DNA barcoding of xeniid soft corals (Octocorallia: Alcyonacea: Xeniidae) from Indonesia: species richness and phylogenetic relationships

We conducted the first ever survey of xeniid octocorals in the Indonesian Archipelago, centre of the highly biodiverse Coral Triangle region of the Indo-Pacific. Among 48 xeniid specimens collected from Lembeh Strait, North Sulawesi, we identified 26 morphospecies belonging to six genera based on assessment of the morphological characters traditionally used for xeniid taxonomy. Multilocus DNA barcodes obtained from 23 morphospecies clustered into 21 molecular operational taxonomic units (MOTUs) separated by average genetic distance values >0.3%. The overall concordance between morphospecies and MOTUs was 91%; just one pair and one trio of morphospecies were not distinguished by the DNA barcodes. A molecular phylogenetic reconstruction of family Xeniidae based on four loci (COI, mtMutS, ND2, 28S rDNA) supported the distinction of Anthelia and Cespitularia+Efflatounaria from all other xeniid genera. Although the remaining genera for which molecular data were available (Asterospicularia, Heteroxenia, Ovabunda, Sansibia, Sarcothelia, Sympodium, Xenia) belonged to a single, well-supported clade, the phylogenetic relationships among them were poorly resolved. Species of Xenia were distributed among three different sub-clades within which they were paraphyletic with Ovabunda (clade X1), Heteroxenia (clade X2) and Sansibia plus Sarcothelia (clade X3). No morphological characters have yet been identified that differentiate these three phylogenetically distinct clades of Xenia. Use of molecular barcodes to discriminate species will facilitate future ecological studies of Xeniidae, a group that has been shown to opportunistically monopolize disturbed reef habitat.


Introduction
Soft corals of the family Xeniidae are common and frequently dominant space-occupying organisms on shallow coral reefs. Small, fleshy xeniid colonies can spread asexually via stolonal growth (Karlson et al., 1996), colony fission or active movement (Benayahu & Loya, 1985) to form extensive monocultures that may carpet tens to hundreds of square metres of reef (Benayahu & Loya, 1981;Dinesen, 1983;Fabricius, , 1998. Xeniids are rapid, opportunistic colonizers of disturbed habitats -particularly coral rubble -and are capable of overgrowing or excluding recruits of scleractinian corals as well as other octocorals (Fabricius, 1998;Fox et al., 2003;Tilot et al., 2008;Wood & Dipper, 2008). With increasing climate change-related and anthropogenic disturbances to coral reefs, there is some concern that diverse communities of reef-building scleractinians could be replaced by low diversity monocultures of xeniids and other soft corals (Fabricius, 1998;Wood & Dipper, 2008). It has also been suggested that over longer time periods soft corals may facilitate the recruitment and recovery of scleractinians, for instance by stabilizing loose substrate (Fox et al., 2003). To date, however, no ecological studies have monitored the long-term population dynamics of octocorals or their interactions with scleractinians in reef communities recovering from disturbance (Fabricius, 1998).
Despite their apparent ecological importance in disturbed reef communities, few ecological studies of xeniid octocorals have been conducted outside of the Red Sea (Benayahu & Loya, 1977, 1981, 1985. We lack even basic information about their diversity and taxonomy throughout much of the Indo-Pacific. Among the notoriously difficult-to-identify octocorals (Fabricius & Alderslade, 2001) xeniids are especially problematic. In Xeniidae, the calcium carbonate sclerites, whose shape and size are the primary taxonomic characters used to Correspondence to: Catherine S. McFadden. E-mail: mcfadden@ g.hmc.edu distinguish species in most other groups of octocorals, are homogeneous and simple in structure, usually consisting solely of minute oval platelets less than 20 mm in diameter (Fabricius & Alderslade, 2001). As a result of this lack of variation in sclerite form, xeniid taxonomy has relied on a handful of soft tissue characters, such as the size of colonies, polyps and tentacles, numbers of rows of pinnules and numbers of pinnules per row on the tentacles, as well as characters such as colour and pulsation behaviour that can only be assessed in live material . All of these characters are variable within species, overlap in range among species, and the degree to which they are influenced by the environment has never been assessed (Reinicke, 1997). There is growing recognition that the crystalline ultrastructure of the sclerite surface may be taxonomically important (Alderslade, 2001;Janes, 2008;Benayahu, 2010;Aharonovich & Benayahu, 2011;Hal asz et al., 2013), but this character can only be evaluated with the use of scanning electron microscopy.
