A spectacular new species of seadragon (Syngnathidae)

The exploration of Earth's biodiversity is an exciting and ongoing endeavour. Here, we report a new species of seadragon from Western Australia with substantial morphological and genetic differences to the only two other known species. We describe it as Phyllopteryx dewysea n. sp. Although the leafy seadragon (Phycodurus eques) and the common seadragon (Phyllopteryx taeniolatus) occur along Australia's southern coast, generally among relatively shallow macroalgal reefs, the new species was found more offshore in slightly deeper waters. The holotype was trawled east of the remote Recherche Archipelago in 51 m; additional specimens extend the distribution west to Perth in 72 m. Molecular sequence data show clear divergence from the other seadragons (7.4–13.1% uncorrected divergence in mitochondrial DNA) and support a placement as the sister-species to the common seadragon. Radiographs and micro-computed tomography were used on the holotype of the new species and revealed unique features, in addition to its unusual red coloration. The discovery provides a spectacular example of the surprises still hidden in our oceans, even in relatively shallow waters.


Summary
The exploration of Earth's biodiversity is an exciting and ongoing endeavour. Here, we report a new species of seadragon from Western Australia with substantial morphological and genetic differences to the only two other known species. We describe it as Phyllopteryx dewysea n. sp. Although the leafy seadragon (Phycodurus eques) and the common seadragon (Phyllopteryx taeniolatus) occur along Australia's southern coast, generally among relatively shallow macroalgal reefs, the new species was found more offshore in slightly deeper waters. The holotype was trawled east of the remote Recherche Archipelago in 51 m; additional specimens extend the distribution west to Perth in 72 m. Molecular sequence data show clear divergence from the other seadragons (7.4-13.1% uncorrected divergence in mitochondrial DNA) and support a placement as the sisterspecies to the common seadragon. Radiographs and microcomputed tomography were used on the holotype of the new species and revealed unique features, in addition to its unusual red coloration. The discovery provides a spectacular example of the surprises still hidden in our oceans, even in relatively shallow waters.

Introduction
The fraction of undescribed biodiversity in the ocean is arguably high [1,2], and new fish species are likely to be described, especially from the deeper continental slopes [3]. Here, we describe a remarkable example, a new species of seadragon. Phyllopteryx dewysea n. sp. is only the third known species of seadragon, and the first to be discovered in 150 years, possibly because it lives in slightly deeper waters than its relatives.
Seadragons (Syngnathidae) are fish of mesmerizing beauty (figure 1), with the leafy seadragon (Phycodurus eques) having more elaborate appendages than the colourful common seadragon (Phyllopteryx taeniolatus). Their ornamentation helps camouflage them among seagrasses and kelp in the shallow coastal waters of southern Australia [4]. The    The range of the leafy seadragon is shown in gold, the common seadragon in blue and the type locality of the new seadragon Phyllopteryx dewysea n. sp. is indicated by the red arrow. The red circle represents the collection sites of the three paratypes. The maximum-likelihood tree is based on the partitioned dataset (4416 bp). Numbers next to the nodes indicate bootstrap (top) and jackknife (bottom) values; full support is indicated by white circles.
spans the entire southern coast from Western Australia to New South Wales and Tasmania, whereas the leafy seadragon has a more restricted distribution from Western Australia to South Australia (map in figure 1). The two known species of seadragons fascinate aquarium visitors around the world, draw many SCUBA divers into the water and are charismatic symbols for conservation of marine species [5,6].
On a biodiversity survey (part of the Marine Futures project 2007; www. marinefutures.fnas.uwa. edu.au/), a male seadragon carrying a brood of offspring was trawled off the coast of Western Australia. The specimen was accessioned at the Western Australian Museum (WAM P33223.002) as a common seadragon, and its uniqueness only subsequently recognized following DNA sequencing. Three additional specimens were located in scientific collections. The holotype of the new seadragon shows a bright red coloration and with transverse body-markings similar to leafy seadragons. However, the body shape more closely resembles the common seadragon. Here, we examine the anatomy and evolutionary history of the new seadragon to determine its systematic placement and formally describe it.

