A new species of freshwater Chaetonotidae ( Gastrotricha , Chaetonotida ) from Obodska Cave ( Montenegro ) based on morphological and molecular characters

Gastrotricha is a cosmopolitan phylum of aquatic and semi-aquatic invertebrates that comprises about 820 described species. Current knowledge regarding freshwater gastrotrichs inhabiting caves is extremely poor and there are no extant data regarding Gastrotricha from Montenegro. We describe a new species from Obodska Cave, which is also the first record of a gastrotrich from this region. Due to its unusual habitat and morphological characteristics, this species may be important when considering the evolution and dispersion routes of Chaetonotidae Gosse, 1864 (sensu Leasi & Todaro 2008). We provide morphometric, molecular and phylogenetic data for the new species, together with photomicrographs and drawings.


Introduction
The eumetazoan meiofauna is considered a significant component of both rocky and soft bottoms of various natural aquatic ecosystems (Giere 2009).The meiofauna is an important source of food for macrofauna, small fish, juveniles of large fish and other epibenthic predators (Danovaro et al. 2007).

Study area
This study was performed in the Obodska Cave (42°21.118′N, 19°0.304′E), which is located west of the Rijeka Crnojevica, in the cadastral municipality of Ljubotinj II, Montenegro (Fig. 1).The area is not cultivated and most of the natural vegetation is still intact.The landscape is mostly covered with deciduous forest.The climate is classified as humid subtropical (no dry season, hot summer), with a temperate warm and wet forest biozone (Bonada et al. 2008).The area is high in leptosol (lp), a weakly developed shallow soil.Cetinje field and its surroundings are inclined to the southeast toward Skadar Lake, which causes gravity flows of groundwater in that direction (Bonada et al. 2008).There are numerous caves and cavities in this region, indicating the degree and depth of karstification.Cavities are vertical or horizontal with an opening on the surface.In most cases they are located in areas of vertical cracks or fracture systems, where extended karst processes occur on the tectonic lines and on the contact zone between limestone and dolomite (Martinović 1964;Lješević 1968;Doderović et al. 2013).
Obodska Cave is a deep cave with the spring-fed Crnojevica River flowing through it (Figs 2-3).The major part of the cave is formed beneath Pecki Hill.The Crnojevica River flows into the cave through a trench that plunges beneath the surface at the foot of Pecki Hill.Obodska Cave was created from stratified limestone, where edges of layers create horns visible on the vault.The sides and bottom of the upper channels are polished, in places, with narrow shelves on the horns of the layers.The cave is situated on three to five morphological levels (Palmer 1991).The total length of the cave is more than 350 meters and comprises three compartments connected by two siphons.The cave was formed through erosion by an underground river flowing along the initial chasm.The river flowed fast through stones and gravel in the whole cave and left holes in the solid rock (Martel 1893).Water in the cave leads to a humid microclimate (Obodska Cave has a precipitation/potential evapotranspiration index higher than 0.65) (Martel 1893;Lješević 1968;Mihavc 1983).Obod spring is characterized by a very variable flow that ranges from a minimum of 0.24 m 3 /s to a maximum of 46 m 3 /s.Typically, the minimum water level occurs in November or December, and the maximum occurs in March or April, or rarely in February.The first post-summer minimum peak in flow is caused by minimal autumn rainfall and high temperatures, while the maximum peak in flow is due to rain with snow and low temperatures.The low precipitation in Cetinje field and its adjacent surroundings results from a lack of water streams on land surface (Martinović 1964;Lješević 1969;Radulović & Radulović 2004).
The entrance to the cave is situated at 375 m above sea level, and the lowest point of the cave reaches an elevation of 244 m (192 m below the cave entrance).The main channel of the cave is divided by two siphons and creates the upper and lower channel.The lower channel is only partially passable (Mihavc 1983).
Obodska Cave was created in several morphological and hydrological stages.The main upper channel with a constant water flow was created during the first stage, when narrow pits were created at the bottom of the channel.The extension of the pits gradually allowed an increase of the amount of water that plunged through them and formed the second channel.This bifurcation of the channel has migrated during the second stage of development from a point close to the cave entrance to a point 45 meters from the entrance to the cave, where it is currently located.The third stage was characterized by the further scouring of the cave by the water flow, so that the upper part of the channel became dry.It is impossible to observe the bifurcation, because the lower water plunges through sinkholes in the bottom European Journal of Taxonomy 354: 1-30 (2017) and flows through cave channels underground.All this shows that there is another, lower cave channel, which today contains warm water.The genesis of this cave and its water flow reflect the typical lowering of the underground water flow in karst.Finally, it should be noted that the regime of the Crnojevic springs, which emerge from the cave, is very uneven.The flow of water fluctuates from 0.03 to 2 m 3 per year (Dinić 1965).A few endemic species of invertebrates are known from this cave, e.g., Amphipoda: Metohija carinata Absolon, 1927;Gastropoda: Plagigeyeria montenigrina Bole, 1961 andColeoptera: Adriaphaenops stirni Pretner, 1959.They occur only there or in a few other caves on the Adriatic coast (Pretner 1972;Pešić 2010;Hou & Sket 2015).

