Molecular phylogenetic evidence supports a new family of octocorals and a new genus of Alcyoniidae (Octocorallia, Alcyonacea)

Abstract Molecular phylogenetic evidence indicates that the octocoral family Alcyoniidae is highly polyphyletic, with genera distributed across Octocorallia in more than 10 separate clades. Most alcyoniid taxa belong to the large and poorly resolved Holaxonia–Alcyoniina clade of octocorals, but members of at least four genera of Alcyoniidae fall outside of that group. As a first step towards revision of the family, we describe a new genus, Parasphaerasclera gen. n., and family, Parasphaerascleridae fam. n., of Alcyonacea to accommodate species of Eleutherobia Pütter, 1900 and Alcyonium Linnaeus, 1758 that have digitiform to digitate or lobate growth forms, completely lack sclerites in the polyps, and have radiates or spheroidal sclerites in the colony surface and interior. Parasphaerascleridae fam. n. constitutes a well-supported clade that is phylogenetically distinct from all other octocoral taxa. We also describe a new genus of Alcyoniidae, Sphaerasclera gen. n., for a species of Eleutherobia with a unique capitate growth form. Sphaerasclera gen. n. is a member of the Anthomastus–Corallium clade of octocorals, but is morphologically and genetically distinct from Anthomastus Verrill, 1878 and Paraminabea Williams & Alderslade, 1999, two similar but dimorphic genera of Alcyoniidae that are its sister taxa. In addition, we have re-assigned two species of Eleutherobia that have clavate to capitate growth forms, polyp sclerites arranged to form a collaret and points, and spindles in the colony interior to Alcyonium, a move that is supported by both morphological and molecular phylogenetic evidence.


Introduction
The anthozoan sub-class Octocorallia comprises a clade of approximately 350 genera and 3400 species of soft corals, gorgonians and sea pens that are found throughout marine environments worldwide (Daly et al. 2007;Williams and Cairns 2013). The current morphology-based taxonomic classification of Octocorallia recognizes three orders, with the majority of families and species belonging to Alcyonacea Lamouroux, 1816 (soft corals, gorgonians and stoloniferans) (Daly et al. 2007). Attempts to further subdivide this very large taxon into smaller orders or sub-ordinal groups have been fraught with difficulty (McFadden et al. 2010). Molecular phylogenetic reconstructions of Octocorallia have confirmed that most of the morphologically defined subordinal groups that have traditionally been recognized (Bayer 1981, Fabricius andAlderslade 2001)-as well as a majority of families-represent polyphyletic assemblages (Berntson et al. 2001;McFadden et al. 2006McFadden et al. , 2010McFadden and Ofwegen 2012). Analyses based on mitochondrial genes (McFadden et al. 2006), nuclear ribosomal genes (Berntson et al. 2001), and both (Brockman andOfwegen 2012) instead support the division of Octocorallia into two major clades, Holaxonia-Alcyoniina and Calcaxonia-Pennatulacea, plus a third, smaller clade, Anthomastus-Corallium (McFadden et al. 2006(McFadden et al. , 2010. The phylogenetic relationships among the family-level clades that comprise the morphologically heterogeneous mix of soft corals (Alcyoniina), gorgonians (Holaxonia, Scleraxonia) and stoloniferous forms (Stolonifera) belonging to Holaxonia-Alcyoniina remain unresolved (McFadden et al. 2006(McFadden et al. , 2010, hindering efforts to revise their taxonomy. Among the many families of octocorals that appear from molecular phylogenetic analyses to be polyphyletic, the soft coral family Alcyoniidae stands out as one of the most phylogenetically heterogeneous (McFadden et al. 2006(McFadden et al. , 2010. Alcyoniidae is also a morphologically eclectic taxon, as it has frequently been the repository for genera that lack the specific diagnostic characters of other, more narrowly circumscribed families (Daly et al. 2007). Genera of Alcyoniidae are distributed across Octocorallia in at least 10 distinct clades, with the majority belonging to the poorly resolved Holaxonia-Alcyoniina clade (McFadden et al. 2006). Although the family clearly requires taxonomic revision, increased phylogenetic resolution along the backbone of Holaxonia-Alcyoniina will be necessary in order to determine which clades of Alcyoniidae should be reassigned to different families, and what the diagnostic morphological characters of those families might be.
