A total of 25 stoloniferous octocoral specimens (Table 1) was collected from eight locations (Figure 1) in and around TARP, off the coast of Kota Kinabalu, Sabah, Malaysia, between 20 and 27 March 2018. Specimens were collected at depths of 4–18 m by means of SCUBA and all voucher material was preserved in 70–80% ethanol and sub-samples in 95% ethanol. High-resolution in situ images were taken with an Olympus Tough TG-4 in an Olympus PT-056 underwater housing. Vouchers and type material have been deposited at the National Museum of Nature and Science (NSMT), Tokyo, Japan (all holotypes) and at the Borneo Marine Research Institute (IPMB), Universiti Malaysia Sabah (UMS), Sabah, Malaysia (all paratypes).

Figure 1. 

Map of eight sampling locations where stoloniferous octocorals were collected in this study, around Udar and Sepangar Islands and within Tunku Abdul Rahman Park (including Gaya, Sapi, Manukan, Mamutik, and Sulug Islands), Kota Kinabalu, Sabah, Malaysia.

Sclerites were isolated by dissolving different parts of the specimens (polyp tentacle, calyx, entire polyp and stolon) in 4% hypochlorite (household bleach). Rinsing and visualisation of sclerites followed the same protocol as described in Lau et al. (2018, 2019). Additionally, sclerites from specimens that needed more detailed examination were mounted on scanning electron microscope stubs and coated with Pd/Au for imaging on a JEOL JSM6490LV scanning electron microscope (SEM) operated at high vacuum at 15 kV.

DNA was extracted from polyps using a DNeasy Blood and Tissue kit (Qiagen, Tokyo). PCR amplification and sequencing of three mitochondrial markers, cytochrome c oxidase subunit I (COI), mitochondrial mutS-like protein (mtMutS), and the ND6 subunit (ND6), and one nuclear ribosomal marker, 28S ribosomal DNA (28S rDNA) followed the protocols in Lau et al. (2019). The amplified products were visualised by 1% agarose gel electrophoresis. PCR products were treated with Exonuclease I and Alkaline Phosphate (Shrimp) and sent for bidirectional sequencing on an ABI 3730XL (Fasmac, Kanagawa, Japan). Sequences were assembled and edited using Geneious R11 (Kearse et al. 2012) and BioEdit (Hall 1999). COI, mtMutS, and ND6 were checked for introns, exons, and stop-codons in AliView (Larsson 2014).

Multiple sequence alignments were performed using MAFFT 7 (Katoh and Standley 2013) and coding markers were aligned using MACSE (Ranwez et al. 2011) under default parameters. To determine the phylogenetic position of the collected specimens, consensus sequences for markers 28S rDNA, COI and mtMutS were aligned to a reference dataset of 121 octocoral genera (n = 127 sequences), including sequences of Cornularia pabloi McFadden & van Ofwegen, 2012 and Cornularia cornucopiae (Pallas, 1766) as outgroup (total n = 134 sequences). Alignments of 893 bp for 28S rDNA, 717 bp for COI and 923 bp for mtMutS were separately run in Maximum Likelihood (ML) analyses (Suppl. material 1: Figures S1–S3), which were highly congruent with the phylogenetic construction generated with the concatenated three-marker dataset (2533 bp).

A separate phylogenetic analysis was made using a concatenated four-marker dataset to further investigate the phylogenetic position of collected clavulariid and arulid specimens. The four separate markers (28S rDNA, 804 bp; COI, 717 bp; mtMutS, 734 bp; ND6, 441 bp) were also run in ML analyses to check for congruency (Suppl. material 1: Figures S4–S7). The concatenated markers resulted in a 2696 bp dataset with a total of nine reference taxa, including Hanabira yukibana Lau, Stokvis, Imahara & Reimer, 2019 as outgroup (total n = 33).

Alignments of the separate markers were all concatenated using SequenceMatrix 1.8 (Gaurav et al. 2011). ML analyses were run with RAX-ML 8 (Stamatakis 2014) using the GTRCAT model. The best ML tree was calculated using the –D parameter. A multi-parametric bootstrap search was performed, which automatically stopped based on the extended majority rule criterion. The Bayesian inference was performed with ExaBayes 1.5 (Aberer et al. 2014) using the GTR substitution model. Four independent runs were run for 10,000,000 generations during which convergence (with a standard deviation of split frequencies <2%) had been reached. Bootstrap supports and posterior probabilities were depicted on the branches of the best ML tree using P4 (Foster 2004). The resulting tree was visualized in FigTree 1.4.2 (Rambout 2014). Additionally, average distance estimations within species and within genera were computed using MEGA X (Kumar et al. 2018) by analysing pairwise measures of genetic distances (uncorrected P) among sequences (Suppl. material 1: Tables S1–S6).

