Fensomea setacea, gen. & sp. nov. (Cladopyxidaceae, Dinophyceae), is neither gonyaulacoid nor peridinioid as inferred from morphological and molecular data

Dinophyte evolution is essentially inferred from the pattern of thecal plates, and two different labelling systems are used for the important subgroups Gonyaulacales and Peridiniales. The partiform hypotheca of cladopyxidoid dinophytes fits into the morphological concepts of neither group, although they are assigned to the Gonyaulacales. Here, we describe the thecate dinophyte Fensomea setacea, gen. & sp. nov., which has a cladopyxidoid tabulation. The cells displayed a Kofoidean plate formula APC, 3′, 4a, 7″, 7C, 6S, 6′′′, 2′′′′, and slender processes were randomly distributed over the echinate or baculate surface. In addition, we obtained rRNA sequences of F. setacea, gen. & sp. nov., but dinophytes that exhibit a partiform hypotheca did not show a close relationship to Gonyaulacales. Character evolution of thecate dinophytes may have progressed from the ancestral state of six postcingular plates, and two more or less symmetrically arranged antapical plates, towards patterns of only five postcingular plates (Peridiniales) or more asymmetrical configurations (Gonyaulacales). Based on our phylogenetic reconsiderations the contact between the posterior sulcal plate and the first postcingular plate, as well as the contact between an antapical plate and the distalmost postcingular plate, do not represent a rare, specialized gonyaulacoid plate configuration (i.e., the partiform hypotheca of cladopyxidoid dinophytes). Instead, these contacts correspond to the common and regular configuration of peridinioid (and other) dinophytes.

. Schematics of hypothecae in armoured dinophytes. (A) Cladopyxidoid, (B) peridinialean, (C) gonyaulacalean: sexiform (modified 8 ). Putatively homologous plates are color-coded, using the AY-model system 17 . For plate labelling, we followed Kofoidean notation 23 (black lettering), while the Taylor-Evitt notation is in grey. Thus, our first antapical plate corresponds to a posterior intercalary plate, and our second antapical plate corresponds to the only antapical plate 3,8,17,18 . Note the (putatively ancestral) connection between the posterior sulcal plate and the first postcingular plate (red triple bar in A and B). The reduction from six to five postcingular plates in Peridiniales may result from a fusion of the third and fourth postcingular plates as they are present today in cladopyxidoid dinophytes including Fensomea setacea, gen. & sp. nov. extant dinophytes exhibiting partiform hypotheca are the Amphidomataceae 24 , but their possible relationship to Cladopyxidaceae has not been clearly worked out at present.
The distinctiveness between Gonyaulacales and Peridiniales lineages has been confirmed by molecular phylogenetics [25][26][27][28][29] . However, taxa with available DNA sequence information and a partiform hypotheca (i.e., Amphidomataceae) do not show clear phylogenetic proximity to either of these two major dinophyte lineages. Thus, extant Cladopyxidaceae may provide a missing link of thecate dinophytes that would enable a better understanding of the first evolutionary transformations from ancestral configurations towards the more abundant and derived patterns in Gonyaulacales and Peridiniales 15,26,[30][31][32] . To investigate this potential link, as well as the taxonomy of the constituent elements of the Cladopyxidaceae, the morphology of the thecal plate pattern and phylogenetic placement must be determined. In this study, we elucidate new steps towards achieving this integrative goal and describe a new cladopyxidoid dinophyte. We present light and scanning electron microscopy (LM and SEM, respectively) results of cultivated material and discuss our findings with regards to previous SEM studies of Cladopyxidaceae 33,34 . We combine these morphological data with the first DNA sequence data of a cladopyxidoid dinophyte.