Several genera of Alcyoniidae fall entirely outside of Holaxonia-Alcyoniina, and belong instead to two small clades located near the base of Octocorallia (McFadden et al. 2006). These include Anthomastus Verrill, 1878 and Paraminabea Williams & Alderslade, 1999, both of which belong to the Anthomastus-Corallium clade (McFad-den et al. 2006, Brockman and. Sufficient molecular phylogenetic evidence now exists to suggest that several species in the alcyoniid genera Eleutherobia Pütter, 1900 and Alcyonium Linnaeus, 1758 comprise a second clade that also lies outside of Holaxonia-Alcyoniina and is well separated from all other genera of Alcyoniidae. Here we present corroborating morphological evidence to support the description of a new family and genus of Alcyonacea to accommodate this unique clade. In addition, we designate a new genus and combination for a species of Eleutherobia that belongs to the Anthomastus-Corallium clade, and reassign two other species of Eleutherobia to Alcyonium.

Collection
Specimens of Eleutherobia and Alcyonium suitable for molecular analyses were collected in South Africa in 2008 and Palau in 2005 and 2010 (Table 1). Following collection using SCUBA, colonies were preserved in 70% EtOH; tissue sub-samples to be used for molecular analyses were preserved in >90% EtOH. Specimens of E. flammicerebra Williams, 2003 were collected by dredge from New Caledonia during the 2008 Terrasses cruise (R/V Alis), conducted as part of the MNHN/IRD Tropical Deep-Sea Benthos cruises, 2003(Bouchet et al. 2008. Vouchers have been deposited at the Naturalis Biodiversity Center (formerly Rijksmuseum van Natuurlijke Historie, Leiden (RMNH)), and the Muséum national d'Histoire naturelle, Paris (MNHN). Additional material was obtained from the Museum and Art Gallery of the Northern Territory, Darwin, Australia (NTM), the Florida Natural History Museum, Gainesville (UF), the Zoological Museum, University of Copenhagen (ZMUC), and the Zoologische Staatsammlung München, Germany (ZSM) ( Table 1).

Morphological analysis
Sclerites were obtained by dissolving tissues from the upper and lower regions of a colony in 10% sodium hypochlorite (household bleach). Sclerites were rinsed well with deionized water, dried, and mounted on stubs for SEM. They were imaged using a JEOL JSM-6480LV scanning electron microscope operated at 10 kV.

Molecular phylogenetic analyses
Extraction of DNA from ethanol-preserved tissue samples, PCR amplification, and sequencing of the mtMutS (msh1) and COI genes followed the protocols published in McFadden et al. (2011). In addition, we sequenced an approximately 810 nt frag-ment of the 28S nuclear ribosomal gene using primers 28S-Far and either 28S-Rar or 28S-Rab (McFadden and Ofwegen 2013). Sequence data were aligned to a previously compiled reference dataset of 130 octocorals and anthozoan outgroup taxa (McFadden and Ofwegen 2012; Appendix) using the L-INS-i method in MAFFT (Katoh et al. 2005). Modeltest 3.0 (Posada and Crandall 1998) was used to select appropriate models of evolution for maximum likelihood analyses that were run for 100 bootstrap replicates using GARLI 2.0 (Zwickl 2006). The 28S rDNA and mitochondrial gene (mtMutS + COI) datasets were analyzed separately, and in a combined analysis with different models of evolution applied to separate data partitions (mt genes: TVM+I+G; 28S: GTR+I+G). Bayesian analyses were run using MrBayes v. 3.2.1 (Ronquist et al. 2012) with the same data partitions; because MrBayes does not support the TVM model, however, a GTR+I+G model was applied separately to each partition. Analyses S. flammicerebra (Williams, 2003) MNHN-IK-2012-12004 New Caledonia 2008 S. flammicerebra (Williams, 2003) ZMUC- ANT-000256 Mauritius 1929 were run for 6,000,000 generations (until runs had converged and standard deviation of split partitions < 0.01) and sampled every 500 generations, with a burn-in of 25% and default Metropolis coupling parameters.

