The Taxonomy and Phylogeny of Echinometra (Camarodonta: Echinometridae) from the Red Sea and Western Indian Ocean

The number of valid species in the genus Echinometra (Echinodermata, Echinoidea) and their associated identification keys have been debated in the scientific literature for more than 180 years. As the phylogeny and dispersal patterns of these species have been widely used as a prominent model for marine speciation, new insights into their taxonomy have the potential to deepen our understanding of marine speciation processes. In this study we examine Echinometra taxonomy, combining morphology and molecular tools. We present the taxonomy and phylogeny of Red Sea and Western Indian Ocean Echinometra. The currently available morphological keys were found to be limited in their ability to delineate all species within this genus. Nonetheless, morphological similarities between the Red Sea and Western Indian Ocean populations were high, and delimited them from the other species. These latter populations together formed a monophyletic clade, genetically distant from any of the other Echinometra species by more than 3%. Combining both traditional taxonomy and molecular evidence, we found that these populations were neither Echinometra mathaei nor E. oblonga, as previously considered. The morphological discrepancies of these populations, and their genetic divergence from the other Echinometra species, suggest that they should be considered as a new Echinometra species.


Introduction
The genus Echinometra currently comprises eight species, two of them still undescribed, with species-level taxonomy of this genus yet to be completed [1][2][3][4][5]. Echinometra are of pantropical distribution and are often among the most prevalent urchins in the reefs they inhabit [2,6,7]. Though early studies of Echinometra suggested that only one species of this genus, E. mathaei, occurred in the Indo West Pacific (IWP) [8,9], later studies revealed the presence of four independent Echinometra species in that region [3,4,6,[10][11][12][13][14][15]. These four closely-related IWP species occur sympatrically in Okinawa [6,11,14,16], and were originally referred to as Echinometra species A, B, C, and D [3]. Studies on both the morphological characteristics [3] and genetics [1,4] of these species, suggest that E. sp. B and E. sp. D in Okinawa are E. mathaei and E. oblonga, respectively; while E. sp. A and E. sp. C have remained unnamed. One of these four species, E. mathaei, has been called the world's most abundant sea urchin [6], with distribution that spans from Hawaii and Tahiti throughout the Indo West Pacific (IWP), to the Western Indian Ocean (WIO) and the Red Sea (where it is the only Echinometra species reported) [2,6,[17][18][19].
Though the taxonomy, phylogeny, and genetic structure of IWP Echinometra have been extensively studied [3,4,[12][13][14]16], and much has been done in the eastern Pacific and tropical Atlantic [7,20,21], little has been done to date in regard to the Red Sea and WIO.
Here we investigated the taxonomy and phylogeny of Red Sea (Eilat) and WIO (Zanzibar) Echinometra. The four Okinawan Echinometra served as reference for delineating species from the other two regions. The phylogeny of Echinometra from the Red Sea and WIO is presented here for the first time. Our updated taxonomy of urchins from the latter two regions and their newly suggested phylogeny demonstrate the advantages of combining both molecular and morphological tools in delineating the boundaries and inferring relations between species of this genus.

Ethics statement
All field research and collection of specimens were approved by the local authorities in the country of collection, and permissions were granted as follows: Permit number 2007/28851, issued by the Israeli Nature and National Parks Protection Authority for collection in Eilat, Israel. Permit number AOl/VolXV/38, issued by the University of Dar Es Salaam, Institute of Marine Sciences for collection in Zanzibar, Tanzania. Sample collection in Okinawa was conducted as part of the 21 st Century Center of Excellence (COE) summer program, conducted at the University of the Ryukyus, Japan. Studies involving Okinawan Echinometra did not involve endangered or protected species and did not require a permit.

Procedures and sample collection
Samples of Echinometra spp. were collected between June 2007 and November 2008 from three locations; Okinawa, Zanzibar, and Eilat (Gulf of Aqaba/Eilat, northern Red Sea) ( Figure 1). A total of 69 individuals were collected from Zanzibar, 86 from Okinawa, and 42 from Eilat (Table 1). Samples were first morphologically identified using Mortensen's criteria [8] and then sequenced and grouped based on the mtDNA phylogenetic tree reconstruction, in order to ratify the current taxonomical attributions. In this work Zanzibarian Echinometra are referred to as ZE and Eilat's Echinometra as EE.