Sampling and documentation
One sample of approximately 50-70 cm 3 from the top layer of the bottom sediment, together with approximately 100-120 cm 3 of ambient water, was collected from each of 4 sites within one locality on 29 July 2015.Each sample was placed in a 200 cm 3 plastic container.The sampling area was situated at the end of the middle compartment approximately 150 meters from the entrance to the cave.Samples were collected in an area not covered by stones and gravel close to the second siphon, connecting the second and third compartments.Four samples were collected from different parts of the river channel: (1) in the main channel at a depth of 0.4 m, (2) from the main channel at a depth of 0.25 m, (3) from shallows at a depth of 0.1 m, and (4) from pockets among stones and gravel.The last sample (4) was collected from a place where water filled the space between stones and boulders which had fallen from the roof of the cave.The sampling locations were separated by no less than 2 m (Fig. 2).
Water parameters, such as temperature, pH, electrolytic conductivity and dissolved oxygen content, were measured in the field with an Elmetron CX-401 multiparametric sampling probe; BOD5 by Winkler's method, and NH 4 , NO 3 , PO 3 with a Slandi LF205 photometer.The collected samples were subsequently placed in an insulated box and transported to the laboratory, where the samples were oxygenated and held at 12°C in order to create conditions similar to those at the sampling site.Within 10 days after collection, a total of 2 cm 3 of sediment from each sample was searched for gastrotrichs.Specimens were extracted with a micropipette under an Olympus SZ51 stereo microscope.All specimens were observed, photographed and documented alive with an Olympus BX53 microscope equipped with phase contrast and an Array Artcam 500 digital camera or a Leica DM 5500 B microscope equipped with

Granulometric analyses
To determine the grain-size properties, the granulometric protocol described by Buchanan (1984) was applied.The grain-size statistics were calculated through Gradistat software using a logarithmic method of moments (Blott & Pye 2001) and sediments were classified according to the Folk and Ward system (Folk & Ward 1957).The organic matter content of the same sediment samples was determined using the combustion method.The main results of the granulometric analyses are provided in Table 1.

