An integrative redescription of the nominal taxon for the Mesobiotus harmsworthi group (Tardigrada: Macrobiotidae) leads to descriptions of two new Mesobiotus species from Arctic

The Mesobiotus harmsworthi group has a global distribution, with localities in polar, temperate and tropical zones. Since the first species of the harmsworthi group was described in the beginning of the 20th century, tens of new species within the group were found and named. However, the diagnosis of the nominal Mesobiotus harmsworthi is insufficient and enigmatic, thus it can be is a serious obstacle in solving the taxonomy of this group. Here, we integratively redescribe the nominal species for the genus Mesobiotus, i.e., Mesobiotus harmsworthi and clarify taxonomic statuses of the two subspecies: M. harmsworthi harmsworthi and M. harmsworthi obscurus that have been recognised as distinct taxa for more than three decades. Traditionally, egg chorion in M. harmsworthi was considered almost smooth and without any traces of areolation, however here we report many misunderstandings that accumulated across decades and we show that, in fact, the chorion in this species exhibits a partially developed areolation. We present an integrative (morphological, morphometric and molecular) diagnosis of the nominal taxon and we confirm that it differs from other species of the harmsworthi group by morphological characters of both animals and eggs. Additionally, we describe two new species of the genus Mesobiotus: M. skorackii sp. nov. from the Kyrgyz Republic (using classical morphological description) and M. occultatus sp. nov. from Svalbard Archipelago (by means of integrative taxonomy). Finally, we also provide the first genetic phylogeny of the genus Mesobiotus based on COI sequences which, together with molecular species delimitation, independently confirms the validity of the analysed taxa.


Introduction
The phylum Tardigrada comprises over 1,200 species [1,2,3] that inhabit both aquatic (freshwater, brackish, marine) and terrestrial environments throughout the world, from the deepest seas to the highest mountain peaks [4,5,6,7]. Tardigrades have been investigated for over two hundred years, but because of old descriptions and insufficient morphometric data, many species currently need revision and redescription, especially those representing nominal taxa for cosmopolitan groups or genera. One of such taxa is Mesobiotus Vecchi, Cesari, Bertolani, Jönsson, Rebecchi & Guidetti, 2016 [8], a cosmopolitan genus comprising ca. 40 known species that inhabit plants, lichens and soil and are considered carnivorous and/or omnivorous [8].
The genus Mesobiotus is divided into two species groups: the harmsworthi and the furciger group, classified within the genus Macrobiotus C.A.S. Schultze, 1834 [9] prior to the erection of Mesobiotus. Species of the harmsworthi group are characterised by three clearly separated macroplacoids in the shape of short, rounded rods and a distinct microplacoid situated very close to them, as well as by conical or hemispherical egg processes [10]. Mesobiotus harmsworthi (Murray, 1907) [11], the nominal species for the harmsworthi group, was described by Murray [11] as Macrobiotus harmsworthi. Since the original description, the problem with the exact characteristics of the species was addressed several times by different authors (e.g. [4,12,13]). Moreover, descriptions of two M. harmsworthi subspecies, M. harmsworthi coronatus (de Barros, 1942) [14] and M. harmsworthi obscurus (Dastych, 1985) [15], did complicate the taxonomy of this group even further. The first subspecies was later elevated to the species level by Pilato et al. [13], whereas the second is considered valid as a subspecies. Although problems with the taxonomy of the nominal species were broadly discussed in literature, a formal redescription of M. h. harmsworthi, based on type material or material from the type locality, has not been published. The only attempt to characterise M. h. harmsworthi in more detail was made by Pilato et al. [13]. However, the work was based solely on light microscope observations, thus some fine morphological details, that are below light microscope resolution, could remain unknown. Also, molecular markers, that are necessary for the accurate characterisation of the species and for the detection of potential cryptic species, are not available. Furthermore, in addition to specimens from loci typici (Spitsbergen and Shetlands), Pilato et al. [13] pooled and analysed together individuals and eggs from Italy and France. Thus, without DNA sequences, it is not possible to verify whether all specimens constituted a single or multiple cryptic or pseudocryptic species. Therefore, in order to clarify the taxonomy of the harmsworthi group, we integratively redescribe M. h. harmsworthi based on typical material of M. h. obscurus (which is now, in fact, a synonym of M. h. harmsworthi s.s.; for more details see Results and Discussion sections below) and we additionally describe two new species of the harmsworthi group: M. occultatus sp. nov. from Spitsbergen and M. skorackii sp. nov. from the Kyrgyz Republic. Our study involved an integrative taxonomy approach by combing morphological and morphometric data from phase contrast light microscopy (PCM) and molecular data in form of DNA sequences of four molecular markers (nuclear: 18S rRNA, 28S rRNA, ITS-2 and mitochondrial: COI). Finally, we provide the first COI phylogeny of the genus Mesobiotus.

