Polyphyly and hidden species among Hawaiʻi’s dominant mesophotic coral genera, Leptoseris and Pavona (Scleractinia: Agariciidae)

Sample collections

Skeletal and tissue samples of Leptoseris and Pavona spp. were collected throughout the Hawaiian Archipelago (n = 109) and Line Islands (n = 3) (Table S1). Collections were authorized via special activity permits by the State of Hawaiʻi Board of Land and Natural Resources. Although this study focused on species from mid- to lower MCEs (i.e., >60 m), shallow-water samples (<30 m) were also collected for comparative purposes. Most samples came from MCEs in the Auʻau Channel, located between Maui and Lanaʻi, and along the nearshore shallow-water reefs off Oʻahu. Collections were made using the Hawaiʻi Undersea Research Laboratory (HURL) Pisces submersibles, technical diving or traditional SCUBA, depending on the depth. This effort included 14 submersible dives and one remotely controlled vehicle dive ranging from 49 to 216 m, 14 technical dives from 40 to 78 m, and eight shore dives from 5 to 24 m. Collected samples included all six Leptoseris spp. known from Hawaiʻi and ranged from 2 to 127 m depth, encompassing most of the accepted bathymetric range for the genus (Dinesen, 1980). (Vaughan’s (1907) record of L. hawaiiensis to 470 m is likely erroneous, see Kahng et al., 2010.) Some samples were too small for morphological analyses (n = 18); others lacked preserved tissue and were not included in molecular analyses (n = 42). The remaining 52 samples were included in both molecular and morphological analyses.

The Leptoseris type collections of the US National History Museum (USNM) and the Natural History Museum, London, UK (BNHM) were also examined (Table S2). Type specimens for a majority of Leptoseris spp. and all species previously known from Hawaiʻi were analyzed morphologically (tissues for molecular analyses were unavailable). Pavona types were not examined because at the time of museum visits Pavona was not thought to be abundant in Hawaiian MCEs.

Molecular analyses

Phylogenetic relationships were determined by analyzing sequences from the cox1-1-rRNA intron, a novel, rapidly evolving mitochondrial intergenic spacer. Preliminary analyses of cox1-1-rRNA showed that it gave much higher resolution between Leptoseris spp. than either the NAD5 5′-intron or the ITS region (DG Luck, unpublished data).

Host genomic DNA was extracted using the E.Z.N.A.^® Tissue DNA Kit (Omega Bio-tek, Norcross, GA, US). The cox1-1-rRNA intron was amplified using Biomix Red (Bioline, Taunton, MA, US), a high-density PCR reaction mix, and the following primers: ZFCOXIF (5′-TCT GGT GAG CTC TTT GGG CTC T-3′) and ZFtrnar (5′-CGA ACC CGC TTC TTC GGG GC-3′). Primers were designed based on Agaricia Lamarck, 1801 and Pavona mitochondrial genomes acquired from GenBank (NC_008160,5). Each PCR totaled 12.5 µl and included reagents in the following proportions: 7.5 µl Biomix Red; 6.51 µl nanopure^® (Thermo Fisher Scientific) H₂O; 0.195 µl of each primer; and 0.6 µl extracted genomic DNA. The thermocycler protocol was as follows: one cycle at 95°C for 2 min; 35 cycles beginning with 95°C for 30 s, followed by 52°C for 30 s, followed by 72°C for 1 min; and one extension cycle at 72°C for 10 min. PCR products were cleaned using shrimp alkaline phosphatase and Sanger sequenced using the forward primer. Only the forward primer was necessary since it produced sufficiently long sequences (up to 900 bp) with many informative sites.

In order to provide higher resolution and to root the mtDNA phylogeny, cox1-1-rRNA sequences were obtained for Agaricia humilis Verrill, 1901, Pavona clavus (Dana, 1846) and Siderastrea radians (Pallas, 1766) from whole mitochondrial genomes available on GenBank (accession numbers: DQ643831, DQ643836, DQ643838). These three taxa were included in all phylogenetic analyses.

Sequence data were manually cleaned with BioEdit version 7.1.3 (Hall, 1999) and imported into MEGA 5.10 for alignment (Tamura et al., 2011). Separate alignments were performed using ClustalW and Muscle algorithms, first using default settings and then with reduced gap opening and extension penalties. The ClustalW alignment (with default parameters) was chosen for final analysis because it produced trees with slightly higher bootstrap support.

