Taxonomic revision and phylogeny of the Ophiocoma brevipes group (Echinodermata, Ophiuroidea), with description of a new subgenus (Breviturma) and a new species

The taxonomy of the genus Ophiocoma was last revised by Devaney in 1970. Recent discoveries of new species and re-instatement of previously synonymized names suggest that we still do not fully understand the species limits in this genus. A recent biodiversity survey of the SW Indian Ocean shallow reefs strongly suggested an unrecognised species in the genus, closely related to O. brevipes/O. dentata. This study examined both the molecular phylogenetic relationships and the morphological characteristics of several species in the genus in order to characterise the unrecognised species. The focal species clusters with O. brevipes, O. dentata, O. doederleini within a monophyletic clade supported by molecular data for the fi rst time. The name Breviturma subgen. nov. is proposed for this clade, previously known as brevipes group. Type material of nominal species that have been synonymized with O. dentata was examined and re-assessed. Ophiocoma marmorata proved not conspecifi c with O. dentata. A rarely used character, dorsal disc granule density, was tested and showed differences between the examined species at similar sizes. In combination with colour pattern, disc granule density, arm spine sequence and maximum disc size, the new species was delimited morphologically and described as Ophiocoma krohi sp. nov.


Introduction
The history of the taxonomy of the genus Ophiocoma L. Agassiz, 1835 is convoluted and confused.Devaney (1970) resolved some of the confusion using external and internal characters such as spine sequences along the arm, colour pattern, and the shape of the dental plate, to delimit four groups, namely the brevipes, scolopendrina and pumila groups earlier proposed by Clark (1921), and the pica group (Devaney 1970).After 1970, few new species of Ophiocoma have been described, namely O. paucigranulata Devaney, 1974 from the Caribbean and O. aegyptiaca Soliman, 1991 from Egypt.Also, O. endeani Rowe & Pawson, 1977 was proposed as a replacement name for O. alternans Endean, 1964 (a homonym) by Rowe & Pawson (1977).Ophiocoma similanensis Bussarawit & Rowe, 1985 was synonymized with O. erinaceus Müller & Troschel, 1842 by Price & Rowe (1996).Recently a new species O. cynthiae Benavides-Serrato & O'Hara, 2008 was described and O. schoenleini Müller & Troschel, 1842 was reinstated as a valid species (Benavides-Serrato & O'Hara 2008).Both are close to O. erinaceus and have previously been mistaken for it, as was suggested by an earlier molecular study (O'Hara et al. 2004).Although Ophiocoma is a well studied genus, consisting of comparatively large, conspicuous, abundant and easily accessible tropical shallow water species, it clearly still holds unrecognized diversity.Currently, 22 species are accepted in the genus (Stöhr & O'Hara 2013).
In a recent large-scale marine biodiversity survey in the Southwestern Indian Ocean (BIOTAS), we identified several specimens as belonging to the brevipes group based on morphology.According to Devaney (1970), the brevipes group currently includes the three species Ophiocoma brevipes Peters, 1851, O. dentata Müller &Troschel, 1842 andO. doederleini de Loriol, 1899.In our material, we found specimens belonging to these three species and a fourth unknown species was suspected and confirmed by subsequent molecular analysis.They formed a distinct genetic lineage when compared to other available species of Ophiocoma (Hoareau et al. 2013).Currently (Devaney 1970;Stöhr & O'Hara 2013), the following nominal species are regarded as synonyms of the species in the brevipes group: O. brevispinosa Smith, 1876 = O. brevipes;O. insularia Lyman, 1862, O. marmorata Marktanner-Turneretscher, 1887, O. ternispina v. Martens, 1870and O. variegata Smith, 1876 = O. dentata.The name Ophiopeza danbyi Farquhar, 1897 was included in the synonymy of O. brevipes by Clark (1915), but not by Devaney (1970).The newly recognized lineage may either belong to a new, not yet described species, or it may correspond to one of the synonymized species.Specimens of the brevipes group are characterized by similar numbers of arm spines on each side of the arm segments beyond the disc, four to six arm spines on each side of proximal arm segments, a dental plate that is roughly twice as long as wide, with wide septum dividing the tooth foramina, broadly oval dorsal arm plates, densely packed, small and spherical disc granules that cover the entire ventral interradius, and two tentacle scales along most of the arm (Devaney 1970).According to Devaney (1970), the comparatively short dental plate with small region for the tooth papillae is an important character to distinguish the brevipes group from the other subgroups of Ophiocoma.Moreover, Benavides-Serrato & O'Hara (2008) found that subtle differences in characters such as the extent of granulation on the ventral disc, tube foot colour and shape of the dental plate had previously been overlooked or misinterpreted as intraspecific variation in species of the scolopendrina group.These criteria might be informative to characterise the specimens of the newly found lineage in the brevipes group.
This study aims to describe that unidentified species in the brevipes group, using both molecular methods and morphological characters.Using two mitochondrial loci, we constructed a phylogeny to assess the molecular status of the brevipes group relative to the scolopendrina group and to evaluate the species relationships within the brevipes group.In addition, we identified a new character, dorsal disc granule density, to distinguish between the species in the brevipes group and re-evaluate type material of the studied species.

