Cyclotella or Discostella? An Evaluation of the Morphological and Molecular Evidence Regarding the Generic Placement of a Centric Diatom from Eastern Asia and the Creation of Discocyclus gen. nov.

Abstract

In this study on diatoms, which are unicellular microalgae with a siliceous cell wall, images obtained by electron microscopy were used to compare three taxa from East Asia that were referred as Cyclotella hubeiana, Discostella guiyangensis and “Cyclotella fottii” in the original studies. Morphometrical and morphological analyses indicate that these three taxa are conspecific. The distinctive morphological features of this species are the following: one rimoportula located on the valve mantle within the ring of marginal fultoportulae, marginal fultoportulae situated on costae with internally two satellite pores positioned circumferentially (laterally) and externally long tubular opening, biseriate striae and branching costae. A comparison of these morphological features with those of Discostella and Cyclotella species and a re-assessment of the existing molecular data suggest that this species could be separated from both these genera. On this basis, we describe the new genus Discocyclus gen. nov. with Discocyclus hubeianus comb. nov. as the type species.

1. Introduction

The Stephanodiscaceae Glezer and Makarova 1986 [1] is a large family of centric diatoms mainly represented by freshwater species. Currently in Algaebase, this family includes 472 species [2]. In the past ten years, new genera of Stephanodiscaceae have been described on the basis of morphological observations. These genera are often described from fossil material such as Strelnikoviella P. Kociolek, G.K. Khursevich & E.C. Theriot [3], Criblionella Jovanovska, Cvetkoska, Tofilovska, Ognjanova-Rumenova & Levkov [4], Cyclocostis Paillès [5], Fascinorbis Stone and Edlund & Alverson [6] but also from modern material (e.g., Edtheriotia Kociolek, Q.M. You, Stepanek, R.L. Lowe et Q-X. Wang [7]). In the same period, thanks to an approach that integrates morphological observations with molecular data, the numbers of genera recognized within the diatom family Stephanodiscaceae has increased further. Examples of such studies include that of Ács et al. [8], who separated the genus Pantocseckiella K.T. Kiss & Ács from the genus Lindavia (Schütt) De Toni et Forti, that of Kulikovskiy et al. [9] who confirmed Stephanocyclus Skabitchebskiy as a separate genus from Cyclotella Brébisson and that of Tuji et al. [10] who created Praestephanos A. Tuji & M. Julius based on Stephanodiscus suzukii A. Tuji & Kociolek.

In two separate recent studies, light and scanning electron microscope images of an apparently identical centric diatom of the family Stephanodiscaceae were published. In one of these papers [11], this diatom was described as a species new to science and with the support of molecular analysis was attached to the genus Discostella Houk & Klee. In the second paper [12], the apparently same diatom species was referred as Cyclotella hubeiana Chen & Zhu, and on the basis of light and scanning electron microscope images, its description was emended and an epitype was proposed. In addition, populations of a similar centric diatom were also reported from South Korean rivers and referred to as “Cyclotella fottii” [13].

This paper presents an account of the investigation of these three “taxa” using morphometric analysis and morphological analysis based on SEM images. The evidence for the generic placement of these “taxa” in either Cyclotella or Discostella are discussed, and the establishment of a new genus, Discocyclus gen. nov., is proposed.

2. Materials and Methods

Material identified as Cyclotella hubeiana comes from two populations, one collected from the sediments of the Xionghe Reservoir in Hubei Province (central China) and one collected from the sediments of Qiandao Lake in Zhejiang Province (eastern China). The full details for this material are given in [12]. Material identified as Discostella guiyangensis comes from Huaxi Park in Guiyang, Guizhou Province, in southern China, as described in [11]. The diatoms from Korea originally identified as “Cyclotella fottii” come from several locations. In the Algal Flora of Korea [13], occurrences of this taxon were reported in the lower reaches of the Nakdong River in the southeastern part of South Korea and from the Namhan River, located in the center of the peninsula. Only light microscope images were published in the Algal Flora of Korea. This taxon was also found in reservoirs (unpublished data) such as Lake Daecheong (a large man-made lake formed by a dam on the Geum River in Daejeon Municipality in the center of South Korea) and in Lake Chungju (formed by a dam on the Namhan River). The SEM images of the valves we analyzed in this study come from these two lakes.