Sequences of arulids Bunga payung gen. nov. et sp. nov. and Laeta waheedae gen. nov. et sp. nov., collected from the north of TARP (Udar, Gaya and Sepangar Islands) formed a completely-supported clade with the arulid genera Arula and Hana; 100%/1.00 for both the three- and four-marker datasets (Figures 6, 7, respectively). In both phylogenies, Bunga was sister to the remaining clade. However, it remains unresolved how Laeta waheedae, Arula, and Hana are related to one another in the four-marker phylogeny reconstruction (21%/0.56). Nonetheless, all genera in the clade have high support values.

Figure 7. 

Combined 28S rDNA+COI+mtMutS+ND6 phylogenetic reconstruction for four Phenganax species and four arulid genera (Hana, Arula, Bunga gen. nov., and Laeta gen. nov.) collected in and around TARP, Kota Kinabalu, Sabah, Malaysia; Hanabira yukibana was used as outgroup (total n = 33). The best Maximum Likelihood tree is shown, with values at branches representing bootstrap probabilities (shown when > 50%; top/left) and Bayesian posterior probabilities (shown when > 0.50; bottom/right). Photograph credit: in situ images NTM C15597 Phenganax parrini, by Daniel Knop (modified from Alderslade and McFadden 2011, reproduced with permission from copyright holder). Sclerite images RMNH Coel. 40188 Arula petunia (modified from McFadden and Van Ofwegen 2012, reproduced with permission from CSIRO Publishing).

Genetic distances (uncorrected P, expressed as percentage) between specimens of Bunga gen. nov., Laeta gen. nov., Arula, and Hana (between 2.6–4.6% for COI, 5.4–6.1% for mtMutS) indicated that the levels of pairwise differences were well above those observed among congeneric species (McFadden et al. 2011), Suppl. material 1: Table S1.

Morphological data supported the molecular data and justified the description of the two new genera Bunga and Laeta. There were distinct differences in sclerite features between Bunga, Laeta, Arula, and Hana, involving the presence and absence of sclerite types in the calyx; three types of rods (branched with high tuberculate processes, club-like rods and table-radiate-like rods), which were only observed in Laeta but not in the other sister genera, and the absence of 6-radiates in Bunga, which were present in all sister genera. There was also a difference in sclerites of the stolon; Arula is the only genus without fused table-radiates, but instead has table-radiates in the calyx. The polyp morphology did not show many obvious differences, although two of them could be observed in the outer margins of the characteristic fused oral membrane; both Bunga and Laeta had polyps with oral discs that were less pronounced at their outer margins than those of other genera, having a smooth transition from oral disc to the connecting tentacles, while the furrows dividing the oral disc into eight lobes were shallower in both of these genera (Figure 2). The combination of molecular and morphological information supports the erection of two new genera within the Arulidae.

Sequences of 18 clavulariid specimens grouped together with Acrossota amboinensis (Burchardt, 1902) and Phenganax parrini Alderslade & McFadden, 2011 in a well-supported clade in both the three-marker and four-marker analyses (Figures 6, 7, respectively), in which Acrossota was sister to the remaining clade (100%/1.00). However, there was disparity in the relationships within the remaining clade in the two phylogeny reconstructions and the relationships within the remaining clade were not clearly resolved. The four-marker phylogeny reconstruction, with more specimens, gave a higher resolution (Figure 7) and here, each species had high support values, although it remained unclear how P. parrini is related to the other three species (50%/-).

On the other hand, within-genus genetic distances for Phenganax were 0.57% for COI and 3.10% for mtMutS, percentages within the range of congeneric species commonly observed in octocorals (Suppl. material 1: Table S3). When comparing pairwise differences between the four species, the values were well above averages for conspecifics; the observed values were in between 0.98–4.09% for COI and between 1.60–4.07% for mtMutS (Suppl. material 1: Table S5), supporting the division into four species (McFadden et al. 2011).