Morphological measurements
The morphological array of characters used to delineate the species of Echinometra comprised: length, width and height of test, color of spines, milled rings (A small flange near the base of the spine marking the distal most limit of muscle attachment onto the spine base), and skin around the peristome, shape of spicules in the tube feet and gonads, the number of pore-pairs, and sperm morphology. The length and width of the tests were measured at the ambitus while height was measured along the oral-aboral axis. Measurements to the nearest 0.5 mm were performed using thin blade Vernier calipers to prevent interference by the spines. Color of the spines was described from live samples. Milled rings were determined as bright, faded, or dark, and the skin around the peristome as bright or dark. Following external examination the urchins were dissected and the gonads removed. Gonad samples were then used for spicule analysis, sex determination, and DNA extraction. In order to obtain tube feet spicules, several tube feet were detached with forceps, while gonad spicules were obtained from a small portion of the gonad. Tube feet and gonad spicule samples were photographed under X20 and X10 magnifications, respectively, using a light microscope, and analyzed using the software ImageJ [22]. The number of porepairs on every ambulacral plate was counted under a dissecting microscope. Pair-wise Kolmogorov-Smirnov tests were performed for every pair of samples to check for similarities in pore-pairs distribution. P-values were adjusted for multiple testing to minimize false-discovery-rate using the Bonferroni correction.
Examination of sperm was carried out using a scanning electron microscope (JEOL JSM5610LV). Sperm samples were obtained as undiluted semen, i.e., "dry sperm", by removing the testes. The samples were then fixed in 2.5% glutaraldehyde in filtered (0.2 µm) seawater (FSW) and stored at 4°C. Further preparation of the samples followed a procedure modified from [3].
The chains were sampled every 100 generations. Convergence was achieved at 3,000,000 generations. After convergence, the sampling continued until the analysis reached 20,000,000 generations. The first 3,000,000 generations, amounting to 15% of the total number of generations, were discarded as burnin. Genetic divergences within and among taxa were calculated as Kimura-2-parameter (K2P) distance [30] using MEGA v5.0 [31].

Morphologically-based identification of Echinometra from Okinawa, Zanzibar, and Eilat
Summary of the morphological set of characters: color of spines, milled rings, skin of peristome, tube feet, and gonad spicules yielded 27 character-states arranged in 60 different character state combinations and summarized in Table 2. Based on these morphological criteria, individuals from Okinawa were assigned to four groups, corresponding to the four known IWP Echinometra; while individuals from Eilat and Zanzibar seemed to differ from these species based on the same characters. However, even among the IWP species, the current morphological keys were sometimes insufficient for conclusive species identification. For example, while E. sp. A is distinctively characterized by having white-tipped spines (found in 100% of the samples), the great diversity of spine color in E. sp. C (from light brown to violet, green, and black), rendered this character uninformative for this species. The unique whitetipped appearance of E. sp. A was never observed in ZE or EE. While the majority of ZE individuals (62.7%) resembled E. oblonga in bearing black spines, the rest of the population included character states shared by both E. mathaei and E. sp. C (i.e. light brown) or a new unique violet color. EE could not be discriminated from either E. mathaei or E. sp. C based on this character. Milled rings presented three character-states: 'bright', 'dark', and 'faded'. While the first two appeared in individuals from all locations, the 'faded' morph was strictly confined to the Zanzibar and Eilat individuals (32.5% and 30%, respectively). The skin around the peristome was described as being either 'dark' or 'bright'. These two character-states were not evenly distributed, as the 'dark' morph predominated in all but one species (E. sp. C). ZE and EE had highly similar proportions of peristomal skin coloration. For both, proportions of the bright-skinned individuals were very low (1.9% and 2.5%, respectively) in comparison to the dark-skinned individuals (98.0% and 97.5%, respectively). These proportions were close to those displayed by E. oblonga (entirely dark-skinned around the peristome), and furthest from E. mathaei and E. sp. A (62.2% and 75%, respectively).
Four dominant spicule types ( Figure 2) were encountered in the gonads and tube feet, arranged in seven different spicule combinations. Two additional spicule arrangements were defined as 'none' (i.e. having no spicules) and 'multiple' (i.e having a mixed array of spicules), bringing the total number of character-states of this character to nine. The tube feet spicules of E. sp. A, ZE, and EE were composed entirely of the bihamate type; all other species contained varying amounts of the triradiate spicule type. E. mathaei presented the greatest variety of tube feet spicules, with five different spicule categories, followed by E. sp. C with four different tube-feet spicule arrangements. In E. oblonga half of the population exhibited triradiate spicules while the other half had none. Gonad spicules were the most variable and diverse of all characters studied. This diversity is evident in the nine different character-states described from the studied populations. E. sp. A presented the most homogenous spicule array, with all spicules belonging to the needle type. In contrast, E. sp. C was the most diverse of all species studied, displaying seven different character states (Table 2). E. mathaei and E. oblonga had an intermediate level of diversity in this character, presenting four character-states each. The gonad spicules of ZE and EE were similar in composition and bore the highest

