Introduction

The Red Sea (RS) is a north–south orientated and semi-enclosed basin at the northwestern corner of the Indian Ocean (IO). It is connected to the IO only at the narrow (18 km) and shallow (137 m) Strait of Bab al Mandab (Sheppard et al. 1992). Despite its modest surface area (438,000 km2), the total coral reef area of the RS (c. 8890 km2) is similar to that of the Caribbean (c. 10,530 km2) and its reef system length (c. 2000 km) is comparable to that of the Great Barrier Reef (GBR) (c. 2300 km) (Berumen et al. 2013). The RS has long been recognized as a region of high marine biodiversity and endemism (Klunzinger 1870, 1877; Ekman 1953). The unique reef fauna of the RS has multiple evolutionary origins (DiBattista et al. 2016a), with some taxa originating within the RS during the Pleistocene glaciations (DiBattista et al. 2015; Coleman et al. 2016) and others diverging from their IO sister groups long before the last glacial cycles (DiBattista et al. 2013; Hodge et al. 2014).

In the RS, 5.5% of the 364 scleractinian coral species are endemic (DiBattista et al. 2016b). whereas levels of endemism for scleractinians is typically less than 2% in the IO (Obura 2012; Veron et al. 2015). However, our knowledge of the coral ecology and biodiversity of the RS is limited compared to other major reef systems, e.g., the GBR and the Caribbean (Berumen et al. 2013). While the Gulf of Aqaba (GofA) has been the source of notable and seminal works in the field of coral reef ecology and taxonomy (Loya et al. 2014), the restricted area of the GofA (<2% of the RS) may preclude extrapolation to the broader RS. RS corals attracted the attention of early taxonomists (e.g., Forskål 1775; Lamarck 1816; Ehrenberg 1834; Klunzinger 1877) and were later the subject of more detailed surveys and larger reference collections (e.g., Scheer and Pillai 1983; Sheppard and Sheppard 1991). These publications, representing milestones for coral taxonomy in the RS, are based on classical morphology-based taxonomy, providing detailed descriptions of coralla and corallite features at the macroscopic scale. Recently, molecular works coupled with analyses of the macro- and micromorphology of the skeletons have allowed the description of new reef corals in the RS (Terraneo et al. 2014; Arrigoni et al. 2015, 2016a, b, 2017) and have showed unexpectedly low genetic diversity for some genera previously thought to have high species richness in this region (Terraneo et al. 2016; Arrigoni et al. 2016c). Moreover, other genetic surveys have revealed several cases of genetic differentiation between Indian and Pacific coral populations previously obscured by traditional taxonomy (Stefani et al. 2011; Ladner and Palumbi 2012; Arrigoni et al. 2012; Kitano et al. 2014; Richards et al. 2016).