Despite the absence of sclerites, which was characteristic for the entire clade, there were visible differences in morphological polyp features. The main differences were found in tentacle pinnules and colony growth (Figure 3b–d). Phenganax marumi sp. nov. was similar to P. parrini Alderslade & McFadden, 2011 in polyp morphology, although differing in colour and polyp size; expanded polyp sizes ranged 4.0–5.5 mm in P. marumi and 2.5–3.0 mm in P. parrini. Additionally, the numerous amounts and densely packed polyps described for a P. parrini colony (Alderslade and McFadden 2011) had not been observed in any of the 11 collected P. marumi colonies; instead the maximum number of polyps observed for one colony was approximately 30–40. However, the density of P. parrini could be due to unnatural conditions (aquarium environment) in which it was kept for scientific research (Alderslade and McFadden 2011).

Phenganax marumi sp. nov. was found on exposed spots on the reef, not sheltered from strong current or light, opposed to P. parrini, which was found in dimly lit, sheltered locations. Phenganax subtilis sp. nov. differs from P. parrini and P. marumi mainly in the shape of the pinnules; P. subtilis does not have the plump, diamond-shaped pinnules as seen in P. marumi and P. parrini. As well, the colour of P. subtilis is different from P. parrini and P. marumi, being brown. Phenganax stokvisi sp. nov. shows similarity to the densely packed P. parrini, with polyps being in the same size range (2.5–3.0 mm width), but showing differences in the number of pinnule pairs and colour of the polyps (possibly due to the presence of zooxanthellae).

Differences in morphology combined with genetic distances and high support values in the four-marker phylogeny reconstruction supported the erection of the three new sister species of Phenganax parrini; P. marumi sp. nov., P. subtilis sp. nov., and P. stokvisi sp. nov.

Clavulariidae sp. had a similar DNA profile to two genera within the Clavulariidae; sequences from the two specimens clustered in a well-supported clade (96%/1.00), together with the genera Azoriella (Lopez Gonzalez & Gili, 2001) and Cervera Lopez-Gonzalez et al., 1995, with Cervera as sister to Azoriella and Clavulariidae sp. (Figure 6). Genetic distances between the two specimens and Azoriella were 1.41% for COI and 5.09% for mtMutS, values within ranges amongst intergeneric octocoral species (Suppl. material 1: Table S2). However, with the current available molecular data it remains unresolved how Clavulariidae sp. and Azoriella are related to one another, as branching was weakly supported (52%/0/70).

There are minor differences in polyp morphology when comparing Bunga payung gen. nov. et sp. nov., Laeta waheedae gen. nov. et sp. nov., Arula petunia, and Hana spp.; Arula and Laeta are similar in colour, having blue colorations in the oral disc and/or tentacles. Hana spp. are similar to Arula petunia in having a very pronounced oral disc with deep furrows dividing the oral disc into eight distinct lobes, as opposed to Laeta and Bunga, which both have an oral disc that fuses into the tentacle base and shallow furrows that divides the eight lobes.

It could therefore be proposed that instead of four different genera within Arulidae, Arula, Hana, Bunga gen. nov., and Laeta gen. nov. could be five distinct species of Arula. However, the erection of a new genus for each new species is justified by the combination of sclerite morphology and molecular information. The four genera of Arulidae are distinguished morphologically by differences in sclerite form and the presence or absence of certain sclerite types in different parts of the colony or being absent completely. The only two sclerite types that all four genera have in common are the rods present in the anthocodiae and the table-radiates in the calyx, which are characteristic to the family Arulidae. Bunga is the only genus that has only these sclerites types in the polyps; anthocodial rods and table-radiates in the calyx. Arula, Hana, and Laeta have an additional sclerite type in the calyx; Arula and Hana have 6-radiates. Laeta has three different ornamented rods in the calyx, which are absent in all other genera.

Hana and Arula are the two most similar genera in terms of morphology. However, Arula is the only genus which has separate table-radiate sclerites in the stolon, and it can also be distinguished from the other genera in having table-radiates that are smaller in size (up to 0.09 mm length) and by not having table-radiates which are elongated in shape; the other three genera all have stolons with fused table-radiates which form a sheet and all have larger table-radiates up to 0.10–0.19 mm length.