Morphological analyses
A morphological feature was measured only if its orientation was conducive to accurate measurement.Ranges of measurements are presented.Each of the specimens was documented by means of a set of photomicrographs.The species description follows the convention of Hummon et al. (1992), in which the longitudinal distance of a morphological character from the anterior end is expressed as percentage units (U) of the animal's total length (i.e., the distance of the character from the anterior end, divided by the total body length, and multiplied by 100).The identification of gastrotrichs, their morphological study and terms follows Kisielewski (1981Kisielewski ( , 1991Kisielewski ( , 1997) ) and Kolicka et al. (2016).The terms describing the shape of furcal branches and furcal indentation follows Roszczak (1969).In this paper, the following formula describing the distribution of scales was used: as well as the pharynx formulae according to Kisielewski (1991): Abbreviations used: D = dorsal; DL = dorsolateral; L = lateral; LV = ventrolateral; V= ventral.
The new species proposed in this paper is described by the first author only.This solution is consistent with chapter 11, article 50 of the International Code of Zoological Nomenclature (The International Commission on Zoological Nomenclature 1999).

Molecular analyses
Total genomic DNA was extracted from three single specimens of the new species using the DNeasy Blood and Tissue Kit (Qiagen GmbH, Hilden, Germany) as described by Dabert et al. (2008).A 628 bp fragment of the mitochondrial cytochrome c oxidase subunit I (COI) gene was amplified with a 1:1 mixture of two forward primers, bcdF08 and bcdF09 (Kolicka et al. 2016), and reverse primer bcdR04 (Dabert et al. 2010).A fragment coding for both internal transcribed spacers (ITS1-5.8SrDNA-ITS2) was amplified with ITS1_18S and ITS2_28S primers (Navajas et al. 1998).A complete sequence of 18S rRNA was amplified in two overlapping fragments using 18Sfw/rev960 and fw390/rev18S primer pairs (Dabert et al. 2010), respectively.A 2381 bp fragment of 28S rRNA was amplified using the primer pair 28SF0001/28SR2850 (Dabert et al. 2016).For primer details see Table 2. PCRs were carried out in 10 µl reaction volumes containing 5 µl of the Type-it Microsatellite PCR Kit (Qiagen), 0.5 µM of each primer and 4 µl of the DNA template using a thermocycling profile with one cycle European Journal of Taxonomy 354: 1-30 (2017) 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 step of 5 min at 72°C for all reactions.After amplification, the PCR products were diluted with 10 µl of MQ water; 5 µl of the diluted PCR reaction was analysed by electrophoresis on 1% agarose gel.Samples containing visible bands were purified with exonuclease I and Fast alkaline phosphatase (Fermentas) and sequenced using the BigDye Terminator v3.1 kit and the ABI Prism 3130xl Genetic Analyzer (Applied Biosystems), following the manufacturer's instructions.Individual sequence reads were aligned and manually assembled into contigs in ChromasPro v. 1.32 (Technelysium) and GeneDoc v. 2.7.000 (Nicholas & Nicholas 1997).Genetic distance among the COI sequences was estimated using the Kimura 2-parameter model as implemented in MEGA 7 (Kimura 1980;Tamura et al. 2013).

Phylogenetic analyses
The nucleotide blast search of COI, 18S and 28S rRNA sequences of Chaetonotus (Chaetonotus) antrumus sp.nov.suggested Chaetonotus sp. 1 (in Kånneby, Todaro & Jondelius 2013), Chaetonotus (Chaetonotus) cf.sphagnophilus Kisielewski, 1981 and Chaetonotus (Chaetonotus) cf.laroides Marcolango, 1910 as the most similar taxa.Therefore, in our phylogenetic analyses we used representatives of the main clades reconstructed in the published molecular phylogeny of Chaetonotidae (Kånneby et al. 2013;Kolicka et al. 2016).As the outgroup we used sequences of Aspidiophorus polystictos Balsamo & Todaro, 1987, which has been reconstructed as the sister group to all other species of Chaetonotidae by Kånneby et al. (2013).In total, our data set consisted of 4881 nucleotide positions for 55 terminals (Table 3) and involved COI+18S+28S markers.ITS sequences were excluded from the data set because of the lack of sequence data for this marker for most species included in the phylogenetic analysis.
Choice of an appropriate model of DNA sequence evolution for 18S and 28S rDNA was made using jModeltest 0.11 (Posada 2008); the GTR + I + G model was appropriate for both markers.For COI DNA sequences the two-rate codon-based model was applied (Goldman & Yang 1994) with invertebrate mtDNA genetic code.Tree inference was performed using Bayesian Inference with Markov Chain Monte Carlo (BI), with the appropriate substitution model for each partition.Four independent chains were run on a parallel version of MrBayes 3.2 (Ronquist et al. 2012).Each run of the BI analyses was performed in 3-10 × 10 6 generations, and the trees were sampled every 1000 th generation.The final 50% majority rule consensus tree was generated after discarding the 25% burn-in fraction of initial trees after assessing the chain convergence in Tracer v.1.(Rambaut & Drummond 2007) judged by the average standard deviation of split frequencies dropping below 0.01.Tree editing was performed using FigTree 1.4.2(Rambaut 2014).

