Circumtropical distribution and cryptic species of the meiofaunal enteropneust Meioglossus (Harrimaniidae, Hemichordata)

Hemichordata has always played a central role in evolutionary studies of Chordata due to their close phylogenetic affinity and shared morphological characteristics. Hemichordates had no meiofaunal representatives until the surprising discovery of a microscopic, paedomorphic enteropneust Meioglossus psammophilus (Harrimaniidae, Hemichordata) from the Caribbean in 2012. No additional species have been described since, questioning the broader distribution and significance of this genus. However, being less than a millimeter long and superficially resembling an early juvenile acorn worm, Meioglossus may easily be overlooked in both macrofauna and meiofauna surveys. We here present the discovery of 11 additional populations of Meioglossus from shallow subtropical and tropical coralline sands of the Caribbean Sea, Red Sea, Indian Ocean, and East China Sea. These geographically separated populations show identical morphology but differ genetically. Our phylogenetic reconstructions include four gene markers and support the monophyly of Meioglossus. Species delineation analyses revealed eight new cryptic species, which we herein describe using DNA taxonomy. This study reveals a broad circumtropical distribution, supporting the validity and ecological importance of this enigmatic meiobenthic genus. The high cryptic diversity and apparent morphological stasis of Meioglossus may exemplify a potentially common evolutionary ‘dead-end’ scenario, where groups with highly miniaturized and simplified body plan lose their ability to diversify morphologically.


Sampling and data gathering
Meioglossus were sampled over 15 years and multiple expeditions from nine localities in the Caribbean and five localities in the Indo-Pacific (Fig. 1, locality details and collection data are specified in Supplementary Table 1).Fine to coarse coralline sand sediment samples were collected by hand, scuba-diving or dredging, from 0.3 to 33 m depth.Sediment samples were anesthetized using a 1:1 isotonic MgCl 2 and seawater solution releasing adhesive animals from the sand grains.Samples were then resuspended and animals decantated through a 63-µm cone-shaped mesh 23 .Animals were revitalized in petri dishes with seawater and Meioglossus sorted out using a dissecting scope.Specimens were fixed in 2% paraformaldehyde (PFA) or glutaraldehyde (GLU) for morphological analyses or stored in 99% ethanol (EtOH) for molecular analyses.
Type material is deposited in the Natural History Museum of Denmark (NHMD) except for one paratype from Jeju Island deposited in the National Institute of Biological Resources (NIBR).

Extraction, PCR and sequencing
DNA was extracted from 38 entire specimens (up to five individuals, when possible, of each population), using the Qiagen Dneasy ® Blood & Tissue Kit (Cat.no.69506) following the manufacturer's instructions.Four common gene markers representing both mitochondrial and nuclear genes, and fast and slow evolving genes, were targeted for resolving the relationships of Meioglossus: mitochondrial 16S ribosomal RNA (16S rRNA, 450 base pairs (bp)), Cytochrome c oxidase subunit I (COI, 650 bp), nuclear 18S ribosomal RNA (18S rRNA, 1800 bp) and Histone 3 (H3, 340 bp).
Polymerase Chain Reactions (PCR) were performed following a previously optimized protocol for interstitial meiofaunal invertebrates 41,42 (check Supplementary Table 2 for list of primers, sequences, and references).Amplified PCR products were visualized on 1% agarose gels stained with GelRed® (Biotium, 41003) (Hames, 1998), and purified using the E.Z.N.A. ® Cycle Pure Kit, following the manufacturer's instructions.They were later shipped to Macrogen Europe (The Netherlands, Amsterdam) for sequencing 43 .