To date, the scleractinian coral family Agariciidae Gray, 1847 comprises seven genera (Kitahara et al. 2012) and 49 extant species (WoRMS 2016). The genus Pachyseris Milne Edwards & Haime, 1849 was also ascribed to the family (Vaughan and Wells 1943; Wells 1956; Veron and Pichon 1980). Nevertheless, recent molecular works provide evidence that Pachyseris is not closely related to the Agariciidae and, instead, is basal to the family Euphylliidae Alloiteau, 1952 (Fukami et al. 2008; Kitahara et al. 2010); the position of the genus remains uncertain. Within the family, the genera Coeloseris Vaughan, 1918, Dactylotrochus Wells, 1954, Gardineroseris Scheer & Pillai, 1974, and Pavona Lamarck, 1801 occur in the Indo-Pacific; Agaricia Lamarck, 1801 and Helioseris Milne Edwards & Haime, 1849 occur in the Western Atlantic; and Leptoseris Milne Edwards & Haime, 1849 occurs in both the Indo-Pacific and the Atlantic Ocean. Agariciids are common in tropical shallow-water coral reefs (<30 m) throughout their distribution range (Veron 2000; Waheed and Hoeksema 2014; Waheed et al. 2015). Moreover, representatives of Leptoseris and Agaricia may be a dominant component of mesophotic coral ecosystems (Fricke et al. 1987; Hinderstein et al. 2010; Rooney et al. 2010; Kahng et al. 2014) occurring in the tropics and subtropics between depths of 30 m to over 150 m (Kahng et al. 2010). For example, Leptoseris fragilis Milne Edwards & Haime, 1849 has been found at 145 m in the GofA (Fricke and Knauer 1986) and Leptoseris hawaiiensis Vaughan, 1907 at 165 m from Johnston Atoll (Maragos and Jokiel 1986). In the RS, species of Gardineroseris and Pavona are commonly found between 1 and 20 m depth, while most species of Leptoseris are more abundant in low-light conditions under reef overhangs or below 30 m, where they can be a major component of coral coverage (Scheer and Pillai 1983; Sheppard and Sheppard 1991). Notably, Leptoseris has been recorded from over 100 m depth (Fricke et al. 1987). Scheer and Pillai (1983) listed 15 species of Agariciidae from the Saudi Arabian RS, Sheppard and Sheppard (1991) listed 16 species from the Saudi Arabian RS, and Veron (2000) listed 19 species from the entire RS (see Table 1 for the complete list of species reported by these three publications). Subsequently, Leptoseris amitoriensis Veron, 1990 was reported for the first time in the RS by Ziegler et al. (2015).

Table 1 List of Agariciidae species recorded from the Red Sea (RS) by Scheer and Pillai (1983), Sheppard and Sheppard (1991), Veron (2000), and this study
Full size table

Despite the ecological importance of this family, its taxonomy and systematics are still almost entirely based on traditional studies of skeletal morphology (Dinesen 1980; Veron and Pichon 1980; Veron 2000). The definition of agariciid genera and species boundaries occurred prior to the introduction of molecular techniques that have revolutionized coral systematics at all taxonomic ranks (Kitahara et al. 2016). A detailed morpho-molecular approach by Kitahara et al. (2012) demonstrated that the solitary and deep-water species Dactylotrochus cervicornis (Moseley, 1881), previously ascribed to the Caryophylliidae Dana, 1846, is the oldest extant representative of Agariciidae. Phylogenetic investigations of the Hawaiian mesophotic Leptoseris and Pavona species based on the mitochondrial intergenic region between COI and 16S rRNA (IGR hereafter) showed that the two genera are not monophyletic and revealed several hidden species (Luck et al. 2013; Pochon et al. 2015). Other integrative works focusing on species delimitation among shallow-water species of Pavona found genetic distinction between Pavona chiriquiensis Glynn, Mate & Stemann, 2001 and Pavona varians Verrill, 1864 from Panama (Maté 2003), between Pavona cactus Forskål, 1775 and Pavona decussata Dana, 1846 from Mauritius (Moothien Pillay et al. 2006), and between Pavona maldivensis (Gardiner, 1905) and Pavona explanulata (Lamarck, 1816) from Malaysia (Waheed et al. 2015). However, the data produced in the above studies are only partially comparable because they are based on different molecular techniques or distinct loci with variable resolutions; they also typically represent single cases and do not address the whole spectrum of diversity within the family.

In this work, we present the first survey of the genetic diversity of shallow-water (<30 m) Agariciidae from the Saudi Arabian RS for the morphospecies encountered during different biodiversity surveys organized by the King Abdullah University of Science and Technology (KAUST). The skeletal macromorphology of the collected material was used to identify the specimens based on existing taxonomic references and, when possible, through comparison with type material. Two molecular markers, i.e., mitochondrial IGR and internal transcribed spacer 1 from nuclear ribosomal DNA (ITS1), were sequenced to define morphospecies monophyly and relationships. This study represents the first molecular phylogeny of Agariciidae from the RS, and in general from the IO, and will allow the comparison of agariciid genetic diversity and connectivity of the RS to other localities of the Indo-Pacific. The results highlight a disagreement between our identifications based on classical morphologic characters and DNA for most species, suggesting the need to include more locations, variable loci, and the integration of micromorphological data to disentangle the systematics of this group.