Morphometrics and morphological nomenclature
All measurements, made with the QuickPhoto Camera 2.3 software, are given in micrometres [μm]. Structures were measured only if their orientation was suitable. Body length was measured from the anterior extremity to the end of the body, excluding the hind legs. The types of bucco-pharyngeal apparatuses and claws were classified according to Pilato & Binda [19] and Vecchi et al. [8]. The terminology used to describe oral cavity armature, and used in differential diagnoses, follows Michalczyk & Kaczmarek [20]. Buccal tube length and the level of the stylet support insertion point were measured according to Pilato [21]. Other buccal apparatus traits and claws were measured according to Kaczmarek & Michalczyk [22]. Macroplacoid length sequence is given according to Kaczmarek et al. [23]. The pt ratio is the ratio of the length of a given structure to the length of the buccal tube expressed as a percentage [21]. Distance between egg processes was measured as the shortest line connecting base edges of the Table 1. The list of Mesobiotus specimens from Svalbard archipelago genotyped for the phylogenetic analysis with their collection details (numbers in brackets indicate number of sequences and number of studied specimens).

Genotyping
For DNA isolation and sequencing, three individuals of M. harmsworthi and five individuals of M. occultatus sp. nov. were used ( Table 1). The list of individuals with their collection details are provided in Table 2. The genomic DNA was extracted from single individuals according to the protocol described in Dabert et al. [40], using Tissue Kit (Qiagen GmbH, Hilden, Germany). In order to obtain tardigrade exoskeletons, a technique adopted by Zawierucha et al. [41] was used. Specifically, individuals were digested for 48 hours (56˚C) in mixture of ATL buffer, proteinase K in Eppendorf vials and then centrifuged at 6000 rpm. From each vial 90 μm of DNA extract was taken for further analyses by carefully removing this volume using a half-automatic pipette. The remaining 10 μm of the DNA extract with the tardigrade exoskeleton was preserved in 96% ethanol. Then, exoskeletons were mounted in Hoyer's medium for PCM analyses. Further molecular analyses involved the amplification and sequencing of four DNA markers, three nuclear (18S rRNA, 28S rRNA, ITS-2) and one mitochondrial (COI). The markers differ in effective mutation rates: the first two are considered conservative whereas the last two are more variable. The 18S rRNA together with 28S rRNA are often used for resolving relationships at higher taxonomic levels such as families and genera (e.g. [25]) whereas COI and ITS-2 are appropriate for examining intraspecific and intrageneric genetic variation (e.g. [42,43,44,45,46,47]).
Amplification of DNA fragments (PCR) for 18S rRNA and 28S rRNA was conducted in the total volume of 5.5 μl including: 3 μl Type-it Microsatellite PCR Kit (Qiagen), 0.5 μl of each primer, 0.5 μl Q-Solution (Qiagen) and 1 μl of the DNA template. For ITS-2 and COI a total volume of PCR mix was 5 μl including: 3 μl Type-it Microsatellite PCR Kit (Qiagen), 0.5 μl of each primer and 1 μl of the DNA template. Each PCR reactions proceeded in the following steps: One cycle of 5 min. at 95˚C, followed by 40 steps of 30 s at 95˚C, 90 s at 50˚C, 1 min at 72˚C, and with a final elongation step of 5 min at 72˚C. After PCRs each products were diluted by 5 μl MQ water. Separation of PCR products were carried out by 1% agarose gel electrophoresis. Samples containing visible bands were purified with exonuclease I and Fast alkaline phosphatase (Fermentas). The fragments were sequenced using the BigDye Terminator v3.1 kit and the ABI Prism 3130xl Genetic Analyzer (Applied Biosystems), following manufacturer instructions.
All four mentioned above molecular markers were sequenced for two additional Mesobiotus species from Norway and Russia following the protocol by Stec et al. [17]. Only for 18S rRNA and COI other primers were used: 18S_Tar_Ff1 with 18S_Tar_Rr1 and LCO1490 with HCO2189, respectively (see Table 2 for details).