Increasingly, molecular data are being used to evaluate species boundaries and to assist in the identification of groups such as octocorals in which morphological characters often do not distinguish species reliably. DNA barcoding has been proposed as a tool to facilitate identification of species (Hebert et al., 2003), and more controversially (e.g. DeSalle et al., 2005;Will et al., 2005;Rubinoff et al., 2006), to discover new species. As a consequence of their unusually slow rates of mitochondrial gene evolution (Shearer et al., 2002;Hellberg, 2006;Huang et al., 2008;Chen et al., 2009), however, the development of species-specific DNA barcodes for anthozoan cnidarians has lagged behind that of other groups (Bucklin et al., 2010). Nonetheless, several recent studies have identified multilocus barcodes that distinguish morphospecies of octocorals with relatively high (70-80%) success Baco & Cairns, 2012). In particular, a barcode that combines cytochrome oxidase I (COI) plus an adjacent intergenic region (igr1), the octocoral-specific mitochondrial mutS gene (mtMutS) and a fragment of the nuclear 28S ribosomal RNA gene has been shown to distinguish molecular operational taxonomic units (MOTUs; Floyd et al., 2002) that are in agreement with >70% of morphospecies identifications across a wide taxonomic range of shallow-water octocorals (McFadden et al., 2014). In most octocoral taxa, we cannot currently determine if the~30% of cases in which there is disagreement between molecular and morphological species identifications are due to inadequacies of the molecular markers or instead to flawed interpretations of morphologically defined species boundaries.
Both the ecology and taxonomy of Xeniidae have been widely studied within the Red Sea, with 29 species belonging to five genera having been recorded there (Reinicke, 1997). In contrast, relatively little is known of xeniid diversity in the Coral Triangle, that area centred on the Indonesian Archipelago within which the species diversity of zooxanthellate scleractinian corals reaches its maximum (Veron et al., 2009). Schenk (1896) described eight species of Xenia Lamarck, 1816 and one Anthelia Lamarck, 1816 from Ternate Is. in eastern Indonesia. Working from existing museum collections, Roxas (1933) later recorded approximately 40 species of xeniids in five genera from Puerto Galera, Mindoro, Philippines. The only modern-day biodiversity surveys from the Coral Triangle region that have identified xeniids to species are from the Bismarck Sea (13 species in five genera; Ofwegen, 1996) and Taiwan (11 species in six genera; Benayahu et al., 2004). Although a few other recent studies have documented the occurrence of particular species of Xeniidae in Indonesia (Verseveldt, 1960;Imahara, 1996;Stemmer et al., 2013), to date no systematic biodiversity surveys of the family have been conducted anywhere within the Indonesian Archipelago. Here we document a diverse assemblage of Xeniidae found in Lembeh Strait, North Sulawesi. We use both morphological and molecular criteria to estimate species richness, discuss cases of non-concordance between the two types of evidence, and present the first molecular phylogenetic reconstruction of the family Xeniidae to include representatives from 10 of the 17 known genera.

Collection and morphological assessment
Xeniid octocorals were collected in May 2009 from Lembeh Strait, between Lembeh Island and the coast of North Sulawesi, Indonesia. Dives were made to a maximum depth of 30 m using SCUBA at 28 stations spanning approximately 5 km of coastline on either side of the northern half of the Strait. Colonies were photographed and measured in situ prior to being collected; they were preserved in 90% EtOH immediately following collection. Vouchers of all material have been deposited at the California Academy of Sciences (CASIZ).