Material and methods
Brittle stars of the genus Ophiocoma were collected at La Réunion Island and off Madagascar (Nosy-Be Island).Specimens of O. doederleini from La Réunion and Polynesia were provided by G. Paulay from the collections of the Florida Museum of Natural History.Total genomic DNA was extracted from a piece of arm following the DNeasy protocol (Qiagen).COI sequences were already available for most of the species (Hoareau et al. 2013; Supplementary file: KC759738-KC759923) except for O. dentata and O. doederleini.A portion of the COI barcoding gene and of the 16S ribosomal gene were amplified using echinoderm-specific hybrid primers (Hoareau & Boissin 2010) and universal primers (Palumbi 1996), respectively.We followed the PCR conditions described in these studies.The amplicons were then sequenced using BigDyeTerminator (Applied Biosystems, Foster City, CA, USA) cycle sequencing reactions, and electrophoresed on an ABI 3730xl DNA Analyzer (Applied Biosystems) at the Interdisciplinary Center for Biotechnology Research (University of Florida).We aligned the sequences using mafft 6 online (http://mafft.cbrc.jp/alignment/server/)with all the default settings.All the sequences are available in GenBank (see Supplementary file).From the concatenated datasets, a Neighbour Joining phylogenetic tree was reconstructed using Mega5.05(Tamura et al. 2011) applying 1000 bootstraps and the Kimura-2-Parameters model of substitution.Additionally, we used maximum likelihood inference using Mega.Because we used two mitochondrial loci, we estimated a single substitution model for the concatenated dataset using Mega (GTR + gamma + invariant site model).Support for each node was assessed with 200 bootstrap replicates.To overcome potential complications introduced by alignment gaps, we applied the Partial-Deletion option available in Mega in both reconstruction methods.Sequences of Ophiocoma erinaceus and O. scolopendrina (Lamarck, 1816) from the scolopendrina group were used as outgroup.
After DNA extraction, some of the individuals were used for morphological examination.Since the new species is smaller than the others, we selected comparative material from other species at similar size (11-13 mm disc diameter), depending on availability.Two of the largest specimens of Ophiocoma sp.nov. at 11 mm and 13 mm disc diameter (dd) and a specimen each of O. doederleini (13 mm dd), O. brevipes (13 mm dd) and O. dentata (11 mm dd) were selected for scanning electron microscopy (SEM).These specimens were first photographed with a digital camera.Then they were treated with diluted (1:1) household bleach (NaOCl) to remove the epidermis.They were mounted on aluminium stubs with spray glue, dorsal side up, and examined in a Hitachi FE-S4300 scanning electron microscope.Then the glue was dissolved with butyl acetate and the animals were remounted, ventral side up, and examined by SEM again.Finally, the specimens of O. doederleini (UF5318), O. brevipes (SMNH-133230), O. dentata (URUN2009-11047) and the 13 mm dd Ophiocoma sp.nov.(SMNH-Type 8535) were disarticulated in concentrated bleach.The skeletal elements were washed in tap water, mounted on stubs and examined by SEM.
Whole animals were measured using a digital calliper, smaller parts were measured on the SEM images using the software ImageJ v.1.47c(Rasband 1997(Rasband -2012)).Disc diameter is defined as the distance from the distal edge of a pair of radial shields (or if concealed the disc edge atop the arm) to the edge of the opposite interradius.Densities of disc granules were measured by overlaying a square frame of 0.5 mm side length (0.25 mm 2 ) to the image in ImageJ and counting all complete granules inside the square.This was repeated at least three times in different places of the image, but always in the central part of the disc to avoid the often bulging, more sparsely granulated disc periphery.The numbers were multiplied by four to give densities for 1 mm 2 for easier comparability with previously published values.Vertebrae were measured in ImageJ across their greatest width and height; oral plates (jaws) were measured by fitting an ellipse around their irregular shape and calculating the major (here height) and minor (here width) axes.Arm spines were counted on 2-5 arms of each specimen and when the numbers at a segment varied between arms the less common number is given in parentheses.The taxonomy follows Stöhr & O'Hara (2013).The morphological terminology follows Stöhr et al. (2012).Field numbers, museum registration numbers and GenBank accession numbers are listed in the Supplementary file.

