The insights into the systematic relationship of Gastrostyla-affinitive genera, with report on a new saline soil ciliate genus and new species (Protozoa, Ciliophora)

Background Hypotrichia are a group with the most complex morphology and morphogenesis within the ciliated protists. The classification of Gastrostyla-like species, a taxonomically difficult group of hypotrichs with a common ventral cirral pattern but various dorsal and ontogenetic patterns, is poorly understood. Hence, systematic relationships within this group and with other taxa in the subclass Hypotrichia remain unresolved. Results 18S rRNA gene sequence of a new Gastrostyla-like taxon was obtained. Phylogenetic analyses based on the 18S rRNA gene sequences indicate that this ciliate represents a new genus that is closely related to Heterourosomoida and Kleinstyla within the oxytrichid clade of the Hypotrichia. However, the position of this cluster remains unresolved. All three genera deviate from the typical oxytrichids by their incomplete (or lack of) dorsal kinety fragmentation during morphogenesis. Morphology and morphogenesis of this newly discovered form, Heterogastrostyla salina nov. gen., nov. spec., are described. Heterogastrostyla nov. gen., is characterised as follows: more than 18 fronto-ventral-transverse cirri, cirral anlagen V and VI develop pretransverse cirri, and dorsal ciliature in Urosomoida-like pattern. Conclusions Similar to the CEUU-hypothesis about convergent evolution of urostylids and uroleptids, we speculate that the shared ventral cirral patterns of Gastrostyla-like taxa might have resulted from convergent evolution.

In April 2015, an undescribed Gastrostyla-like species was isolated from saline soil within the Longfeng Wetland Nature Reserve, a district of Daqing, northern China. Analyses of its morphology and cell division, as well as the small subunit ribosomal DNA (SSU rDNA) sequence, indicate that it represents a new species and a new genus. Phylogenetic analyses of all available Gastrostyla-like species were performed.

Results
SSU rDNA sequence and phylogenetic analyses ( Fig. 1) The SSU rDNA sequence of Heterogastrostyla salina nov. spec. was deposited in the GenBank database with the accession number MT739409. The length and GC content of the SSU rDNA sequence are 1687 bp and 46.00%, respectively. Phylogenetic trees inferred from the SSU rDNA sequences using two different methods, i.e., maximum likelihood (ML) and Bayesian inference (BI), show similar topologies, therefore we present only the ML tree with bootstraps and posterior probabilities from both algorithms (Fig. 1).
Cladistics relationship and morphological patterns of Gastrostyla-like species (Figs. 2 and 3) A cladogram of Gastrostyla-like species was constructed based on the presence/absence of dorsomarginal/dorsal fragmentation, the fate of the old dorsal kineties, the number of caudal cirri and whether anlage V contributes to pretransverse ventral cirri (Fig. 2). We also provide illustrations showing the morphology of Gastrostyla-like genera for clarity (Fig. 3). Heterogastrostyla salina, Kleinstyla dorsicirrata, Pseudogastrostyla flava, and Neogastrostyla aqua form one clade, because they have dorsomarginal rows. Other Gastrostyla spp. are separated from this clade due to their incomplete, or the complete absence of, dorsal kinety fragmentation. Neogastrostyla aqua is distinguished from H. salina, K. dorsicirrata and P. flava by its anlage V not contributing to pretransverse ventral cirri. Pseudogastrostyla flava is distinguished from H. salina and K. dorsicirrata by the number of caudal cirri. In H. salina, the dorsal fragmentation is absent, whereas K. dorsicirrata has incomplete fragmentation. As concerns Gastrostyla-like taxa without dorsomarginal rows, Hemigastrostyla differs from Apogastrostyla rigescens and Protogastrostyla pulchra in exhibiting multiple dorsal kinety fragmentation, whereas A. rigescens and P. pulchra are distinguished from each other by the retention/resorption of the parental dorsal kineties.

Etymology
Composite of the Greek adjective heteros (different) and the well-known genus name Gastrostyla. This indicates that Heterogastrostyla has a similar ventral ciliature to Gastrostyla but differs in the dorsal side. Feminine gender.