Comparative molecular analysis
For molecular comparisons, all published sequences of the four abovementioned markers for the genus Mesobiotus were downloaded from GenBank (listed in Table 3). The sequences were aligned using the default settings (for COI) and the Q-INS-I method (for ribosomal markers: 18S rRNA, 28S rRNA and ITS-2) of MAFFT version 7 [48,49] and manually checked against nonconservative alignments in BioEdit. Then, the aligned sequences were trimmed to: 721 (18S rRNA), 751 (28S rRNA), 497 (ITS-2), 565 (COI) bp. All COI sequences were translated into protein sequences in MEGA7 version 7.0 [50] to check against pseudogenes. Genetic distances were calculated using MEGA7 as suggested by Srivathsan & Meier [51], i.e., as uncorrected pairwise distances instead of K2P distances. The matrices with calculated uncorrected p-genetic distances for each of the analysed DNA fragment are provided as supplementary materials (S1 Table).

Phylogenetic and species delimitation analysis
To establish phyletic relationships of M. harmsworthi and M. occultatus sp. nov. and to molecularly delineate the species, we constructed a phylogenetic tree based on all COI sequences for the genus Mesobiotus available from GenBank (Table 3) together with sequences obtained in the present study. The COI sequences were aligned using the default settings of MAFFT version 7 [48,49] COI is a protein coding gene, before partitioning, we divided our alignment into three data blocks constituting separated three codon positions. Using PartitionFinder version 2.1.1 [54] under the Bayesian Information Criterion (BIC), the best scheme of partitioning and substitution models were chosen for posterior phylogenetic analysis. We ran the analysis to test all possible models implemented in the program. As best-fit partitioning scheme, PartitionFinder suggested to retain three predefined partitions separately. The best fit-models for these partitions were: SYM+I+G for the first codon position, GTR+I for the second codon position and TVM+I+G for the third codon position. Bayesian phylogenetic tree obtained from this data set was highly polytomous, thus we used only for molecular species delimitation analysis with the PTP method, which uses a non-ultrametric phylogenetic tree as the input data, based on which, the switch from speciation to coalescent processes is modelled and then used to delineate species [55]. For the purpose of the PTP analysis, we discarded the outgroup to protect against eventual biases caused by the distant relationship between the outgroup and the ingroup taxa. The calculations were conducted on the bPTP webserver (http://species.h-its.org/ptp), with 100,000 MCMC generations, thinning the set to 100, burning at 10% and performing search for Maximum Likelihood and Bayesian solutions. Then, in order to reduce polytomy and search for more resolved relationships, we concatenated the COI data set with three nuclear markers, 18S rRNA, 28S rRNA and ITS-2, using SequenceMatrix [56]. Since the molecular data on the genus Mesobiotus are limited not all taxa have been represented in all added data sets (see Table 3 for details). Before concatenation, we aligned sequences of each nuclear marker using the Q-INS-I method of MAFFT version 7, which considers the secondary structure of ribosomal genes [48,49]. Next, we manually checked against non-conservative alignments in BioEdit. The aligned sequences were trimmed to: 721 (18S rRNA), 751 (28S rRNA) and 497 (ITS-2). Before partitioning, we divided our alignment into 6 data blocks constituting three separate blocks of ribosomal markers and three separate blocks of three codon positions in COI data set. Using PartitionFinder under the Bayesian Information Criterion (BIC), the best scheme of partitioning and substitution models were chosen for posterior phylogenetic analysis. We ran the analysis to test all possible models implemented in the program. As best-fit partitioning scheme, PartitionFinder suggested to retain six predefined partitions separately. The best fit-models for these partitions were: K80+G for 18S rRNA, SYM+G for 28S rRNA, K80+G for ITS-2, GTR+I +G for the first and the third codon position, and HKY+G for the second codon position. Bayesian inference (BI) marginal posterior probabilities were calculated for both the COI and the concatenated (COI+18S rRNA+28S rRNA+ITS-2) data set using MrBayes v3.2 [57]. Random starting trees were used and the analysis was run for ten million generations, sampling the Markov chain every 1000 generations. An average standard deviation of split frequencies of <0.01 was used as a guide to ensure the two independent analyses had converged. The program Tracer v1.6 [58] was then used to ensure Markov chains had reached stationarity, and to determine the correct 'burn-in' for the analysis which was the first 10% of generations. The ESS values were greater than 200 and the consensus tree was obtained after summarising the resulting topologies and discarding the 'burn-in'. The consensus tree was viewed and visualised by FigTree v.1.4.3 available from http://tree.bio.ed.ac.uk/software/figtree.