Materials and methods

Sampling

Between 2012 and 2016, 91 samples belonging to the family Agariciidae were collected from several localities along the Saudi Arabian RS coast, from the GofA to the Farasan Islands (Table 1). Prior to collection, living coral colonies were photographed in the field using either a Canon G9 or G11 PowerShot digital camera (Canon Inc., Tokyo, Japan) in an Ikelite underwater housing (Ikelite Underwater Systems, Indianapolis, IN, USA). Specimens were sampled during SCUBA diving between 1 and 30 m depth using hammer and chisel. A small fragment (<1 cm2) of each coral colony was preserved in 96% ethanol for molecular analyses. The rest of the corallum was tagged and bleached in sodium hypochlorite for 48 h in order to remove tissue parts, rinsed with fresh water, and air-dried. Images of cleaned skeletons were taken with a Canon G9 or G11 PowerShot digital camera.

Species identification

Specimens were identified to the species level based on macromorphological features of the coralla and the corallites. For the genus Leptoseris, we referred to the revision by Dinesen (1980), including descriptions and illustrations of type material. For Leptoseris yabei (Pillai & Scheer, 1976), originally described in Pavona, we referred to the original description and holotype illustration (pl. 16, figs. 1, 2) by Pillai and Scheer (1976). For the other genera, Veron and Pichon (1980), Scheer and Pillai (1983), Sheppard and Sheppard (1991), Nishihira and Veron (1995), and Veron (2000) were referenced. The type material of Agariciidae species deposited at the National Museum of Natural History (MNHN), Paris, France and at the Smithsonian Institution, National Museum of Natural History (formerly known as United States National Museum of Natural History) (USNM), Washington, USA, was also examined by FB and imaged. These images were used as reference for the identification of a number of species, namely Leptoseris fragilis Milne Edwards & Haime, 1849 (MNHN 468), Leptoseris scabra Vaughan, 1907 (USNM 20885; USNM 20886), Leptoseris hawaiiensis Vaughan, 1907 (USNM 20874; USNM 20876), Leptoseris mycetoseroides Wells, 1954 (USNM 44805; USNM 44807), Pavona diffluens (Lamarck, 816) (MNHN IK-2010-587), Pavona danai (USNM 136), Pavona duerdeni Vaughan, 1907 (USNM 21630; USNM 21631), and P. decussata (USNM 176).

DNA extraction, amplification, and sequence analyses

Genomic DNA was extracted from samples using the commercial DNAeasy® Blood & Tissue Kit (Qiagen Inc., Hilden, Germany), following the manufacturer’s protocol. Primers AGAH and AGAL (Terraneo et al. 2014) were used to amplify IGR locus, while 1S and 2SS primers (Wei et al. 2006) were used to amplify ITS1. All of the polymerase chain reaction (PCR) products were purified with Illustra ExoStar (GE Healthcare, Buckinghamshire, UK) at 37° for 60 min, followed by 85° for 15 min, and sent for direct sequencing in both forward and reverse directions using an ABI 3730xl DNA analyzer (Applied Biosystems, Carlsbad, CA, USA). All the sequences generated from this work were deposited in EMBL (Table 1).

Forward and reverse sequences were viewed, edited, and assembled using Sequencher 5.3 (Gene Codes Corp., Ann Arbor, MI, USA). Nuclear sequences were phased using SeqPHASE (Flot 2010; available online at http://seqphase.mpg.de/seqphase/) and PHASE was used (Stephens et al. 2001; available online at http://stephenslab.uchicago.edu/software.html) when alleles showed the same length; Champuru (Flot 2007; available online at http://seqphase.mpg.de/champuru/) was used if the two predominant alleles were of different lengths. In the former case, the two alleles with the highest probability (an order of magnitude greater than the other sequence pairs) were chosen whenever there were multiple possible phases. No obvious or significant differences of genetic diversity and haploweb inference were obtained using alternative phases (results not shown). Phased heterozygotes were represented by both alleles in the further alignments and population genetic analyses. Final alignments were performed using the E-INS-i option in MAFFT 7.130b (Katoh and Standley 2013) under default parameters and manually checked using BioEdit 7.2.5 (Hall 1999). Ambiguously aligned regions were removed from the alignments using Gblocks 0.91b (Castresana 2000) using the “more stringent” selection (alignment data are available from the corresponding author upon request). Invariable, polymorphic, and parsimony-informative sites were detected using DnaSP 5.10.01 (Librado and Rozas 2009).