The plate formula was APC, 3′, 4a, 7″, 7C, 6S, 6′′′, 2′′′′ and is schematically drawn in Fig. 3. The epitheca (Fig. 4A,B) consisted of three apical plates, four anterior intercalary plates, seven precingular plates, and an apical pore complex (APC). The first apical plate was broad, heptagonal, and rectangular in the centre. The second www.nature.com/scientificreports/ apical plate was heptagonal and larger than the hexagonal plate 3′. The four anterior intercalary plates formed a series (dorsal to ventral) on the cell's right side. Plate 1a was square, whereas the larger and irregularly shaped plate 2a was heptagonal. Both plates 3a and 4a were pentagonal, with plate 4a being more elongated and located ventrally next to plate 1′. Each of the precingular plates was in contact with four other epithecal plates (including plate Sa), except for plate 4″, which was adjacent to three other epithecal plates. The first two precingular plates were slightly broader compared to the remaining plates of the series; plate 7″ was the smallest precingular plate (Fig. 4A,B). The APC (Fig. 4C-F) consisted of a rounded pore plate with a straight or slightly curved ventral suture with plate 1′. The pore plate was bordered on its dorsal and lateral sides by a minute elevated rim formed by the sutures of plates 2′ and 3′. In the centre of the pore plate, there was a small and slightly raised tubular or globular structure with a horseshoe-shaped cover plate (cp), which was generally obscured by mucus and difficult to observe. Internal views of the pore plate (Fig. 4F) revealed a crescent-shaped apical pore opening. www.nature.com/scientificreports/ When disregarding the sulcal plates, the partiform hypotheca consisted of six postcingular and two antapical plates (Fig. 4G). Each postcingular plate was in contact with three other hypothecal plates, with the exception of plate 3′′′ that was adjacent to both antapical plates and thus to four other hypothecal plates. All postcingular plates were similar in width, except for plate 1′′′, which was narrower. The right side of this plate bore a distinct, curved flange that partly covered the sulcal area. The antapical plates were dissimilar in size; plate 1′′′′ was smaller, pentagonal, and positioned slightly more ventral compared to the hexagonal plate 2′′′′ (Fig. 4G). www.nature.com/scientificreports/ The cingulum (Fig. 4H,I) was composed of seven cingular plates of almost equal size; plate C7 was distinctly smaller. The posterior cingular suture was zigzag-shaped (most obvious on pentagonal dorsal plates C3 and C4, less distinctive on plate C1). Six sulcal plates were identified (Figs. 3E, 5). The anterior sulcal plate (Sa) was part of the epitheca (Fig. 4A,B). The posterior sulcal plate (Sp) was large, at least twice as long as wide, and extended posteriorly almost to the antapex (Figs. 2, 4G). Anterior to plate Sp, there were two small sulcal plates (right and left posterior sulcal plates: Sdp and Ssp, respectively). Plate Ssp formed the posterior, emarginate area, where presumably both flagella emerged (Fig. 3E), and was adjacent to both the first (C1) and last (C7) cingular plates. Two small anterior plates (Sda and Ssa) were arranged lateral to the flagellar opening. These two plates as well as plate Ssp (but not plate Sdp) did not bear ornamentation (Fig. 5). In strain GeoB*184, deviations from this plate pattern were occasionally observed, including the fusion or subdivision of epithecal and/or hypothecal plates (Fig. S2).