For the scanning electron microscope (SEM) analysis of the Cyclotella hubeiana populations, 10 drops of cleaned suspensions were filtered and rinsed with deionized water through a 3 µm Isopore™ polycarbonate membrane (Merck, Darmstadt, Germany). For Discostella guiyangensis, as the material was issued from a culture and very clean, 10 drops of the suspension were taken with a pipette and directly deposited on a 12 mm diameter glass coverslip. All the preparations from China (membranes and coverslips) were mounted onto aluminum stubs using double-coated carbon conductive adhesive tape and coated with gold using a Quorum Q150R S plus sputter coater (Quorum Technologies, Laughton, UK). SEM images were taken using a Phenom XL desktop SEM (Nanosciences Instruments, Phoenix, AZ, USA) at the Institute of Geology and Geophysics (Chinese Academy of Sciences, Beijing, China) operated at 10 kV. Regarding the material collected from South Korea, that from Lake Daecheong was filtered through a Millipore membrane filter while that from Lake Chungju was simply dried on a small glass coverslip. In both cases, platinum was used for coating, and the images were captured using a Hitachi S-4800 (Japan). For the Lake Daecheong material, the SEM images were captured at the Center for Research Instrument and Experimental Facilities located in Chungnam National University in Daejeon City, while the images from Lake Chungju were obtained at the Busan Center of the Korea Basic Science Institute (KBSI).

For the morphometric analysis of the three “taxa”, we selected SEM images of specimens in valve view. In total, 122 valves of Cyclotella hubeiana (28 from Xionghe Reservoir and 94 from Qiandao Lake, i.e., the same dataset as in [12]), 132 valves of Discostella guiyangensis and 12 valves from the Korean material were analyzed. The diameters of the 266 valves included in the dataset were measured using the software ImageJ [14]. The number of costae and the number or marginal fultoportulae (mfp) were also enumerated along the circumference of each valve. Following the method described on the “Diatoms of North America” website [15], these numbers of stria and mfp were used to calculate the densities in 10 µm circumference that are reported in

Table 1. List of the populations of centric diatoms investigated and morphometric characteristics measured from SEM images obtained for this study.

Taxon (as Given in Original Publication)	Locality	Collector	Nb of Images	Valve Diameter (µm)	Stria Density (in 10 µm)	mfp Density (in 10 µm)
“Cyclotella fottii”	Lakes Daecheong and Chungju (S. Korea)	Gyeongje Joh	12	12.0–31.4	11.6–13.5	4.2–5.1
Cyclotella hubeiana	Xionghe Reservoir (China)	Mengna Liao	28	8.8–29.5	10.3–14.4	3.9–5.5
Cyclotella hubeiana	Qiandao Lake (China)	Mengna Liao	94	5.6–35.8	9.1–12.6	2.8–5.3
Discostella guiyangensis	Huaxi Park, Guiyang (China)	Qi Liu 1	132	15.4–27.7	10.7–12.9	3.2–5.2

¹ See acknowledgments.

. Bivariate analysis, based on scatter plots drawn using Microsoft Excel, was employed to investigate the ranges in the tree morphometric characters in relation to each other. The terminology used in this study follows [16].

3. Results and Observations

3.1. Morphometric Analysis

Following the method used in [12], we used bivariate plots to display the numbers of striae and numbers of marginal fultoportulae plotted against the valve diameter as well as the numbers of marginal fultoportulae against the numbers of striae (Figure 1). There are some differences between the three “taxa” in their ranges in valve size and in their values of stria and mfp densities (Figure 1 and Table 1 and Table S1). The valves of Cyclotella hubeiana, which come from two natural populations, display the wider range. The relatively narrow ranges measured for the Korean material is likely due to the small number of valves measured (only 12), while those for Discostella guiyangensis are likely related to the fact that the population of this material was derived from a culture. However, the trajectories observed in these bivariate plots are almost identical in the combined populations of Cyclotella hubeiana, in “Cyclotella fottii” and in Discostella guiyangensis.