The two loci were analyzed separately, and phylogenetic relationships between species were assessed using Bayesian inference (BI), maximum likelihood (ML), and maximum parsimony (MP), as implemented in MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003), PhyML 3.0 (Guindon and Gascuel 2003), and PAUP* 4.0b10 (Swofford 2002), respectively. We used the Akaike information criterion (AIC) implemented in MrModeltest 2.3 (Posada and Crandall 1998) in conjunction with PAUP* 4.0b10 to determine the best-fit substitution model of sequence evolution. The AIC selected the General Time Reversible (GTR) + invariable sites + gamma model of nucleotide substitution for both markers (Pinv = 0.28, α = 0.51 for IGR, and Pinv = 0.78, α = 0.55 for ITS1). The BI analysis consisted of four independent Markov chain Monte Carlo (MCMC) runs for 6,000,000 generations for IGR (10,000,000 for ITS1), saving a tree every 1000 generations and discarding the first 25% of the trees as burn-in, on the basis of parameter estimations and convergence examined by Tracer 1.6 (Rambaut and Drummond 2007). Clade support was based on posterior probability. The ML analysis was run using the default parameters, the previously selected model, and 1000 bootstrap replicates to verify the robustness of the internal branches of the tree. For MP, a heuristic search was performed using starting trees obtained by random stepwise addition with 10 replicates and the tree bisection and reconnection (TBR) branch-swapping-algorithm, generating a strict consensus tree. The robustness of the internal branches of the tree was verified using 1000 bootstrap replicates. The two alignments and phylogenetic trees are available in TreeBASE (http://purl.org/phylo/treebase/phylows/study/TB2:S20870).

Results

The 91 Agariciidae colonies collected in the Saudi Arabian RS were identified based on colony and corallite macromorphology, amounting to a total of 20 species ascribed to three genera, namely Pavona, Leptoseris, and Gardineroseris (Table 1) (Figs. 1 and 2). One Leptoseris specimen was identified as Leptoseris cf. hawaiiensis. The uncertainty about the identification of this distinct specimen stems from the substantial morphological differences between the actual type material and specimens and colonies identified by other authors as L. hawaiiensis (Sheppard and Sheppard 1991; Veron 2000).

Fig. 1
figure 1

In vivo colony morphology of the Agariciidae collected for this study in the Saudi Arabian Red Sea (RS): a SA0076 Pavona venosa; b SA0857 Pavona varians; c SA0368 Pavona maldivensis; d SA0105 Leptoseris fragilis; e SA0269 Pavona decussata; f SA0810 Gardineroseris planulata; g SA0369 Pavona clavus; h SA0943 Leptoseris sp.; i SA0942 Pavona explanulata; j SA0781 Leptoseris mycetoseroides; k SA0884 Leptoseris cf. hawaiiensis; l SA2654 Leptoseris scabra; m SA2608 Leptoseris incrustans; n SA0097 Leptoseris explanata; o SA0893 Pavona diffluens; p SA0088 Leptoseris glabra; q SA0424 Leptoseris yabei; r SA0052 Pavona danae; s SA0205 Pavona cactus; t SA2653 Pavona frondifera

Full size image
Fig. 2
figure 2

Skeletal macromorphology of the Agariciidae species collected for this study in the Saudi Arabian Red Sea (RS): a SA0076 Pavona venosa; b SA0857 Pavona varians; c SA0368 Pavona maldivensis; d SA0105 Leptoseris fragilis; e SA0269 Pavona decussata; f SA0810 Gardineroseris planulata; g SA0369 Pavona clavus; h SA0943 Leptoseris sp.; i SA0942 Pavona explanulata; j SA0781 Leptoseris mycetoseroides; k SA0884 Leptoseris cf. hawaiiensis; l SA2654 Leptoseris scabra; m SA2608 Leptoseris incrustans; n SA0097 Leptoseris explanata; o SA0893 Pavona diffluens; p SA0088 Leptoseris glabra; q SA0424 Leptoseris yabei; r SA0052 Pavona danae; s SA0205 Pavona cactus; t SA2653 Pavona frondifera