In SEM preparations of strain GeoB*184, slender processes ('setae') were scattered over the cell's surface. The setae proximal surfaces were striate, and distal surfaces were glabrous (Figs. 2, 6A-E). Their diameter was consistent (0.36 ± 0.02 µm, n = 32), but length varied from 1.9 to 6.3 µm (mean 3.8 ± 1.2, n = 27). In most cases, these structures seemed to be randomly arranged and not regularly attached to the cells but occasionally, they appeared to be attached to small pores (Fig. 6C,D). Moreover, depressions with round outer edges were visible on both epithecal (including the pore plate) and hypothecal plates, which were characteristically surrounded by three or four tiny granules (Fig. 6B,E,F). It was unclear whether these structures penetrated the plates like true pores. Internally, the position of these structures in the hypo-and epitheca (Figs. 4F, 6G,H) was visible not as www.nature.com/scientificreports/ openings but as small bumps or margined by minute granular depositions. The slender processes and the echinate cell surface were also seen under LM (Fig. 7). The presence of growth bands, interior thecal views (e.g., Figs. 4B, 5G,H), and thecate cells with slightly disarranged plates allowed for identification regarding the overlap pattern of plate margins (Fig. 3C,D). Keystone plates (i.e., those plates overlapping all of their neighbours) from the cingular, precingular, and postcingular series

Discussion
Diversity of thecal processes. Planktonic cells with elongated processes are rare among extant dinophytes, and three types can be readily distinguished: (a) delicate, unbranched, and filiform setae of Micracanthodinium Deflandre; (b) robust, striated, and intratabular (unbranched and branched) processes present in Acanthodinium Kof. and Cladopyxis; (c) slender processes (visible even under LM) that are randomly distributed over the cell surface and not associated with particular thecal plates. To the best of our knowledge, F. setacea, gen. & sp. nov., is unique in exhibiting this final characteristic (c), but whether these setae are present in all stages of life-history, as well as their precise function, remains to be determined. The setae described here were present on cells derived from cultivated material, but they have previously been documented in field samples 34 ; thus, a culture artefact appears unlikely. In a few cases, ambient conditions, such as temperature, salinity, or turbulence, have been shown to modify surface features and process length in coccoid cells of extant dinophytes [35][36][37] . However, this has not been demonstrated in thecate cells. Additionally, there are no data that show long, robust, and divided processes can become short, fine, and unbranched as a result of culture conditions. Thus, we consider the slender processes of F. setacea, gen. & sp. nov., to be a stable feature and the most striking diagnostic trait to delimit our new taxon from previously described species (see "Taxonomic activity"). This characteristic also delimits F. setacea, gen. & sp. nov., from other extant cladopyxidoid taxa without processes, such as Palaeophalacroma J.Schiller (= Epiperidinium Gaarder) and Sinodinium D.S.Nie 12,38 (which also have no partiform hypotheca).
The presence of thecal plates was not noted in any of the original descriptions of species currently assigned to Micracanthodinium. In a subsequent SEM study 34 that claimed to illustrate the tabulation pattern of Micracanthodinium for the first time, no rigorous explanation for the identification of Micracanthodinium setiferum (Lohmann) Deflandre was provided, and two different organisms may have been studied (compare his Figs. 2 and 6). As a result, there is still no published study that reliably shows the filiform setae of true Micracantho-dinium together with a dinophyte plate pattern. Thus, it remains unclear whether the SEM plates of John D. Dodge 34 include cells assignable to F. setacea, gen. & sp. nov.
There is an open question whether all such setae and processes that are variously slender, robust, or branched are homologous among dinophytes. In particular, it is not even known at present whether the setae of true Micracantho-dinium (distributed mainly along the cingulum margins) conform with 'skeleton'-based structures (as in Acanthodinium and Cladopyxis, in which they appear associated with specific thecal plates) or 'membrane'based structures, such as pseudopodia. It is likely that at least the robust and smaller processes of C. hemibrachiata and Balech F. setacea, gen. & sp. nov., respectively, are homologous because their plate patterns are very similar (see "Epithecal configurations"). Thus, Acanthodinium, Cladopyxis, and F. setacea, gen. & sp. nov., may appear as integral elements of the Cladopyxidaceae, but the taxonomic identity of Micracanthodinium from its type locality in Sicily 39 , and its relationship to cladopyxidoid 40,41 or other (last not least unarmoured) dinophytes, remains elusive.