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub:86D87D80-5398-4D89-9767-C8E2A160ADFE. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Animals (morphometrics in
Bucco-pharyngeal apparatus of the Macrobiotus type ( Fig 3A and 3B), with the ventral lamina and ten peribuccal lamellae. Mouth antero-ventral. The oral cavity armature well developed and composed of three bands of teeth (Fig 3C-3F). The first band of teeth is composed of numerous small granules arranged in a several rows situated anteriorly in the oral cavity, just behind the bases of the peribuccal lamellae (Fig 3C and 3F; arrowhead). The band is hardly detectible under PCM in small specimens and clearly visible in large individuals. The second band of teeth is situated between the ring fold and the third band of teeth and comprises ridges parallel to the main axis of the buccal tube and additional teeth between and below them, larger than those in the first band ( Fig 3C, 3E and 3F; arrow). The teeth of the third band are located within the posterior portion of the oral cavity, between the second band of teeth and the buccal tube opening (Fig 3C-3F; indented arrowhead). The third band of teeth is divided into the dorsal and the ventral portion. Under PCM, both dorsal and ventral teeth are visible as two lateral and one median transverse ridges (Fig 3C-3F; indented arrowhead). Pharyngeal bulb spherical, with triangular apophyses, three rod-shaped macroplacoids and a triangular microplacoid. Macroplacoid length sequence 2<3 1. The first macroplacoid narrower anteriorly, the second without constrictions and the third with a small, subterminal constriction (Fig 3G and 3H; empty arrowhead).
Claws of the Mesobiotus type, robust (Fig 4A-4D). Primary branches with distinct accessory points. Accessory point on claws IV are larger and more protruding than in most macrobiotids (Fig 4B and 4D; arrowheads). Lunules under claws I-III smooth and slightly dentated under claws IV (Fig 4B and 4D; empty arrowhead). Thin cuticular bars under claws I-III present ( Fig  4A, arrow). Other cuticular structures on legs absent.
Eggs (morphometrics in Table 5). Laid freely, white, spherical and ornamented, with processes and delicate areolation (Figs 5-7). Egg processes in the shape of wide cones (Fig 6A-6H). The cones can be slightly concave (Figs 5B, 5C, 6B and 6C) or sigmoidal, i.e., with a slightly swollen base and a narrowed apex ( Fig 6D). The processes with a single sharp (Fig 6A  and 6E) or slightly blunt (Fig 6B and 6F) apex, only occasionally bifurcated ( Fig 6D, 6G and 6H). In PCM, processes reticulated with mesh size 0.5-2.0 μm in diameter, evidently larger near the process base and apex (Fig 6A-6D). Sometimes, instead of several large meshes, a single very large bubble is present in the apex (Fig 6A-6D, arrows). In SEM, processes smooth, but with well visible small pores at the bases and inside the areoles close to the processes (Figs 6E-6H, 7C and 7D, arrows). Each process surrounded by five or six areolae delimited by thin brims (Fig 7A-7D). The brims are very often discontinuous, thus areolae are not always fully formed ( Fig 7A and 7C, arrowheads). Surface inside the areolae with clearly visible wrinkles, both in PCM (Fig 7A and 7B) and in SEM (Fig 7C and 7D). Occasionally, the wrinkles may form a small whirl in the areola centre ( Fig 7C, empty arrowhead).
DNA sequences. We obtained sequences for all four analysed genetic markers from the three sequenced syngenophores collected from Spitsbergen and an additional COI sequence from a syngenophore collected from Phippsøya (see Table 1 for details). All nuclear markers were represented by a single haplotypes whereas COI exhibited two distinct haplotypes (separated by the p-distance of 0.7%): The 18S rRNA sequence (GenBank: MH197146), 829 bp long: The 28S rRNA sequence (GenBank: MH197264), 751 bp long: The ITS-2 sequence (GenBank: MH197154), 415 bp long: The COI haplotype 1 sequence (GenBank: MH195150), 621 bp long: The COI haplotype 2 sequence (GenBank: MH195151), 621 bp long: Etymology. Although Murray [11] did not explain the choice of the species name, it seems reasonable to assume that M. harmsworthi was named after Cape Mary Harmsworth in  [11,15,64].