Molecular phylogenetic analyses
The separate maximum likelihood analyses of the 28S rDNA and mitochondrial gene alignments generated phylogenies that were congruent with one another (with the sole exception of some internal relationships within Pennatulacea), but were poorly resolved overall. The partitioned analyses of the combined mt + 28S dataset had much higher support values for many of the deeper nodes within the tree, and it is these analyses that we present ( Fig. 1) and discuss here. As has been demonstrated previously based on analyses of similar datasets (Brockman andOfwegen 2012), the combined gene analyses supported the division of the majority of octocorals among two major clades, Holaxonia-Alcyoniina and Calcaxonia-Pennatulacea ( Fig.  1). In addition, there was strong support from both maximum likelihood and Bayesian methods for three small clades that fell outside of Holaxonia-Alcyoniina but whose relationships to Calcaxonia-Pennatulacea and to one another remain unresolved. These include the previously recognized Anthomastus-Corallium clade (McFadden et al. 2006, Brockman and, a clade comprising a heterogeneous mix of scleraxonians plus the stoloniferan genus Telestula Madsen, 1944, and a clade consisting of several species of the alcyoniid genera Eleutherobia and Alcyonium (Fig. 1). The stoloniferan genus Cornularia Lamarck, 1816 was recovered as the sister taxon to all other octocorals. Members of family Alcyoniidae were distributed throughout Holaxonia-Alcyoniina in nine distinct clades ( Fig. 1). Some of these clades included only alcyoniid genera, while others supported close relationships among alcyoniids and genera classified in different families. There was insufficient resolution of the deeper nodes within Holaxonia-Alcyoniina to infer the phylogenetic relationships of any of these nine clades to one another. In addition, four genera of Alcyoniidae fell entirely outside of Holaxonia-Alcyoniina (Fig. 1). These included Paraminabea and Anthomastus, both of which belonged to the Anthomastus-Corallium clade, as well as several species of Eleutherobia and Alcyonium that formed a distinct, well-supported clade whose relationship to the Calcaxonia-Pennatulacea, Anthomastus-Corallium and Scleraxonia-Telestula clades was poorly resolved. This clade included Eleutherobia aurea Benayahu & Schleyer, 1995, E. rotifera (Thomson, 1910), E. aff. grayi (Thomson & Dean, 1931), and Alcyonium valdiviae Kükenthal, 1906. Two of the species of Eleutherobia we sequenced were not included in this clade: E. variabile (Thomson, 1921) fell into a clade with northern hemisphere members of the genus Alcyonium, while E. flammicerebra Williams, 2003 belonged to the Anthomastus-Corallium clade, phylogenetically distinct from both Paraminabea and Anthomastus (Fig. 1).

Alcyonacea Lamouroux, 1816 Alcyoniidae Lamouroux, 1812
Alcyonium Linnaeus, 1758 Alcyonium has long served as a repository for species that lack characters to support their placement in other more narrowly circumscribed genera. Over time the diagnosis of the genus has been broadened to include almost every possible colony growth form observed within Alcyoniidae (Williams 1988) as well as a diversity of different sclerite types and arrangements. In recent years the heterogeneity of this genus has been acknowledged (Alderslade 2000;Williams 2000a), and a number of new genera have been erected to accommodate species whose characters clearly differ from those of the type species, the northern hemisphere A. digitatum Linnaeus, 1758. Molecular phylogenetic analyses have supported the taxonomic distinction of new genera such as Klyxum Alderslade, 2000, Rhytisma Alderslade, 2000, Lampophyton Williams, 2000, and Discophyton McFadden & Hochberg, 2003 (Fig. 1), all of them established to accommodate species formerly placed in Alcyonium.
Phylogenetic evidence suggests that genus Alcyonium should be further restricted to species in which the polyp sclerites are arranged as a collaret and points and the sclerites of the coenenchyme are in two distinct layers, a surface layer consisting of predominantly radiates or clubs, and an inner layer of spindles or rods (Alderslade 2000). The colony growth form may be lobate, digitate, capitate or encrusting, and the sclerites are usually colored. Adoption of this restricted, phylogenetically supported diagnosis will necessitate not only the removal of additional species from Alcyonium, but also the inclusion of species currently placed in several other genera. For example, despite their placement in three different families, the genera Bellonella Gray, 1862 (Alcyoniidae), Eleutherobia Pütter, 1900 (Alcyoniidae), Anthothela Verrill, 1879 (Anthothelidae) and Gersemia Marenzeller, 1878 (Nephtheidae) all include species with sclerite characters that suggest a close affinity with Alcyonium. Molecular phylogenetic evidence supports a paraphyletic relationship between Alcyonium, Gersemia, and Anthothela (McFadden et al. 2006), as well as the inclusion of a species of Eleutherobia in Alcyonium sensu stricto (Fig. 1). Williams (2003) reassigned Alcyonium variabile (Thomson, 1921) to Eleutherobia subsequent to his modification of that genus to accommodate the capitate E. flammicerebra Williams, 2003. The capitate growth form of E. variabile differs from the lobate to digitate forms characteristic of most northern hemisphere species of Alcyonium (McFadden et al. 2001). Similar to other species of Alcyonium sensu stricto, however, E. variabile has polyp sclerites arranged to form a distinct collaret and points, radiates and club-like forms in the surface layer of the polyparium and stalk, and long spindles in the interior of the polyparium (Williams 1986(Williams , 1992. Molecular phylogenetic analyses strongly support the inclusion of E. variabile in Alcyonium sensu stricto (Fig. 1), therefore we transfer this species back to that genus and re-instate the combination Alcyonium variabile (Thomson, 1921) comb. n.