Molecular sampling
A piece of the tail of the holotype (preserved in ethanol) was used for DNA extraction as the dermal appendages were missing in this species; the specimen was a male carrying its brood, and we also obtained DNA sequences for one of its brood of embryos in order to obtain the mitochondrial haplotype of the mother. A DNA extraction protocol for formalin-preserved tissues [7] was unsuccessful for two of the paratypes (ANFC C2269 and C2270) and not attempted for the third paratype (WAM P660) because of its advanced age (collected 1919). For comparison with the other two seadragon species, we included genetic samples of wild-caught common seadragons and leafy seadragons from across their respective ranges (table 1). A small tissue clip was taken from the ventral dermal appendages of live animals that were subsequently released. This non-lethal sampling was considered less destructive than finclipping as the dermal appendages do not seem to have a locomotory function and are often damaged in otherwise healthy seadragons. We included three specimens of the common seadragon from Western Australia (

DNA extraction and PCR amplification
DNA was isolated with the Qiagen DNeasy blood and tissue kit (Qiagen Sciences, Germantown, MD). Seven marker regions (four mitochondrial, three nuclear) were amplified in a reaction volume of 12.5 µl with the following primer pairs: -12S: L1091 and H1478 [11], modified following reference [           -16S: L2510 and H3058 [12], which amplify 522 bp of the 16S ribosomal RNA gene, with the same thermal profile as 12S. -ND4: L11424-ND4 [13] and H12293-Leu [14], amplified an 838 bp fragment of the NADH dehydrogenase subunit four, including transfer RNA histidine and serine, with a thermal profile of 94 • C for 9 min, 34 cycles of 94 • C for 45 s, 52 • C for 45 s, 72 • C for 60 s, and a final extension at 72 • C for 6 min. -Control region: L15995 L-PRO [15] and H693.5 12S5r.5 [16], amplified a 1036 bp fragment of the mitochondrial control region, including fragments of the 12S ribosomal RNA gene. The fragment has a repetitive region in the middle of the fragment that impaired the sequencing reaction. To obtain the full-length fragment, we developed an additional primer set for sequencing the flanking regions of the mitochondrial control region (CR-FlanksF 5 -CCCCCACCCCCTTTAAAGAC-3 and CR-FlanksR 5 -CGAGTCGTATGTGTCCCACC-3 ). All fragments were amplified via a thermal profile of 94 • C for 3 min, 40 cycles of 94 • C for 60 s, 50 • C for 30 s, 72 • C for 60 s, and a final extension at 72 • C for 5 min. -S7: S7RPEX1F and S7RPEX2R [17] amplified up to 674 bp of the intron 1 and exon 2 of the nuclear S7 ribosomal protein gene with the thermal profile specified in reference [18]. -Aldolase: we used forward and reverse primers as specified in reference [19], amplifying a 382 bp long fragment of a nuclear aldolase-like gene containing both coding and non-coding regions with a thermal profile as outlined in reference [19]. -Tmo-4c4: Tmo-4c4F and Tmo-4c4R [20] amplified a 559 bp long coding fragment of the nuclear Tmo-4c4 gene with the thermal profile specified in reference [18].

Solegnathus dunckeri
PCR products were purified with Exo-SAP-IT (USB Affymetrix) and sequencing was completed with the corresponding PCR primers (Eurofins MWG Operon, Inc.). Overlapping fragments were edited and then merged into consensus sequences in GENEIOUS v. 6.0.3 (Biomatters Ltd.). All sequences were deposited on GenBank (table 1). Sequences for each marker were aligned on the MAFFT v. 7 server [21,22]. Most markers aligned without gaps, but gaps were inserted into the control region fragment resulting in an aligned length of 1084 bp, and S7-like protein aligned to a length of 685 bp. Alignments were deposited on TreeBASE (S15971).

Distance-based analysis
Uncorrected and model-corrected genetic distances were calculated in PAUP* v. 4b10 [23]. The appropriate model selected by JMODELTEST v. 2.1.4 [24] employing the Akaike information criterion was GTR + G for the mitochondrial dataset (control region, 16S, 12S, ND4) with a gamma parameter of 0.283. For S7 and aldolase, the chosen model was HKY, and K80 + I for Tmo-4c4 with a proportion of invariant sites of 0.723.