3.2. Morphological Analysis

We first checked the identity of the material from South Korea originally identified as “Cyclotella fottii” by [13]. The examination of the SEM images (Figure 2) clearly indicates that this population does not correspond to Cyclotella fottii Hustedt from Lake Ohrid (a large, ancient lake located at the border between North Macedonia and Albania in Europe) as illustrated in several publications [17,18,19]. Cyclotella fottii, which is probably endemic to this ancient lake, has some distinctive characters. Externally, its valves show a ring of long spines at the valve face/mantle junction, multiseriate striae and simple openings of the marginal fultoportulae. Internally, there are three to eight rimoportulae positioned in the marginal area of the central area, clearly detached from the alveolae. By contrast, the valves from the Korean material do not possess long spines and have biseriate striae, and the external openings of the marginal fultoportulae can be simple or with more or less developed tubular projections (Figure 2A–I). Internally, the Korean valves display a single rimoportulae located on a costa and within the ring of marginal fultoportulae (Figure 2J–L). The characteristics of the Korean population are therefore identical to those of Cyclotella hubeiana from China as shown in Figure 3 (more SEM images were published in [12]). Similarly, the SEM images of Discostella guiyangensis (Figure 4) share the same morphological features. Together, the trajectories on the bivariate plots (Figure 1) and the examination of SEM images (Figure 2, Figure 3 and Figure 4) indicate that the valves of these three “taxa” belong to the same species.

Among the set of characters used to understand the phylogeny of the Stephanodiscaceae and define the genera within this family, the position and structure of the rimoportula(e) and fultoportulae are considered very important [6,20]. The taxon studied in this paper possesses one rimoportula and marginal fultoportulae that are positioned on costae. This combination of characters precludes this taxon from belonging to the genus Discostella, which was distinguished from all other cyclotelloid genera by having one rimoportula and marginal fultoportulae all located between costae [21]. On the other hand, as already discussed in [12], the arrangement of the marginal fultoportulae, with only two satellite pores and in lateral position, does not fit with the clades currently assigned in Cyclotella sensu stricto in [21], which are characterized by mfp with three satellite pores or two satellite pores in radial position. Besides this morphological difference, the molecular results published by Xiang et al. [11] suggest that this taxon should not be included in the genus Cyclotella. Therefore, the combination of characters and molecular evidence distinguishes this taxon from all currently described genera within the Stephanodiscaceae, and we propose the establishment of a new genus: Discocyclus gen. nov.

3.3. Systematic Emplacement and New Genus Description

• The classification is following Algaebase [22].
• Phylum: Heterokontophyta Moestrup, R.A. Andersen & Guiry.
• Subphylum: Bacillariophytina Medlin & Kaczmarska.
• Class: Mediophyceae Medlin & Kaczmarska.
• Subclass: Thalassiosirophycidae Round and Crawford.
• Order: Stephanodiscales Nikolaev & Harwood.
• Family: Stephanodiscaceae Makarova.
• Genus: Discocyclus Rioual & Schultz gen. nov. (Figure 2, Figure 3 and Figure 4).

Diagnosis: A new genus of the Stephanodiscaceae. Cells solitary, frustules cylindrical in girdle view, circular in valve view. Initial cells lenticular. Valve face with two different morphological patterns, with radially striated marginal part that surrounds a distinctly different central area that can vary from almost flat to concentrically undulate, convex or concave. The central area has no central fultoportula. Internally, the marginal zone consists of radial simple alveolae with hyaline laminae covering the internal surface of the alveolae on the valve face. Alveolae open internally by oval foramina located at the face/mantle junction. Radiate alveolae are separated internally by costae. A single ring of fultoportulae occurs on the valve mantle. Mantle fultoportulae bearing two satellite pores in circumferential (lateral) position. One rimoportula located on the mantle on a costa within the ring of marginal fultoportulae. Externally, mantle/face junction broadly sloped, striated margin spans from central area to mantle edge, striae biseriate and separated by costae. Mantle fultoportulae outer openings are either simple or with well-developed tubular projections.

Etymology of Discocyclus gen. nov.: this genus name is a combination of the genus names Discostella and Cyclotella, indicating that Discocyclus shares some morphological features of both genera.

Type species: Discocyclus hubeianus (Chen & Zhu) Rioual comb. nov.