Bellonella studeri Thomson, 1910, a species with a clavate to capitate growth form similar to that of A. variabile comb. n., was reassigned to Eleutherobia by Verseveldt and Bayer (1988). The sclerites of E. studeri are very similar to those of A. variabile comb. n. Both species have polyps with spindles arranged to form a collaret and points, capstan-like radiates in the surface of the polyparium and stalk, and sclerites in the colony interior that are predominantly slender spindles (Williams 1992). Material of E. studeri was not available for molecular phylogenetic analysis, but based on its morphological similarity to A. variabile comb. n. we suggest that this species also belongs in Alcyonium, and propose the new combination Alcyonium studeri (Thomson, 1910) comb. n. Verseveldt and Bayer (1988) suggested that Thomson's (1910) Metalcyonium clavatum (non Pfeffer, 1889) might be a synonym of A. studeri comb. n. Whether or not that is the case, M. clavatum also appears from Thomson's (1910) description to belong to Alcyonium.
Our molecular phylogenetic analyses included a specimen of Alcyonium roseum Ofwegen, Häussermann & Försterra, 2007, a species recently described from Chile. We note that that name is pre-occupied by A. roseum (Tixier-Durivault, 1954), and hereby designate the Chilean species Alcyonium varum nom. nov. Etymology: from the Latin varus, crooked or bow-legged, denoting the shape of the sclerites in the polyps (Ofwegen et al. 2007). Diagnosis. Soft corals with a capitate growth form, with a distinct, spherical polyparium raised on a sterile stalk. Polyps monomorphic. Anthocodial sclerites absent. Sclerites of colony surface and interior are large, tuberculate spheroids and smaller radiates. Sclerites permanently colored. Azooxanthellate.
Etymology. From the Latin/Greek sphaera-meaning a sphere or ball and Greek sclero-, hard, denoting the large, spheroidal sclerites that characterize this genus. Gender: fem.
Remarks. Williams's (2003) assignment of E. flammicerebra to Eleutherobia necessitated emending the diagnosis of that genus to include capitate growth forms. E. flammicerebra does share other character states-such as monomorphic polyps that lack sclerites and tuberculate spheroids and radiates in the surface of the polyparium and stalk (Figs 2-3)-with some species of Eleutherobia. Molecular phylogenetic analyses suggest that E. flammicerebra is, however, not closely related to morphologically similar members of Eleutherobia such as E. rotifera (Thomson, 1910) but instead falls into the Anthomastus-Corallium clade of octocorals (Fig. 1). Based on its distinctive colony growth form and unique phylogenetic position, we hereby designate a new genus, Sphaerasclera, and new combination, Sphaerasclera flammicerebra (Williams, 2003) for this species.
Unlike all other members of the Anthomastus-Corallium clade, Sphaerasclera gen. n. appears to have monomorphic rather than dimorphic polyps. As discussed by Williams (2000b), however, siphonozooids may be difficult to detect in contracted, preserved material. The large, densely packed, spherical sclerites in the coenenchyme of S. flammicerebra comb. n. obscure the finer details of the colony surface, and also make this species very difficult to section for histological study of the polyps. For now we concur with Williams (2003) that the species is monomorphic, but entertain the possibility that future observation of living specimens might reveal the presence of siphonozooids.
