Sample collections for molecular analysis

One specimen of Astroclon suensoni was collected by a fishing line, off Kashino-zaki, Kushimoto, Wakayama, central Japan, 100–140 m depth, 9 January 2014 (NSMT E-10717), and an arm tip specimen of Astroclon propugnatoris was collected by ROV Hakuyo 2000 of M/S Shinsei-maru (Fukada Salvage and Marine Works Co. Ltd.), off Tarama-jima Island, Okinawa, southwestern Japan, 24°29.72′N, 124°33.78′E. 284–290 m depth, 8 March, 2005 (NSMT E-6274). These two specimens were directly immersed in 99% ethanol for molecular analysis.

DNA extraction, PCR amplification, and DNA sequencing

We sequenced nuclear 18S rRNA and mitochondrial COI and 16S rRNA genes for Astroclon suensoni and mitochondrial COI for Astroclon propugnatoris. The method of DNA extraction and PCR parameters followed that by Okanishi and Fujita (2013). Total DNA solution of Astroclon was diluted by distilled water 50 times for PCR of 18S rRNA, 100 times for COI and 16S rRNA. Primer sets of SSU001 (5′-GCTTGTCTTAAAGACTAAGCCATGC-3′) and SSU002 (5′-CCGTGTTGAGTCAAATTAAGCCGC-3′) (Okanishi et al., 2011) was used for PCR of 18S rRNA, COI005 (5′-TTAGGTTAAHWAAACCAVYTKCCTTCAAAG-3′) and COI008 (5′-CCDTANGMDATCATDGCRTACATCATTCC-3′) (Okanishi and Fujita, 2013) was used for PCR of COI, and 16Sar (5′-CGCCTGTTTATCAAAAACAT-3′) and16Sbr (5′-CCGGTCTGAACTCAGATCACGT-3′) (Palumbi, 1996) was used for PCR of 16S rRNA. The PCR products were separated from excess primers and oligonucleotides using Exo-SAP-IT (GE Healthcare), following the manufacturer's protocol. All samples were sequenced bidirectionally and sequence products were run on a 3730xI DNA Analyzer (Thermo Fisher Scientific). The accession numbers of DNA Data Bank of Japan (DDBJ) of the Astroclon are LC272069 (18S rRNA), LC272067 (COI), and LC272070 (16S rRNA) for A. suensoni, and LC272068 (COI) for A. propugnatoris.

Table 1.

Sampling locality and voucher depository of the examined specimens. Asterisks indicate type species.

Phylogenetic analysis

We analyzed sequences of both species of Astroclon, with the sequence data of other 38 gorgonocephalid species obtained by Okanishi and Fujita (2013). For outgroups, we selected five euryalid species with the shortest genetic distance from Gorgonocephalidae, two from Asteronychidae and three from Euryalidae, to avoid long branch attraction (Bergsten, 2005; Okanishi and Fujita, 2013). Additionally, we selected Ophiopholis aculeata (Ophintegrida; Ophiactidae) as a representative of Ophintegrida. This species is also used in previous molecular phylogeny (Okanishi and Fujita, 2013) with sequences of 18S rRNA, COI, and 16S rRNA genes.

All sequences were aligned using the Clustal W algorithm in MEGA5 (Thompson et al., 1994; Tamura et al., 2011). Gene regions where the alignment was ambiguous, including ribosomal loops, were excluded by eye, and all missing sequences were scored as gaps. Overall average of substitution rate was computed using MEGA5 according to the Kimura 2-parameter model (Kimura, 1980) to compare the evolutionary rate of each gene. The Kimura 2 parameter model (Kimura, 1980) with a gamma distribution was selected as best-fit model of 18S rRNA gene, and general time reversible model (GTR; Yang, 1994a) assumed to be invariable (I; Hasegawa et al., 1985) with a gamma distribution approximated by discrete categories (γ Yang, 1994b) being selected as the best-fit model of nucleotide substitution with COI and 16S rRNA genes, by using the “Find best fit models” option of MEGA5. BioEdit ver. 7.0.5.3. (Hall, 1999) and Seaview ver. 4.3.0 (Gouy et al., 2010) were used in preparing the data matrices in FASTA and PHYLIP format, respectively. Seaview ver. 4.3.0 was also used in concatenating sequences of three gene sequences. Tree View for Win 16 (Page, 1996) was used in exploring tree files, in preparing NEWICK format and exploring alternative tree topologies. The phylogenetic tree was constructed with MrBayes ver. 3.1.2 (Ronquist and Huelsenbeck, 2003) to obtain Bayesian posterior probabilities (BPP) and RAxML ver. 8.1.20 (Stamatakis, 2014) for maximum likelihood analysis (ML) to obtain bootstrap support values. The three gene sequences were placed in separate partitions. We set parameters in MrBayes as follows; two substitution types (nst = 2, statefreqpr = fixed; K2P) for 18S rRNA and six substitution types (nst = 6; GTR) for COI and 16S rRNA were employed; rate variation across sites was modeled using a gamma distribution (rate = gamma) for 18S rRNA and using a gamma distribution with a proportion of sites being invariant (rate = invgamma) for COI and16S rRNA; the shape, proportion of invariable sites, state frequency, and substitution rate parameters were estimated for each partition separately. The Markov-Chain Monte-Carlo (MCMC) process was run with four chains for 4,000,000 generations, with trees being sampled every 100 generations. The first 10,000 trees were discarded as burn-in. Data sets were partitioned by gene region for the maximum likelihood analysis to allow for separate optimization per-site substitution rates. The best-supported likelihood tree was found by performing 1000 replications. Nodes in the phylogenetic trees were considered as supported if BPP values were larger than 0.98 and bootstrap value larger than 85%. BPP values lower than 0.97 and bootstrap lower than 74% for each node were considered as not significant (Fig. 1).