Results
The concatenated sequences of the 17 specimens contained 1125 sites (1069-1119 bp) of which 303 (26.9 %) were parsimony informative.For most of the specimens the complete target sequence was obtained.For the specimens of O. doederleini, 614 bp were recovered from the original 659 bp.The 16S sequences varied depending on the presence of multiple indels (450-456 bp).For the COI locus, no indels, distinct double pics or stop codons were detected which rule out the presence of pseudogenes.Subgenus Breviturma subgen.nov.urn:lsid:zoobank.org:act:DA2BAD13-0E03-45B3-9D7A-78796329BC3C

Diagnosis
Subgenus of Ophiocoma, previously known as brevipes group.Characterized by equal (non-alternating) numbers of arm spines; disc granules uniform in size, low and blunt; dorsal arm plates rounded, almost oval (in adult specimens, more rounded triangular in younger individuals); dental plate about twice as long as wide and thus the shortest in the genus, with tooth papillae region occupying 25 % of the dental plate length.

Etymology
The name Breviturma refers to the original name brevipes group, combined with turma = troop, here meaning group.
MorphologiCal data.12.5 mm dd.Colour pattern on dorsal disc consists of thicker, brown, reticulating lines, instead of a thin net-like pattern as in the other specimens; otherwise similar to the others (Fig. 2C).

Remarks
Ophiocoma doederleini occurs in several colour morphs, with various reticulated patterns (Fig. 2A, C) or spotted disc (Devaney 1970), which is confirmed by our molecular data (Boissin & Hoareau, unpublished).It can easily be distinguished from the other species of Breviturma subgen.nov.by its dark and light strongly annulated, more slender, and longer arm spines.It also has the widest dorsal arm plate of all species of Breviturma subgen.nov.The high granule density in the smallest specimen was surprising and may need further investigation.Maximum size at least 26 mm according to the studied material.Peters, 1851 Figs

Diagnosis
Species of Ophiocoma (Breviturma) subgen.nov.characterized by a light colour pattern, cream to white with some brown to grey markings; five arm spines from segment 4 or 5 on large specimens for more than 10 segments; disc granule densities up to about 250 mm -2 .Maximum size at least 20 mm dd according to the examined material.

Remarks
Ophiocoma brevipes is clearly distinguished from the other species of Breviturma subgen.nov.by its consistently white and light brown to grey colour pattern, up to five arm spines far out on the arms, and dorsal disc granule densities up to 250 mm -2 .According to Devaney (1970), five spines occur in some specimens from 3 mm dd.Müller & Troschel, 1842 Figs

Diagnosis
Species of Ophiocoma (Breviturma) subgen.nov.characterized by rarely possessing more than four arm spines and if present only at few segments; arm spines never annulated; variable in colour pattern (uniformly or marbled brown, reticulated of spotted); dorsal disc granule density less than 150 mm -2 .Maximum size at least 28 mm dd (Devaney 1970).
ColleCting data.Hawaii, 1860, 1 spm (originally donated to Stockholm by Museum of Comparative Zoology at Harvard).
MorphologiCal data.13.3-16.6mm dd, all dark brown to black on dorsal disc and arms, including spines, ventral side slightly lighter brown.Both middle spines about equal, longer than dorsal and ventral spines.(Photos provided by Adam Baldinger, MCZN).