Although Sphaerasclera gen. n. differs by having monomorphic polyps, it does share other morphological characters with Paraminabea and Anthomastus, the two genera with which it is most closely allied phylogenetically (Fig. 1). Anthomastus likewise includes species with capitate growth forms, but in that genus the autozooids have sclerites, and the sclerites in the surface and interior of the colony include rods and needles in addition to radiates (Bayer 1993). Like Sphaerasclera gen. n., Paraminabea lacks sclerites in the polyps and its coenenchymal sclerites are predominantly radiates and spheroids (Williams and Alderslade 1999). The colony growth form of Paraminabea, however, is digitiform, hemispherical or lobate rather than capitate. Moreover, Paraminabea has a unique molecular synapomorphy, a mitochondrial gene rearrangement that distinguishes it from all other genera of octocorals (Brockman and McFadden 2012). Screening of mitochondrial gene junctions suggests that S. flammicerebra comb. n. and Anthomastus both retain the ancestral octocoral mt gene order, and do not share the derived state found in Paraminabea (Brockman and McFadden 2012).
Paraminabea and Anthomastus are both classified in family Alcyoniidae, and for convenience we have also assigned Sphaerasclera gen. n. to that family. All other genera of Alcyoniidae, however, belong to the Holaxonia-Alcyoniina clade of Octocorallia, far removed phylogenetically from the Anthomastus-Corallium clade (Fig. 1). In addition to Paraminabea, Anthomastus and Sphaerasclera gen. n., the Anthomastus-Corallium clade also includes all members of Coralliidae Lamouroux, 1812 and Paragorgiidae Kükenthal, 1916, two families of gorgonians that have historically been assigned to the sub-ordinal group Scleraxonia. Although it is clear from the phylogenetic evidence that the soft coral taxa that fall within this clade should be removed from Alcyoniidae, we defer their reassignment to another family pending an in-depth analysis of the morphological character states shared among the seemingly heterogeneous genera and families that are united within Anthomastus-Corallium.
Our records extend the known geographic distribution of S. flammicerebra comb. n. from Palau (Williams 2003) to New Caledonia and Mauritius. The colony growth forms and sclerites of specimens from these widespread localities closely match that of the type material from Palau (Figs 2-3). Included genera. Parasphaerasclera gen. n. Diagnosis. Soft corals with a digitiform, digitate or lobate growth form, usually with a sterile stalk although this may be indistinct. Polyps monomorphic. Permanent calyces absent, although retracted polyps may remain visible as small mounds on the polyparium surface. Anthocodial sclerites absent. Sclerites of colony surface and interior predominantly radiates and tuberculate spheroids, occasionally rods and crosses. Sclerites permanently colored. Azooxanthellate.   Verseveldt and Bayer (1988) and modified by Williams (2003), the alcyoniid genus Eleutherobia Pütter, 1900 encompasses species with a diversity of sclerite forms and arrangements. Species within this genus are united primarily by their digitiform to lobular colony growth forms, although the diagnosis was recently emended to include capitate forms (Williams 2003). A subset of the species within Eleutherobia have in common a complete lack of sclerites in the polyps, and the sclerites in the surface and interior of the colony are predominantly small radiates and spheroids (Figs 4-9). Molecular analyses unite these species in a clade far removed phylogenetically from other genera of Alcyoniidae (Fig. 1), thus we describe a new family to accommodate them. Parasphaerascleridae fam. n. falls outside of the large Holaxonia-Alcyoniina clade of Octocorallia to which the majority of soft corals belong (McFadden et al. 2006). Although several other genera of Alcyoniidae also lie outside of Holaxonia-Alcyoniina (e.g., Paraminabea, Anthomastus, Sphaerasclera gen. n.), they are united with the scleraxonian families Coralliidae and Paragorgiidae in the Anthomastus-Corallium clade (McFadden et al. 2006). Paraphaerascleridae fam. n. does not belong to that clade.
Diagnosis. As for the family. Etymology. From the Greek root para-, meaning beside or alongside of Sphaerasclera gen. n. These two genera share similar sclerite complements and forms, and both were previously considered to belong to Eleutherobia. Gender: fem.
We also reassign two species of Alcyonium to Parasphaerasclera gen. n., P. morifera (Tixier-Durivault, 1954) comb. n. andP. valdiviae (Kükenthal, 1906) comb. n. The inclusion of P. valdiviae comb. n. in Parasphaerasclera gen. n. is supported by both molecular phylogenetic (Fig. 1) and morphological evidence. P. valdiviae comb. n. lacks polyp sclerites, the sclerites found in the surface of the colony are compact radiates and spheroids (Figs 6-7), and it shares with P. rotifera comb. n. a growth form in which a conspicuous stalk gives rise to either branched or digitate lobes (Verseveldt and Williams 1988).