Number of pore pairs
Pore-pairs ratios differed significantly between some species while in other species these ratios were statistically insignificant ( Figure 3). Pore-pairs distribution in EE differed significantly from ZE (K-S test, D = 0.2534, P* < 0.01), Okinawan E. oblonga (K-S test, D = 0.2507, P* < 0.01), and Mauritian E. oblonga (K-S test, D = 0.2562, P* < 0.01) ( Table 3). No significant differences were observed between EE and the other Echinometra studied. ZE was significantly different from E. mathaei of both Okinawan (K-S test, D = 0.2096, P* < 0.05) and Mauritian (K-S test, D = 0.231, P* < 0.01) origin. Nonetheless, apart from EE and ZE, statistically significant, Bonferroni corrected differences in pore-pairs distribution were only detected between E. mathaei and E. oblonga and not among any of the other Echinometra species (Table 3). * Bonferroni corrected P values.

Sperm head morphology
The morphology of sperm heads varies greatly among species of Echinometra. In general, two types of sperm head morphologies are noted: compact, with a length over width ratio of about three to one; and elongated, with a length over width ratio of about six to one. At the time of observation, ZE sperm samples were found to be severely degraded, possibly due to inadequate preservation or exposure to excessive heat, consequently, data from these samples could not be trusted and was removed from further analyses. EE displayed the compact sperm morphology with an average length of 3.90 ± 0.34 µm and an average width of 1.27 ± 0.08 µm. The shape (i.e. proportion of length to width) averaged at 3.08 ± 0.29 µm. In that sense EE resembles E. sp. C and E. mathaei, yet significantly differs (student t-test, p < 0.05) from E. sp. A and E. oblonga (Figure 4,

Phylogenetic relationship
Both Maximum Likelihood (ML) and Bayesian Inference (BI) analyses yielded similar tree topologies and reflected no conflicts with respect to any of the major clades. Therefore, we only present the BI tree ( Figure 5). Ingroup taxa were monophyletic (BSP = 100; PP = 1) and clustered into nine main clades. These nine clades correspond to the eight known species of Echinometra and a ninth clade that comprises both EE and ZE. These clades clustered partially based on geographic region: clade A comprises all the species from the Indo West Pacific (IWP) and western Indian Ocean (WIO), including the Red Sea and a species from the eastern Pacific (BSP = 100; PP = 1); and clade B comprises western Atlantic and the other eastern Pacific species (BSP = 100; PP = 1). Among the six taxa of clade A, E. sp. A, E. mathaei and E. oblonga together form a group of entirely IWP species (BSP = 100; PP = 1). The other taxa in clade A are comprised of taxa from various geographic regions, including the IWP (E. sp. C), eastern Pacific (E. insularis), Red Sea (EE) and WIO (ZE) (BSP = 100; PP = 1). The clade containing EE and ZE is sister to the clade containing E. sp. C and E. insularis, albeit weakly supported (BSP < 50; PP = 0.79). Intra-and inter-specific Kimura-two-parameter (K2P) genetic divergence among taxa is summarized in Table 5. Intraspecific divergence in EE+ZE (0.70%) was similar to the values observed in E. mathaei and E. oblonga (0.68% and 0.63%, respectively). These levels of divergence were higher than the intermediate levels present in E. sp. C (0.53%) and more than three-fold that of E. sp. A (0.22%). Interspecific divergence values between EE+ZE and the IWP species showed the greatest distance from E. sp. C (4.39%), followed by E. oblonga (3.58%). The distance between E. mathaei to E. sp. A (2.90%) was smaller than the distances between EE+ZE and the latter two species (3.03% and 3.19%, respectively). EE+ZE distance from E. oblonga (3.58%) was similar to E. oblonga's divergence from E. sp. A (3.59%).