Syntypes MCZN 1666
Syntypes of Ophiocoma insularia.MorphologiCal data.19.5 and 31.6 mm dd, dark brown disc and arms, with lighter abraded patches on dorsal disc, dark brown on ventral disc and arms; a note among the labels reads "4-5 spines, 3 upper about equal".(Photos provided by Adam Baldinger, MCZN).

Remarks
The types of O. insularia are clearly all conspecific.Their colouration and large size suggest that they are also conspecific with O. dentata.Ophiocoma dentata occurs in several colour morphs, from uniformly dark brown to brown marbled discs with banded arms, and reticulated or spotted forms that resemble O. doederleini.It can always be distinguished from the latter species by the absence of annulations on the arm spines, and shorter arm spines.In addition, only the largest individuals sometimes have more than four spines on few arm segments.The conspecificity of the various colour morphs is also confirmed by molecular data (Boissin & Hoareau, unpublished).Another species currently believed to be conspecific with O. dentata is O. variegata with type locality Rodriguez Island.According to the original description (Smith 1876), the type measured 28 mm dd and had only four arm spines, which agrees with O. dentata.Devaney (1970)

Diagnosis
Species of Ophiocoma (Breviturma) subgen.nov.with variegated to uniform dark brown colour pattern; arm spines may be weakly annulated, but not as clearly light and dark banded as in O. doederleini; up to four arm spines (in large specimens beyond segment 15), sometimes five at segments 6-9 (rarely 5-8); relatively coarse disc granulation.Maximum size about 16 mm dd.

Description
Holotype 12.7 mm dd, radial shields completely obscured by granules, granule density 104 mm -2 .Arm spine sequence 3(2), 3(2), 3,4,4,4,4,4,4,4,4,3,3,3,3. Dorsal disc with dark and lighter brown lines radiating from a dark central patch, some white spots at disc edge (Fig. 6A).Dorsal arms medium brown with dark and white transverse bands (Fig. 6B).Ventral disc dark brown mottled, ventral arms and oral frame lighter brown, oral shields dark.Ventral interradii completely, but sparsely granulated (Fig. 6D, E).Dorsal spines brown, ventral spines weakly annulated brown on white.DAPs oval, W:L 1.8:1, contiguous.Dorsalmost spine a little longer than a segment, middle spines about 2 segments long, ventralmost spine slightly shorter.Arm spines flattened, slightly tapered with truncated tips, dorsalmost spine wider than other spines.Oral shields trapezoid, distally almost as wide as long, proximally just over half as wide as long, lateral edges straight (except madreporite), proximal and distal edges convex (Fig. 6D).Madreporite larger than other oral shields, more rounded, all edges convex.At tip of jaw a cluster of 8-10 small, round tooth papillae.Along each jaw edge 4 rectangular, wider than high, oral papillae, and distally a smaller papilla at an angle to the others (Fig. 6E).Adoral shields uneven triangular, framing the lateral edges of the oral shield, not meeting proximally.Bursal slits extend from oral shield to disc edge.First ventral arm plate outside mouth-slit, small, winglike pentagonal, middle of distal edge notched.Following plates pentagonal, about as long as greatest width, with obtuse proximal angle, lateral edges notched, distal edge convex, contiguous (Fig. 6C). 2 oval scales at a slightly open angle at the lateral arm plate of each tentacle pore proximally; the inner scale decreases gradually and is absent on the distalmost segments.Tube feet expanded, smooth, with distal constrictions and terminal bulb.
Oral plates, adradial distal part with strong folds, abradial face with large muscle flange with horizontal striations that end in deep incisions in the distal edge; 1.9 mm wide, 2.3 mm high (Fig. 5C).Dental plate short, with two tooth foramina on external dorsal half, dorsalmost one oval, wider than long, other small, round; internally a wide septum divides each foramen into two pairs of vertical oval holes, upper ones larger than lower ones (Fig. 5F).Radial shield angular, with large round condyle at internal distal end (Fig. 5I).Adradial genital plate long, flat, thin, thickened at distal end, with deep pit; abradial genital plate thin, blade-like, curved distally (Fig 5L).Proximal vertebrae 2.0 mm wide and 1.5 mm high (Fig. 5Q).