Full size image

Phylogenetic analyses

The final mitochondrial alignment consisted of 1081 bp and included 187 polymorphic sites (71 of which were singleton sites) and 116 parsimony-informative sites. Two hundred and fifty-seven mutations were recovered from the dataset when evaluating synonymous and non-synonymous substitutions. The ITS alignment consisted of 468 bp and 54 polymorphic sites (14 of which were 14 singleton sites), 40 parsimony-informative sites, and a total of 49 mutations.

Two sequences of Siderastrea radians (Pallas, 1776) were downloaded from GenBank and used as the outgroup for both mitochondrial and nuclear datasets because of the known divergence of this species from Agariciidae (Fukami et al. 2008; Luck et al. 2013; Terraneo et al. 2014; Pochon et al. 2015). Topologies derived from BI, ML, and MP were mostly in agreement, resolving the same well-supported molecular clades without any conflicting partition. In Figs. 3 and 4, we report the Bayesian phylogram with branch support indicated by Bayesian posterior probability (BBPI), ML bootstrapping support (BTML), and MP bootstrapping support (BTMP).

Fig. 3
figure 3

Phylogenetic reconstruction of the family Agariciidae from the Saudi Arabian Red Sea (RS) inferred from Bayesian inference (BI) analysis of the mitochondrial intergenic spacer between COI and 16S rRNA (IGR). All the different morphospecies are color coded. Node values are posterior Bayesian probabilities, maximum likelihood (ML) bootstrap values, and maximum parsimony (MP) bootstrap values. Posterior Bayesian probabilities below 0.7, ML bootstrap values below 50%, and MP bootstrap values below 50% are indicated by a dash (−). Siderastrea radians was selected as the outgroup

Full size image
Fig. 4
figure 4

Phylogenetic reconstruction of the family Agariciidae from the Saudi Arabian Red Sea (RS) inferred from Bayesian inference (BI) analysis of the internal transcribed spacer 1 from nuclear ribosomal DNA (ITS1). All the different morphospecies are color coded as in Fig. 3. Node values are posterior Bayesian probabilities, maximum likelihood (ML) bootstrap values, and maximum parsimony (MP) bootstrap values. Posterior Bayesian probabilities below 0.7, ML bootstrap values below 50%, and MP bootstrap values below 50% are indicated by a dash (−). Siderastrea radians was selected as the outgroup

Full size image

The mitochondrial phylogenetic reconstruction demonstrates that genus-level boundaries of Leptoseris, Pavona, and Gardineroseris are not genetically supported (Fig. 3). In fact, Leptoseris and Pavona as currently defined are intermixed between each other, and the monotypic genus Gardineroseris nests together with Leptoseris and Pavona.

According to the mitochondrial reconstruction, species-level boundaries are also unclear for many of the analyzed taxa. Gardineroseris planulata forms a monophyletic group (CLADE V) strongly supported by the three analysis criteria (0.94/100/100). Similarly, Leptoseris fragilis (CLADE III), L. yabei (CLADE XIII), and Leptoseris sp. (CLADE VII) are monophyletic and grouped in highly supported clades (node values 1/100/100, 0.99/−/100, and 1/99/100, respectively). Sequences of L. mycetoseroides are split into two distinct clades (CLADE IX and CLADE XI), showing cryptic genetic divergence within the morphospecies. Leptoseris cf. hawaiiensis, L. scabra, L. incrustans, L. explanata, and L. glabra are genetically indistinguishable based on this marker and grouped together in CLADE XII, which also includes all the analyzed samples of P. diffluens. Pavona maldivensis (CLADE II) forms a monophyletic group strongly supported by the three phylogenetic analyses (0.98/96/84). All representatives of P. decussata (CLADE IV) cluster together in a highly supported lineage (1/100/100). The only examined sample of P. clavus is highly divergent within the tree (CLADE VI), but one specimen is not sufficient to evaluate its monophyly. Based on mitochondrial sequence data, P. venosa and P. varians are indistinguishable from a molecular point of view, and are nested together in CLADE I, while P. explanulata is comprised of two distinct lineages (CLADE VIII and CLADE X). Finally, the frondose Pavona danai Milne Edwards, 1860, Pavona frondifera (Lamarck, 1816), and P. cactus form a single clade (CLADE XIV) basal to the entire phylogeny (Fig. 3, Table 3).