Results
Gastrotrichs, belonging to only one species, were present in two of the four sampled sites: Site 1 (9 specimens) and Site 3 (27 specimens) in Fig. 2.
The physicochemical parameters of the water on the investigated cave river did not vary between sampling sites and were the following: temperature: 12°C; conductivity: 3.18 µS/cm; dissolved oxygen: 10.66 mg/dm 3 ; biochemical oxygen demand over 5 days (BOD 5 ): 1.98; NO 3 : <0.10 mg/dm 3 ; NH 4 concentration: 0.207 mg/dm 3 ; PO 3 concentration: 0.170 mg/dm 3 .The granulometry varied among sites (Table 1).Sites 1 and 3 had a high organic content with moderately to poorly sorted fine sediments, while Sites 2 and 4 had lower organic content and more coarse sands.In addition to Gastrotricha in the examined material, we found protozoa (mainly Ciliata), nematodes and rotifers (Bdelloidea as well as Monogononta).

Type area
Denmark.

Diagnosis
Slender body, length from 91.2 to 129.7 μm.Head five-lobed, cephalion and pleuria weakly marked in the head outline.Hypostomium small and rhomboidal.Ocellar granules absent.Scales small, threelobed and with strong keels.One pair of one-lobed, keeled scales on the dorsal side of the posterior part of the trunk, and on the dorsal and dorsolateral sides of the furcal appendages.Two pairs of three-lobed, spined scales present on the ventral side of the furcal appendages.Scales distributed in 29-35 total longitudinal rows (11-13D+6-8DL+6L+4-6LV+2V) with 23-27 scales in the central row, and differing morphologically in the areas of the head, neck and trunk, respectively.Spines thick, simple, with blunt ends.Spine lengths strongly vary: spines of the head are longer than those of the neck; spines of the neck are short, but become progressively longer to the widest body region, after which they gradually shorten towards the furcal base.The spines gradually increase in length from lateral to ventral body side towards the ciliary bands.Ventral scale spines longer than the others and hair-like.Last pair of parafurcal spines longer and stronger.Ventral interciliary field naked, except for the posterior trunk region.A pair of ventral terminal scales long, oval with shallow posterior notches, keeled and spineless.Pharynx narrow with two weakly marked dilatations.Straight intestine with a distinct, short anterior section appearing as a narrow band.

Etymology
From Latin 'antrum', cave, referring to the habitat where the species was found.

