In his revision of the soft coral genus Sinularia, Verseveldt (1980) mentioned three stalked Sinularia species with digitiform lobules, leptoclados-type surface clubs and unbranched interior spindles. These are Sinularia firma Tixier-Durivault, 1970, Sinularia leptoclados (Ehrenberg, 1834), and Sinularia maxima Verseveldt, 1971. Subsequently, comparison of material collected from New Caledonia (RMNH Coel. 10447-10449) to type material proved Sinularia firma to be an encrusting species. The original description of Sinularia firma was based on a tiny fragment that obscured the colony growth form. In the key of Verseveldt (1980: 12) Sinularia leptoclados and Sinularia maxima were separated from each other by colony growth form, Sinularia maxima with robust lobes, up to 120 mm high, and Sinularia leptoclados with shorter ones. Verseveldt based his description of Sinularia leptoclados on a RMNH specimen from the Red Sea, without mentioning its catalogue number. He certainly did not have the type specimen, as he states that he failed to find that in museum collections (Verseveldt: 9). Additionally, in his revision Verseveldt synonymized Sinularia leptoclados var. gonatodes Kolonko, 1926, with Sinularia leptoclados and stated the species exhibited an Indo-Pacific distribution.

The first two authors have based their identifications of Sinularia leptoclados on the microscope slides of Verseveldt at their disposal, and following the Sinularia revision of Verseveldt (1980), have considered Sinularia leptoclados specimens to be stalked with finger-like lobules and variable leptoclados-type clubs in the surface layer of the colony. The results have been published in a series of studies (see below) that have further supported Verseveldt’s (1980) statement that the species is widespread in the Red Sea and in the Indo-West Pacific area.

Alderslade and Shirwaiker (1991) were the first after Verseveldt’s (1980) revision to describe another species with characterssimilar to Sinularia leptoclados, their Sinularia kavarattiensis from the Laccadive Archipelago, India. They compared Sinularia kavarattiensis with the holotype of Sinularia leptoclados var. gonatodes and considered the many small spindles present in the surface layer of the lobes of the latter asa major difference between the two species. Later on, Manuputty and Ofwegen (2007) described three species from Ambon (Indonesia) which resembled Sinularia leptoclados: Sinularia acuta, Sinularia corpulentissima and Sinularia longula. In that studythey used for comparison a specimen from Ambon (RMNH Coel. 38426), considered by them to be Sinularia leptoclados.

McFadden et al. (2009), the first molecular study of the genus Sinularia ever conducted, discovered that specimens from Australia identified as Sinularia leptoclados by P. Alderslade (NTM C5421) and the first author (NTM C14492, 14519-21) differed genetically from Red Sea specimens identified as Sinularia leptoclados by the second author (ZMTAU CO 34095). This unexpected finding prompted us to re-examine the Sinularia leptoclados collections of the RMNH and ZMTAU and to search for the type material of this widespread species (e.g., Verseveldt). Fortunately, we discovered the type specimen of Sinularia leptoclados still exists in the ZMB, probably overlooked by Verseveldt, while revising the genus, most likely because it was labelled as Lobularia leptoclados Ehrenberg, 1834. After examination of its sclerites and comparison to RMNH and ZMTAU material identified as Sinularia leptoclados it became obvious that this species does not exhibit an Indo-West Pacific distribution as stated by Verseveldt (1980), but is rather limited to the Red Sea and eastern Indian Ocean. Material wrongly assigned to Sinularia leptoclados from other parts of the Indo-Pacific by the two first authors proved to be a mixture of misidentifications and as yet undescribed species. Interestingly, the specimens from Australia that were erroneously identified as Sinularia leptoclados have sclerites and a colony morphology that closely resemble that species. However, certain small morphological differences, its unique genetic haplotype, and the now disjunct distribution (Red Sea and western Indian Ocean vs. Pacific Ocean, Australia), convinced us to describe this material as a new species.

While collecting new material of Sinularia leptoclados at Eilat, northern Gulf of Aqaba, Red Sea, we unexpectedly found two other species with leptoclados-type clubs and leptoclados-like colony shape: Sinularia verseveldti Ofwegen, 1996 (Fig. 5f), so far only known from the Pacific, and a yet undescribed species which is described here.