Maximum Likelihood (ML) and Bayesian phylogenies were constructed using evolutionary models selected by jModelTest 2.1.1 (Darriba et al., 2012). ML analysis was performed in PhylML (Guindon & Gascuel, 2003) using a transversional model with a gamma distribution of substitution rates across sites and 1000 bootstrap replicates. Bayesian analysis was performed in MrBayes 3.2.1 (Ronquist et al., 2012) using a three-parameter model with unequal base frequencies and a gamma distribution of substitution rates across sites. The Bayesian model was run over 1.75 million generations at which point the standard deviation of split frequencies fell below 0.01.

Taxonomy, sample identification, and morphological terminology

With respect to the taxonomy of the genus Leptoseris, the synonymies of Dinesen (1980) were followed except that L. tubulifera Vaughan, 1907 was treated as a valid species based on molecular and micromorphological evidence. Following the suggestion of Dai & Lin (1992), Pavona yabei Pillai & Scheer, 1976 was kept in the genus Pavona (rather than reassigned to Leptoseris as proposed by Veron & Pichon, 1980). The septal and corallite morphologies of P. yabei are more similar to those of other Pavona spp. (e.g., septocostae that always alternate in length, rows of granules rather than continuous meninae on the lateral faces of septocostae, high corallite density, predominance of radial collines, etc.) and there is no molecular evidence that it should be grouped with Leptoseris spp. All samples included in this study were identified to the species-level by comparison to type material and by consultation with taxonomists ZD Dinesen and JE Maragos.

In reference to the features of the coral skeleton, the terminology of Dinesen (1980) was employed except that ‘meninaes’ (Gill & Russo, 1980) was used rather than ‘lateral ridges’ for the paired lists or ridges along the lateral surfaces of septocostae. For detailed definitions of skeletal terms, consult Wells (1956) and Budd & Stolarski (2009).

Morphological analyses

Skeletal morphological analyses were conducted on 94 samples (52 with paired tissue vouchers) in order to (1) provide additional support for molecular-based species groups and (2) to identify diagnostic characters for species identification when molecular analyses are not possible. Emphasis was placed on identifying discrete characters because they are easily mapped onto phylogenies.

Consistent with other integrated systematic studies of scleractinians (Budd & Stolarski, 2009; Benzoni et al., 2010; Budd et al., 2010), morphological analyses focused on septocostal and costal micromorphology, although macromorphological observations of the corallum and corallites were also made. Where samples had free growth margins, costal ornamentation (‘costae’ being extensions of septocostae on the non-calicinal surface and ‘ornamentation’ being the granules/spines on costae) were also analyzed. Because of their small size (typically <100 µm in diameter) and location on the non-photosynthetic side (i.e., the underside) of the corallum, costal granules/spines were selected for closer examination since they may be subject to reduced environmental selection and therefore show less ecotypic variation within a given species.

In order to identify diagnostic characters, skeletal samples were examined under a Nikon SMZ 800 light microscope. Skeletons were bleached with sodium hypochlorite to remove tissue, rinsed and then air dried. Examinations were made first with each skeleton intact. Visible features were recorded and digital photographs of the corallites, septocostae and costae were taken using a Nikon P-6000 camera mounted to the microscope. When samples were large enough (n = 40), longitudinal sections were cut throughout the coenosteum using a Dremel^® tool or wet saw to investigate the arrangement of meninaes. A subset of skeletons was also analyzed using scanning electron microscopy (SEM). Longitudinal sections or chips of these samples were mounted on metal stubs using carbon glue or tape, sputter-coated with gold-palladium for approximately 2 min, and then viewed with a Hitachi S-4800 SEM.

Phylogenetic results

Sequence data from the cox1-1-rRNA intron were obtained from 70 coral colonies (Table S1) and three sequences were obtained from previously published mitochondrial genomes. The final alignment of 73 sequences was 743 bp long and included 517 variable sites, 321 of which were parsimony informative. Bayesian and ML trees had similar topologies and placed most taxa in well-supported clades (Fig. 1; only Bayesian tree shown). Pairwise nucleotide distances between significant clades were always an order of magnitude higher than within group distances (mean uncorrected p-distance of 0.193 ± 0.025 vs. 0.015 ± 0.008).

The Bayesian phylogeny shows that the genus Leptoseris, as currently defined, is polyphyletic (Fig. 1). A majority of Hawaiian Leptoseris spp. belong to clades I–II including L. hawaiiensis, L. incrustans (Quelch, 1886), L. papyracea (Dana, 1846), L. tubulifera and L. sp. 1. However, L. scabra, which forms the sister clade (VII) to Agaricia humilis, is distantly related to clades I–II. The mean pairwise nucleotide distances between clade VII and clades I–II were among the highest distances between all groups compared (uncorrected p-distances: I vs. VII = 0.294 ± 0.018; II vs. VII = 0.284 ± 0.018). Three haplotypes of L. mycetoseroides Wells, 1954 are scattered throughout the phylogeny, but all three clearly fall outside the group formed by clades I–II.

