A new genus of Atlantic octocorals (Octocorallia: Gorgoniidae): systematics of gorgoniids with asymmetric sclerites

Gorgonian corals from the families Gorgoniidae and Plexauridae are among the most abundant and diverse shallow‐water cnidarians found along tropical and temperate Atlantic coasts and reefs. Some morphological traits in the Gorgoniidae comprise a taxonomic challenge due to their morphological plasticity as well as a lack of discrete character states among most genera. Some gorgoniids have asymmetrical sclerites in the surface coenenchyme, a character trait that consists of a continuum among several genera and can cause confusion with members of the family Plexauridae. The collection of a unique gorgoniid in Tobago (Trinidad and Tobago), which contains this trait but also maintains affinities to many different genera, stimulated a wider morphological comparison of gorgoniids using scanning electron microscopy (SEM). Using the morphology of asymmetrical sclerites, gorgoniids could be grouped in four monophyletic clades that were compatible with recent molecular phylogenies and provide a provisional subfamilial classification: (1) a derived clade including gorgoniids with scaphoid sclerites, comprising the reef fauna from the western Atlantic (e.g. Gorgonia, Pseudopterogorgia [ = Antillogorgia], Pterogorgia, and Phyllogorgia); (2) a related group with taxa containing capstan sclerites including genera with short capstans and/or disk‐spindles sclerites such as Pacifogorgia and Leptogorgia ( = Lophogorgia); (3) a clade of gorgoniids with asymmetrical spiny sclerites including the Atlantic genus Muriceopsis and the Indopacific Pinnigorgia together with the distant single species from the new genus and species Tobagogorgia hardyi (it is worth noting that Muriceopsis was previously classified in the Plexauridae, but there is overwhelming molecular support that groups it with the Gorgoniidae and morphological evidence that suggests affinity with Pinnigorgia); and finally (4) a basal clade with long and spiny spindles, which includes some African species such as Filigorgia angolana and F. schoutedeni as well as other new combinations (all species previously classified as Leptogorgia but not phylogenetically related). The combined set of sclerites of Tobagogorgia (e.g. both asymmetrical and irregularly bent spindles) separates it from the diagnostic characters of related genera such as Muriceopsis and Pinnigorgia. The latter two genera are also profusely branched, including pinnate patterns, whereas T. hardyi has only one or two branches. Externally, T. hardyi looks like some Leptogorgia colonies that branch minimally but it is clear that the sclerites are very different and thus also their phylogenetic divergence.


Introduction
Gorgonian coral systematics relies on a combination of characters such as axial skeleton (if any), colonial form, polyp arrangement, and sclerite (microscopic magnesium calcite formations) morphology (Bayer 1961). Excluding sclerite morphology, comparisons of these traits against molecular phylogenies have shown a great deal of homoplasy (Bayer 1961;Sánchez et al. 2003aSánchez et al. , 2003bSánchez 2004). In addition, there are no landmarks in any of these traits (Sánchez and Lasker 2003). Unwieldy ancient literature with no or neglected illustrations and lost or badly preserved type material have played a major role concerning the current taxonomic confusion in octocoral systematics. Critical examinations of octocoral taxonomy have not been frequently used due to the absence of independent sources of characters. Opportunely, however, there has been a rebirth in invertebrate systematics due to the ever-increasing availability of DNA sequence characters, which can turn taxonomic observations into testable hypotheses (Mallet and Willmott 2003).
The Gorgoniidae is comprised of important genera of shallow-water octocorals distributed worldwide, and is characterized by having a thin coenenchyme and reduced sclerites that are regularly sculpted or ornate (Bayer 1961;Grasshoff and Alderslade 1997). However, some taxa include diagnostic characters that are not well defined at the generic level, and based on the family diagnosis some species could be classified in either the Gorgoniidae or Plexauridae. For instance, the axis of plexaurids can have a ''cortex commonly (but not always) loculated'' and in gorgoniids it can be ''loculated little or not at all'' (Bayer 1961, p 91, 192) and both can have a cross-chambered core. Axial characters were considered the key character for family-level distinction within the Holaxonia, but it is clear that there is no discrete differentiation in these two families and molecular results have strongly suggested multiple origins of gorgonian axes (McFadden et al. 2006). Some generic descriptions include traits such as tubercles arranged in whorls, but irregular in some species (Grasshoff and Alderslade 1997), but this irregular trait may correspond to some plexaurids. The thickness of the coenenchyme and sclerite size is also quite confusing between these two families. Muriceopsis spp. (Plexauridae), for example, has sclerites of intermediate size among the two families and a thin coenenchyme as in gorgoniids (Sánchez et al. 2003b). The genus Plexaurella, also Plexauridae, has reduced (,0.5 mm), regularly ornate sclerites yet has the thickest coenenchyme in the family. Both Plexaurella spp. and Muriceopsis spp. have sclerite types different from those of other plexaurid species. Interestingly, the molecular phylogeny of the Plexauridae and Gorgoniidae, according to three coding genes from mitochondrial DNA (ND2, ND6, and MSH1), clearly show how Muriceopsis and Plexaurella group with the Gorgoniidae (Sánchez et al. 2003b) suggesting that suprageneric taxa can be classified within precise morphological limits (e.g. sclerites ,0.5 mm length) with a combination of molecular and morphological data. Similarly, the Plexauridae is a complex family, including multiple lineages and remarkably derived morphologies (Sánchez et al. 2003b;Sánchez 2005;Wirshing et al. 2005). Some gorgoniids have asymmetrical sclerites in the surface coenenchyme, which comprises a continuum among several genera and generates confusion with the family Plexauridae. The collection of a unique gorgoniiid in Tobago (Trinidad and Tobago), which contains this trait but also maintains affinities with many different genera, stimulated a wider morphological comparison within the Gorgoniidae using scanning electron microscopy (SEM). Aguilar and Sánchez (2007a) present several new phylogenetic hypotheses of most gorgoniid genera using ITS2 sequences and their predicted RNA secondary structures. Although there are several competing hypotheses for the relationships of basal nodes depending on the type of analysis of whether the source of characters were sequence-or structure-based, support for many genera and some intrafamilial clades differ from the current taxonomy. This study describes a new octocoral, and compares its sclerite morphology, in the light of the recent molecular phylogenies, against a number of gorgoniids with an emphasis on species with asymmetrically spiny sclerites. Four reciprocally monophyletic clades were observed.