Morphological observation

For morphological comparison between the two species of Astroclon (A. propugnatoris and A. suensoni) and other gorgonocephalids, 19 species in 16 genera covering the three subfamilies of Gorgonocephalidae were examined (Table 1). Specimens identifiable as the type genera of the three subfamilies, Astrothamnus, Astrotoma, and Gorgonocephalus, were included. From Astrothamninae and Astrotominae, we examined one or more species from each of the five genera (Astrothamnus, Astrothrombus, Astrocrius, Astrohamma, and Astrotoma). Since we did not obtain any dissectable specimens of Astrotoma drachi and A. agassizii (which were only used for obtaining molecular data), we examined a specimen of A. manilense as a representative of this genus; according to O'Hara and Harding (2014), A. drachi and A. manilense are potentially synonymous. In Gorgonocephalinae, some genera were recovered to be non-monophyletic (Okanishi and Fujita, 2013). Therefore, we selected species so that they represent all the gorgonocephaline subclades recognized by molecular analysis in this study (see Fig. 1, nodes S1–S7). We examined one species of Astrochele, one species of Astrodendrum, and two species of Gorgonocephalus from clade S1, and one species each from clades S2–S7 (Astrosierra from S2, Asterporpa from S3, Astrothorax from S4, Astrodendrum S5, Astroboa from S6, Astroglymma from S7). Because Asteroporpa muricatopatella and three species of Astrocladus formed a clade despite with the relatively lower support value of ML, we selected Astrocladus coniferus as a representative of this species group (see Fig. 1, node marked by a star).

Fig. 2.

Oral views of disc. Skins and external ossicles were partly removed. (A) Astroclon suensoni (NSMT E-10755); (B) Astrothamnus echinaceus (NSMT E-2207-A); (C) Astrothrombus chrysanthi (NSMT E-6715); (D) Astrotoma manilense (NSMT E-3148); (E) Gorgonocephalus eucnemis (NSMT E-1563); (F) Astrosierra amblyconus (NSMT E-1943); (G) Asteroporpa annulata (NSMT E-1767); (H) Astrochlamys bruneus (NSMT E-697). Black arrow heads indicate supplementary oral plates. Abbreviations: AP, adoral plate; EO, external ossicle; LAP, lateral arm plate; OP, oral plate; VAP, ventral arm plate.

In morphological observation, we focused on supplementary oral plates, adoral plates, and the first ventral arm plates, all recognized as diagnostic characters of subfamilies in Gorgonocephalidae since Matsumoto (1917). The presence/absence of supplementary oral plates have been observed only for Astrocrius sobrina (Astrotominae) (Matsumoto, 1917). Adoral plates and the first ventral arm plates have been observed for a limited number of species (e.g., Astrochele pacifica, Mortensen, 1933), but never for Astroclon. We also examined the following external characters used in generic diagnoses in Gorgonocephalidae: presence/absence of tubercles, spines, special calcareous plates, and concentric transverse ridges on disc; maximum number of arm spines; maximum numbers of secondary teeth on hooklets; and patterns of arm branching (see Matsumoto, 1917; Döderlein, 1927; Mortensen, 1933; Fell, 1960; Turner and Boucher, 2010).