Assembly, alignment, and outgroup selection
Chromatogram visualization and contig assembly were performed using either Sequencher v4.8 (Gene Codes Corporation, Ann Arbor, MI USA) or Geneious prime v2021.2.2 (Dotmatics) 44 .Each consensus sequence was verified for contamination on the NCBI Standard Nucleotide Blast online platform, using the BLAST tool (Basic Local Alignment Search Tool) 45 .All sequences were deposited in GenBank ® .
Sequences were aligned using MAFFT v7.450 55,56 as implemented in Geneious.The E-INS-i algorithm was selected for ribosomal markers (16S and 18S rRNA), and the G-INS-i one for protein coding genes (COI and H3).Default parameters were selected for all alignments (Gap open penalty: 1.53, Offset value: 0.123, Scoring matrix: 200PAM/k = 2).Ribosomal gene datasets were re-aligned with the 'nwildcard' option selected on the MAFFT v7 online platform 57,58 , to ensure that missing data were not designated as gaps.As they show no variation in length, protein-coding gene alignments were trivial.However, to verify the presence of stop codons and indels, they were translated into amino-acid and re-aligned in Geneious.Single-gene datasets were concatenated using the 'Concatenate Sequences or Alignments' tool in Geneious.A total of 137 sequences were produced from the Meioglossus individuals.37 sequences were generated for 16S rRNA, 33 sequences for 18S rRNA anterior fragment and 23 for the posterior fragment, 27 sequences for COI and 17 sequences for H3.The final combined dataset (16S rRNA, 18S rRNA, COI and H3) alignment is 3506 nucleotides in length (16S rRNA = 617 nucleotides, 18S rRNA = 1825 nucleotides, COI = 687 nucleotides and H3 = 377).

Phylogenetic reconstructions
Phylogenetic reconstructions were performed on single and combined gene datasets using both Maximum Likelihood (ML) and Bayesian Inference (BI) methods.
ML analyses were conducted using RAxML v.8.2.11 (Randomized Axelerated Maximum Likelihood) 59 , as implemented in Geneious.RAxML only implements the General Time Reversible model (GTR).Given that and by performing the ModelTest program in Geneious using the Akaike Information Criterion (AIC), a GTR model with corrections for discrete gamma distribution (GTR + G + Ґ) was specified for the 16S rRNA, 18S rRNA and concatenated genes datasets, while a GTR + G model was selected for the protein coding genes.Nodal support estimations were generated using non-parametric bootstrapping with 1000 replicates.
Bayesian analyses were conducted using the MrBayes v3.2.6 plugin in Geneious 60 .Prior to analyses, optimal evolutionary models were inferred using JModelTest 61 under the Schwartz Bayesian Information Criterion (BIC) on the CIPRES Sciences Gateway 62 and were as follows: GTR + G for 16S rRNA; Hasegawa, Kishino and Yano (HKY + G) for 18S rRNA; HKY + G + Ґ for COI; GTR + G + Ґ for H3 and the concatenated dataset.Independent analyses were run twice on single gene and concatenated datasets, for 15 million generations with trees sampled every 1000 generations and using four heated chains.One quarter-corresponding to 3,750,000-of the generations were discarded as burn-in.Haplotype networks (only shown in Supplementary Figs. 12, 13) were generated on PopART v1.7 63 using the Minimum Spanning Network 64 .

Species delineation
Four methods widely used in species delineation were conducted; two based-tree methods; Generalized Mixed Yule Coalescent approach (GMYC) 65 and Poisson Tree Process including a bayesian implementation (bPTP) 66 and two genetic distances methods; Automatic Barcode Gap Discovery (ABGD) 67 and Assemble Species by Automatic Partitioning (ASAP) 68 .Prior to analyses, outgroups were removed from all the datasets.No delineation analyses were conducted using either the 18S rRNA or H3 datasets, due to the small variability between the sequences, or the lack of data.
For GMYC analyses, ultrametric trees were generated using BEAST v2.6.7 69 .Parameters for the BEAST runs were defined with bEAUti v2.6.7,generating xml files.For all analyses, tree parameters were chosen based on a Yule model with a constant clock evolution.Nucleotide substitution models were estimated under AIC using JModelTest as implemented on CIPRES.16S rRNA dataset was run under a Generalized Time Reversible model with a proportion of invariable sites (GTR + I), and HKY + G was selected for the COI dataset.Single and concatenated datasets underwent independent Markov Chain Monte Carlo (MCMC) analyses comprising 10 million generations with tree subsampling occurring every 1,000 generations.Convergence verification of all MCMC runs was done with Tracer v1.2.7 70 .A maximum clade credibility (MCC) consensus tree was obtained for each BEAST dataset with TreeAnnotator v2.6.7 to summarize Bayesian results.GMYC analyses were performed with RStudio 2022.02.3 (R Core Team, 2022) using the SPLITS v1.0-20 package 71 .bPTP analyses were carried out on the bPTP online server (https:// speci es.h-its.org).Default parameters (MCMC generations of 100,000 and subsample of 100) were set, except for the burn-in set at 25%.