Etymology
The species-group name salina refers to the saline habitat where the type specimen was discovered.

Stomatogenesis
In the opisthe, the formation of membranelles commences left of the anterior end of the oral primordium ( Fig. 6 a). As the formation of adoral membranelles proceeds posteriad, the undulating membranes anlage (= FVT-anlage I) contributes the leftmost frontal cirrus and splits longitudinally into two streaks from which the endoral and paroral are formed (Figs. 5 l and 6 e, g, i). In the proter, the parental undulating membranes gradually dedifferentiate into an undulating-membrane anlage. The differentiation of the undulating membranes anlage follows a similar pattern to that in the opisthe ( Fig. 6 c, e, g, i). Interestingly, on one middle stage specimen, we found a small patch of densely distributed kinetids posterior to frontoventral anlage I in the proter (Figs. 5 n and 6 d), but this small patch disappeared in the next stage ( Fig. 6 g). We deduce that it is a remnant of the FVT-anlagen. The parental adoral zone of membranelles is retained intact during the morphogenetic process (Figs. 6 a, e, g, i and 7 a).

Development of frontoventral ciliature
The FVT-anlagen II-VI develop as primary primordia and then divide into two groups transversely, one for each daughter cell (Figs. 5 l-n and 6 a, e, g). Subsequently, anlagen I-VI of each group segregate new cirri in the normal pattern: 1:3:3:4:5:5 (Figs. 5 o q and 6 i). After migration and differentiation, three frontal, one buccal, six to eight frontoventral, four or five postoral ventral, two pretransverse and five transverse cirri are formed. The origination of ventral ciliature is as follows: (i) the leftmost front frontal cirrus comes from anlage I; (ii) the middle frontal cirrus and buccal cirrus come from anlage II; (iii) the rightmost frontal cirrus originates from anlage III; (iv) the six frontoventral cirri come from anlage III (×1), anlage IV (×2) and anlage V (×3); (v) the four postoral ventral cirri come from anlage IV (×1) and anlage V (×3); the two pretransverse cirri come from anlage V (×1) and anlage VI (×1); and the five transverse cirri derive from the posterior end of anlagen II-IV, respectively (Figs. 6 i and 7 a).

Marginal rows
In each divider, the marginal rows anlagen develop intrakinetally. These anlagen then increase in size by adding basal bodies on the right side of the parental structure (Fig. 6 e). Meanwhile, the parental rows are gradually resorbed (Fig. 6 g, i).

Dorsal kineties
In the earliest stage, several patches of the dorsal-kinety (DK) anlagen appeared intrakinetally in the middle of each old structure without a clear separation for the proter and the opisthe (Fig. 6 b); Whether the DK-anlagen are primary primordia is not clear, however, since their early development is not known. Dorsal morphogenesis proceeds in Urosomoida-like pattern, i.e. the new dorsal kineties develop intrakinetally as three anlagen each in the proter and opisthe without fragmentation, and each dorsal kinety produces one caudal cirrus in the late stage (Figs. 5 s, 6 b, f, h, j and 7 b). It is noteworthy that a short dikinetid-row appears anterior of the right marginal anlagen, more or less distinctly separated from the right marginal anlagen (Figs. 5 p, r and 6 i). This is the dorsomarginal kinety anlage. It is unclear whether the dorsomarginal kinety anlage derives from the anterior portion of the right marginal anlage and later moves to the dorsal side.

Division of nuclear apparatus
The nuclear apparatus divides in the usual way, i.e., the two macronuclear nodules fuse to form a single mass during the mid-divisional stage which then divide twice prior to cytokinesis (Figs. 6 b, f, h, j and 7 b).

Physiological reorganization
Only one early stage of physiological reorganization was observed (Fig. 7 c, d), which indicated that the early process of cortical development in reorganizers is similar to morphogenesis.