Remarks
Ophiocoma krohi sp.nov.differs from O. brevipes and O. dentata in maximum size, granule size and density, and in colour pattern.Although quite variable in colour, it has never been observed with reticulated or spotted pattern as O. dentata and O. doederleini.We examined 12 specimens and half of them showed up to five arm spines on 2-4 segments, the other half had only four.Thus they all differ in spine number from O. brevipes, but some are similar to O. dentata in lacking a fifth spine, although they may have four spines farther out on the arm than O. dentata.
The specimens reported by Stöhr et al. (2008) are most likely conspecific with O. krohi sp.nov.based on colour pattern, the presence of five spines and their small size.The species has a wide geographic distribution on coral reefs across the Indian and Pacific Oceans at depths of 0-30 m (Hoareau et al.

Remarks
The original description of O. marmorata mentions only a single specimen of 9 mm dd (Marktanner-Turneretscher 1887), which should be regarded as the holotype.The remaining three specimens cannot be regarded as type material since they are not mentioned in the description.The locality of O. marmorata is most likely incorrect as it lies in the middle of the Atlantic Ocean at great depth.Ophiocoma is a shallow water tropical genus, rarely reported below 100 m depth.Taking into account that the determination of coordinates, in particular longitude, was difficult and often inaccurate in the 19 th century, we think it possible that the correct locality may be the Saint Peter and Saint Paul Archipelago, which is situated at 0.9°N, 29°W, assuming that the data are based on the Greenwich Meridian.Instead, the Ferro Meridian may have been used, which would put the locality at 40°40'-42°40'W today, even closer to Brazil.The specimens were bought by director Eichhorn of Graz (Austria) from the trader Hugo Schilling in Hamburg in 1882, together with several other echinoderms, and transferred to the natural history museum in Vienna (A. Kroh, pers. comm.).
The alternating spine sequence of the holotype of O. marmorata may be interpreted as an unusual variation, if we accept that it is conspecific with the three smaller specimens in the lot.This character is typical for species in the scolopendrina group though and therefore the holotype of O. marmorata may belong to one of the species in that group, possibly O. echinata (Lamarck, 1816) or O. wendtii Müller & Troschel, 1842, which are the only two species of that group known from the Atlantic Ocean.The smaller specimens, however, may belong to a different species, possibly O. pumila Lütken, 1856, a member of the pumila group, which has a spine sequence similar to the smaller O. marmorata  and is known from the Atlantic.Devaney (1970) does not seem to have examined O. marmorata and did not know about the existence of five spines on one of the specimens in the lot.He also did not consider the significance of geographic distribution when he suggested that O. marmorata (an Atlantic species) may be conspecific with O. dentata (an Indo-Pacific species).Although the fifth arm spines in the small O. marmorata are in exactly the same position, arm segments 6 and 7, as in O. krohi sp.nov., the granules in these small specimens are high and pointed, different from those found in the species of Breviturma subgen.nov.Granules like that are known from the pumila group.The holotype of O. marmorata has low round granules, which also supports that it is a different species from the three smaller specimens.
The taxonomic status of O. marmorata could not be completely resolved by this study and requires further investigation, but we strongly doubt that it is conspecific with O. dentata.We propose that the synonymy of O. dentata includes only O. ternispina, O. insularia and O. variegata.