Specimens were identified to morphospecies based on assessment of the following character set typically used for taxonomy of Xeniidae (e.g. Reinicke, 1997;Hal asz et al., 2013): overall colony size and growth form; length of polyps and tentacles, numbers of rows of pinnules on the tentacles, and numbers of pinnules in the aboral row; size, shape and crystalline ultrastructure of sclerites in the polyps and colony stalk; and colour in life (Table 1). To facilitate counting pinnules, polyps were stained with acid fuchsin and mounted on glass microscope slides using Durcupan AMC mounting medium (Fabricius & Alderslade, 2001). Sclerites were obtained by dissolving tissue 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 6480LV Scanning

Molecular phylogenetic analyses
Extraction of DNA from ethanol-preserved tissue samples, PCR amplification and sequencing of the mitochondrial mtMutS (msh1), COI + igr1 and ND2 genes followed the protocols published in McFadden et al. (2006,2011). In addition, we sequenced an approximately 750 nt fragment of the 28S nuclear ribosomal gene using primers 28S-Far and 28S-Rar (McFadden & Ofwegen, 2013). Sequences were aligned using the L-INS-i method in MAFFT (Katoh et al., 2005), and pairwise genetic distances (Kimura 2-parameter) among specimens were calculated using the DNADist program in PHYLIP v. 3.69 (Felsenstein, 2005). MOTHUR v. 1.29 (Schloss et al., 2009) was used to cluster sequences into MOTUs based on an average neighbour distance threshold of 0.3%, a value that has been shown previously to yield a high concordance between molecular and morphological identifications in other octocoral families (McFadden et al., 2014). The concordance of species identifications was estimated as the percentage of specimens for which molecular and morphospecies classifications were in agreement. For example, if a MOTU included three specimens of morphospecies A and one specimen of morphospecies B it would be counted as three concordant identifications and one non-concordant identification for a total concordance of 75%. If a fourth individual of morphospecies A belonged to a different MOTU it would also be counted as a non-concordant identification, for an overall concordance of 60% among the five specimens. GARLI 2.0 (Zwickl, 2006) was used to construct Maximum Likelihood trees following selection of appropriate models of evolution using Modeltest 3.0 (Posada & Crandall, 1998). Because of a lack of congruence between mitochondrial and nuclear gene trees, we constructed two separate trees, one for 28S rDNA (GTR+I+G model) and another for the three mitochondrial genes (mtMutS, COI +igr1, ND2) concatenated (HKY+I+G model). Analyses were run for 1000 bootstrap replicates. Bayesian analyses of both datasets were conducted separately with MrBayes v. 3.2.1 (Ronquist et al., 2012) using either the GTR+I+G (28S) or HKY+I+G (mt genes) models of evolution; analyses were run for 2 million generations (until standard deviation of split partitions < 0.005) with a burn-in of 25% and default Metropolis coupling parameters (i.e. 2 runs, 4 chains (3 heated), sample frequency ¼ 500 generations). Several xeniid sequences published previously by the first author (McFadden et al., 2006) were included in the phylogenetic analyses to increase the representation of xeniid genera (Appendix S1, see online supplemental material, which is available from the

Morphospecies identifications
Among the 48 xeniid specimens collected, 26 distinct morphospecies belonging to six genera were identified (Table 1). These included 16 morphospecies of Xenia, four morphospecies of Anthelia, three morphospecies of Heteroxenia K€ olliker, 1874, and one morphospecies each of Cespitularia Milne-Edwards & Haime, 1857, Sansibia Alderslade, 2000 and Sympodium Ehrenberg, 1834. Fifteen morphospecies were tentatively identified to nominal taxa by comparison to original species descriptions or, in a few cases, to type material (Table 1). Eleven morphospecies, however, could not be matched to existing species accounts, and are provisionally considered to represent new, undescribed taxa. A complete taxonomic treatment of the material will be published elsewhere (Janes et al., pers. comm.).