Phylogenetic position of the new genus Heterogastrostyla and related taxa
The present phylogenetic analyses show that Heterogastrostyla nov. gen. is most closely related to Heterourosomoida and Kleinstyla. However, the systematic position of this group is far from being resolved, as indicated by the variable statistical support in the SSU rDNA tree (Fig. 1).
The grouping of Heterogastrostyla, Heterourosomoida, and Kleinstyla was supported by their morphological similarities in that all these three genera exhibit deviation from the typical oxytrichid fragmentation of dorsal kinety 3. The former two genera share the same Urosomoida-like pattern in which fragmentation of dorsal kinety 3 is lost, whereas Kleinstyla exhibits incomplete fragmentation of dorsal kinety 3 [18]. Neogastrostyla aqua, resembles H. salina in terms of the dorsal ciliary pattern, however, they are not closely related in the SSU rDNA tree as N. aqua nests robustly within the Oxytricha granulifera clade. Similarly, Gastrostyla is distinctively placed within the oxytrichid clade, which is consistent with assertion of Wirnsberger et al. (1986) that G. steinii is a stylonychine oxytrichid [31]. Other Gastrostyla-like genera, i.e., Pseudogastrostyla, Apogastrostyla, Protogastrostyla, and Hemigastrostyla, are consistently placed outside the oxytrichid clade, as shown in the previous studies [17,23].
The cladogram based on the dorsal ciliary pattern (Fig.  2) of Gastrostyla-like species is broadly consistent with the molecular tree (Fig. 1). With the presence of the dorsomarginal row, Heterogastrostyla salina shows a close relationship with Kleinstyla dorsicirrata and Pseudogastrostyla flava. Together with Gastrostyla spp., they are closely related to oxytrichids, whereas Apogastrostyla-Protogastrostyla-Hemigastrostyla are distinctly separated from the oxytrichid clade since they lack a dorsomarginal row. The presence/absence of dorsomarginal kineties plays a significant role in the classification of hypotrichs, supporting the Dorsomarginalia hypothesis [32].

Heterogastrostyla salina can be distinguished from
Heterogastrostyla salina resembles Urosomoida, Paraurosomoida, Hemiurosomoida, and Heterourosomoida in terms of its dorsal ciliature, i.e., fragmentation of dorsal kinety 3 is lost. However, H. salina can be distinguished from these taxa by its Gastrostyla-like ventral cirral pattern.
Morphogenetic characteristics of the new species basically correspond with that of Neogastrostyla aqua Kaur et al., 2019, except for development of the pretransverse ventral cirri [20]. In the latter, all cirri (except for a single transverse cirrus) generated from cirral anlage V, move anteriad to form the postoral ventral cirri, that is, cirral anlage V does not develop any pretransverse cirri. However, in H. salina, some cirri generated from cirral anlage V move anteriad to form postoral ventral cirri, while others move posteriad to form a pretransverse and a transverse cirrus.
Although morphogenesis of the new species closely resembles that of Gastrostyla spp., it differs in that: (1) fronto-ventral-transverse cirral anlagen are formed from primary primordia (vs. in secondary primordia), and; (2) the dorsal kineties anlagen are in a Urosomoida-like (vs. an Oxytricha-like) pattern [4,33,42] A comparison of ontogenesis of Gastrostyla-like species is summarized in Table 2. The ventral development of Heterogastrostyla salina proceeds basically as in Hemigastrostyla, Apogastrostyla, and Protogastrostyla. Specifically, six primary FVT-anlagen generate an increased number (>18) of fronto-ventral-transverse cirri, with frontoventral cirri not regularly grouped but arranged in a more or less continuous, slightly oblique (frontoventral) row. Heterogastrostyla salina, however, differs significantly from the above mentioned three genera in: (i) the fate of the parental adoral membranelles in the proter (completely retained vs. only apical part of old adoral zone retained, combining the newly built membranelles formed from the proter's oral primordium), and; (ii) the dorsal development (Urosomoida-like pattern vs. Gonostomum-like pattern or Hemigastrostylalike pattern). It should be noted that in Protogastrostyla, the dorsal kinety anlagen are unique since the primary primordia and old dorsal kineties are retained, resulting in a higher number (9-11) of dorsal kineties [28].