Sequences for mtMutS and COI (including igr1) were available or newly obtained for 31 specimens representing 19 morphospecies of Sinularia belonging to clade 5C; 28S rDNA sequences were obtained for all but four specimens (Table 1). mtMutS (735 nt) and COI (888 nt) sequences were concatenated for a total mitochondrial gene alignment of 1623 nt. 28S sequences ranged from 797–799 nt in length for a total alignment length of 801 nt. Maximum likelihood and Bayesian analyses resulted in identical tree topologies for all three data sets (mt genes only, 28S only, all three genes combined). Support values were generally somewhat stronger for Bayesian analyses, however, and several nodes that were not supported by maximum likelihood (bootstrap values <50%) nonetheless had Bayesian posterior probabilities >0.9 (Figs 16, 17). All alignments and trees have been submitted to TreeBASE (www.treebase.org ).

Within Sinularia species with leptoclados-type clubs (clade 5C), genetic distances (uncorrected p) among recognized morphospecies range from only 0–1.7% for mtMutS, 0–0.8% for COI and 0–1.4% for 28S rDNA. Despite these relatively low levels of genetic differentiation among taxa, several moderately- to well-supported clades appear in both the mitochondrial and 28S gene trees (Fig. 16). Sinularia maxima and Sinularia corpulentissima share identical mt and 28S haplotypes with one another, but are well differentiated from all other species in clade 5C. Sinularia acuta, Sinularia longula and Sinularia molesta are also very similar to one another genetically (Sinularia molesta and Sinularia acuta share identical mt and 28S haplotypes), and form a well-supported clade in both trees. Finally, Sinularia erecta is genetically distinct, separated from all other species by genetic distances of >0.8% at mtMutS (28S was not available for Sinularia erecta).

Two additional clades are moderately supported by the combined analysis of the mt and 28S genes (Fig. 17); the species in these clades also group together in the separate analyses, but with low bootstrap support (<50%) (Fig. 16). Sinularia penghuensis, Sinularia bisulca, Sinularia robusta, Sinularia digitata and Sinularia slieringsi comprise one of these moderately-supported clades (Fig. 17); these species share identical or nearly identical 28S sequences (28S was not available for Sinularia slieringsi) (Fig. 16b). Within the mt gene tree (Fig. 16a) they constitute two distinct clades, one comprised by Sinularia robusta, Sinularia digitata and Sinularia slieringsi and the other by Sinularia penghuensis, Sinularia bisulca and Sinularia daii. The latter is, however, distinct from all other species at 28S, and falls outside of this clade in the combined analysis (Fig. 17). Sinularia leptoclados, Sinularia abrupta, Sinularia australiensis sp. n. and Sinularia densa also form a moderately-supported clade in the 28S tree (Fig. 16b) and in the combined tree (supported by Bayesian but not maximum likelihood analyses; Fig. 17), but their relationship is unresolved in the mt tree (Fig. 16a). Sinularia australiensis sp. n. and Sinularia abrupta share identical 28S haplotypes, but differ from Sinularia leptoclados by 0.3%. Sinularia australiensis sp. n. differs from both Sinularia leptoclados and Sinularia abrupta by 0.1% and 0.1-0.2% at mtMutS and COI respectively.

The relationships among the remaining species in the clade — Sinularia verseveldti, Sinularia wanannensis and Sinularia eilatensis sp. n. — were poorly resolved and exhibited some incongruence between the mitochondrial and 28S gene trees. Sinularia wanannensis, all four specimens of Sinularia verseveldti, and two specimens (ZMTAU Co 35303, ZMTAU Co 35304) that were tentatively assigned to Sinularia eilatensis sp. n. share identical or nearly identical mtMutS and COI haplotypes, and cluster together within the mt tree (but with bootstrap values <50%). Two specimens of Sinularia eilatensis sp. n. (ZMTAU Co 35305, ZMTAU Co 35260) fall outside of that group, and differ from it by >0.5% at mtMutS (Fig. 16a). At 28S, however, ZMTAU Co 35303 and ZMTAU Co 35304 are genetically identical to both individuals of Sinularia eilatensis sp. n., and those four specimens form a moderately-supported clade together with Sinularia verseveldti ZMTAU Co 35309 (Fig. 16b). Two additional specimens of Sinularia verseveldti share identical 28S haplotypes with Sinularia wanannensis. The combined tree reflects the topology of the mt gene tree, and shows the separation of Sinularia eilatensis sp. n. (ZMTAU Co 35305, ZMTAU Co 35260) from ZMTAU Co 35303, ZMTAU Co 35304 and all other species (Fig. 17).