Description
This new species has a slender body.Its head is wider than the neck and separated from the trunk by a distinct neck constriction.The neck extends into the trunk, which gradually widens towards its widest region beyond its midpoint (ca U61), after which it gradually tapers towards a distinct furcal base at U84.The branches of the furca are set wide apart.The furcal indentation is V-shaped, and the adhesive tubes diverge posteriorly.They measure 10.1-11.3μm, are straight and thin and do not taper towards their blunt ends (Figs 4-6).
The head is five-lobed and semi-circular.All plates are visible in the dorsal head outline.The cephalion (U1-U5) adheres to the head along its entire length, is narrow and widens at the dorsal edge.The epipleuria (U4-U6) are small and slightly convex.The hypopleuria (U7-U13) are more than twice as large as the epipleuria.The hypopleuria are not visible from the dorsal side; only their outline is marked in body shape (Figs 4A,7).The hypostomium (U5-U8) is short and rhomboidal with slightly rounded edges and a strong anterior edge (Fig. 4C).Two pairs of cephalic ciliary tufts are present.The anterior tufts have four cilia each that emerge from between the cephalion and epipleuria and are arranged in lines around the lateral edges of the cephalion.The beginning of these lines (the first two cilia) is clearly visible on the dorsal side.The anteriormost cilium in both anterior tufts is fairly short (the shortest).The second cilium is longer than the first.The third cilium is very long and is the longest of all cilia in the tuft.The posteriormost cilium is shorter than the third and similar in length to the second cilium.The posterior tufts have five cilia each and emerge ventrally at the anterior edge of the hypopleuria.The length of the cephalic cilia in the posterior tufts increases from the anteriormost to the fourth cilium.The posteriormost cilium is similar in length to the first (see Appendix).Ocellar granules are not present.The mouth ring is narrow, located sub-terminally at U2-U3 and has weakly marked, granular reinforcements.Short inner hairs are present inside the mouth ring and suboral hairs are located around it.
The body is covered by small, three-lobed scales and single one-lobed scales that adhere over their entire surface to the cuticle (Figs 5-9).Scales are distributed in 29-35 total longitudinal and alternating rows (11-13D+6-8DL+6L+4-6LV+2V) with 23-27 scales in the dorsal central row.Each scale has a strong keel and triangular with a deeply-notched posterior edge.On the head, neck and anterior and middle parts of the trunk, the longitudinal rows of scales run parallel to each other, while in the posterior part of the trunk and on the furcal base the scales gradually converge towards the central longitudinal row (Figs 4,(7)(8)(9).The scales on the head, neck and trunk are arranged close to one another, but their edges do not adhere or overlap.The longitudinal rows of scales begin on the head beyond the posterior edges of the cephalion, epipleuria and hypopleuria.The scales show a strong morphological diversity in posterolateral lobe distinctness, edge roundness and size throughout the different surfaces of the head, neck and trunk regions.On the head, there are deeply-notched, rounded, wide triangles with rounded posterolateral lobes that are weakly differentiated from the central lobe.On the neck, scales are narrower, and KOLICKA M. et al., New species of Chaetonotus from cave their postero-lateral lobes show a weaker separation from the central lobe (see Appendix; Figs 4A, 5).On the trunk, scales are shaped like narrow, deeply-notched triangles with their postero-lateral lobes clearly differentiated from the central lobe and with edges that are less rounded (Figs 4-5, 8A, C, E, 9A).The size of the scales decreases rapidly at the beginning of the neck, after which it gradually increases towards the beginning of the trunk (head: 1.6-3.8μm length × 1.8-4.3μm width vs neck: 1.3-3.3μm length × 1.4-3.9μm width vs trunk: 2.5-5.6 μm length × 1.6-4.3μm width).The dorsal head and neck scales differ most from one another (Figs 4A, 5, 7A), whereas in the other areas of the head and neck, the differences in shape and size between the scales are gradual.The size of the scales of the trunk increases from the anterior end towards the widest trunk region, after which it decreases towards the furcal base (see Appendix).On the posterior part of the trunk, scales are clearly smaller than those at the widest point of the body.A fine (1.9-2.7 μm length × 1.3-1.8μm width) one-lobed, keeled and spineless scale is located at U78 on the dorsal part of the trunk, anterior to each scale bearing a sensory bristle (Figs 4A, 8C) and is shaped like a strongly-rounded triangle.Medially, on the dorsal side of the furcal appendages (U86-U88), three three-lobed scales are present that are slightly narrower than the other scales of the trunk (see Appendix; Figs 4A, 8G).Lateral to these scales, on the furcal appendages (U85-U88), there are two pairs of elongated one-lobed scales shaped like narrow ovals with a weakly-notched posterior edge.The anterior one of each pair of scales is located dorsally, has a strong keel and a long, straight spine, whereas the posterior one is located dorsolaterally and has a long keel, but no spine.The lateral edges of these scales are partially overlapping.On the lateral side of the furcal appendages (U86-U90), two pairs of three-lobed scales, of the same type as the scales of the trunk, with spines are present (Figs 6, 9A).The dorsal, dorsolateral, lateral, ventrolateral and ventral scales do not strongly vary in size, except that the scales in the area of the neck and in the longitudinal rows next to the ciliary bands are considerably smaller than the others.The ventral scales of the longitudinal rows located closest to the ventral ciliary bands are about half the size of the scales in the other rows (head: 1.6-2.9μm length × 1.8-2.3μm width; neck: 1.3-1.8μm length × 1.4-2.0μm width; trunk: 2.5-3.0 μm length × 1.6-2.0μm width) and have their central lobe rotated about 20° towards the bands (see Appendix; Figs 4C, 8C, 10C).
The spines arising from the posterior scales region are thick and straight, taper very slightly towards their blunt ends and have no lateral denticles (Figs 4, 7B, D, 8B, D, F, 9B, 10E).The spines that adhere directly to the cephalion and pleuria are the shortest of the head spines (Fig. 7B).The spines on the head increase in length (1.1-3.1 μm), whereas those on the neck decrease rapidly in length.The spines on the neck are much shorter than those in the head area; the spines are merely vestigial halfway down the neck, after which they gradually lengthen towards the trunk (0.5-2.7 μm) (Figs 4, 7B).The spines on the trunk gradually and slightly lengthen from the beginning of the trunk (ca U30) up to the widest body region (ca U61), after which they gradually shorten towards the furcal base at U84 (1.3-4.1 μm) (Figs 4, 8B, D, F).The pair of posteriormost lateral trunk spines is slightly longer and thicker than the surrounding spines (3.4-5.9 μm).Parafurcal spines emerging from two lateral scales per side on the furcal appendages are slightly thicker and longer than the other spines of the body, the posteriormost pair is longer, thicker and stronger than those of the preceding pair (see Appendix; Fig. 4).These spines taper slightly towards their blunt distal ends.The dorsal and dorsolateral spines do not vary substantially in length (Table 4).The spines lengthen gradually and slightly from the lateral side towards the ciliary bands (Fig. 9B).The spines arising from the ventral longitudinal row of scales located closest to the ciliary bands are much longer than those of the body, curved, and hair-like along their entire length (head: 5.1-8.6 μm; neck: 5.0-9.0 μm; trunk: 7.1-14.0μm).This species has three pairs of dorsal sensory bristles (Fig. 4).The first pair is located on the head, directly posterior to the cephalion, near the lateral edges of the epipleuria (U5), and each bristle emerges from a small, round papilla.The second pair is located on the neck (U25) and each bristle emerges from a small, rounded papilla.The third, posterior pair, which emerges from double-keeled scales located in European Journal of Taxonomy 354: 1-30 ( 2017 4A, 5).
On the ventral side, the longitudinal ciliary bands begin at U8 and run back to U84 (Fig. 4C).They are wider in the area of the head than in the other parts of the body.Most of the ventral interciliary field is naked: fine, keeled, spined scales are present only in the posterior part of the trunk (from ca U78 to U82) ( Figs 4C, 8H).Their differentiation increases towards the posterior body region: the anterior scales are weakly delineated and partially submerged in the cuticle.These scales are shaped like narrow triangles with a very weakly notched posterior edge.The scales of the ventral interciliary field increase in size towards the posterior end of the body (1.4-3.7 μm length × 1.1-1.9μm width).The terminal scales are located at U82-U85 and are shaped like long, narrow ovals with a very weakly notched posterior edge and have a long keel running along their entire length, but are spineless (Figs 4C, 5E).
The pharynx (U2-U28) is relatively narrow and has weak anterior and posterior dilatations, with the posterior dilatation wider than the anterior one (Figs 4B, 7C; Appendix).The pharynx connects through the small and narrow pharyngeal intestinal junction (U30) to a straight intestine (running from U29 to U86).The intestine has a distinct, short (U29-U31) anterior section marked as a narrow band (Fig. 4B).