P. morifera comb. n., another species for which we lack molecular data, shares many morphological characters with other species of Parasphaerasclera gen. n. Verseveldt and Bayer (1988) synonymized Nidalia morifera Tixier-Durivault, 1954 with Eleutherobia rotifera, but Williams (1992) later maintained the distinction between them. He reassigned N. morifera not to Eleutherobia but rather to Alcyonium, based on its lack of permanent calyces. Although Verseveldt and Bayer's (1988) diagnosis of Eleutherobia stated "Anthocodiae retractile within calyces" (p. 27), Benayahu and Schleyer (1995) noted that some of the species included in the genus lacked permanent calyces. Williams and Little (2001) subsequently emended the diagnosis of Eleutherobia to "calyces absent, although retracted polyps often form low rounded to conspicuous protuberances" (p. 198). P. morifera comb. n. is strikingly similar to P. aurea comb. n.; both species have a digitiform to lobular growth form with a short, indistinct stalk, and the sclerites in the colony surface are compact radiates and spheroids (Figs 4A, 5A). The only difference between the two species is the presence of "double-deltoids" in the colony interior of P. aurea comb. n. (Figs 4B, 5B) (Benayahu and Schleyer 1995).
It is possible that another South African species of Alcyonium, A. distinctum Williams, 1988, may also belong to this genus. Like other species of Parasphaerasclera gen. n. it lacks sclerites in the polyps, and the sclerites in the stalk surface are tuberculate spheroids and radiates (Williams 1988(Williams , 1992. The colony growth form is lobate, with lobes arising from a short, thick stalk, somewhat resembling P. valdiviae comb. n. Unlike other species of Parasphaerasclera gen. n., however, in A. distinctum the sclerites are restricted to the stalk surface: there are no sclerites in the lobes (polypary) and interior of the colony (Williams 1988). In addition, the sclerites are not colored, and the bright purple color of living colonies is the result of an alcohol-soluble pigment. Material is not currently available for molecular analysis, so we cannot yet confirm the placement of A. distinctum in Parasphaerasclera gen. n.
A number of specimens of Parasphaerasclera gen. n. are known that differ somewhat from the descriptions of any of the nominal species listed above, and may represent additional, undescribed species of the genus. For example, there is considerable variation among those specimens of P. grayi comb. n. that have been described and illustrated in the literature. Thomson and Dean's original description (1931: 37) is rather confusing and sclerites were not depicted. The lectotype of P. grayi that was designated and described by Verseveldt and Bayer (1988, figs. 24, 25) has sclerites in the colony surface that consist of 8-radiates (0.06-0.08 mm in length), crosses, and rods with tuberculate processes. The sclerites of the colony interior in-      clude particularly distinctive rod-like forms with smooth waists and high processes, up to 0.18 mm in length. Williams (2001) subsequently re-described P. grayi based on specimens from the Solomon Islands. His specimens include 7-and 8-radiates (0.06-0.08 mm long) and crosses similar to those of the lectotype, but the rod-like sclerites in the colony interior are considerably smaller (most 0.07-0.08 mm long) and lack smooth waists (Williams 2001, figs 5-8). In contrast, specimens of P. grayi from Vietnam that were later re-described and illustrated by Dautova and Savinkin (2009, figs. 5-7) include "rather smooth" (p. 10) rods that more closely resemble those depicted by Verseveldt and Bayer (1988), but are somewhat smaller, up to 0.14 mm long. The specimens from Palau that we have sequenced and identified here as P. aff. grayi have tuberculate rods that lack a smooth waist (Figs 8-9), similar to those of Williams's specimens. In our specimens, however, the rods are considerably larger (to 0.16 mm long). In addition, although the radiates in the colony surface of P. aff. grayi from Palau are similar in size to those of other P. grayi specimens, they are more compact and some approach ovals in form (Figs 8A, 9A). This range of variation in sclerite form and size observed among the different specimens attributed to P. grayi suggests that more than one species may be involved. It remains to be determined if P. aff. grayi from Palau is the same as P. grayi sensu Williams (2001) from the Solomons, and if either of these forms is conspecific with Dautova and Savinkin's (2009) material from Vietnam. The latter most closely matches the P. grayi lectotype of Verseveldt and Bayer (1988).