Remarks on species delimitations
Prior to Devaney's (1970) work, the three species previously recognized in the brevipes group were considered mere variations of a single species.Devaney (1970) was able to separate them by spine sequence, colour pattern and arm width.In addition to the annulated colour pattern of the arm spines, spine numbers seem to be a critical character to distinguish between O. dentata and O. doederleini.According to Devaney (1970), O. dentata has five arm spines only at disc diameters above 20 mm and spine numbers decrease to three on the middle arm (beyond segment 17 at 13-14 mm dd, beyond segment 25 at 24 mm dd), whereas four spines are found on far more distal arm segments in O. doederleini of similar disc sizes (to segment 32-38 at 13-14 mm dd, to segment 55 at 24 mm dd).In light of our findings (Table 1) we cannot exclude though that the specimens regarded as O. dentata by Devaney actually included individuals of O. krohi sp.nov..All of his data must therefore be treated with caution.The type of O. dentata (18-19 mm dd) has up to four spines on the proximal arm segments, decreasing to three from segment 13, which contradicts Devaney's (1970) observations.Our largest specimen of O. dentata has two occurrences of five spines among all ten sides of the arms, and three spines from segment 16.
In the type material of O. brevipes, spine numbers appear to be uncorrelated with disc size; five spines are present from the fourth, fifth or sixth segment and for a variable number of segments before the number drops to four and finally three again on the distalmost part of the arm.The smallest type, at 12 mm dd, shows 12 segments with five spines, whereas the largest, at 17-18 mm dd shows only six segments with five spines; in both animals there are five spines from the sixth segment.Similarly, spine numbers vary between specimens on the types of the three nominal species currently regarded as conspecific with Fig. 7. Granule densities in relation to disc diameter of the examined nominal species of Ophiocoma, from Table 1.
O. dentata.In O. krohi sp.nov., the smallest specimen has five spines on two segments, while the largest one has only up to four spines.These variations are common in the genus Ophiocoma as was observed already by Devaney (1970).We also found significant differences in granule density among species as shown in Fig. 7 (Kruskal-Wallis test, K=8.9055, d.f.=3, P =0.0306).Likewise, granule sizes and/or densities may be subject to growth changes as suggested by the inverse correlation observed between the granule density and the disc diameter in O. krohi sp.nov.(Spearman test, R=-0.8012,P=0.0006).The smallest O. krohi sp.nov.had 152 granules/mm 2 , whereas the largest one had only 64-96 granules/ mm 2 .Small O. dentata have three times higher granule densities as large ones, which suggests that granule numbers do not increase with growth.Somewhat contradictory are the data for O. brevipes, where specimens from La Réunion have greater granule densities than similar size specimens from Madagascar.
We decided to examine a rarely used character, granule density, that has occasionally been mentioned in species descriptions and revisions, but never been used in a systematic way to differentiate Ophiocoma species.Devaney (1970) used granule densities as a character in his key to the species of the scolopendrina group, but only for few of the species.Granule densities were compared between light images and SEM images of the same animal and the small observed differences can probably be attributed to uneven distribution of the granules on the disc, rather than to a real difference in accuracy between these methods.Our method of counting granules on digital images resulted in considerably higher numbers than what has been published before.However, published values are not always related to size and as we see in our results, the largest specimens have larger and fewer granules.It is also possible that counts on images are more accurate than counts under the dissecting microscope.In any case, since we used the same method for all specimens, the differences between them are valid and important.Devaney (1970) examined arm span on the 10 th free segment, "composed of the length of the longest arm spines on each side of a segment, the breadth of the dorsal arm plate and the breadth of the lateral arm plates" (cited from Devaney 1970), and found larger arm spans in O. doederleini than in O. dentata.However, his values overlap and vary uncorrelated to size, and we consider this a weak character that is difficult to assess.

Phylogeny
The molecular phylogeny suggests monophyly of the brevipes group with respect to the scolopendrina group, which supports the establishment of Breviturma subgen.nov.This is the first time that the molecular basis of the species groups of Ophiocoma described by Devaney (1970) can be demonstrated.The phylogeny also revealed four deep lineages within Breviturma subgen.nov.corresponding to the three known species and the new species described here.Within this subgenus, O. doederleini is the most distantly related, and the new species O. krohi seems to be sister taxon to the O. brevipes/O.dentata clade, even though the data are not strongly supported.Devaney (1970) showed that the dental plate can be used to divide the genus Ophiocoma into groups.All our specimens concur with his description of the dental plate in the brevipes group, about twice as long as wide with small part for tooth granules, and all tooth foramina divided by a wide septum.The relationships within Breviturma subgen.nov.are difficult to resolve with morphological means.Benavides-Serrato & O'Hara (2008) showed that the dental plate in the scolopendrina group differs between species, but that does not seem to hold true within Breviturma subgen.nov.There appear to be slight differences in the geometry of the oral plates, but it is currently unclear if these are significant.The overall shape of the oral plate is the same in all species, but it seems to be lower and shorter in O. dentata than in O. brevipes and O. krohi sp.nov.Similarly, the geometry of the vertebrae seems to differ slightly, with O. dentata having the smallest vertebra.However, these were not taken from exactly the same arm segment and size differences may correlate with ontogenetic age and overall size of the animal.