Remarks
Chaetonotus (Chaetonotus) antrumus sp.nov. is an interstitial species which was recorded in the lotic system in a cave.The gastrotrich fauna in interstitial freshwater habitats is relatively rich, but composed  of fewer species than in epibenthic or periphytic habitats (e.g., Balsamo et al. 2015).Interstitial communities are composed of not only taxa specific to them, but also of eurytopic species.Out of ca 40 species found in sandy biotopes, fewer than 10 seem to constitute exclusively interstitial taxa (Balsamo et al. 2015).All of the species share certain morphological traits, e.g., a small body size, a poorly ornamented cuticular covering, a well developed locomotory ciliature and adhesive organs (Balsamo & Fregni 1995;Balsamo et al. 2015).Entire sets of these characteristics also  3).obscured by the cilia or erroneously interpreted as one of the cilia.Thus, the presence of dorsal sensory bristles on the head cannot be considered as a good diagnostic characters at the species level.
The presence of a developing egg was observed in 21 out of 32 adult specimens of C. (C.) antrumus sp.nov.However, sperm and an X-organ were not observed.
Chaetonotus (Chaetonotus) antrumus sp.nov.showed very quick locomotion under microscopic observation.The species often remained in motion even after its integument had burst from compression under a cover glass, thus producing a visible deformation.