Species discrimination using DNA barcodes
DNA sequences were obtained from 44 specimens representing 24 of the 26 morphospecies collected in Lembeh; complete 4-locus genotypes (~2996 bp) were obtained for 35 specimens (Appendix S1, see supplemental material online). No DNA sequences were obtained from the morphospecies identified as Xenia felicianoi Roxas, 1933 and Anthelia sp. 2, and only ND2 was obtained from Sympodium caeruleum Ehrenberg, 1834. ND2 exhibited less variation than the other three loci, and consequently was less informative as a species-specific marker ( Table 2). For example, eight morphospecies belonging to three different genera all shared an identical ND2 haplotype. Among the various combinations of markers and genetic distance thresholds we tested (Table 2), both species richness accuracy (MOTUs: morphospecies ratio) and concordance were highest using the combined mt + 28S barcode (mt ¼ COI + igr1 + mtMutS; McFadden et al., 2011) at an average genetic distance threshold of 0.3%. This DNA barcode distinguished 21 MOTUs among the 23 morphospecies for which we had complete or partial sequence data. Morphospecies identified as Xenia sp. 1, X. sp. 2 and X. sp. 6 belonged to the same MOTU, as did Xenia sp. 3 and X. sp. 4. One of the seven specimens identified as X. membranacea Schenk, 1896 (CASIZ 184563) belonged to a separate MOTU from the remaining six. In all other cases, morphospecies identifications were concordant with MOTUs, for an overall concordance of 91% (Table 2).

Molecular phylogenetic analysis
Phylogenetic reconstructions using maximum likelihood and Bayesian methods yielded similar tree topologies, and differed only in relative support for some of the deeper nodes within the tree (Figs 1, 2). In general, support values from maximum likelihood tended to be lower than those from Bayesian analyses. The separate mitochondrial and 28S rDNA gene trees supported most of the same clades, and differed only in the phylogenetic placement of a few species. All analyses supported three well-defined, major clades within Xeniidae: a monophyletic Anthelia, further distinguished from all other genera by a unique 6-bp insertion in the intergenic region just upstream of ND2 and a unique amino acid insertion in mtMutS (Fig. 1); Cespitularia plus Efflatounaria Gohar, 1939; and a third large clade comprising all of the remaining genera. Within the latter clade, five distinct sub-clades were resolved, although levels of support for these sub-clades  (Figs 1, 2). Clade X1 included X. fisheri Roxas, 1933, X. lepida Verseveldt, 1971, X. viridis Schenk, 1896, X. hicksoni Ashworth, 1899 and X. sp. 7 along with members of genus Ovabunda Alderslade, 2001. Strong support for this clade was provided by a unique 24-bp deletion in the intergenic region upstream of the ND2 coding region that was shared only by these species (Fig. 1). Both mt and 28S gene trees positioned the small clade comprising Xenia sp. 3 and X. sp. 4 as the sister to clade X1, although the 28S tree supported that relationship only weakly and neither of those two species shared the 24-bp ND2 deletion with clade X1 (Fig. 1). The two gene trees also differed with respect to the position of X. ternatana Schenk, 1896, which lay within clade X1 as a sister to X. fisheri in the mt gene tree, but outside of that clade in the 28S tree. Xenia ternatana does share the 24-bp deletion with other clade X1 species, supporting its position within rather than outside of that clade (Fig. 1).
Clade X2 comprised all four species of Heteroxenia, plus X. puerto-galerae Roxas, 1933 and Xenia sp. 1, X. sp 2 and X. sp. 6. The 28S analysis also included X. lillieae Roxas, 1933 within this clade, but with lower overall support values for the clade. All nine of these species, however, shared a unique single amino acid deletion in mtMutS, a synapomorphy that supports the inclusion of X. lillieae in clade X2 (Fig. 2). The mt gene tree provided strong support for the monophyly of Heteroxenia as the sister to the Xenia species within the clade, while the 28S tree only supported the monophyly of three of the four Heteroxenia species, excluding H. elizabethae K€ olliker, 1874 (Fig. 2).