Morphology
As with the molecular data, Ophiocoma doederleini is clearly distinguished from the other three species by morphology.The most conspicuous difference is the presence of annulated arm spines.The spines are also longer and thinner than in the other species.Granule density of the smallest specimen concurs with both O. dentata and O. brevipes, while the larger specimens have densities similar to O. dentata.It is most similar in arm spine numbers to O. krohi sp.nov.which also exhibits weakly annulated spines.However, the new species reaches a much smaller size (16 mm dd vs. 26 mm dd) and differs in disc colour patterns (Fig. 2G-L).
Ophiocoma brevipes and O. krohi sp.nov.are similar in having up to five arm spines at small sizes, but differ in the number of segments that have five spines; they are similar also in the shape of the radial shield with a large ball-like articulation condyle that is less pronounced in O. dentata and O. doederleini.The genital plates of the three most closely related species O. brevipes, O. dentata and O. krohi sp.nov.are similar, but in O. dentata the adradial genital plate is relatively shorter and the abradial genital plate has a wider proximal edge, similar to O. doederleini.In granule density, O. dentata is close to O. brevipes, but both differ greatly from O. krohi sp.nov., which again mirrors the assumed phylogenetic relationships.

Conclusion
All four species in Breviturma subgen.nov.are similar to each other in some characters.In granule density and in the presence/absence of annulations on the arm spines, O. krohi sp.nov.seems to be related to O. doederleini.The higher number of arm spines separates O. brevipes from the other three species.Ophiocoma dentata and O. doederleini share several colour morphs.Ophiocoma krohi sp.nov. is the smallest of these species with about 16 mm dd, a difficult character that only clearly distinguishes significantly larger specimens.It is currently impossible to deduce the ancient states of any characters or infer a phylogeny within the subgenus by morphology.However, a comprehensive phylogenetic analysis of the whole genus may offer better clues to solving this problem.
The identity of the new species was difficult to reveal, since several nominal species have been synonymized with O. dentata and each one could potentially be conspecific with the new species.The spine number and granule density appear to be the most important characters to differentiate these species.However, the ophiuroid skeleton grows continuously and the number, shape and proportions of various parts are known to change during growth (Stöhr 2005).For instance, we show that in Ophiocoma krohi sp.nov. the granule density decreases with size.This complicates comparisons with type material that is naturally only available at limited size ranges and usually not available for dissection.
The recent description of two new species (Benavides-Serrato & O'Hara 2008 and this study) in a relatively easily accessible (shallow water), conspicuous and well-studied group (ophiocomid brittle stars) suggests that a considerable part of marine biodiversity may remain to be discovered.

Fig. 1 .
Fig. 1.Phylogenetic reconstruction of Ophiocoma (Breviturma) subgen.nov.(two species of the scolopendrina group as outgroup), estimated by maximum likelihood and neighbour-joining analysis of concatenated sequences of the mitochondrial cytochrome c oxidase subunit I and rDNA 16S (in total 1125 bp).Values at branches indicate bootstrap values for maximum likelihood (above branches) and neighbour-joining (below branches).Specimen codes are resolved in the Supplementary file.

Table 1 .
Only three amino-acids showed variation in the COI sequences; one site involved a change from Serine to Alanine in all the specimens of O. krohi sp.nov.Morphometric data for the species of Ophiocoma (Breviturma) subgen.nov.Type material from Zoological Museum Berlin (ZMB), Swedish Museum of Natural History (SMNH) and Natural History Museum Vienna (NHMW).Specimen NHMW 104.66a is regarded as the holotype of O. marmorata, the other specimens of that lot most likely belong to a different species.DD, disc diameter, MNHN, Muséum national d'Histoire naturelle (Paris), UF, Florida Museum of Natural History, URUN, University of La Réunion.