Sequence diversity and phylogenetic relationships
The COI alignment for the distance calculations comprised 628 bp of unambiguous sequence data for three specimens of Chaetonotus (Chaetonotus) antrumus sp.nov.We found two haplotypes that differed in eight nucleotide positions (K2P = 0.013; SD = 0.005).All substitutions were located at synonymous sites, i.e., both haplotypes coded for the same amino acid sequence.The nuclear data, including 4809bp of the DNA region coding for 18S rRNA, ITS1, 5.8S, ITS2, and 28S rRNA, showed no intraspecific variation.All sequences are deposited in GenBank under accession numbers KU705230, KU705231 (COI); KX538804 (18S); KU705232 (28S); KU705233 (ITS).

Discussion
The presence of Chaetonotus (Chaetonotus) antrumus sp.nov. in only two of four sampling sites may reflect the habitat selectivity by the species.The species was observed only in the samples with a high organic matter content, a minimum diameter spectra of the sand grain, and moderately to poorly sorted sediments.This result confirms previous observations regarding the habitat preferences of many freshwater gastrotrichs, which appear to prefer finer sediments with high organic matter content (e.g., Kisielewski 1997;Balsamo & Todaro 2002;Balsamo et al. 2014).However, the occurrence and European Journal of Taxonomy 354: 1-30 (2017) composition of the Gastrotricha fauna inhabiting caves may depend not only on the physicochemical habitat properties (temperature, sunlight, granulometry of the sediments, the water flow rate) (e.g., Balsamo et al. 2014), but also on the possibility of colonization and food availability.Gastrotrichs, as other meiobenthic invertebrates, have limited locomotory abilities.Dispersal in freshwater gastrotrichs is probably limited to migration through aquatic sediments and dispersal of resistant eggs (see below).
The result of our molecular phylogenetic analysis is for the greater part congruent with the previous molecular phylogenies of Chaetonotidae (Kånneby et al. 2013;Kolicka et al. 2016) and gave further evidence to the high degree of polyphyletism of chaetonotid genera.Freshwater species of the order Chaetonotida, which may produce opsiblastic, resting eggs, are known for passive migration with water currents, watercourses, surface run-offs, wind and being carried by more mobile animals, e.g., birds, bats, amphibians, insects or annelids (e.g., Gerlach 1977;Kolicka et al. 2014).One of those vectors was most probably responsible for the transport of gastrotrichs into Obodska Cave.Because of their wide range of ecological tolerance and parthenogenetic reproduction, Gastrotricha may have persisted there and developed a population.When analyzing the fauna of cave ecosystems, it must be remembered that those habitats are not influenced by sunlight.Thus, there are no photoautotrophs and the ecosystem is based mainly on the influx of organic matter from outside the cave.Most of the currently known species of Gastrotricha feed on bacteria, detritus, and small algae (e.g., Bennett 1979;Balsamo & Todaro 2002;Todaro & Hummon 2008).The lack of unicellular autotrophs may not be a limiting factor for the majority of Gastrotricha species that feed mainly on bacteria (e.g., Balsamo et al. 2014).
Further studies will show whether Chaetonotus (Chaetonotus) antrumus sp.nov. is endemic to Obodska Cave, or more widely distributed and not associated with only one specific type of environment.The application of methods typical for integrative taxonomy (i.e., combined morphological, morphometric and molecular analyses), will allow verification of future reports regarding the presence of this species and accurate determination of its biogeography.

Fig. 1 .
Fig. 1.Map of Montenegro with the location of Obodska Cave.
Fig. 11.Phylogenetic relationships of Chaetonotidae inferred from the Bayesian analysis of 18S rRNA, 28S rRNA and COI sequence data (for details on the species names see Table3).
et al., New species of Chaetonotus from cave

Table 1 .
Main results of sediment sample granulometric analyses.All measurements are given in micrometres (μm); indicators are given as a percentage (%) and italicized.
KOLICKA M. et al., New species of Chaetonotus from cave

Table 2 .
PCR primers used in this study; A and S refer to amplifying and sequencing, respectively.

Table 3 .
DNA sequences of the gastrotrichs species used in phylogenetic analysis.
Balsamo & Todaro, 1995, both of which were originally reported from Italian mountain pools.These two species were selected from all the Chaetonotus representatives for the comparison of the new species due to a similarity in terms of (1) the alignment of scales whose edges are juxtaposed; the presence of a different type of scales on the dorsal and dorsolateral sides of the furcal base; (2) the type of scales on the ventral interciliary field; (3) the number of terminal scales of the ventral interciliary field; (4) two longitudinal bands of ventral locomotor cilia wider on the head region.Despite the fact that from all of the hitherto known species, C. (C.) naiadis and C. (C.) daphnes have the highest number of common features with the newly described species, they are significantly different from C. (C.) antrumus sp.nov., most strikingly in the scale type and shape.All the differences between the new species and the most morphologically similar taxa have been summarized in Table4.
) recovered, with maximum support, C. (C.) antrumus sp.nov.ascloselyrelated to the undetermined Chaetonotus sp. 1 TK-2012 (Kånneby et al. 2013); both grouped with C. (C.) cf.laroides.This clade was in a sister relation to a clade consisting of some representatives of Chaetonotus (Chaetonotus) Ehrenberg, 1830 and Aspidiophorus (Voigt, 1903) and including Chatonotus (Chaetonotus) daphnes Balsamo & Todaro, 1995, which is morphologically most similar to the new species.Differential diagnosisEven if the new species was recorded from a sandy habitat, it is not similar to any other species considered as exclusively interstitial taxa.Chaetonotus Ehrenberg, 1830 is a polyphyletic genus, containing varied species.Of all the 165 currently known nominal freshwater representatives of this genus, C.(C.) antrumus sp.nov. is morphologically closest to C. (C.) naiadisBalsamo & Todaro, 1995  and C. (C.)daphnes