Discussion
Identification based on traditional taxonomy IWP species. The type locality of Echinometra mathaei is Mauritius [32]. Consequently, Echinometra from adjacent localities such as Zanzibar and even the Red Sea have long been regarded as such [2,8,18,33], despite a lack of sufficient taxonomical background. Color is a prominent external taxonomic feature in many organisms, including sea urchins from the genus Echinometra [8]. However, it is clear that spine color alone is insufficient to determine species affiliation in most Okinawan Echinometra, since all but one species share spine colors. The exception is E. sp. A, in which a single color morph was observed. The high plasticity of spine color and its subjective terminology thus limit the effectiveness of this character as a delineating feature in Echinometra. Another character prone to ambiguity is the skin around the peristome. This character is as variable in color as the spines themselves. However, unlike the spines in which seven distinct characterstates are used to define their variety, in the skin around the peristome, only two are used. When applied to a plastic feature such as the color of this skin, this may lead to conflicting decisions. Milled rings are also only partially informative. In E. sp. A and E. oblonga all individuals presented a single character-state (Table 2). While findings from these two species correspond well with the literature [3], in E. sp. C and Table 3. Pair wise comparisons of the number of pore-pairs between species of Echinometra.  E. mathaei great differences were found. Arakaki et al. [3] describes E. sp. C and E. mathaei as having completely dark and completely bright milled rings, respectively, making this character useful in delineating Okinawan species. However, in this study we found that both species feature bright and dark rings simultaneously ( Table 2). The reason for these differences may be attributed to the low number of samples obtained by Arakaki et al. (a total of n = 21 for both species), in comparison to the 72 samples collected in the current study. In contrast to coloration, spicules of the tube feet and gonads were clearly distinguishable and unambiguous. Here too sample size was relevant: in the current study, the 45 E. mathaei and 27 E. sp. C sampled yielded five and four tube feet spicules character-states, respectively; whereas Arakaki et al. [3] found only two tube feet spicule character-states in E. mathaei and three in E. sp. C, for their substantially lower sample sizes of 11 and 10, respectively. Clearly, a larger number of samples are more likely to represent the proportion of natural states, and offer a better prospective for identifying subtle morphological differences in closely-related species, such as IWP Echinometra. Nonetheless, even with thorough sampling Echinometra may present a high plasticity of morphological characters. In contrast to previous studies [34], we found lower proportions of bihamate spicules and higher proportions of triradiate spicules (29% and 38%, vs. 69%-85% and 15%-31%, respectively) in the tube feet of E. mathaei. Moreover, while Rahman and Uehara [34] report no bihamate gonad spicules in E. oblonga from Okinawa, we found these spicules in 20% of the population from the same location. Hence, large intraspecific variations and inconsistencies in morphology and even reproduction [34], are present in IWP Echinometra. WIO and Red Sea species. ZE and EE appear to be the same species based on their external morphological features, as they share some unique morphological features and express a high resemblance in others. For example, a unique character-state only found in populations from these two locations is the 'faded' milled rings, which are not apparent in any of the Okinawan urchins. In other characters too, ZE and EE show highly similar proportions of character-states. One example of such similarity is seen in the character 'skin of peristome', in which both populations present, in highly similar proportions, predominantly 'dark'-skinned individuals with only a few 'bright'-skinned ones ( Table 2).

EE ZE A [Ok] B [Ok] C [Ok] D [Ok] B [Ma] D [Ma
The spine coloration of urchins from these two localities may provide some explanation as to why these populations were mistaken for E. mathaei. Although ZE are generally darker than EE (Table 2), urchins from both localities also share the light brown color morph, similar to E. mathaei. Moreover, ZE and EE presented the 'bihamate' spicule type in the tube feet in all of the samples studied. E. mathaei, though also presenting this spicule type, did so in less than 30% of its population. In the gonads of both EE and ZE, 'needle' type spicules were always present, either solely (70% and 60.8% of samples, respectively) or in combinations with other spicule types. These proportions were twice as high as in E. mathaei (only 35.5%). Furthermore, in almost 60% of E. matheai's population the needle spicule was entirely absent. Finally, in the gonads of both EE and ZE, the 'figure-eight' shaped spicules were eminent in 20% and 27.45% of the samples, respectively, in contrast to the other Echinometra species were these spicules where nearly absent. In fact, the only time these spicules were observed in a species other than EE or ZE, was in E. sp. C, and only in the 'multiple' spicule combination that occurred in no more than 3.7% of the samples taken. The differences in spicule types and arrangements between ZE and EE, and the rest of the IWP Echinometra, together with the stability of spicules as a morphological marker, suggest that urchins from the two former populations may differ from the IWP species. Though originally oversimplified: "Four pore pairs to the arc is the normal number in Echinometra mathaei, whereas the other species of Echinometra have a greater number, 5-8" [8], this character was considered stable even in more recent detailed Echinometra studies [3]. However, the number of pore-pairs was limited in its ability to differentiate species, as some species shared similar pore pair distributions (Table 3). In fact, Sperm size in sea urchins can be polymorphic among populations within species [3]; however, a large variation in length (i.e. more than two-fold) usually represents differences in species [4,[35][36][37]. The primary difference lies in the tip of the sperm head and the shape of the nucleus [3]. In the current study, the sperm morphology of EE resembled that of two other Echinometra species described in the literature, E. mathaei and E. sp. C. That no significant difference in sperm morphology was found between these species (Table 4) implies a certain degree of relatedness between them. Although E. oblonga type locality was never described [32] its presence in the WIO was reported several times in the scientific literature, including some populations in Mauritius [3,38]. However, considerable differences in sperm morphology ( Figure 4, Table 4), tube feet and gonad spicules, reject the assumption that EE and ZE could in fact be E. oblonga.
Since spicule assembly, the number of pore-pairs, and the morphology of the sperm in the Okinawan Echinometra species complex reflect species level differences [3,6], it is possible that Echinometra from Eilat and Zanzibar should be regarded as a new species.