Clade X3 encompassed X. membranacea Schenk, 1896, X. kusimotoensis Utinomi, 1955 and Xenia sp. 5, a group whose monophyly was well supported by the 28S analyses (Fig. 2). The mt gene tree weakly supported the inclusion of these species in a larger clade with members of Sansibia and Sarcothelia, two genera for which 28S sequences were unavailable (Fig. 1).
In addition to the phylogenetic positions of X. ternatana, X. lillieae and H. elizabethae, the relationship between the family Xeniidae and the outgroup taxa also differed between 28S and mt gene trees. The mt tree strongly supported the monophyly of Xeniidae (Fig. 1). In the 28S tree, however, Coelogorgia palmosa (family Coelogorgiidae) fell within Xeniidae as the sister taxon to Cespitularia, albeit with low maximum likelihood bootstrap support (Fig. 2).

DNA barcoding in Xeniidae
The concordance between morphospecies identifications and molecular classification into MOTUs obtained here using the mt + 28S DNA barcode is the highest yet achieved in any application of DNA barcoding to octocorals. McFadden et al. (2011) were able to distinguish 69% of morphospecies using the mt barcode (COI + igr1 + mtMutS) in a biodiversity study of Red Sea octocorals. Concordance between morphospecies and molecular identifications was a slightly higher 74% when the combined mt + 28S barcode was used to discriminate species in a taxonomically comprehensive survey of octocorals in Palau (McFadden et al., 2014). Baco & Cairns (2012) resolved 83% of morphospecies belonging to the deepwater octocoral genus Narella using an extended mitochondrial gene barcode of mt + ND2. The 91% concordance we obtained here using the mt + 28S barcode is not, however, directly comparable to our earlier estimates using the mt and mt + 28S markers (McFadden et al., , 2014 which may have underestimated the concordance between morphospecies and MOTUs. Both of those studies were conducted blind (morphospecies were assessed independently of MOTUs by separate researchers) and no attempt was made to subsequently reconcile any disagreements between methods. In the present study, however, we both re-evaluated morphospecies assignments and re-sequenced some specimens in an attempt to reconcile the two different types of evidence for species boundaries. Only in the cases of Xenia sp. 1, 2 and 6, and Xenia sp. 3 and 4 were we unable to bring the morphological and molecular data into agreement.
A recent study by Stemmer et al. (2013) tested the ability of two different loci, a fragment of the mitochondrial ND6/ND3 region and the nuclear SRP54 gene, to discriminate species of xeniid octocorals. Their tree-based analysis is difficult to compare directly to our results, but if their 'distinct clades' are considered to represent MOTUs, they found that ND6/ND3 discriminated only six MOTUs and SRP54 discriminated nine MOTUs among the 14 morphospecies they identified. For SRP54, it appears that only two morphospecies were fully concordant with MOTUs -all other morphospecies were either divided among multiple MOTUs or multiple morphospecies belonged to the same MOTU. The much higher concordance between morphospecies and MOTUs we found among the samples from Lembeh suggests that mt + 28S is a more reliable species-specific barcode for xeniids than the more variable SRP54 gene.
Three unidentified morphospecies of Xenia from Lembeh, designated X. sp. 1, X. sp. 2 and X. sp. 6, belonged to the same MOTU, sharing identical mtMutS and 28S sequences, and differing from one another by genetic distances of only 0.1% at COI. These three morphospecies are, however, morphologically quite distinct: X. sp. 6 has two rows of pinnules plus a partial third row, whereas X. sp. 1 and X. sp. 2 both have only a single row of pinnules. X. sp. 1 and X. sp. 2 in turn differ from one another in average numbers of pinnules (18-20 vs. 28-31), and in the density as well as the ultrastructure of sclerites in the colony (Table 1). Two additional unidentified morphospecies, Xenia sp. 3 and X. sp. 4, also belonged to a single MOTU, sharing identical genotypes at all loci. Although these two morphospecies both have two rows and similar numbers of pinnules, they differ markedly in sclerite complement: X. sp. 4 has sclerites distributed densely throughout the colony, whereas X. sp. 3 lacks sclerites entirely (Table 1). Based on the morphological characters traditionally used for xeniid taxonomy, each of these five morphospecies would be considered distinct. Further integrative taxonomic studies are needed to determine if these are indeed different species that have not yet diverged genetically at the particular loci we sequenced, or, alternatively, if the morphological differences we observed could be the result of morphological plasticity or ontogenetic variation within species. Support for the latter case will necessitate a radical reinterpretation of 150 years of xeniid taxonomy.