Basionym: Cyclotella hubeiana Chen & Zhu 1985. Acta Hydrobiologica Sinica, 9: 82, figs 3–4 [23].

Synonyms: Discostella guiyangensis Y.D. Xiang, Q. Liu & S.L. Xie, Journal of Oceanography and Limnology, 42(4): p.1316, figs 1–4, 2024. [11]; Cyclotella fottii Hust. apud Joh [13] fig. 41.

4. Discussion

In order to justify the placement within the genus Discostella of the population they investigated, ref [11] argued that several species of Discostella have mfp positioned on coastae and thus do not fulfill the main criterion used in [21] to define this genus. We think that this statement should be re-assessed. The first species they mentioned is Discostella hellae (Chang & Steinberg) Chang, a species that is poorly characterized as discussed in [24], because no SEM image of the internal view of the valve was provided in the original description by Chang and Steinberg [25]. In addition, as mentioned by Chang and Steinberg who originally described this species as Cyclotella hellae and again by [26] in their discussion on Discostella, “Discostella hellae” is very close morphologically to Cyclotella nana Hustedt, a species that has been confirmed to belong to the genus Cyclotella by molecular analyses [27]. The second species mentioned by Xiang et al. [11] is Discostella mascarenica (Klee, Houk & Bielsa) Houk & Klee. This species possesses thin finger-like costae that are fused into strongly embossed ribs (see Tab. 329 in [19]). Therefore, the placement of the mfp on the costae or between costae is ambiguous in this species due to the structure of the costae themselves. There is no molecular data for this taxon, so far only observed on the Réunion Island (western Indian Ocean). The third and last species is Discostella asterocostata, which possesses short costae between the long costae. The mfp are located on the top of the short costae but never on the long costae. Thus, in that case too, it can be argued that the placement of the mfp on costae is ambiguous. Interestingly, D. asterocostata is also molecularly distant from all other species of Discostella that were investigated by Schultz et al. [24].

The placement of the population from Guiyang within Discostella is therefore mostly supported by molecular evidence (see Fig. 5 in [11]). Yet, “Discostella guiyangensis” is notably distant from the other clades of Discostella, and its position as a sister clade to D. asterocostata has only moderate support as acknowledged by Xiang et al. [11] who reported bootstrap Maximum Likelihood values of 51.0 and Bayesian Inference values of 0.76. In addition, in the divergence time estimation phylogram proposed by Xiang et al. (see Fig. 6 in [11]), “Discostella guiyangensis” (i.e., Discocyclus hubeianus comb. nov.) appears as a sister clade of the clade regrouping all the other Discostella species they investigated, but not within the genus.

Differences between Cyclotella hubeiana (i.e., Discocyclus hubeianus comb. nov.) and species of Pantocsekiella and Cyclotella with similar appearance under the light microscope have already been discussed in Liao et al. [12], so we will not repeat them here. At the generic level, however, it is worth emphasizing that in Discocyclus, the rimoportula is positioned among the ring of marginal fultoportulae, above an alveola, while in Pantocsekiella/Lindavia, the rimoportula(e) are positioned outside the ring of mfp, generally just below the alveola(e) or on the valve face, as illustrated in Nakov et al. [28] and Stone et al. [6].

As discussed by Williams [29], DNA barcoding should be considered as just another line of evidence and morphological data, especially when well established like in the case of Discostella (see discussion in [26]), should not be ignored or only considered when fitting with the “sequence data cladogram”. In our view, the combination of morphological and molecular data supports the creation of a new genus to accommodate the peculiar taxon originally described as Cyclotella hubeiana, which does not belong to Cyclotella according to molecular data and does not belong to Discostella according to the morphological data.

Discocyclus hubeianus nov. comb. is for the moment the only representative of the genus Discocyclus gen. nov. Genera thought to be monotypic at the time of their creation are routinely found to include more than one species with the ongoing discoveries of species new to science and with the constant re-assessment of the literature. Examples of genera thought to be monotypic also exist among the Stephanodiscaceae, as shown by Schultz et al. [30] for the genus Cyclotubicoalitus Stoermer, Kociolek & Cody described by Stoermer et al. [31].