Phylogenetic analysis
To assess the phylogenetic affinities of the seadragons, we conducted maximum-parsimony (MP) and maximum-likelihood (ML) analyses. To check for data congruence, we analysed each nuclear marker individually, while we treated the mitochondrial markers as a single locus (2790 bp), and finally concatenated the datasets (4416 bp) in MESQUITE v. 2.75 [25].
Maximum-parsimony analyses were conducted in PAUP*, with unordered and equally weighted characters, heuristic searches with 1000 random addition-sequence replicates and tree bisection and reconnection (TBR) branch swapping and with zero-length branches collapsed. To assess clade confidence, parsimony jackknifing was conducted with 1000 jackknife replicates, each with 100 random addition replicates, 37% character deletion, heuristic search option and TBR branch swapping.
Maximum-likelihood analyses were conducted in RAXMLGUI v. 0.93 [26,27] employing the GTR + G model for each of the partitions. The dataset was partitioned for the different loci, for introns and exons of aldolase and S7, and additionally for the codon positions for the protein-coding regions of Tmo-4c4 and the exons of aldolase and S7, and the protein-coding regions and tRNA segments of ND4 (partitions given in the electronic supplementary material, file S2). One thousand replicate ML inferences were performed and clade confidence was assessed with 1000 bootstrap pseudo-replicates employing the thorough bootstrapping algorithm (option '-b'). The analyses were rooted with Syngnathoides biaculeatus, with the three Solegnathus species unconstrained. Sequence matrices and resulting trees were deposited on TreeBASE (S15971).

Morphology and skeletal structure
Morphometric measurements were taken of the holotype and three paratypes, and compared with scaled photographs from field observations of leafy (n = 25) and common seadragons (n = 20; for details, see electronic supplementary material, file S1). The descriptive morphology follows [4], and proportional measurements were taken as in reference [28].
Micro-computed tomography (µCT) of the holotype was conducted at the Cartilage Tissue Engineering laboratory at the University of California San Diego. The specimen was imaged in a Skyscan 1076 (Kontich, Belgium), wrapped in ethanol-soaked gauze, and positioned in a sealed LDPE container. Imaging was conducted at 36 µm isotropic voxel size, applying an electrical potential of 100 kVp and a current of 100 µA, with a 1.0 mm aluminium filter. A beam-hardening correction algorithm was applied during image reconstruction. The reconstructed file size was 1000 × 1000 × 5482 pixels (xyz).
DICOM stacks were processed and segmented in AMIRA v. 5.4 (Visage Imaging, Inc.), and visualized in MAYA 2014 service pack 4 (Autodesk, Inc.). X-ray radiographs of all seadragon species were obtained at Scripps Institution of Oceanography.

Genetic distance
The new species showed clear genetic divergences in mitochondrial and nuclear sequences of a comparable magnitude to those between the established species of seadragons, both in uncorrected and model-corrected distances. As an example, the mitochondrial markers of the new species differed by 7.4% to Phyllopteryx taeniolatus and by 13.1% to Phycodurus eques, which is similar to the 11.6% distance between the two known species (uncorrected pairwise distances). Model-corrected distances (GTR + G) of the mitochondrial data showed 9.9% divergence between the new species and Phyllopteryx taeniolatus, and 22.3% to Phycodurus eques, similar to the two known species that diverged by 18.9%. Comparisons for the three nuclear loci showed a similar pattern (electronic supplementary material, tables S2-S4). The mitochondrial haplotype of the male was 6 bp different from its brood, which represented the mother's mitochondrial haplotype. The father and his offspring were genetically identical for the three nuclear markers.

Phylogenetic placement
All analyses retrieved the seadragons as a well-supported clade with respect to Syngnathoides biaculeatus and Solegnathinae (figure 1; for individual gene trees see electronic supplementary material, figure S3). Among seadragons, the new species was supported as the sister-species to the common seadragon, and the leafy seadragon as the sister to this clade (figure 1; electronic supplementary material, figure S3a,c,d). The nuclear S7 was incongruent with the other markers, supporting the leafy seadragon as the sister to the new seadragon (electronic supplementary material, figure S3b). MP and ML analyses gave similar topologies within the seadragons.

Morphology
The new species has morphological features that clearly distinguish it from the other two seadragons (table 2; see also electronic supplementary material, file S1). In life, the holotype is easily distinguished by its red colour (figure 2a), and pink vertical bars that extend halfway up the body to the lateral trunk ridges (figure 2b). The more uniform coloration differentiates it from the common seadragon, which has numerous colourful spots and blotches (figure 1). These spots are always distinct, even in preserved specimens (see electronic supplementary material, figure S1) allowing differentiation of the two species even when the pigmentation is faded. The patterning also differs from the tan-yellow leafy seadragon, which has vertical bars that extend all the way up the body and are white with purple rims ( figure 1). The X-ray radiographs and µCT showed that the new species has 18 trunk vertebrae with the corresponding outer bony plates (see electronic supplementary material, videos S1 and S2      . dorsal spine on trunk ring 11 orients backward (figure 3a) backward (figure 3b) forward ( figure 2b, figure 3c) .