Phylogenetic relationships within Xeniidae
The molecular phylogeny presented here is the first to be published for Xeniidae that includes more than one or a small number of species from more than just a few genera (McFadden et al., 2006Stemmer et al., 2013). The results from the mitochondrial gene analyses strongly support the monophyly of the family. Although the earlier analyses of McFadden et al. (2006) using just mtMutS and ND2 placed Anthelia outside of Xeniidae, the one Anthelia specimen sequenced for that study is now suspected to have been either misidentified or a contaminant. Neither the ND2 nor mtMutS sequences from that individual fall within the clade of Anthelia presented here (C.S. McFadden, unpubl. data). The molecular phylogeny also supports the monophyly of the genera Anthelia and Cespitularia+Efflatounaria, and positions them as sister clades to all remaining xeniid genera. The morphological distinctions between Cespitularia and Efflatounaria are unclear (Fabricius & Alderslade, 2001;, therefore it is not surprising that there also appears to be little or no phylogenetic distinction between them. The phylogenetic relationships among the six other genera of xeniids included in our analyses are poorly resolved, although both Sympodium and Asterospicularia appear to be well differentiated genetically from other genera (Fig. 1, 2). Sympodium is unique among xeniids in having fully retractile polyps, while the stellate sclerites of Asterospicularia are so different from those typical of other Xeniidae that until recently it was placed in its own family (Alderslade, 2001).
Neither the mitochondrial nor the 28S rDNA gene trees support the monophyly of Xenia, with both analyses instead suggesting that this genus comprises at least three distinct clades that may be paraphyletic with other genera. The most strongly supported clade, X1, is paraphyletic with Ovabunda, a genus that was recently split from Xenia on the basis of its distinctive sclerite microstructure (Alderslade, 2001). There is, however, relatively little genetic differentiation between Ovabunda species and several species of Xenia from the Red Sea, represented in our analyses by X. hicksoni (Haverkort-Yeh et al., 2013). A second clade of Xenia, X2, appears to be paraphyletic with Heteroxenia, a genus that differs from Xenia by having dimorphic polyps (i.e. siphonozooids in addition to autozooids). This trait may, however, vary seasonally or ontogenetically (Gohar, 1940;Achituv & Benayahu, 1990;Fabricius & Alderslade, 2001), and when siphonozooids are absent in Heteroxenia the two genera are indistinguishable. A third clade of Xenia, X3, may be paraphyletic with Sansibia and Sarcothelia, two similar genera in which polyps arise from a membrane rather than a stalk. The relationships among clades X1, X2, X3, Sympodium, Asterospicularia and several additional species of Xenia that do not fall cleanly into any one of the clades (e.g. X. sp. 3, X. sp. 4) are, however, poorly resolved in both gene trees. Additional evidence will be necessary to confirm and explore further the apparent polyphyly of Xenia, as well as the relationships among Xenia, Heteroxenia, Ovabunda, Sansibia and Sarcothelia. At present, no obvious morphological differences distinguish the three clades of Xenia from one another; they all include stalked colonies with monomorphic polyps situated on a distinct capitulum, whose sclerites are typical oval forms with a microstructure composed of dendritic rods ( Table 1).