Molecular analyses of the concatenated four marker dataset (28s rDNA, COI, mtMutS, and ND6 sequences) resulted in similar phylogenetic positions for Bunga payung gen. nov. et sp. nov. and Laeta waheedae gen. nov. et sp. nov. when compared to our analyses of the concatenated three marker dataset (28s rDNA, COI, and mtMutS). It should be noted that the two analyses including more specimens for Bunga and Laeta placed the four genera in slightly different phylogenetic positions. In the three-marker dataset (Suppl. material 1: Figure S8), Bunga and Laeta formed a clade, although it was not very well supported (74/-). This clade was sister to Hana and Arula, which also formed a clade. In the four-marker dataset (Figure 7), Bunga was sister to all remaining clades. Nonetheless, branches of the four-marker dataset had better support values. Therefore, we again recommend the use of the extra gene region (ND6) next to the conventional DNA markers (COI, mtMutS, 28S rDNA) commonly used in octocoral phylogenetic analyses (Lau et al. 2018, 2019).

New discoveries and descriptions of stoloniferous octocorals outside of the Coral Triangle have shown surprising octocoral novelty, at the species, genus, and family levels (Alderslade and McFadden 2011; McFadden and Ofwegen 2012; Benayahu et al. 2017; Lau et al. 2018, 2019). Arulidae, the most recently described family of stoloniferous octocorals (McFadden and Ofwegen 2012), is a good example. Arulids have unique morphological features that had not been seen before in octocorals, including an expanded fused oral disc, and a new type of sclerite, table-radiates. The species Bunga payung gen. nov. et sp. nov. and Laeta waheedae gen. nov. et sp. nov. increase the total number of species within the family Arulidae to five. Moreover, they represent the first confirmed records of arulids in and around the shallow waters of the TARP area in Sabah, Malaysia. The discovery of these two monospecific taxa demonstrates that the recently erected family Arulidae is considerably much more diverse than was originally known. For example, there is photographic evidence of an arulid from the reefs of Bali, Indonesia (McFadden and Ofwegen 2012), and it is possible that other regions in or close to the Coral Triangle harbour unknown stoloniferan diversity.

The distinctive feature of the fused oral membrane seen in arulid polyps and its function is not yet fully understood. It has been proposed by Porter (1976) and Sebens (1979) that an increase in size of the capture surface area of polyps could be positively related to the amount of prey captured (Lasker 1981; Lewis 1982), as has also been suggested for scleractinian corals (Todd et al. 2004; Todd 2008). On the other hand, a large polyp size in scleractinians has been observed to be beneficial for the capture and consumption of large prey items (Alamura et al. 2009; Hoeksema and Waheed 2012; Mehrotra et al. 2014, 2016, 2019; Musco et al. 2018). A study on octocorals by Lasker (1981) showed that there were differences in the number of nematocysts, bigger polyps having relatively less nematocysts, which could influence prey capture. The expanded oral disc seen in arulids could perhaps be an adaptation for feeding and compensation for their relatively small polyp size and lower polyp number per colony, and it would be worthwhile to conduct a study of their nematocysts.

Another compelling hypothesis involves the possibility of these zooxanthellate taxa being heterotrophic, in which an increased light-gathering surface for photosynthesis of symbionts could explain the development of the expanded oral disc. Based on currently available information, the expanded oral disc does not appear to have evolved across the Octocorallia radiation, and is unique to the family Arulidae.

Multiple factors made it initially difficult to establish with certainty if the currently described species were either congeners or possibly already described. The most important factor in this matter is the small number of morphological features that represent the new Phenganax species. The absence of sclerites makes it especially challenging to make confident judgements about evolutionary relationships between species within this genus. Additionally, the available historical literature on described stoloniferous species in the Indo- and northwestern Pacific is in a poor state. There are a number of Clavularia species listed by Alderslade and McFadden (2011) that could possibly belong to Phenganax, mainly due to the fact that these species also lack sclerites and the described distinguishing characters are alike: namely Clavularia reptans Hickson, 1894, Clavularia reptans sensu Thomson & Henderson (1906: 402), Clavularia celebensis Hickson, 1894, Clavularia pregnans Thomson & Henderson, 1906.