Parasphaerasclera gen. n. is most similar morphologically to the alcyoniid genera Paraminabea and Sphaerasclera gen. n. All three genera lack sclerites in the polyps and have spheroids or radiates in the colony surface and interior. The polyps of Paraminabea, however, are dimorphic, while those of Parasphaerasclera gen. n. are monomorphic. The unique capitate growth form of Sphaerasclera gen. n. distinguishes it from all species of Parasphaerasclera gen. n., which are digitiform to digitate or lobate. Parasphaerasclera gen. n. is also easily distinguished from the alcyoniid genera Eleutherobia and Alcyonium sensu stricto, both of which have sclerites arranged to form a collaret and points in the polyps, and spindles or rods in the colony interior.

Discussion
Following the taxonomic changes we have made here, eleven species remain in Eleutherobia. The morphological heterogeneity of these species and their similarities to some other genera suggest that further taxonomic revisions are likely to be necessary. Six of the remaining species of Eleutherobia have polyps with a distinct collaret and points of spindles; radiates, spindles and club-like sclerites in the colony surface; and spindles in the interior coenenchyme. These species likely belong to Alcyonium sensu stricto. Included among them is the type species of Eleutherobia, E. rigida (=E. japonica, Pütter, 1900), as well as E. grandiflora (Kükenthal, 1906), E. rubra (Brundin, 1896), E. somaliensis Verseveldt & Bayer 1988, E. splendens (Thomson & Dean, 1931), and E.
unicolor (Kükenthal, 1906). Another four species of Eleutherobia likewise have a collaret and points of spindles but have tuberculate radiates, spheroids and irregular forms in the colony surface, and spindle-like sclerites with narrow pointed ends and thick waists in the colony interior (Verseveldt and Bayer 1988). Included in this group are E. dofleini (Kükenthal, 1906), E. duriuscula (Thomson & Dean, 1931), E. flava (Nutting, 1912), and possibly E. sumbawensis Verseveldt & Bayer 1988. Whether these species also might belong to Alcyonium or to a different genus cannot be determined at present. E. vinadigitaria Williams & Little, 2001, a species that has needle-like spindles in the polyps and no sclerites in the colony interior, is unlike any other species of Eleutherobia. Acquisition of material suitable for molecular phylogenetic analysis will greatly facilitate future efforts to determine the appropriate taxonomic placement of these species.
Until such time as the phylogenetic relationships among the species that remain in Eleutherobia can be determined, we modify the most recent diagnosis of the genus (Williams 2003) as follows: Alcyoniid soft corals, usually digitiform (conical to cylindrical), sometimes digitate to lobate or clavate; polyparium arising from a common unbranched stalk. Polyps monomorphic. Calyces absent, although retracted polyps may form low rounded or mound-like protuberances of the coenenchyme. Anthocodial sclerites present, arranged in points or collaret and points. Coenenchymal sclerites mostly derived from radiates, although spindles, barrels, tuberculate spheroids, rod-like forms or crosses sometimes present. Color permanent and contained in the sclerites. Azooxanthellate.

Conclusions
Molecular phylogenetic analyses of a number of species belonging to the alcyoniid genera Eleutherobia and Alcyonium have highlighted the heterogeneity of these two taxa as well as some overlap between them. As an initial effort to align the morphologybased taxonomic classification of species in these genera with molecular phylogenetic evidence of their evolutionary affinities we have reassigned two species of Eleutherobia with capitate growth forms, polyp sclerites arranged as a collaret and points, and spindles in the colony interior to Alcyonium; assigned a phylogenetically unique species with monomorphic polyps and a capitate growth form to a new genus of Alcyoniidae, Sphaerasclera gen. n.; and designated a new family, Parasphaerascleridae fam. n., and genus, Parasphaerasclera gen. n., to accommodate species with monomorphic polyps that lack sclerites in the polyps and have predominantly radiates and spheroids in the colony surface and interior. These are the first of many taxonomic revisions that will be required to reconcile the classification of genera currently assigned to Alcyoniidae with their phylogenetic relationships. Whereas the revisions we have made herein are well supported by both morphological and molecular phylogenetic evidence, considerably more evidence of both types will be necessary before we can begin to make taxonomic sense of the poorly resolved relationships among the many additional alcyoniid genera that belong to the Holaxonia-Alcyoniina clade of octocorals.