Molecular taxonomy and phylogenetic relationships
Both ML and BI methods produced tree topologies that reaffirm the phylogenetic relationships among the eight known Echinometra species, with the results from the current study reflecting Echinometra phylogenies published elsewhere [13,39]. To truly represent Echinometra diversity at the sites we studied, and in order to avoid bias due to small sample size [28,29], we aimed at acquiring a large data set from each of the study sites. Thus, our data facilitates the detection of intraspecific variation of each of the species studied. In E. mathaei, for example, three well-defined sub-clades could be distinguished. These sub-clades (referred to as E, F and G; Figure 5), suggest the recent divergence of IWP E. mathaei into two or perhaps even three mitochondrial lineages and possibly new species. However, genetic divergence of these sub-clades was low (0.2%-1.4%), well within the intraspecific divergence boundaries of Echinometra. Nonetheless, as the current analysis is based on a single mitochondrial gene, incorporation of nuclear genes are thus needed to check if interbreeding still exists or not between these mitochondrial lineages.
The deepest split of IWP Echinometra occurred between E. sp. C and its sister species, the endemic Easter Island E. insularis, and the cluster of E. sp. A, E. mathaei and E. oblonga (cluster C). These findings are supported by previous phylogenetic studies of Echinometra [4,13,39]. E. mathaei is similarly distant from E. sp. A as it is from E. oblonga (2.90%), while the latter two are further diverged from one another (3.59%). These genetic distances are identical to those suggested by Palumbi [40], and make E. mathaei and E. sp. A the most closely-related pair of Echinometra species. Moreover, our findings support the claim of sister-species relationship between E. mathaei and E. sp. A, as suggested by Landry et al. 2003. The poor bootstrap support for the branching of these three species (i.e. E. mathaei, E. sp. A, and E. oblonga) thus implies that their genetic structure is not completely resolved and requires further clarification, in particular by adding data from nuclear genes.
The phylogeny of Echinometra from Eilat and Zanzibar is presented here for the first time ( Figure 5). EE and ZE form a monophyletic clade (clade H) that branched off from the cluster of E. sp. C and E. insularis (clade D). Degrees of divergence of clade H from the rest of Echinometra (3.03%-4.39%) are well within the interspecific range for this genus [39], and are higher than the distance between many of the other IWP species. Thus mtDNA data suggest that Echinometra from the Red Sea and WIO differ from the other Echinometra species. That no clear pattern of genetic structure, neither between nor within the latter two populations was found, despite our thorough sampling at these sites, suggests they are comprised of the same species. Whether a single or multiple species exist in the entire WIO and Red Sea regions nonetheless needs further clarification, as our sampling was limited to only two locations (i.e. Eilat in the Red Sea and Zanzibar in the WIO). The numerous possible habitats and the vastness of the Indian Ocean hold great prospects for identifying other Echinometra species in that region. Expanding our search to new sites, utilizing both meticulous morphological analysis combined with molecular evidence, will deepen our understanding of Echinometra diversity and distribution and of marine speciation processes as a whole.

Conclusions
The morphological keys available for identifying species of the sea urchin Echinometra were tested on four closely-related Okinawan Echinometra and were generally in agreement with the molecular taxonomy. However, an in-depth examination of the characters used for identification of these species revealed a much higher intraspecific diversity than previously reported. The COI phylogenetic reconstruction of Okinawan Echinometra strongly reconfirmed their current ascription as four distinct species. While Echinometra from Eilat and Zanzibar had been previously mistaken for E. mathaei, current data suggest that Echinometra from the latter two regions may in fact be a new, undescribed, species. Yet more work is needed to elucidate the long list of synonyms currently subsumed in Echinometra. Thus, an integrated approach that combines both traditional and molecular taxonomic methodologies should be applied in order to satisfy the intricate demands of assigning correct scientific names to taxa.