The phylogenetic tree based on nuclear ITS1 is less resolved than the mitochondrial phylogeny reconstruction, though no contrasting patterns are found between the two analyses, supporting the stability of relationships among species. All agariciids form an unresolved clade due to a large basal polytomy. Similar to the IGR tree topology, Leptoseris, Pavona, and Gardineroseris are closely related and nest together based on ITS1 (Fig. 4), which raises questions regarding the validity of these three genera. Five molecular clades identified by IGR are also resolved in the ITS1 phylogeny, including P. maldivensis (CLADE II), G. planulata (CLADE V), P. clavus (CLADE VI), L. yabei (CLADE XIII), and P. danai, P. frondifera, and P. cactus (CLADE XIV). Pavona decussata (CLADE IV) and P. varians and P. venosa (CLADE I) group together in a molecular lineage (−/79/−) and, interestingly, representatives of P. decussata form a group genetically isolated from the other two species. Leptoseris fragilis (CLADE III) and Leptoseris sp. (CLADE VII) are unresolved, forming part of the large basal polytomy, but they do not cluster with any other morphospecies. Leptoseris mycetoseroides (CLADES IX and XI) is unresolved and representatives of these two clades are mixed together. Similar to the placement of L. mycetoseroides, P. explanulata (CLADES VIII and X) is also unresolved, though sequences ascribed to these two clades mix together. Lastly, P. diffluens, L. explanata, L. glabra, L. incrustans, and L. scabra (CLADE XII) are unresolved in the large basal polytomy of the ITS1 phylogenetic tree.

Discussion

In the present study, we provide a preliminary genetic survey of the scleractinian coral family Agariciidae from the Saudi Arabian RS. After intensive sampling over 1200 km from the GofA in the north to the Farasan Islands in the south, we recorded a total of 20 agariciid morphospecies belonging to the genera Gardineroseris, Leptoseris, and Pavona (Table 2). Taxa were investigated at two molecular regions: the mitochondrial IGR and the nuclear ITS1. The resulting phylogenetic reconstructions showed agreement in resolving six species, G. planulata, P. clavus, P. maldivensis, P. decussata, L. fragilis, and L. yabei, while the boundaries among the remaining 15 species were unclear, with three main species complexes detected (Figs. 3 and 4, Table 3). The IGR and ITS1 sequences represent the first molecular information of Agariciidae from the RS and only the second one from the entire IO, adding to the examination by Moothien Pillay et al. (2006) of the genetic differences between P. cactus and P. decussata from Mauritius. Additionally, seven species (P. venosa, P. danai, P. diffluens, L. fragilis, L. explanata, L. glabra, and L. yabei) were included in a molecular phylogenetic study for the first time in this study.