Morphological observations
To test for potential morphological diagnostic traits, specimens were photographed and measured using an Olympus DP73 camera mounted on an Olympus IX70 inverted light microscope.Moreover, when material was available, we used immunohistochemical staining and Confocal Laser Scanning Microscopy (CLSM) to visualize  3).These new morphological assessments complement the many previous morphological studies conducted on Meioglossus specimens from Belize and Bermuda 11 .CLSM studies were done on material fixed in 4% paraformaldehyde in phosphate-buffered saline buffer (PBS) with 7% sucrose.Nine specimens of Meioglossus from six localities were examined (Cuba Chivirico and Miramar, Belize station 4, Eilat station 32, Maldives and South Korea station 19).Fixed specimens were rinsed twice in PBS and pre-incubated 2 h in PTA buffer (PBS with 1% Triton X-100, 0.25% bovine serum albumin and 7% sucrose).They were later incubated for 36 h in two primary antibodies, either monoclonal mouse anti-tyr-tubulin (SIGMA, T9028) or monoclonal mouse anti-acetylated α-tubulin (SIGMA, T6793) together with either polyclonal rabbit anti-serotonin (Sigma-Aldrich, S5545) or polyclonal rabbit anti-FMRF, with a final concentration of 1:400 in PTA.Specimens were thereafter rinsed in PBS six times over 2 h and incubated for 48 h in the secondary antibodies anti-rabbit-CY3 (Sigma-Aldrich, T5268) and anti-mouse-CY5 (Jackson ImmunoResearch, 115-175-146) with a final concentration of 1:400 in PTA.After the incubation, specimens were rinsed 3 times in PBS over 2 h and transferred through a graded series of increasing concentration of Vectashield ® mounting medium with DAPI (Vector Laboratories, VTECH-1200), before being whole-mounted on slides.Specimens were scanned using an Olympus FluoView FV-1000 CLSM.Sections, slices, and maximum intensity z-stack projection images were generated using the Imarisx64 v7.6.5 software (Oxford Instrument 2022).
Images and plates were arranged using Photoshop and Illustrator (Adobe Illustrator CS4 v14.0.0 and Adobe Photoshop CS4 v11.0).General morphology of Meioglossus is presented in Fig. 4. Supplementary Figs.5-11 present detailed plates for each of the species, except for Meioglossus from Eilat st.16 where the limited material was not adequately preserved for morphology.
The Bayesian inference (BI) and Maximum Likelihood (ML) reconstructions of the combined gene dataset present a similar topology with four major clades; three containing Caribbean individuals and one clade exclusively containing individuals from the Indo-Pacific area.Subclades within these are consistently found with high support, and each of the 14 sampled populations represent monophyletic clades except for mixing among the geographically close locations in South Korea.The only difference between the BI and ML reconstructions is the exact position of two specimens within their respective population clade, due to the low genetic variation within these populations.
The first of the four major clades is consistently found as sister branch to the remaining Meioglossus, cautioning that sequences were only obtained from one individual from Bermuda.A second larger fully supported clade branches off as sister to the remaining two clades and comprise individuals from Cuba (Chivirico) and Belize (station 4).Of the last two sister clades, one clade consists of three distinct subclades: (1) specimens from Curaçao gathering as a fully supported subclade, (2) specimens from Northwest and Southwest Cuba (Miramar and Punta Perdiz) and Belize (station 9) (subclade PP = 1.00 and BS = 99%), and (3) a fully supported subclade regrouping specimens from Northeast Cuba (Gibara) and Turks and Caicos Islands.The remaining major and exclusively Indo-Pacific clade, positioned among the Caribbean clades, brings together Meioglossus from Israel, South Korea, and the Maldives.Although moderately supported (PP = 0.98 and BS = 73%) this clade is consistently found in all concatenated genes tree analyses.The fully supported subclade from the Maldives is sister group to a fully supported subclade of specimens of Eilat station 32.A subclade of Meioglossus from the other Eilat locality (station 16) surprisingly nests next to a subclade of South Korean specimens from two localities (all clades fully supported) (Fig. 2).
Among all single-gene datasets analyses, the 16S rRNA dataset generates the most resolved and robust tree with similar topology to that of the combined-gene dataset and among the BI and ML reconstructions.The analyses of the COI dataset were also informative at the distant nodes although with less resolution and support of the major clades.Both 18S rRNA and H3 datasets were less complete and showed less variation among and within populations hereof yielding less resolved trees though with well supported major clades (single gene trees are shown in Supplementary Figs.1-4).