Conclusions
Similar to the CEUU-hypothesis in urostylids and uroleptids (Foissner et al., 2004), we speculate that the shared ventral cirral patterns of Gastrostyla-like species might result from the convergent evolution from four major groups: (i) true oxytrichids with both a dorsomarginal row and complete fragmentation of dorsal kinety 3 (e.g., Gastrostyla steinii); (ii) those with a dorsomarginal row but without, or with incomplete, dorsal fragmentation (e.g., Heterogastrostyla nov. gen., Kleinstyla, and Pseudogastrostyla); (iii) those without a dorsomarginal row but with multiple fragmentation of dorsal kineties 1 and 2 (e.g., Hemigastrostyla); and (iv) those in which both the dorsomarginal row and dorsal kinety fragmentation are absent (e.g., Protogastrostyla and Apogastrostyla). We assume that the "Gastrostyla-like ventral cirral pattern" evolved at least twice independently in the above-mentioned groups. Since the ventral ciliature is linked with motility, foraging and food uptake, the evolutionary pressure on the ventral ciliature is much stronger than on the dorsal ciliature, which is possibly sensoric and therefore is more conservative [25]. This might explain why the Gastrostyla-like species are scattered throughout the SSU rDNA tree and not in a single group. This supports the CEUU-hypothesis in that it is insufficient to determine the systematic positions of hypotrichs solely by their ventral cirral pattern [43]. Future studies should combine dorsal patterns with molecular analyses to obtain a more robust phylogeny.

Methods
Saline soil samples (0-10 cm; salinity of soil percolate about 20‰; pH 10.0) were collected in the Longfeng Wetland Nature Reserve (lat. 46°35′30″N, long. 125°13′ 08″E), Daqing, northern China, on 16 April 2015. Samples were malodorous (very likely due to hydrogen sulphide), and included a large proportion of rotten leaves and branches. For preservation and future isolation, samples were dried at room temperature (about 24°C ) immediately after collection.
Ciliates were stimulated to excyst by applying the nonflooded Petri dish method [22]. They were then isolated and non-clonal cultures were established at room temperature (about 23°C) in Petri dishes containing filtered soil percolate and squeezed rice grains to enrich the bacterial food.
Living specimens were observed using bright field and differential interference contrast microscopy [31]. Protargol preparation was used to reveal the ciliature and the nuclear apparatus [44]. Counts and measurements of prepared specimens were performed at a magnification of 1,000×. Drawings of protargol-prepared cells were made with the aid of a drawing device (camera lucida). To illustrate the changes occurring during morphogenesis, old (parental) structures were depicted by contour whereas new ones were shaded black. Terminology and systematics basically follow Lynn (2008) [45]; for terms specific for hypotrichs, see references [16,25,26,32] DNA extraction, PCR amplification and sequencing Genomic DNA was extracted from single cells using DNeasy Tissue kit (Qiagen, CA) following the manufacturer's instructions, with the modification that 25% of the volume suggested for each reagent solution was used. The SSU rRNA gene was amplified according to [46] and [9], using the primers 18S-F (5'-AAC CTG GTT GAT CCT GCC AGT-3') and 18S-R (5'-TGA TCC TTC TGC AGG TTC ACC TAC-3') [47]. analysis was performed, using RAxML-HPC2 v8.2.12, on XSEDE [51] on the online server CIPRES Science Gateway [52] with the GTR + G + I model as the optimal choice. Support for the best ML tree came from 1,000 bootstrap replicates with the GTR + CAT model. Bayesian inference (BI) analysis was performed with MrBayes v3.2.6 on XSEDE [53] on the online server CIPRES Science Gateway, using the GTR + I + G model as selected by MrModeltest v.2.0 [54]. Markov chain Monte Carlo (MCMC) simulations were run with two sets of four chains for 2,000,000 generations with a sampling frequency of 100 and a burn-in of 5,000 trees (25%). All remaining trees were used to calculate posterior probabilities using a 50% majority rule consensus. TreeView v1.6.6 [55] and MEGA 4.0 [56] were used to visualize the tree topologies. For interpretation of bootstrap values we follow Vd'ačný and Rajter (2015); that is, we consider values ≥95 as high, from 70-94 as moderate, from 50-70 as low, and <50 as representing no support [57]. Bayesian posterior probability values <0.95 are considered as low and values ≥0.95 as high [58].