Table 2 List of samples analyzed in the present study
Full size table
Table 3 List of Agariciidae morphospecies collected in the Red Sea (RS) and relative genetic assignment from IGR and ITS1 molecular markers
Full size table

The morphospecies diversity that we reported from the Saudi Arabian RS is compatible with that reported in previous works from the region (Scheer and Pillai 1983; Sheppard and Sheppard 1991; Veron 2000), though with some distinct differences (Table 2). Notably, we recorded L. glabra from the RS for the first time. However, it is possible that previous authors recorded it as L. explanata due to its similar morphology. According to Dinesen (1980), the material identified as L. explanata by Veron and Pichon (1980; figs 71–82) actually belongs to L. glabra. Dinesen (1980) also indicated that the type material of L. explanata shown and described in Yabe and Sugiyama (1941: pl. 63, fig. 3e) is morphologically distinct from L. glabra, as she described in her revision of Leptoseris (Dinesen 1980). In our identifications, we followed the original descriptions of these two morphospecies and considered them as morphologically distinct following Dinesen (1980). Nevertheless, the recovery of specimens ascribed to these two species in the same lineage with other Leptoseris merits for further investigation into the relationship between these species. While the existing confusion in the literature concerning L. explanata and L. glabra is easily explained based on their undeniable similarities in terms of calice size and arrangement, they are both markedly distinct from the other morphospecies recovered in the same clade having smaller calices and more numerous septa (Fig. 2). Easily recognizable and common shallow-water species such as G. planulata, P. cactus, P. decussata, P. explanulata, P. maldivensis, P. varians, L. explanata, L. hawaiiensis, L. mycetoseroides, L. scabra, and L. yabei were recorded by Scheer and Pillai (1983), Sheppard and Sheppard (1991), Veron (2000), and this work. Compared to Sheppard and Sheppard (1991) and Veron (2000), we did not record P. duerdeni and L. foliosa during our collection. The former species is rare in the RS, although it can form conspicuous colonies in low-energy waters (Veron 2000), whereas the latter species is uncommon and preferentially found in deep-water habitats (>30 m) (Sheppard and Sheppard 1991). This work did not investigate mesophotic coral ecosystems, potentially explaining why these species were likely missed. The family Agariciidae is among the most abundant coral families in deep-water oceanic habitats (Lesser et al. 2009). Therefore, more species are likely to occur in deeper waters of the RS. Leptoseris often shows an increasing abundance with depth in the IO (Bouchon 1981) and in Hawaii (Kahng and Kelley 2007), where it dominates hard substrates between 60 and 120 m and is, instead, rare in shallow-water reefs. Luck et al. (2013) also found cryptic lineages of Leptoseris across depth gradients in Hawaii, and several studies showed limited larval dispersal, genetic segregation, and low gene flow between shallow and mesophotic agariciid reef corals (Bongaerts et al. 2013, 2015; Pochon et al. 2015). Bongaerts et al. (2013) showed a division of Agaricia species and associated Symbiodinium into two main genetic lineages that matched a clear bathymetric distribution. For these reasons, we caution others to consider our presented survey as fully representative of only the shallow-water genetic diversity of Agariciidae from the Saudi Arabian RS; we strongly encourage further investigations of mesophotic reefs.

Regarding the molecular information, our work further highlighted the hypothesis that some regions of the mitochondrial DNA of corals can be informative in defining boundaries among closely related taxa. Despite the well-documented slow evolution of Scleractinia mitochondrial DNA (Shearer et al. 2002; Hellberg 2006; Huang et al. 2008), recent works have found that some mitochondrial non-coding regions show rates of interspecific polymorphisms comparable to these of nuclear loci and can be phylogenetically informative at low taxonomic levels (Kitahara et al. 2016). The IGR sequenced in the present work has been previously used for delimiting species boundaries in several scleractinian families and genera, including the families Merulinidae Verrill, 1865 and Agariciidae (Kitahara et al. 2016), the genus Pachyseris Milne Edwards & Haime, 1849 (Terraneo et al. 2014), and the genera Sclerophyllia Klunzinger, 1879, Homophyllia Brüggemann, 1877, and Micromussa Veron, 2000 (Arrigoni et al. 2015, 2016b). Ribosomal DNA is widely used for low-level taxonomic comparisons in corals, particularly due to high interspecific divergence between the two internal transcribed spacers regions, ITS1 and ITS2 (Kitahara et al. 2016). In the present study, the phylogeny reconstruction based on ITS1 was poorly resolved, as it was characterized by a large basal polytomy and several species were not differentiated. The absence of genetic differentiation among the analyzed species may be explained by insufficient resolution of the ITS1 locus. Hence, we encourage the investigation of further molecular markers and modern genome-wide sequencing techniques that could provide strong phylogenomic analysis to help better determine species boundaries (Pante et al. 2015).