Species delineation
All delineation analyses throughout the various datasets consistently yielded at least nine distinct phylogenetic entities (Table 2, Fig. 3), reflecting possible unique species.
Species delineation analyses ABGD and ASAP were only calculated for the single gene datasets of 16S rRNA and COI since the nuclear gene 18S rRNA presented poor variation between sequences, and the H3 dataset had a high level of missing data compared to the other datasets.Partitions with the lowest ASAP-score were chosen (2.5 for 16S and 2 for COI).
Species delineation analyses with bPTP and GMYC were performed using the single gene 16S rRNA and COI datasets as well as the combined four-genes dataset.All bPTP and GMYC analyses were statistically significant (P ≤ 0.05) (Table 2) and always recovered a minimum of nine entities.These entities were highly supported in the BEAST trees (Fig. 3) and congruent with the similarly highly supported clades in the phylogenetic analyses (Fig. 2).

Morphological observations
To the extent preserved material allowed for, the external morphology and internal anatomy of Meioglossus was investigated using light microscopy and a combination of immunostaining and CLSM (Fig. 4, Supplementary Figs.5-11).
No new diagnostic traits could be defined based on LM observations, as specimens of all populations presented identical body-organization and detailed internal and external morphology.Even when comparing the specific location and numbers of serotonin-LIR somata (Table 3) among the five Meioglossus species examined with immunostaining and CLSM, no significant differences could be detected in the neural architecture.Measurements of body length (and relative length of body regions) did vary among the holotypes of the delineated species (Table 3).However, measurements of paratypes documented overlapping ranges between species and high intraspecific variations, due to different degree of anaesthetization and fixation protocols as well as different states of asexual reproduction.Morphometric observations could therefore not be used to establish new diagnostic traits.
Light microscopy observations confirmed the presence of sperm in many individuals, with similar position and overall structure as in M. psammophilus.It is remarkable that among all populations examined throughout this study females were never observed, similar to what was reported by Worsaae et al. (2012) 11 .Moreover, asexual reproduction by paratomy was likewise observed in all newly recorded populations.   .DNA extracted from one individual.
Distribution and habitats.Sand patched in middle reef with sheltered part of finer sand, infralittoral, Northeast Bermuda.
Remarks.The presence of a single individual in the analyses led to debate about whether to assign a name to this species discovered in Bermuda.However, all four markers sequenced for this specimen, consistently showed it to be a unique entity, branching off as sister group to all other Meioglossus spp.Its validity is further supported by the species delineation analyses and from showing high interspecific differences.
Interspecific similarities: similarity matrix of 16S rRNA shows  Etymology.Named after the type locality near the city 'Chivirico' , and the Latin root -ensis ('of ').
Distribution and habitats.Coralline sand patches among coral reef, infralittoral.Southeast Cuba and East Belize.
Remarks.The validity of this species is supported by both phylogenetic and species delineation analyses, and from showing less interspecific than intra-specific similarity in 16S and CO1.