The molecular phylogeny presented here does not include representatives of an additional seven genera of Xeniidae for which DNA samples are not currently available, including Bayerxenia Alderslade, 2001, Fasciclia Janes, 2008, Funginus Tixier-Durivault, 1978, Ixion Alderslade, 2001, Ingotia Alderslade, 2001, Orangaslia Alderslade, 2001and Yamazatum Benayahu, 2010. Six of these genera have been described only recently to accommodate species that have unique sclerite microstructure and/or multiple sclerite forms within a colony. Their phylogenetic relationships to other genera remain unknown. Ixion, Ingotia and Orangaslia all share a similar colony growth form with Anthelia but harbour unique sclerite forms (Alderslade, 2001), Fasciclia has a stalked growth form similar to Xenia but with sclerites resembling those of Anthelia (Janes, 2008), and Yamazatum also resembles Xenia but has a unique sclerite form (Benayahu, 2010). Finally, Bayerxenia shares a stalked growth form and polyp dimorphism with Heteroxenia, but differs from that genus by having multiple distinct sclerite forms within a colony (Alderslade, 2001). Stemmer et al. (2013) included three morphospecies they identified as Bayerxenia in their molecular analyses, and could not distinguish them phylogenetically from other species of Heteroxenia and Xenia. Because they sequenced a different set of genes, we were unable to include their material in our analyses for comparison. The species we identified as Heteroxenia sp. 1 has sclerites with microstructure similar to that of Bayerxenia but lacks the multiple sclerite forms that are an important diagnostic character of that genus (Table 1). Heteroxenia sp. 1 was not distinguishable phylogenetically from other species of Heteroxenia. Although sclerite ultrastructure, which can be observed only by using SEM, has been proposed recently to be an important genus-level taxonomic character in Xeniidae (Alderslade, 2001;Janes, 2008;Benayahu, 2010;Aharonovich & Benayahu, 2011;Hal asz et al., 2013), the phylogenetic distinction of species with unique sclerite ultrastructures has yet to be demonstrated.

Xeniid biodiversity in the Coral Triangle
To date, information on the xeniid fauna of the central Indo-Pacific has been limited, especially compared with the extensive studies that have been carried out on this family in the Red Sea (Gohar, 1940;Benayahu, 1990;Reinicke, 1995Reinicke, , 1997Hal asz et al., 2013) and the western Indian Ocean (Benayahu et al., 2003;Janes, 2008). The results of our survey of Lembeh Strait suggest that this one very small area within the Indonesian Archipelago supports as many species as the entire Red Sea. Although 29 species of Xeniidae have been recorded within the Red Sea (and unverified records exist for another five species; Reinicke, 1997), recent taxonomic work has synonymized several species in genus Ovabunda (Hal asz et al., 2013), thereby reducing the estimated species richness of that region by four. Our estimates of the richness of xeniids in Lembeh Strait based on a combination of morphological and molecular data range from 23 to 26 species. It is likely that more extensive surveys over a broader geographic scale within the Coral Triangle will discover many additional species of xeniids. For instance, only two of the nine species described by Schenk (1896) from nearby Ternate Is. and only 10 of the~40 species recorded by Roxas (1933) from the Philippines were found in Lembeh Strait. Eleven of the 26 morphospecies we collected could not be matched to any nominal species of Xeniidae and may represent yet undescribed species. Taken together, the results of our survey plus previous records of Xeniidae suggest the potential for a high species richness within this understudied family in the Coral Triangle.
The high diversity of xeniid species on Indo-Pacific reefs in turn suggests the need for more detailed taxonspecific ecological studies. Studies that have reported instances of xeniids opportunistically recruiting to disturbed reef habitat and either inhibiting or facilitating the recovery of scleractinians have typically only identified the family (Wood & Dipper, 2008;Tilot et al., 2008) or genus (Fox et al., 2003). It is unknown, therefore, if a single opportunistic species of xeniid is responsible for monopolizing space on disturbed reefs, or if a diverse assemblage of ecologically similar species contributes to the high cover of xeniids in such situations. DNA barcoding using the mt + 28S marker promises to greatly facilitate the discrimination and identification of species in this taxonomically confusing group, and provide needed insights into the community dynamics of disturbed reef habitats.