Remarks. In our opinion the presence of this species in Perm Krai needs to be confirmed, especially in the light of the description of a new species Mesobiotus skorackii sp. nov. from the Kyrgyz Republic. Specimens designated as paratypes by Dastych [15] were collected in few localities. According to International Commission on Zoological Nomenclature type specimens should be collected from the same locality. However, since all specimens appear to represent a single species, we decided not to change original designations of type specimens.  (Pilato & Claxton, 1988 [27]), M. nuragicus (Pilato & Sperlinga, 1975 [26]), M. ovostriatus (Pilato & Patanè, 1998 [28]), M. peterseni (Maucci, 1991 [29]), M. pseudoliviae (Pilato & Binda, 1996 [31]) and M. skorackii sp. nov., but differs from all of them by large and protruding accessory points on claws IV. Additionally, M. harmsworthi differs specifically from: 1. M. barbarae (known only from type locality in the Dominican Republic [32]) by: the presence of additional teeth in the second band of teeth, an undivided ventro-median tooth in     Table 6). Body white in living specimens and transparent after fixation (Fig 8A). Eyes present. Cuticle smooth, i.e., without gibbosities, papillae, spines, sculpture or pores. However, under SEM microgranulation is visible on the entire dorso-lateral cuticle (Fig 8B). Granulation present on the external surface of all legs (Fig 8C-8E).
Bucco-pharyngeal apparatus of the Macrobiotus type (Fig 9A), with the ventral lamina and ten peribuccal lamellae. Mouth antero-ventral. The oral cavity armature well developed and composed of three bands of teeth (Fig 9B and 9C). The first band of teeth is composed of numerous small granules arranged in a several rows situated anteriorly in the oral cavity, just behind the bases of the peribuccal lamellae (Fig 9B and 9C, arrowhead). The band is hardly detectible under PCM in small specimens but clearly visible in large individuals. The second band of teeth is situated between the ring fold and the third band of teeth and comprises of ridges parallel to the main axis of the buccal tube (Fig 9B, arrow). The teeth of the third band are located within the posterior portion of the oral cavity, between the second band of teeth and the buccal tube opening (Fig 9B and 9C, indented arrowhead). The third band of teeth is divided into the dorsal and the ventral portion. Under PCM, both dorsal and ventral teeth are visible as two lateral and one median transverse ridges (Fig 9B, indented arrowhead). Pharyngeal bulb spherical, with triangular apophyses, three rod-shaped macroplacoids and a triangular microplacoid. Macroplacoid length sequence 2<3 1. The first macroplacoid narrower anteriorly, the second without constrictions and the third with a small, subterminal constriction (Fig 9D, empty arrowhead).
Claws of the Mesobiotus type (Fig 10A-10C). Primary branches with distinct accessory points. Lunules under all claws smooth (Fig 10A-10C). Thin cuticular bars under claws I-III present (Fig 10A, arrow). Other cuticular structures on legs absent.  Table 7). Laid freely, white, spherical and ornamented (Fig 11A  and 11B). Egg processes in the shape of wide cones. The cones are sometimes bifurcated on the top and often with one to few short apical filaments (Fig 12A-12H). In PCM, processes reticulated with mesh size 0.5-1.6 μm in diameter, slightly larger near the process base ( Fig  12A-12D). In some processes larger meshes are present in the apical part of the process (up to 3.5 μm in diameter; Fig 12A, arrow). At the base of each process a crown of small but well visible thickenings visible both in PCM and SEM (Fig 11C, arrows). In SEM, processes smooth, but with well visible small pores at the processes bases (Fig 12E and 12F and 12H, arrows). Very fine granulation present at the apical part of each process (Fig 12E-12H, empty arrowhead). Egg areolation absent (Fig 11A-11D). In PCM, the surface between processes with irregular and rather poorly visible dots (Fig 11C), in SEM visible as large ridges (Fig 11D).

Eggs (morphometrics in
DNA sequences: We obtained sequences for three of the above mentioned molecular markers from five of six analysed paragenophores (see Table 1 for details). All markers were represented by single haplotypes: The 18S rRNA sequence (GenBank: MH197147), 811 bp long: The ITS-2 rRNA sequence (GenBank: MH197155), 419 bp long: The COI sequence (GenBank: MH195152), 575 bp long: Etymology. The name "occultatus", from Latin = hidden, is given to the new species, because the species remained unrecognised until the nominal M. harmsworthi was accurately characterised.

. Measurements and pt values of selected morphological structures of Mesobiotus occultatus sp. nov. mounted in Hoyer's medium (N-number of specimens/structures measured, RANGE refers to the smallest and the largest structure among all measured specimens; SD-standard deviation
Bucco-pharyngeal apparatus of the Macrobiotus type (Fig 14A), with the ventral lamina and ten peribuccal lamellae. Mouth antero-ventral. The oral cavity armature well developed and composed of three bands of teeth (Fig 14B-14D). The first band of teeth is composed of numerous small granules arranged in a several rows situated anteriorly in the oral cavity, just behind the bases of the peribuccal lamellae (Fig 14C, arrowhead). The band is hardly detectible under PCM in small specimens and clearly visible in large individuals. The second band of teeth is situated between the ring fold and the third band of teeth and comprises of ridges parallel to the main axis of the buccal tube larger than those in the first band (Fig 14B, arrow). The  teeth of the third band are located within the posterior portion of the oral cavity, between the second band of teeth and the buccal tube opening (Fig 14B and 14D, indented arrowhead). The third band of teeth is divided into the dorsal and the ventral portion. Under PCM, both dorsal and ventral teeth are visible as two lateral and one median transverse ridges (Fig 14B  and 14D, indented arrowhead). Pharyngeal bulb spherical, with triangular apophyses, three rod-shaped macroplacoids and a triangular microplacoid. Macroplacoid length sequence 2<3<1. The first macroplacoid narrower anteriorly, the second without constrictions and the third with a small, subterminal constriction ( Fig 14E, empty arrowhead).