Hypothecal configurations. The two major branches of dinophytes, Gonyaulacales and Peridiniales, present a mosaic combination of ancestral and derived character states. Despite the small number of extant species, cladopyxidoid protists are important for evolutionary interpretations because their seemingly rare plate pattern allows character polarity to be identified 15,26,31,32 . The precise systematic position of Cladopyxidaceae within the Dinophyceae, their internal taxonomic delimitations, and the phylogenetic relationships between their constituent elements have not sufficiently worked out. Accurate interpretations of the thecal plate pattern, and homologies between plates, are key to the development of a consistent evolutionary scenario. Charles A. Kofoid was the first person to interpret a cladopyxidoid tabulation for Acanthodinium spinosum Kof. and C. brachiolata (= Acanthodinium caryophyllum Kof.) 42 , but the plates were not yet labelled using his Kofoidean system because he developed that later 22 . However, the drawings show good congruence with later interpretations, particularly C. brachiolata 12 . Based on such studies Cladopyxis and putative relatives, including F. setacea, gen. & sp. nov., have thecal series comprising three apical, three or four anterior intercalary 12 , seven precingular, six postcingular, and two antapical plates.
Despite his erection of the Gonyaulacales, Frank J.R. 'Max' Taylor regarded Cladopyxidaceae as phylogenetically closer to peridinioid dinophytes 17 . In contrast, William R. Evitt considered their plate pattern to be derived from sexiform gonyaulacoids rather than peridinioids 18 . However, F. setacea, gen. & sp. nov., is not an integral part of the Gonyaulacales based on the DNA tree and instead represents an independent lineage within the dinophytes whose closest relatives cannot be ascertained reliably at present. This agrees with phylogenetic sketches (Fig. 192 8 , Fig. 22 30 ), in which cladopyxidoid dinophytes are not part of the Gonyaulacales. In any case, the inferred phylogenetic position of F. setacea, gen. & sp. nov., challenges the assumption that a partiform hypotheca would be a distinctive configuration of gonyaulacoid dinophytes 8,18 . Thus, contacts between plates Sp and 1′′′ (Z and Iu in Taylor-Evitt notation), as well as between plates 2′′′′ and 6′′′ (Y and VI in Taylor-Evitt notation), do not represent a rare, specialized gonyaulacoid plate pattern, but correspond to the common and regular configuration of peridinioid dinophytes 31 . It should be noted that while the peridinioids display only five postcingular plates, this is probably a derived character state. Furthermore, this combination of plate contacts and six postcingular plates has also been found in thecate suessialean dinophytes 43 24,46 , which were, like Cladopyxidaceae, formerly included within Gonyaulacales 8,14,33 . Dinophytes with a partiform hypotheca exhibit a combination of ancestral conditions, such as six postcingular plates (traditionally associated with Gonyaulacales) and the symmetrical arrangement of plates (traditionally associated with Peridiniales). Plesiomorphic traits are unsuitable to support close relationships and thus, the assemblage of dinophytes with a partiform hypotheca does not represent a monophylum in the DNA tree.
Character evolution of thecate dinophytes may take place (in a top-down approach, not considering fossils) from ancestral conditions towards thecal patterns with only five postcingular plates (in Peridiniales) and more asymmetrical arrangements (in Gonyaulacales) as derived character states, respectively (Fig. 9). The reduction of postcingular plates in Peridiniales (probably a fusion of plates 3′′′ and 4′′′ that are still present in extant cladopyxidoid dinophytes; Fig. 1) is most likely an independent evolution of some gonyaulacalean dinophytes [47][48][49] , in which size reduction of the proximate postcingular plate may also lead to fewer elements in this series. In any case, there is no such thing as a newly created posterior intercalary plate in Gonyaulacales, but there are rather two antapical plates (one of which is shifted towards the ventral side 50 ) that appear homologous to those of the Peridiniales 23 (Fig. 1B). The continuous but incorrect systematic placement of dinophytes exhibiting the partiform hypotheca into the Gonyaulacales may have prevented an easier but even more parsimonious interpretation of the data.
Epithecal configurations. If a symmetrical hypotheca configuration is an ancestral character state in thecate dinophytes, then the question arises whether the epitheca also displays ancestral characters. The almost symmetrical epithecal plate pattern of Acanthodinium 42 and C. brachiolata 12 that have three anterior intercalary plates is reminiscent of the Peridiniales and likely plesiomorphic. Cladopyxis hemibrachiata 12 shares an asymmetric epithecal plate pattern and four anterior epithecal plates with F. setacea, gen. & sp. nov., which are probably derived states, but the species differ in their thecal processes (see "Diversity of thecal processes").