To observe the supplementary oral plates, adoral plates, and the first ventral arm plates, skin and external ossicles were removed using domestic bleach (~5% sodium hypochlorite solution). The aboral plates and the first ventral arm plates were isolated, washed in deionized water, observed and photographed with Keyence VHX 600. The terms used in this study to describe brittle stars follow Matsumoto (1917), Martynov (2010), Stöhr et al. (2012), Okanishi and Fujita (2013), and Okanishi (2016). All the specimens with isolated ossicles are deposited in the National Museum of Nature and Science, Japan (NSMT).

Table 2.

Morphological characters of examined species. Abbreviations: AS, arm spine; SOP, supplementary oral plate; SCP, special calcareous plate; ST, secondary teeth of hooklet; VAP, ventral arm plate. Adoral plates are not included because no morphological differences were observed between species.

Molecular phylogeny

After removal of ambiguously aligned sites, we were left with 949 bp of 18S rRNA, 1511 bp of COI, and 429 bp of 16S rRNA. Overall averages of nucleotide substitution rates within nuclear 18S rRNA, mitochondrial COI and 16S rRNA genes were 0.019, 0.261, and 0.186, respectively.

The Bayesian tree of concatenated sequences of the three genes, 18S rRNA, COI, and 16S rRNA is shown in Fig 1. The ML tree also showed the same topology. Both Bayesian and ML analyses supported monophyly of Gorgonocephalidae (Fig. 1, node 1, bootstrap value 85%, BPP 1.00).

Two species of Astroclon formed a clade (Fig. 1, node 2, bootstrap 99%, BPP 1.00). The Gorgonocephalidae clade was divided into the following four clades: two species of Astroclon (Fig. 1, node 2, bootstrap 99%, BPP 1.00), Astrothamninae (Fig. 1, node 3, bootstrap 100%, BPP 1.00), Astrotominae (Fig. 1, node 4, bootstrap 99%, BPP 1.00), and Gorgonocephalinae (Fig. 1, node 5, bootstrap 88%, BPP 1.00). In Gorgonocephalinae, monophyly of seven subclades were well supported (Fig. 1, node S1–S7, bootstrap 85–100%, BPP 1.00), whereas support value of a subclade of three species of Astrocladus and A. muricatopatella was lower in our ML analysis (Fig. 1, node star).

Morphological observation

Supplementary oral plates were present on 19 species including Astroclon but were lacking on Asteroporpa annulata Örsted and Lütken, 1856 (in Lütken, 1856) and Schizostella sp. (Fig. 2; Table 2).

Fig. 3.

Lateral view of first ventral arm plate. (A) Astroclon suensoni (NSMT E-10755); (B) Astrothamnus echinaceus (NSMT E-2207-A); (C) Astrotoma manilense (NSMT E-3148); (D) Astroglymma sculptum (NSMT E-10718). Orientations: top oral side and bottom aboral side. Arc indicates wedge-like structure.

Fig. 4.

Oral view of basal portion of an arm. (A) Astroclon propugnatoris (NSMT E-3564); (B) Astrothamnus ehchinaceus (NSMT E-2207-A); (C) Astrotoma manilense (NSMT E-3148); (D) and Gorgonocephalus eucnemis (NSMT E-1563). (A) large pits present adjacent to tentacle sheath; (B–D) no pits adjacent to the sheath. Abbreviation: TS, tentacle sheath.

Adoral plates were flat and plate-shaped for all examined species, with no distinct morphological differences being confirmed between Astroclon and the other gorgonocephalids. However, the first ventral arm plates were different in shape between Astroclon and the other gorgonocephalids. The oral surface of the first ventral arm plates of Astroclon propugnatoris (Fig. 3A) and A. suensoni were polygonal and flat, with the wedge-like projections protruding from the aboral surface (Fig. 3A). The first ventral arm plates of the other 19 gorgonocephalid species were plate-shaped, and both oral and aboral surfaces were polygonal and flat with no projections (Fig. 3B–D; Table 2).

Tentacle pores of Astroclon propugnatoris and A. suensoni had a large pit beside tentacle sheath (Fig. 4A), while no such pit was observed for the other 19 gorgonocephalid species (Fig. 4B–D).

We observed no distinctive features for Astroclon in the six external characters used as generic diagnoses in Gorgonocephalidae: presence/absence of tubercles, spines; presence/absence of special calcareous plates; presence/absence of concentric transverse ridges on disc; maximum number of arm spines; maximum number of secondary teeth on hooklets; and patterns of arm branching (Table 2).