Claws of the Mesobiotus type (Fig 15A and 15B). Primary branches with distinct accessory points. Lunules under claws I-III smooth and slightly dentated under claws IV (Fig 15B,  empty arrowhead). Thin cuticular bars under claws I-III present (Figs 13C and 15A, arrow). Other cuticular structures on legs absent. Eggs (morphometrics in Table 9). Laid freely, white, spherical and ornamented, with processes and delicate areolation (Fig 16A-16C). Egg processes in the shape of short and wide sharpened cones (Figs 16A, 16B and 17A-17F). In PCM, processes reticulated with mesh size 0.3-1.8 μm in diameter, slightly increased in size from the base to the top (Fig 17B). In SEM processes smooth, but with well visible small pores present mainly at the processes bases ( Fig  17D-17F, arrows). Each process surrounded by six areolae delimited by thin brims (Fig 16B  and 16D-16F). The brims are very often discontinuous, thus areolae are not always fully formed (Fig 16D and 16E). Surface inside the areolae with clearly visible wrinkles, in PCM (Fig 16D and 16E), and small pores and wrinkles in SEM (Fig 16F).
Etymology. We dedicate this species to our friend and a distinguished and prominent Polish acarologist, Professor Maciej Skoracki, a discoverer of many new species of Syringophilidae mites.   Table 3 and the "Phylogenetic analysis" subsection in Material and Methods for details on species and sequences used in the analysis. Scale bar represents substitutions per position.
https://doi.org/10.1371/journal.pone.0204756.g018 Remarks. Eyes were present in in all specimens in freshly made microscope slides (prepared less than 6 months ago) but are absent in all specimens mounted over 8 years ago. This probably means that eyes disappear with time in specimens mounted in Hoyer's medium.
Differential Phylogeny and molecular species delimitation. The PTP analysis of the COI phylogenetic tree revealed the presence of 14 highly supported species, represented as terminal nodes of ingroup taxa in Fig 18. Although the presented phylogeny was based on four molecular markers, the branch support was generally weak which resulted in some polytomies. Thus, currently, only some preliminary conclusions on the relationships within the genus Mesobiotus can be made. The taxa of both species groups recognised within this genus, the harmsworthi and the furciger group, do not cluster in two evolutionary linages but they are intermixed. Mesobiotus harmsworthi redescribed herein is a distinct taxon and, together M. occultatus sp. nov. and an undetermined M. harmsworthi group species from Russia, the three species form a polytomous clade. The phylogenetic and species delimitation analysis also revealed that an unpublished COI sequence (GU113140) of a specimen from China designated in GenBank as "M. harmsworthi", clearly represents a distinct unknown species. Finally, the analysis also showed that three COI sequences of specimens designated in GenBank as "M. furciger" from three Antarctic islands (JX865306, JX865308 and JX865314), represent three rather than a single species.