In the APC, F. setacea, gen. & sp. nov., only has a pore platelet and no canal plate X, whereas other dinophytes with a partiform hypotheca, such as Amphidoma F.Stein and Azadinium Elbr. & Tillmann, show the peridinioid configuration with both plates present 24 . One of the most intriguing traits is the presence of anterior intercalary plates in many Peridiniales that are notably rare in Gonyaulacales; if present, they are in unusual dinophytes, such as Pyrophacus F.Stein, or in early lineages, such as †Lingulodiniaceae 21 . Fensomea setacea, gen. & sp. nov., also has intercalary plates, which is reminiscent of the Peridiniales. The topology may provide evidence of homology between the intercalary plates of F. setacea, gen. & sp. nov., Amphidomataceae, and Peridiniales, but whether they correspond to small and rare plates in Gonyaulacales 8,21 is an area for future research.
Fossils assigned to Cladopyxidaceae are found in Early Jurassic through Palaeocene marine strata 8,16,30 and had the highest diversity in the Mesozoic. Notably, no representative fossils are known from the Quaternary, including modern deposits 14 Figure 9. Summary cladogram of thecate dinophytes, excluding specialized forms such as dinophysalean, prorocentralean and suessialean dinophytes, with apomorphies indicated as red boxes and ancestral state as a grey box. Note that only the ancestral sexiform stage is considered here from which the corniform and quinqueform states derived within Gonyaulacales 21 . Taxonomic activity. We cannot discount that the original drawings 10 (Fig. S1) represent more than a single species, and F. von Stein himself tentatively associated his new species C. brachiolata with Xanthidium furcatum Ehrenb. and †Xanthidium ramosum Ehrenb., respectively. The first (non-fossil) name is accepted today for a desmidiacean green algae, namely Staurastrum furcatum (Ehrenb.) Bréb., to which Christian G. Ehrenberg erroneously assigned also Cretaceous fossils 53 (the misinterpretation of the name was sustained over a long period of time 54 ). However, Figs. 12 and 13 10 may in fact represent the gonyaulacacean †Spiniferites mirabilis (M. Rossignol) Sarjeant or a similar species. Below, we lectotypify C. brachiolata based on one of the original illustrations. The cell selected for this taxonomic act corresponds to the most widely applied concept of the species 8,12,33 . The lectotype designated here should be substantiated by epitypification based on newly collected material and investigated with contemporary methods, such as electron microscopy and molecular sequence diagnostics, to ensure unambiguity of the name's application.
It is noteworthy that F.  Zacharias) and/or protists from other organismal lineages (e.g., pl. II 12-13, see above) and are to be disregarded (ICN Arts 8. 2, 9.14). Figure S1. This taxonomic act has been registered in PhycoBank under http:// phyco bank. org/ 102641. Notes: Delimitation from other cladopyxidoid dinophytes (i.e., a diagnosis) is provided in the "Discussion" (particularly "Diversity of thecal processes"). We consider the slender processes (here referred to as setae) to be the most striking diagnostic trait to delimit our new taxon from previously described ones. The generic name honours Robert A. Fensome, who contributed enormously to the knowledge of extant and fossil dinophytes and who accentuated the phylogenetic importance of cladopyxidoid dinophytes as a link between the Gonyaulacales and the Peridiniales 15 .

Methods
Collection, strain establishment, and morphology. Strain GeoB*184 was established from a single cell recovered from material from the South Atlantic (R/V Meteor cruise 46/4 58 ; wheel pump 3/7/a; ca 31° 25′ S, 37° 31′ W; 23 °C surface temperature; salinity: 35.5) collected on 7 March 2000. The phototrophic strain was maintained in a Percival I-36VL climate chamber (CLF Plant-Climatics; Emersacker, Germany) at 23 °C, 80 µmol photons m −2 s −1 , and a 12:12 h light:dark photoperiod, using K-Medium without silicate 59 in 35 psu artificial seawater at pH 8.0-8.2 60 . The strain decayed several months after its establishment so living material could not be inspected; unfortunately, this decay occurred before its taxonomic features were fully clarified. Therefore, we studied material, using our standard laboratory procedures, that had previously been fixed (almost 20 years before) with formaldehyde (2% final concentration).