This work raises several questions regarding the classical definition of genera and species boundaries within the Agariciidae that will require more in-depth observations of skeletal micromorphology before a formal taxonomic revision of the family is proposed. As previously reported by Luck et al. (2013), our mitochondrial phylogeny reconstruction confirmed that Pavona and Leptoseris are not monophyletic. Our molecular analyses (based on IGR and ITS1) revealed for the first time that the monotypic Gardineroseris is closely related to both Pavona and Leptoseris and it nests within them. The genetic information does not support the morphology-based distinction among these three genera (Scheer and Pillai 1974; Dinesen 1980) and suggests the presence of a single, species-rich genus. However, any taxonomic reclassification is not recommended until a comprehensive phylogenetic reconstruction of the Agariciidae, including all seven extant genera and sequencing of slow-evolving loci such as the mitochondrial COI and 16S rDNA genes and the nuclear 28S region, is conducted (Fukami et al. 2008; Huang et al. 2008). Indeed, non-coding regions are usually considered to be poorly informative at high taxonomic levels because homoplasy caused by repeated mutations in saturated positions is thought to undermine the phylogenetic signal (Kelchner 2000). At the species level, the IGR and ITS1 phylogeny reconstructions revealed the presence of three species complexes (Fig. 3, Table 3). On one hand, the recovery of the two species complexes composed of the frondose and highly plastic P. cactus, P. danai, and P. frondifera and of the encrusting and highly carinated P. varians and P. venosa (Veron and Pichon 1980) could be morphologically plausible. Indeed, these species display similar colony morphology, related collines and carinae, comparable corallite dimension and organization, and previous taxonomists have identified these morphological affinities (Vaughan 1918; Crossland 1952; Veron and Pichon 1980; Scheer and Pillai 1983; Veron 2000). The recovery of a molecular clade including L. explanata, Leptoseris cf. hawaiiensis, L. scabra, L. incrustans, L. glabra, and P. diffluens was unexpected. Although P. diffluens may be mistaken for massive parts of P. explanulata (Sheppard and Sheppard 1991), it does not resemble any Leptoseris species ascribed to this molecular group. Moreover, as mentioned earlier in this discussion, the Leptoseris morphospecies recovered in this one cohesive lineage in the RS do display remarkable differences in terms of colony growth form, calice size and organization, and septocostae thickness, with the notable exception of L. glabra and L. explanata (Fig. 2). A similar situation has been described in the genus Stylophora from the RS (Arrigoni et al. 2016c), where different morphospecies, two of which are endemic to the RS, could not be differentiated based on genetic results. Several hypotheses can be suggested to explain this finding, such as taxonomic misidentification, uninformative morphological characters, non-suitable molecular loci, or incipient speciation.

In conclusion, we investigated the genetic diversity of Agariciidae from the isolated Saudi Arabian RS and provide a regional genetic database for future molecular comparisons between the RS and other localities. The partial disagreement between morphology-based taxonomy and genetic results highlights that agariciid systematics is poorly understood and that the current family taxonomy does not truly represent the evolutionary relationships at the genus or species levels. A combined morpho-molecular approach with a strong focus on skeletal micromorphology is likely necessary to better clarify the boundaries within the family and revise its taxonomy, as previously demonstrated for the agariciid Dactylotrochus (Kitahara et al. 2012) and several other scleractinian families (Kitahara et al. 2016). The molecular phylogeny of Agariciidae is still incomplete because only a few regions and species have been genetically characterized to date (Maté 2003; Moothien Pillay et al. 2006; Benzoni et al. 2012; Luck et al. 2013; Pochon et al. 2015; Waheed et al. 2015). Therefore, a more global sampling effort with a larger number of taxa and more genetic data, in addition to detailed comparative micromorphological analysis, are needed and necessary before a robust molecular hypothesis can be produced and the agariciids fully resolved.