Remarks:
The validity of this species is supported by both phylogenetic and species delineation analyses, and from showing less interspecific than intra-specific similarity in 16S and CO1.
Intraspecific similarities: similarity matrix of 16S rRNA shows 100% similarity among M. eilatensis sp.nov.CO1 only obtained from one individual.
Interspecific similarities: similarity matrix of 16S rRNA shows 87.58% similarity between M. eilatensis sp.nov.specimens and those of its sister group, M. jejuensis sp.nov, and 86.15-87.44%similarity for COI.
Paratype.One specimen as permanent wholemount (NHMD-1731268).Same sampling data as holotype.Etymology.Named after the type locality 'IUI' , the Inter-University Institute for Marine Sciences of Eilat, and from the Latin root -ensis, meaning 'originated in' .
Remarks.The validity of this species is supported by both phylogenetic and species delineation analyses, and from showing less interspecific than intra-specific similarity in 16S and CO1.
Distribution and habitats.Fine to coarse well sorted coral sand, infralittoral, Southern Maldives.
Remarks.The validity of this species is supported by both phylogenetic and species delineation analyses, and from showing less interspecific than intra-specific similarity in 16S and CO1.
Prior to this study only one meiofaunal species of enteropneust was described, the Caribbean Meioglossus psammophilus 11 .We not only document that species of Meioglossus can be found worldwide, but that these larvallooking microscopic enteropneusts constitute a monophyletic genus.Whether it is with nuclear or mitochondrial sequences, single or concatenated datasets, this study corroborates the monophyly of this genus.
None of the newly discovered Meioglossus species have revealed female individuals.Instead, they look morphologically identical to M. psammophilus and likewise contain sperm and reproduce asexually through paratomy 11 .The lack of specimens carrying eggs or embryos in any of the newly discovered populations questions the function of the sperm described in Worsaae et al. (2012) 11 .The seeming absence of females and sexual reproduction put relevance to the previously mentioned hypothesis in Worsaae et al. (2012) 11 , suggesting that the sperm-like structure could instead function as an energy reserve.These structures repeatedly found in most populations should be further investigated with advanced microscopy and staining methods to help address their detailed structure and correct interpretation as sperm.The repetitive finding of paratomy in most of these populations may also indicate that these species mainly (or only) reproduce asexually.If Meioglossus reproduces only by paratomy, this would to our knowledge be a unique example in the animal kingdom, where entirely asexual reproduction in a free-living solitary species otherwise normally involve parthenogenesis (including eggs) 80,81 .All these new findings have unfortunately not solved the enigma of the Meioglossus life cycle.However, our analyses showed a high cryptic species diversity and broad distribution of the genus.Furthermore, the absence of females in any of the newly discovered localities, indirectly support that species of Meioglossus are not just larvae or dwarf males of a macroscopic female but truly represent a broadly distributed genus of permanent meiofauna species.In the more densely sampled Caribbean area we find several populations constituting one species.Examples are M. psammophilus gathering relatively distant populations from Belize (station 9), Northwest and Southwest Cuba, and M. chiviricoensis sp.nov.found in both Belize (station 4) and Southeast Cuba.Meioglossus turkensis sp.nov., comprises specimens from closer localities in Turks and Caicos and Northeast Cuba, respectively.Meioglossus jejuensis sp.nov.groups animals from even closer localities.Single species comprising several populations, and in particular the latter two distribution patterns, can be easily explained by transport of sediment and animals by shared currents across shallow waters not hindered by the lack of a larval or dormant dispersal stages 20,41,[82][83][84] .Such passive dispersal mechanisms are also known to act on microscopic organisms over larger distances, potentially aided by rafting or drifting and by human activities such as ballast and transport sediment, or aquaculture 16,[85][86][87][88] .
On the other hand, several of the Meioglossus populations represent individual species, even when sampled geographically close such as the two species in the Red Sea near Eilat.These overlapping distribution patterns of species upholding genetic disparity but having seemingly similar morphology, may reflect specific physiological adaptations to the abiotic properties of their respective environment (e.g., granulometry, current, temperature, salinity, oxygen, pH) 89,90 .For instance, the environmental conditions of the two Red Sea populations are strikingly different with one locality exhibiting very fine sandy sediment (housing M. eilatensis) and the other coarse coralline sand (housing M. iuiensis), possibly explaining the presence of two different species within close geographical range.The apparent lack of genetic mixing between some of these geographically close populations, e.g., in Eilat and Belize (haplotype networks are shown in Supplementary Figs. 12 and 13), could also reflect distinct reproductive properties, or even be the result of Meioglossus only reproducing asexually 91 .The Everything is Everywhere (EIE) hypothesis 92 asserts an ubiquitous distribution of microscopic organisms 16 .None of the newly described Meioglossus species are globally distributed.These on the contrary, tend to be specific to particular geographical regions and hereby go against the EIE hypothesis.However, the broad and somewhat overlapping distribution of some of the Caribbean species (e.g., including specimens from Belize and Cuba) does indicate that Meioglossus can be dispersed relatively far-hereby partly supporting the EIE statement that 'the environment selects' whether they will successfully colonize these new areas 16,93,94 .
Morphological stasis is noteworthy in Meioglossus, showing genetic diversity without any distinct morphological differentiation.This phenomenon of morphological stability has been noticed in other meiobenthic groups, e.g., copepods, segmented worms, or sea slugs 79,95,96 .It has been suggested to reflect the harsh selection by the physical constraints of the interstitial habitat, such as the restricted space available between sand grains or the habitat instability due to currents or waves actions 21,79,97 .Morphological stasis might also be reinforced by irreversible gene loss 98 .Microscopic animal may not always possess small, nor compacted genomes, but miniaturization events could influence the genome composition.In this case, genes or gene families might be lost, pathways modified, and transposable elements vastly reduced.In the miniaturized, compacted genome of the meiofaunal annelid Dimorphilus gyrociliatus (Schmidt, 1857) 16 , the loss of a few developmental genes could be related to morphological losses of e.g.chaetae (post1 and FGFligand) and reduced mesodermal derivatives such as coeloms (VEGF ligands) 22 .Moreover, despite our limited understanding of the effect of transposable element loss, it may have the potential to influence a species' capacity for morphological diversification.
Findings of a total of 14 populations of Meioglossus from the West Atlantic, the Red Sea, the Indian Ocean, and the East China Sea, documents the broad distribution of Meioglossus and points to a circumtropical distribution of the genus.It is intriguing that Meioglossus exhibits such a wide distribution despite its simplified morphology and apparent absence of larval stages.The broad distribution might be attributed to Meioglossus being an old genus which potentially dispersed with tectonic plate movements and/or movement of shallow sediments with storms.As being the case for some other meiofaunal genera and families (e.g., Refs. 99,100), Meioglossus seems limited to tropical and subtropical regions, since it has never been found in temperate or polar regions.Our phylogenetic investigation demarks an Indo-Pacific clade nested among the Caribbean clades, indicating that Meioglossus may have originated in the Caribbean/West Atlantic Ocean and later dispersed into the Indo-Pacific Oceans.However, it is cautioned that the Indo-Pacific clade is poorly supported, and that many tropical and subtropical areas, especially in the East Pacific, have never been explored for Meioglossus.Thus, potential new findings of Meioglossus may change our current understanding of the intrageneric relationship.