Our findings indicate that specimens of the same species generally shared identical or nearly identical sequences at all three loci. The only exceptions were the two distinct mitochondrial haplotypes of Sinularia eilatensis sp. n. discussed above, and the four specimens of Sinularia verseveldti. All Sinularia verseveldti shared identical or nearly identical mtMutS and COI sequences, but differed at 28S. Most of these differences, however, reflected polymorphic nucleotide positions at which one or more specimens exhibited heterozygosity. For example, at position 533 of the 28S alignment, ZMTAU Co 35309 and Coel. 40842 had C, Coel. 40843 had T, and Coel. 40844 had both C and T. A total of 8 such heterozygous nucleotide sites among the four Sinularia verseveldti specimens contribute to their disjunct distribution within the 28S and combined trees.

The two new species described here are supported both by morphological characters and by the molecular analysis. Although Sinularia australiensis sp. n. is similar genetically to Sinularia leptoclados and both belong to the same sub-clade within Sinularia clade 5C, they differ at all three of the loci sequenced here. Furthermore, the 28S and combined analyses suggest that Sinularia leptoclados and Sinularia australiensis sp. n. are not sister taxa, but that Sinularia australiensis sp. n. is closer to Sinularia abrupta, a species with which it shares a 28S haplotype. The disjunct geographical distribution between Sinularia leptoclados, which occurs in the Red Sea and western Indian Ocean, and Sinularia australiensis sp. n. from Australia, further supports their distinction.

Although sympatric with Sinularia leptoclados in the Red Sea, Sinularia eilatensis sp. n. is clearly distinct from that species, both morphologically and genetically. Within clade 5C, Sinularia eilatensis sp. n. is most similar genetically to the geographically widespread Sinularia verseveldti and to Sinularia wanannensis, a species recently described from Taiwan (Ofwegen and Benayahu 2012). Morphologically, however, Sinularia eilatensis clearly differs from Sinularia verseveldti and Sinularia wanannensis by its long polyp sclerites, up to 0.25 mm long in Sinularia eilatensis vs up to 0.15 mm long in the other two species. Undoubtedly, the phylogenetic relationships among these three species need further investigation. In particular, the conflicting phylogenetic signals obtained from the mitochondrial and nuclear genes suggest the possibility of past hybridization events between Sinularia verseveldti and Sinularia eilatensis sp. n. Two specimens from the Red Sea (ZMTAU Co 35030 and ZMTAU Co 35304) appear morphologically to belong to Sinularia eilatensis sp. n. and have the same 28S sequence as that species but share a distinct mitochondrial haplotype with Sinularia verseveldti. This observed mito-nuclear discord could reflect a hybrid origin of these specimens, as has been suggested for some other octocorals (reviewed in McFadden et al. 2010). In addition, the polymorphism observed at the 28S locus in Sinularia verseveldti could be indicative of recent hybridization events involving this species, although it could also be the result of incomplete lineage sorting following recent speciation. The possible hybrid origin of ZMTAU Co 35303 and ZMTAU Co 35304 should be investigated further using single-copy nuclear gene markers.

Previous molecular systematic work on Sinularia and other octocoral genera has highlighted the inadequacies of mitochondrial gene markers for species discrimination and species-level phylogenetic analyses in the group (McFadden et al. 2009, 2011). Although both mtMutS and COI effectively distinguish genera and distinct clades within genera, neither gene is variable enough to distinguish all congeneric species pairs unequivocally. The region of the nuclear 28S rDNA gene we sequenced exhibits somewhat greater variability than mtMutS in some genera of the family Alcyoniidae (Benayahu et al. in press), but did not distinguish among all of the morphospecies of Sinularia examined in the current study. Despite the relatively small genetic distances separating morphospecies and the low resolution of the resulting phylogenies, we believe the analysis presented here adequately supports the distinctions of the new species that are the focus of this study. Development of additional, more variable molecular markers, will be necessary in order to fully resolve the relationships among morphospecies in Sinularia clade 5C and to address the possibility of hybridization among them.