Paratypes
Three paratypes are designated based on morphological similarity to the holotype (table 3 and

Etymology
Named for Mary 'Dewy' Lowe, for her love of the sea and her support of seadragon conservation and research, without which this new species would not have been discovered.

Diagnosis
Eighteen trunk segments. Head crest without appendage. Pectoral area enlarged (pa in figure 2b,e). Dorsally enlarged spine on trunk ring 11 pointed forward (ds in figure 2b) Table 3. Morphometric measurements for the three seadragon species. Values for Phyllopteryx dewysea are given separately for the holotype and as ranges for the three paratypes. Proportions are given as % of the total length (TL), head, or snout length. The tail length of the holotype was measured from the photograph (figure 1a) as the tail is incomplete in the preserved specimen.

Distribution and bathymetry
The type locality (arrow in map of figure 1) at 50+ m shows a complex habitat of a mixed reef and sandy habitat [29,30]. Two paratypes were collected from the continental shelf near Perth (72 m, circle in figure 1); the habitat at this site is unknown. The fourth paratype, collected at Cottesloe, Perth, does not have depth data, as it was washed up on the beach.

Remarks
The placement of this species within Phyllopteryx was informed by the phylogeny of seadragons ( figure 1). The new species lies as the sister group to Phyl. taeniolatus and this does not allow for its inclusion within the other seadragon genus, Phycodurus. We made the decision to place it within Phyllopteryx rather than erect a new genus, as this provides information about the closer relationship of Phyl. dewysea with Phyl. taeniolatus than with Phyc. eques. An expanded description of the type series of Phyl. dewysea, of Phyl. taeniolatus and Phyc. eques, including detailed comparisons, can be found in the electronic supplementary material, file S1.

Discussion
The discovery of a new species of seadragon represents an important finding. It is genetically distinct from the two known seadragons, with supporting evidence from morphology and habitat. Although the two previously known seadragon species were described in the nineteenth century, Phyl. dewysea remained unrecognized for almost a century, after being first collected in 1919. This seadragon may occasionally be washed ashore, and be subjected to trawling, but may not have been recognized (as was the case with the samples located for this study) owing to a superficial similarity to common seadragons. Nevertheless, it seems surprising that this species escaped recognition until now given the public interest in seadragons. A possible explanation is that it may occur in depths beyond recreational SCUBA range: three of our specimens of Phyl. dewysea came from relatively deep water (51 and 72 m) a few kilometres offshore. The bathymetric range of common and leafy seadragons is insufficiently known, but they seem to be mostly restricted to inshore areas [4,31]. Clearly, two sampling localities in offshore waters do not make Phyl. dewysea an exclusive deeper water species, but the coloration of the holotype may be explained by a deeper habitat: although the red colour is conspicuous when out of the water (figure 2a), red light is rapidly absorbed with depth and so being red may effectively render the seadragon cryptic [32].
Basic biological information on habitat preferences and distribution are needed to assess the status of Phyl. dewysea in the wild, and underpin any future management plans. The few records of the new seadragon from southwestern Australia that we currently have cannot give a satisfactory picture of the distributional range. Furthermore, the records off the coast of Perth are almost 60 years old and it is unclear if Phyl. dewysea still occurs in this region. The urban development surrounding the Perth region has been significant in the past century [33], including changes in water quality and vegetation [34,35] that may have also affected the offshore habitat of Phyl. dewysea.
Only the retention of the unrecognized Phyl. dewysea specimens in museum collections allowed for its discovery as a new species. This highlights the significance of natural history collections as a treasury of new species [36], even decades after their collection [37]. Another invaluable component of biodiversity discovery consists of exploratory expeditions such as the Marine Futures project that trawled the holotype and mapped its habitat [29]: the on-board documentation provided the only available photograph of a fresh Phyl. dewysea, a currently unique representation of its natural coloration (figure 2a).
Southwestern Australia hosts a large number of marine fish [38], and is considered a hotspot for future discovery [3]. Phyl. dewysea is an illustration of the surprises that are waiting in the ocean, but many open questions remain regarding its range, habitat and biology. The other two seadragons are considered Near Threatened by the IUCN [6,39], and their conservation is of public interest [5,40]. The new seadragon is no less charismatic and deserves to become a flagship species for the discovery of biodiversity.
Ethics statement. Non-lethal clipping of the dermal appendages with subsequent release of the seadragons was authorized by the following authorities: Department of Primary Industries and Regions South Australia (PIRSA),