Type and neotype locality of M. harmsworthi
The type locality of M. harmsworthi is problematic because Murray [11] did not designate a single locale for this species, which was a usual practice in his times, and he stated that he found M. harmsworthi in Franz Joseph Land, Spitsbergen and Shetland. The first two localities are in the High Arctic and the distance between Cape Mary Harmsworth in Franz Joseph Land and Phippsøya in the Svalbard Archipelago (the most northern locality with a molecularly verified population of M. harmsworthi) is only ca. 435 km. Thus, finding the same species in these two Arctic localities should not be unlikely. The Shetlands, however, are located further south, in a warmer climate zone (Subarctic). Moreover, the distance between Spitsbergen (the most southern locality with a molecularly verified population of M. harmsworthi) and the Shetland Islands is ca. 2000 km. Nevertheless, several tardigrade species were found in localities that were several hundred or a few thousand kilometres apart [74,75,76,77]. However, without the molecular verification, potential cryptic species cannot be ruled out. Moreover, at the time of Murray, the concept of species in tardigrade taxonomy was much different from the current understanding of species. Tardigrade species were thought to exhibit considerable morphological variation and wide (often global) geographic ranges. However, the available integrative data on intra-and interspecific variability, although still scarce, suggest that tardigrade species may exhibit limited morphological variability and geographic distribution, e.g. [76]. Therefore, the presence of M. harmsworthi in the UK should be treated as a hypothesis rather than as an established fact. This, in turn, implies that Murray's slides with specimens from the Shetlands, deposited at the National Museum of Scotland in Edinburgh, should not be considered as M. harmsworthi type material, especially that the name of the species (referring to the Cape Mary Harmsworth) suggests the Arctic to be the primary type locality. Moreover, we examined a specimen deposited at the National Museum of Scotland in Edinburgh, but unfortunately claws IV in that specimen are not visible and it is not possible to verify if the accessory points are in the shape typical for M. harmsworthi (Fig 19A-19E). Thus, it is not possible to state whether the specimen represents M. harmsworthi or another species of the harmsworthi group. In other words, the specimen should be designated as M. aff. harmsworthi (see also below).

Comparison of the neotype with the original description of M. harmsworthi
The original description of M. harmsworthi is brief and many taxonomic characters that we currently consider important were described vaguely or were not described at all [11]. Nevertheless, the description states that M. harmsworthi animals are of a moderate size (up to 500 μm), pale yellow and have very small eyes. The pharynx is oval, with short apophyses, three rod shaped macroplacoids (the third is the longest) and a microplacoid, which is not mentioned in the text, but clearly marked in the drawings. Claws are of the hufelandi type, 24 μm long and with large and protruding accessory points on claws IV (also clearly drawn in Fig 7c in [11]; compare with Fig 4B and 4D herein). Lunules under claws I-III are smooth whereas lunulae IV are crenate (dentate). The eggs were not described verbally, but Murray [11] presented a simplistic drawing (Fig 7a in [11]). The number of processes on the circumference of the drawn egg is ca. 10-12. Processes are in the shape of sigmoidal cones, with short, narrowed and sharpened apices. The width of process base is similar to process height. Egg processes are sometimes in contact and sometimes they are separated and the morphology of egg surface between the processes is not detailed (Fig 7a in [11]).
The morphology of neoparatypes and additional animals fits the original description well and the trait that distinguishes the species from all other known harmsworthi group taxa, the protruding accessory points on hind claws, are clearly marked in the original drawing. The only discrepancy between the description and our redescription is the claw height: whereas the maximal value for the posterior primary branch (including accessory points) in the neotype population was 18.5 μm (with lunules), the original description states 24 μm (in both cases the maximal body length is nearly 500 μm). We suspect, however, that Murray [11] probably measured claws including lunules, which could explain the discrepancy.
In contrast to animals, establishing whether neotype eggs morphology conforms to the type material is more challenging. The original drawing (Fig 7c in [11]) suggests that processes are tightly packed on the egg surface, which does not leave much space for areolation that we observed in the neotype material (Figs 5C and 7A-7D herein). Moreover, the areolation, although often weakly developed, is clearly visible under PCM at least in some M. harmsworthi eggs. Suspiciously, however, eggs of M. occultatus sp. nov. seem to fit the drawing in [11] much closer than neotype M. harmsworthi eggs. Processes in M. occultatus sp. nov. are more densely arranged (Fig 11A-11D) and there is no areolation between the processes (Fig 11C  and 11D). Moreover, we frequently encountered both M. harmsworthi and M. occultatus sp. nov. in the same moss samples. Thus, taking all this into consideration, we suspect that Murray [11] most likely linked M. harmsworthi animals with M. occultatus sp. nov. eggs. If this was indeed the case, we think that animals should be given priority over eggs in establishing which species should be considered M. harmsworthi. Therefore, individuals exhibiting hind claws with large and protruding accessory points, and with dentate (crenate) lunules should be considered M. harmsworthi s.s. whereas eggs with tightly arranged processes and no areolation should be classified as M. occultatus sp. nov. Comparison with other records of M. harmsworthi Dastych [15] described M. h. obscurus and differentiated it from M. h. harmsworthi based on the presence of large accessory points, appendices at the bases of egg process and differences in the morphology of the oral cavity armature. Dastych [15] compared his new subspecies with "M. harmsworthi", also abundantly represented in his material from Svalbard. He identified individuals and eggs as "M. harmsworthi" following the characteristics proposed by Argue [78] and the oral cavity armature morphology according to Pilato [79]. However, Argue [78] studied and described specimens from New Brunswick (Canada) and Pilato [79] described the oral cavity of specimens mainly from Italy. Taking into consideration that these specimens were collected far from loci typici and from different climate zones, it cannot be ruled out that both Argue [78] and Pilato [79] examined different species of the harmsworthi group. In fact, it is not possible to verify their identifications as these authors did not present detailed descriptions of their specimens. Moreover, Argue [78] compared his specimens with the descriptions presented in Marcus [12] and Petersen [80], which are contradictory. Specifically, Marcus [12] described crenate lunules and large accessory points whereas Petersen [80] mentioned that claws were "completely similar to those of M. hufelandi", i.e., with smooth lunules and without protruding accessory points.