Methods
The subject specimens collected in Tobago (Trinidad and Tobago) and representative species from the Gorgoniidae, with and without asymmetrical spiny sclerites in the surface coenenchyme, as well as a few members of the Plexauridae as outgroups, were examined. Asymmetrical spiny spindles are defined as those structures with spines or different ornaments along only one side of the sclerite independent of whether the sclerite is bent. SEM analyses were done following the methods of Sánchez et al. (2003b) and Sánchez (2005) at the National Museum of Natural History (SEM-Leica Stereoscan, Laboratories of Analytical Biology, Smithsonian Institution, Washington, DC, USA). Specimen museum sources and catalogue numbers from species used for comparisons are presented in the figure legends.
The affinities of the questioned specimens with respect to other gorgoniids were based on the congruency of surface layer sclerites, considered one of the most phylogenetically informative characters in gorgonian corals (Sánchez et al. 2003b;Sánchez 2005;Wirshing et al. 2005), compared against the best supported molecular phylogenetic hypotheses obtained by Sánchez (2007a, see also 2007b).
Results and discussion Figure 1 shows the best supported phylogenetic hypothesis from sequence alignments corrected with the RNA secondary structures according to maximum likelihood including common nodes and support values from maximum parsimony and Bayesian inference of phylogeny. Surface sclerite characters were mapped on this phylogenetic hypothesis ( Figure 1). Gorgoniids could be grouped in four reciprocally monophyletic clades and provide a provisional subfamilial classification.

Genus Pinnigorgia Grasshoff and Alderslade, 1997 (Figure 7)
Pinnate colonies ( Figure 5C-E). Similar diagnostic characters to Muriceopsis (Figures 5A,  B, 6). Spiny spindles from the surface always bent; spines flattened (Figures 7, 8) Diagnostic characters. The genus and species can be distinguished by the presence of both enlarged ornamentation with multiple spines on the convex side of the asymmetrical sclerites from the surface layer ( Figure 9B), and irregularly bent spindles in the inner layer ( Figure 9D). Cylindrical branches and polyps placed throughout the branch (no rows or   grooves) can distinguish this genus from Leptogorgia, which have analogous sclerite forms. Type species: Tobagogorgia hardyi spec. nov.; by monotypy and original designation.
Description. Tobagogorgia hardyi has overall morphological features as other gorgoniid octocorals, which include spindles ,0.3 mm and complex (many internal geometrical/ fractal subwarts) warts (or tubercles) disposed in transverse girdles, usually ,0.02 mm wide (Sánchez et al. 2003b). The holotype has slimy, whip-like, and sparsely branched colonies (one main branch and two to three daughter branches) up to 300 mm in length ( Figure 9A). Its branches are thin and cylindrical (2-4 mm width). Living colonies are yellow ( Figure 10) with even brighter polyps (grey when dry, black in alcohol). The polyps are placed throughout the branches and have robustly ornamented rods 0.12-0.16 mm long ( Figure 9C) but do not form any particular structure. No rows or series of polyps are observable. Sclerites are colourless. Surface layer sclerites are composed of bent (asymmetrical) and ornamented spindles 0.1-0.17 mm long (Figures 2A-D, 9B). The inner sclerite layer harbours two different forms: small radiate capstans (0.05-0.16 mm: Figure 9D) and acute irregularly bent ornamented spindles (up to 0.2 mm long: Figure 9E).
Etymology. The new genus is named for the island Tobago and the suffix gorgia commonly used for gorgoniid genera (feminine gender). The species is named in honour of Jerry D. ''Dave'' Hardy, Jr who has dedicated his career to the study of biodiversity in Tobago.
Species comparisons. The molecular results by Aguilar and Sánchez (2007a), which showed that Tobagogorgia did not group close to any gorgoniid genera or species, validated a separation of this species as a new genus in spite of its colonies and sclerites resembling those of a number of other valid genera. Tobagogorgia hardyi is a gorgoniid species that does not have symmetrically sculptured spindles (e.g. Figure 12). The combined set of sclerites of Tobagorgia (e.g. both asymmetrical and irregularly bent spindles) makes it different from the diagnostic characters of homologous genera such as Muriceopsis and Pinnigorgia (Figures 5-9). The latter two genera are also profusely branched, including pinnate patterns, whereas T. hardyi has one or two branches only. Externally, T. hardyi looks similar to some Leptogorgia colonies that also barely branches but it is clear that the sclerites are very different and thus their phylogenetic divergence.