Our LM work used an Axioskop 2 microscope (Zeiss, Göttingen, Germany) with differential interference contrast (DIC) and 1000× magnification. Cells were documented with a digital camera (MRC5, Zeiss www.nature.com/scientificreports/ SEM observations, cells were collected on polycarbonate filters (Millipore, 25 mm diameter, 3 µm pore size) in a filter funnel, in which all subsequent washing and dehydration steps were carried out. Eight washing steps (2 ml of MilliQ-deionized water each) were followed by a series of dehydration steps in ethanol (30%, 50%, 70%, 80%, 95%, and 100% at 10 min for each step). Filters were dehydrated with hexamethyldisilazane (HMDS) in 1:1 HMDS:EtOH and then twice in 100% HMDS and stored in a desiccator under a vacuum. Finally, the filters were mounted on stubs, sputter-coated (SC500, Emscope, Ashford, UK) with gold-palladium and viewed with a Quanta FEG 200 SEM (FEI, Eindhoven, Netherlands). Some SEM micrographs were presented on a black background using Adobe Photoshop 6.0 (Adobe Systems, San Jose, California, USA). The labelling of dinophyte thecal plates was performed according to the Kofoidean system 22,23 . Molecular phylogenetics. For DNA isolation, fresh material was processed using the the Nucleo Spin Plant II Kit (Machery-Nagel; Düren, Germany). Various loci of the rRNA operon (i.e., SSU, ITS, LSU) were amplified using primer pairs specified previously and following standard protocols 18,49 . Gel electrophoresis yielded single bands that were purified. PCR products were sequenced directly in both directions using the ABI Big-Dye dye-terminator technique (Applied Biosystems; Foster City, USA-CA), according to the manufacturer's recommendations, and a ABI 3730 capillary sequencer (Applied Biosystems). Sequences were edited and assembled using Sequencher™v5.1 (Gene Codes; Ann Arbor, USA-MI). For visual comparison of the edited sequences, the alignment editor 'Se-Al' (http:// tree. bio. ed. ac. uk/ softw are/ seal/) was used.
To compute a dinophyte reference tree inferred from a concatenated rRNA alignment 29,61 , we compiled a systematically representative set comprising 152 dinophytes (plus nine outgroup accessions; Table S1). For alignment constitution, separate matrices of the rRNA operon were constructed, aligned using 'MAFFT' v6.502a 62 , and the −qinsi option to consider the secondary structure, and concatenated afterwards. The aligned matrices are available as Fensomea.nex file in the supplement.
Phylogenetic analyses were carried out using Maximum Likelihood (ML) and Bayesian approaches, as described 63 , using the resources available from the CIPRES Science Gateway 64 . Briefly, the Bayesian analysis was performed using 'MrBayes' v3.2.7a 65 (freely available at http:// mrbay es. sourc eforge. net/ downl oad. php) under the GTR + Γ substitution model and the random-addition-sequence method with 10 replicates. We ran two independent analyses of four chains (one cold and three heated) with 20,000,000 generations, sampled every 1000th cycle, with an appropriate burn-in (10%) inferred from evaluation of the trace files using Tracer v1.7.1 66 . For the ML calculations, the MPI version of 'RAxML' v8.2.4 67 (freely available at http:// www. exeli xislab. org/) was applied using the GTR + Γ substitution model under the CAT approximation. We determined the best-scoring ML tree and performed 1000 non-parametric bootstrap replicates (rapid analysis) in a single step. Statistical support values (LBS: ML bootstrap support; BPP: Bayesian posterior probabilities) were drawn on the resulting, best-scoring tree.