The limited original description with a simplistic drawing of the egg allowed to attribute a wide range of animal and egg morphotypes to M. harmsworthi. In fact, any typical Mesobiotus (Macrobiotus prior to year 2016) with eggs equipped with conical processes and no areolation on egg surface used to be identified as "M. harmsworthi" (e.g. see [10]). However, it is clearly visible, from various drawings and/or photos (especially those of eggs) that these probably represent various species of the harmsworthi group rather than a single species and most certainly not M. harmsworthi s.s. (e.g. see [4,12,78,81,82,83]).
The lack of an accurate diagnosis of M. harmsworthi resulted in increasing difficulties in the taxonomy of the harmsworthi group. Thus, in attempt to clarify the diagnosis of the nominal species for the genus Mesobiotus, Pilato et al. [13] analysed animals and eggs from Italy, France and Spitsbergen identified by them "in agreement with tradition" as M. harmsworthi. Moreover, Pilato et al. [13] examined a slide with an individual identified by Murray as M. harmsworthi, deposited in the National Museum of Scotland. Importantly, however, it is not clear whether the specimen was designated as a paratype or only a regular specimen that was identified by Murray as M. harmsworthi. Moreover, it is also not clear if Pilato et al. [13] examined animals or eggs or both from the Shetland Islands. We hypothesise, however, that eggs were not examined because they were not deposited in the National Museum of Scotland. Pilato et al. [13] also presented minimal and maximal values for some morphometric traits of M. harmsworthi individuals and eggs and provided a drawing of the buccal apparatus, claw and egg, but again, they did not specify which specimens they measured and illustrated. Importantly, hind claws depicted in Fig 3b in Pilato et al. [13] have typical accessory points, which is in disagreement with the original description M. harmsworthi that clearly states and depicts large accessory points [11]. Moreover, the egg shown in Fig 3c in Pilato et al. [13] has no signs of areolation. This is in contrast to our findings that individuals of M. harmsworthi exhibit protruding accessory points and lay (at least partially) areolated eggs. Therefore, the drawings in Pilato et al. [13] do not show M. harmsworthi but a different species of the harmsworthi group (i.e. M. aff. harmsworthi, see below).
The doubts and conflicting characterisations of M. harmsworthi described above show explicitly that a modern, integrative redescription of the species based on new material from locus typicus was very much needed and may allow the discovery of species that otherwise would have been classified as "M. harmsworthi".

Geographic distribution of M. harmsworthi
Similarly to redescriptions of other nominal taxa with uncertain original diagnoses, e.g. [84,85], the present redescription of M. harmsworthi resets the geographic distribution of the redescribed species. Prior to the redescription, "M. harmsworthi" was reported from all continents [5,13,68,69,70,86]. However, comparing neotype material with other records identified as "M. harmsworthi" (see above for details), we propose that depending on the type of available data, the following identifications may be achieved: M. aff. harmsworthi-when individuals and/or eggs fit the general M. harmsworthi morphotype (= an unidentified species of the M. harmsworthi group).
• M. cf. harmsworthi-when qualitative traits fit the redescription but incomplete quantitative data and/or the lack of DNA sequences do not allow a full verification of the record against the neotype series (= a probable record of M. harmsworthi).
• M. harmsworthi-when qualitative and quantitative traits fall within the ranges described in this study and/or DNA sequences show immediate relatedness to the neotype sequences provided here (= a certain record of M. harmsworthi).

Conclusions
Based on our analyses, we postulate that all the specimens exhibiting morphology identical with M. h. obscurus should be now considered as the nominal M. harmsworthi s.s., whereas specimens reported from Svalbard by Dastych [15] as M. h. harmsworthi represent M. occultatus sp. nov. The remaining specimens reported as M. harmsworthi throughout the world should be treated as unidentified species of the harmsworthi group (i.e., as M. aff. harmsworthi).
Supporting information S1