The genus Pavona is also polyphyletic at the cox1-1-rRNA intron and consists of at least two clades (V and VIII). Clade V is sister to the group formed by A. humilis and L. scabra, whereas clade VIII is more closely related to the Leptoseris spp. of clades I–II.

Morphological results

Morphological analyses revealed that most molecular-based species groups could be independently identified using skeletal characters (Fig. 2; Table S3). Although no single character could distinguish between all Leptoseris or Pavona spp., combinations of discrete characters were effective in pairwise comparisons. Among the more important characters were the costal ornamentation, the shape and continuity of the upper septocostal margin, the development of meninaes, and species-specific macro-projections.

Several Leptoseris spp. had clear diagnostic characters. For example, L. incrustans consistently had meninaes four to five times wider than the upper septocostal margin, which was always a narrow and continuous ridge (Fig. 2B). This width ratio was constant throughout the corallum (mean = 0.19, standard deviation = 0.09) except in the vicinity of corallites or where insertions occurred. L. tubulifera was clearly diagnosed by the presence of corallites on tubes, a characteristic absent from Pavona? sp. 2, which may also form tubes.

In cases where a species lacked a single diagnostic character, it could usually be distinguished from species with similar gross morphologies by analyzing combinations of discrete micromorphological characters. For example, L. hawaiiensis and L. scabra both grow as thin plates in lower MCEs (Figs. 3D and 3J). The costal ornamentation of the two species, however, is clearly different (Figs. 2C–2D). L. hawaiiensis has conical spines predominantly aligned in a single row down the center of each costa, whereas L. scabra has cylindrical or beady granulations in three rows, one central row and one row on each margin. Similarly, Pavona sp.1 and L. papyracea, both of which form branching colonies (Figs. 3E and 3H), are easily distinguished by their costal ornamentation. L. papyracea has exsert, conical spines arranged in a single row down the center of each costa while P. sp. 1 has triangular or bead-like granulations scattered along costae. A final example is that of L. scabra versus L. sp. 1. Both species grow as crateriform vases with irregular, bumpy surfaces in mid-mesophotic depths (∼60–80 m), but the two differ in their development of septal teeth (Figs. 2A and 2D). L. scabra has peg-like teeth formed by discontinuities of the septocostae, including their meninaes. In L. sp. 1, teeth are absent or visible only as discontinuities of the upper septocostal margin above the meninaes. L. sp. 1 also has septocostae that form numerous abortive branches at proximal cushions, with some branches turning antiparallel from their parent septocosta. In fact, L. sp. 1 is distinct from any species described in the literature and is putatively a new species in need of formal description.

Some aspects of micromorphology were also concordant with generic differences at the cox1-1-rRNA region. For example, members of the “Leptoseris” clades (I–II) had more or less continuous meninae, whereas most Pavona spp. had granules arranged in rows along their lateral surfaces but lacked continuous meninae. The position of L. mycetoseroides outside of the clades I–II is also supported by micromorphology as all three haplotypes of L. mycetoseroides have rows of granules rather than continuous meninae. Continuous meninae, however, are apparently not synapomorphic as they are sometimes found in L. scabra (clade VII). In fact, while morphological analyses revealed many useful characters, few synapomorphies were found. Most of the patterns of costal ornamentation that differentiate between Leptoseris spp. can also be found in Pavona spp. (Fig. 2: B vs. E) and L. scabra (Fig. 2: A vs. D).

Depth distribution of Hawaiian Leptoseris and Pavona

Leptoseris and Pavona clades sorted into two major groups according to their depth distributions (Fig. 4). Most clades occurred in shallow-water (<30 m) down to mid-mesophotic depths (60–80 m). Only clades Ib (L. hawaiiensis) and VII (L. scabra) were restricted to depths >60 m. These two clades were observed deeper than any others, with colonies of L. hawaiiensis observed at 141 m and L. scabra observed to 127 m. One Leptoseris species, L. incrustans, was never observed in mesophotic depths but was among the more common agariciids observed in dimly-lit, shallow-water caves and undercuts. Of the nominal Pavona spp., only P. sp. 1 (clade V) was observed from lower MCEs (89 m); all other Pavona spp. were restricted to depths <80 m.