealed that F. fungites is a species complex that contained one other species. As shown in the photographs of the neotype, the type specimen of F. fungites has a density of septal dentations of seven  to 12 teeth per cm ( fig. 10). Hoeksema (1989) showed that the intraspecific range in F. fungites for density of septal dentation was 8-25, which is more similar to those of clade A (8-22) than to clade B (12-33). In addition, the septal teeth shape of the neotype is much more similar to that of specimens of clade A ( fig. 7) than that of clade B ( fig. 8). Thus, morphologically, specimens of clade A are identified as true F. fungites.
Our molecular phylogenetic analysis showed that outgroups D. scruposa and H. pileus were genetically more closely related to clade A than clade B. The morphological characteristics of D. scruposa and H. pileus are distinct from both clades A and B of F. fungites. For instance, H. pileus is largely different in colony shape (polystomatous and therefore with a much larger maximum corallum size: > 600 mm) (Hoeksema, 1991) than clades A and B (monostomatous and smaller size: < 310 mm) although the shape of costal spines of H. pileus is similar. The septal teeth of H. pileus are also nearly similar, although more protruding around the mouths, but this cannot be said of its sister species, Halomitra clavator Hoeksema, 1989, which shows club-shaped septal teeth that are more or less uniform, also around the mouths (Hoeksema, 1989;Hoeksema & Gittenberger, 2010). Danafungia scruposa differs from them by showing rudimentary (poorly developed) costal spines on their higher order costa. Furthermore, the shape of costal spines is also different -D. scruposa has spindlier spines whereas specimens in clades A and B have more triangular or club-like spines. In fact, the morphological differences between two genera Danafungia and Fungia consist predominantly of the shape and development of their costal spines.
Hence, based on molecular and morphological data, we conclude that specimens in clade A are true F. fungites, and that those in clade B are of a yet unidentified species belonging to a different genus than Fungia. So far, F. fungites contains over 30 junior synonyms (see Hoeksema, 1989;Hoeksema & Cairns, 2019b). Therefore, to clarify whether this unidentified species, Fungiidae sp., has been described previously, we need to check all of the type specimens of those synonyms, which will be done in another paper with more detailed morphological comparisons.
The COI-ITS (supplementary fig. S1) tree showed that "F. fungites" (one sample from Indonesia) used in Gittenberger et al. (2011) was included in clade B. We also confirmed that the specimen had the typical morphological characteristics of Fungiidae sp. (clade B). Thus, this result suggests that Fungiidae sp. could be widely distributed in the western Pacific. Table 3 Pairwise comparison of density of septal dentation between immature type and two morphs

Morphs
Our molecular data showed that two morphs (attached and unattached morphs) were observed in both clades (i.e., two species), indicating that these two morphs represent intraspecific phenotypic differences. Although the two morphs result from intraspecific variation, the proportions of both morphs were different in each clade. The full-grown attached morph of clade B (Fungiidae sp.) was a single specimen with a diameter of 53.2 mm, which is nearly immature in size (less than 50 mm). In contrast, for clade A (F. fungites), 10 specimens of the full-grown attached morph were included, in which four specimens were over 70 mm in diameter. Considering these results, the attached morph most commonly found in the field would be F. fungites. "Fungia sp. (Sessile)" was the unidentified species reported for an attached morph in Nishihira & Veron (1995). In verifying the morphological characteristics based on photographs of "F. sp. (Sessile)" shown in Nishihira & Veron (1995), we identified it as the full-grown attached morph of clade A (F. fungites). This identification is also supported by the fact that the shape of septal teeth is lobate and the septal face looks course, although the exact number of septa could not be counted from the photos. This is also consistent with the identification for the large, attached specimens with late detachment that were reported from the Gulf of Thailand (Hoeksema & Yeemin, 2011).

Ecological features of two species
In general, immature specimens of unattached morphs dissolve their stalk during growth in order to detach more easily from the substrate (Yamashiro & Yamazato 1996;Hoeksema & Yeemin 2011;Hoeksema & Waheed, 2012). Therefore, the existence of the attached morph in both F. fungites and Fungiidae sp. could be caused by the delay of such a skeletal-dissolving mechanism. At this time, we do not know the mechanism but the attached morph looks like a neotenic characteristic because it retains the same form as the anthocaulus stage (= immature). The occurrence of the character states of attached vs. unattached in full grown mushroom corals used to be distinctive at genus level (Wells, 1966;Cairns, 1984;Hoeksema, 1989Hoeksema, , 2009), but since the application of molecular methods this distinction has only remained at species level (Gittenberger et al., 2011;Benzoni et al., 2012). The present study shows that this distinction has also become less clear within a single species.

Conclusion
The present study reveals that F. fungites has large phenotypic variation, including attached and unattached forms. This kind of intraspecific morphological variation has never been reported for mushroom corals but is not unique among scleractinian corals since the opposite pattern-an unattached form shown by an otherwise attached species-has been observed (Hoeksema, 2012;Hoeksema & Wirtz, 2013). In mushroom corals, studies of intraspecific morphological variation with molecular data are limited (Hoeksema & Moka, 1989;Hoeksema, 1993;Gittenberger & Hoeksema, 2006), but are important for understanding the complexity of their morphology. We also found that F. fungites is a species complex including one more species (Fungiidae sp.) belonging to a different genus. We are now describing that taxon as a new genus and a new species. We have also demonstrated the utility of molecular phylogenetic analysis using COI and ITS for the exploration of species complexes. Although new species of scleractinian corals have been discovered recently by more detailed phylogenetic analysis using four or more markers (Arrigoni et al., 2016a(Arrigoni et al., , b, 2019 or microsatellite loci (Warner et al., 2015), it would be possible to explore for species complexes at relatively low cost using only these two markers. We expect this simple method of analysis to emerge as the primary method used in the search for species complexes among scleractinian corals.