Figure 1 .
Figure 1.Maps marking the 14 Meioglossus sampling localities (red dots).(A) World map highlighting all the localities.(B) Zoom on Belize and Cuba.(C) Zoom on Turks and Caicos Islands.(D) Zoom on Bermuda.(E) Zoom on Curaçao.(F) Zoom on Israel.(G) Zoom on the Maldives.(H) Zoom on South Korea.Map was generated using QGIS 3.24.2software (https:// www.qgis.org).

Figure 2 .
Figure 2. Phylogenetic relationships of Meioglossus using the concatenated gene datasets (16S rRNA, 18S rRNA, COI, H3).Topology based on Maximum Likelihood (ML) analyses of concatenated gene datasets.Nodal support is indicated with both Maximum Likelihood Bootstrapping (BS) and Bayesian Posterior Probabilities of the consensus tree (PP).Only nodal supports above BS > 50% or PP > 0.5 are shown.Those falling below this threshold are represented by a dash (-).Asterisks indicate maximum support in either BS = 100% or PP = 1.Diamond (◆) shapes indicates full support in both analyses.

Figure 3 .
Figure 3. Ultrametric tree generated with BEAST, using the combined four-genes datasets (16S rRNA, 18S rRNA, COI, H3).Results of species delineations are indicated by vertical blue bars under each method used.Species hypotheses supports for both GMYC and bPTP are indicated (GMYC/bPTP).Asterisks indicate maximum support in one of the analyses and Diamond (◆) shapes indicate full support in both analyses.

Table 1 .
Molecular specimens used in this presented study, including GenBank accession number.New sequences for this paper are shown in bold.*Only first ~ 900 bp obtained.**Only last ~ 900 bp obtained.

Table 2 .
Species delineation results of Meioglossus using ABGD and ASAP methods for single datasets of 16S and COI (18S and H3 are omitted due to incomplete coverage and limited variation), bPTP and GMYC methods for both single and concatenated datasets.Abbreviations: C.I., confidence interval; ent, entities; Est, estimated; ML, maximum likelihood; na, not applicable.***P ≤ 0.001.

Table 3 .
Measurements of Meioglossus species and comparison of the number of ser-LIR somata observed in five species of Meioglossus.Measurements of holotypes; measurement ranges of paratypes given in parentheses.
Interspecific similarities: similarity matrix of 16S rRNA shows 95.59-96.67%similarity between M. turkensis sp.nov.specimens and those of its sister group, and 96.37-97.35%similarity for COI.The sister group of M. maldivensis sp.nov.comprises M. curacaoensis sp.nov.and M. psammophilus specimens.