The main characteristic morphological difference between the Phenganax spp. described in this study and the above mentioned Clavularia spp. are the “numerous densely packed pinnules”, which are sometimes as many as 30 pinnules per row; all Phenganax spp. in the current study have six to 15 pinnules on either side of the tentacle rachis. Not all measurements (for example expanded polyp width) were included in the Clavularia spp. (Alderslade and McFadden 2011). Clavularia reptans was described as having polyps 7.0–10.0 mm width when expanded, whereas expanded polyp widths of Phenganax spp. are within the size range of 2.5–5.5 mm. Further information regarding the colouration of the tentacles and polyp body in combination with colony drawings make it doubtful that these Clavularia spp. belong to Phenganax. Additional arguments supporting the description of the Phenganax species in this study involve ecological aspects. The preferred habitat of P. parrini is dimly lit and in sheltered locations, below 10 m depth. The described Phenganax spp. in the current study, however, were found at exposed locations above 10 m depth.

Despite the fact that the relationships between the Phenganax spp., with emphasis on the position of P. parrini, seem to be unresolved, and all species lack sclerites to aid in their identification, the described species all had high support values and could be distinguished based on some features in colony growth form and polyp morphology, namely tentacle pinnules, the density of polyps, polyp size and spacing between polyps in a colony.

Similar to the new genera within Arulidae, the results for the Phenganax spp. also demonstrate that more specimens and the extra gene region (ND6) resulted in obtaining higher phylogenetic resolution and increased branch support values, and therefore should represent a more accurate reconstruction of evolutionary relationships. Future octocoral taxonomy will unquestionably build upon a combination of molecular and more traditional morphological techniques (Conti-Jerpe and Freshwater 2017); however, the lack of sclerites has important implications for future studies, as these skeletal parts are one of the major diagnostic features of the group. This emphasizes the need for improved molecular techniques, such as finding an optimum between concatenated marker datasets towards whole-genome mapping, combined with next-generation sequencing to allow more reliance on molecular information when morphological information is scarce.

The current and other recent works show that octocoral species without sclerites are not rarities; the absence of sclerites or other calciferous skeletal parts are known from within seven octocoral families (Acanthoaxiidae, Acrossotidae, Cornulariidae, Dendrobrachiidae, Alcyoniidae, Clavulariidae, Xeniidae) (Alderslade and McFadden 2007, 2011; López-González et al. 1995; Benayahu et al. 2017). Sclerites provide protection against predation, provide support and rigidity to soft colony parts, and are critical in octocoral classification (West 1997; Aharonovich and Benayahu 2012). Sclerites in different species differ in density as well as spatial organisation in the connective tissue, and thus they also differ in the degree to which they offer support and protection (Benayahu et al. 2017). It could be hypothesized that small size and inhabiting small cracks on reefs protects small stoloniferous species from exposure and has eliminated the need for sclerites in some of these species. Alternately, perhaps the absence of predation (e.g., by ovuliids or sea slugs) has made the possession of sclerites unnecessary. However, there is no obvious pattern to be found for this character in the octocoral phylogenetic radiation amongst its smallest members (Figure 6), and more work is needed to confirm this hypothesis. The absence of sclerites is an example of the gaps in our present knowledge about octocorals and indicate how much basic work in alpha taxonomy and documentation remains for this group.

Our results demonstrate that the TARP area, off the coast of Kota Kinabalu, in Sabah, Malaysia, harbours stoloniferous octocoral diversity that was previously unknown to science, which is obvious since it represents one of the first studies focusing on Stolonifera in this region. However, the newly described sclerite-free Phenganax spp. as well as Clavulariidae sp. are only small pieces of the puzzle in the needed thorough investigation of the polyphyletic family Clavulariidae (McFadden and Ofwegen 2012; Benayahu et al. 2017; Conti-Jerpe and Freshwater 2017; Lau et al. 2019), and an even smaller step towards a full understanding of the morphological and molecular distinctions amongst clades of Stolonifera and ultimately Octocorallia. Molecular data indicate that the entire Octocorallia classification needs re-structuring, but for now researchers still rely on the traditional morphological classification system (Fabricius and Alderslade 2001; Conti-Jerpe and Freshwater 2017), as evolutionary relationships are still far from being fully resolved (McFadden and Ofwegen 2012; Conti-Jerpe and Freshwater 2017). Integrative taxonomy combining not only morphology and molecular biology but also including ecology and biogeography data appears to be the best way to try to better understand the systematics of such a diverse and important group in marine benthic communities (Perez et al. 2016). To this end, marine parks should include in their goal not only the preservation of commercially important faunal diversity (primarily reef fish), but also of other reef organisms, and in particular, small and inconspicuous cryptic species, as it is these small reef inhabitants that make up the vast majority of the diversity of coral reef communities (Hoeksema 2017).