Clade 4. Gorgoniids with long and spiny spindles
Surface coenenchyme with long spindles with multiple (regular or irregular) rows of ornaments, occasionally spiny towards one side. This clade includes the Leptogorgia ridouri and L. sanguinolenta groups proposed by Grasshoff (1988).
Genus Filigorgia Stiasny, 1937 Profusely branched pinnate colonies. Long, slim, and mostly straight asymmetrically spiny sclerites in the colony surface ( Figure 11). Surface coenenchyme with straight to slightly bent slim or robust spindles ornate with warts irregularly to zoned around the spindle (Figures 11, 12E, F).
Since surface sclerites modify colony flexion by preventing compressibility of the outer cortex when they contact each other and help the colony to resist drag forces and flow (Lewis and Von Wallis 1991), a characteristic on which natural selection can act. Their interaction with the environment has produced a great array of forms and these structures are more variable than polyp or axial sheath sclerites (Sánchez et al. 2003b;Sánchez 2005). It is important to note, however, that the function of surface sclerites can be homoplasious and similar trends exist in many genera (e.g. Figure 12E, G). Nonetheless, surface sclerites can be helpful for supra familial classification and provide a better understanding of octocoral phylogeny in general.
It has been demonstrated that colonial characters such as branching and overall colony form are highly homoplasious, providing multiple analogous forms within the family Gorgoniidae, which should not be considered as characters for classification (e.g. whip like, sea fan, pinnate, etc.). For example, Bayer (1953) first proposed convergent evolution for gorgonian colony architectures such as sea fans (Pacifigorgia spp. and Gorgonia spp., Figure 2) and sea leaves (Phycogorgia spp. and Phyllogorgia spp., Figure 2). Bayer's hypothesis for sea fans has been corroborated with mitochondrial (Sánchez et al. 2003b;Sánchez 2004) and nuclear DNA sequences Sánchez 2007a, 2007b). Pinnate morphologies such as Pseudopterogorgia spp. and Pinnigorgia spp. seemed to have evolved their branching morphologies separately (e.g. Figure 1). Surface layer sclerites under the resolution of SEM provided a good character to differentiate gorgoniid clades with very similar external characters.

The molecular contribution
The corroboration of Filigorgia as a genus apart from Leptogorgia is another example where recent molecular results have improved our understanding of some morphological characters (Fukami et al. 2004). The validation of the genus Filigorgia is also supported with previous molecular studies. Mitochondrial DNA sequences from the mut-S homolog gene (MSH1) show additional classification problems within the Gorgoniidae (A. LePard and S. France, personal communication). Although focusing on Leptogorgia, they found that supposed Leptogorgia ''outgroups'' such as Pseudopterogorgia, Pacifigorgia, and Eugorgia were intermingled with various clades of Leptogorgia, which clearly show that some of the African fauna from Leptogorgia sensu lato are not reciprocally monophyletic. Williams and Lindo (1997) provided a detailed comparison of spindles of Leptogorgia spp. with several gorgoniid genera showing that scaphoid sclerites are not the best character to differentiate gorgoniid genera due to their high similarity, particularly with Indo-Pacific species of Pseudopterogorgia. Although it was not a goal of this paper, it was clearly noticed that Pacific and Atlantic Pseudopterogorgia spp. surface sclerites were not homologous (e.g. Figures 3A  versus 4E), and deserve further study.
Finally, it is important to note that some genera assigned to the Gorgoniidae need careful revision (hopefully both taxonomical and molecular) in order to be unambiguously assigned to this family, such as in the case of Hicksonella and Rumphella (Grasshoff and Alderslade 1997), as well as Adelogorgia and Eunicella, whose sclerites are not likely homologous to the species presented here. In addition, there are a number of plexaurid genera such as Swiftia (Goldberg 2001) and Plexaurella that, with recent molecular analyses (Sánchez et al. 2003b;Wirshing et al. 2005;McFadden et al. 2006), have been found to resemble some gorgoniids.
The results presented here are not by any means definitive. Additional taxa and further tests of relatedness and monophyly can easily be done using the approach used here. Octocorals are often difficult to identify or classify because of a lack of reliable sets of diagnostic characters. The inclusion of molecular characters coupled with SEM provides a replicable approach.