A Description of Biremis panamae sp. nov., a New Diatom Species from the Marine Littoral, with an Account of the Phylogenetic Position of Biremis D.G. Mann et E.J. Cox (Bacillariophyceae)

Here we present a formal description of Biremis panamae Barka, Witkowski et Weisenborn sp. nov., which was isolated from the marine littoral environment of the Pacific Ocean coast of Panama. The description is based on morphology (light and electron microscopy) and the rbcL, psbC and SSU sequences of one clone of this species. The new species is included in Biremis due to its morphological features; i.e. two marginal rows of foramina, chambered striae, and girdle composed of numerous punctate copulae. The new species also possesses a striated valve face which is not seen in most known representatives of marine littoral Biremis species. In this study we also present the relationship of Biremis to other taxa using morphology, DNA sequence data and observations of auxosporulation. Our results based on these three sources point to an evolutionary relationship between Biremis, Neidium and Scoliopleura. The unusual silicified incunabular caps present in them are known otherwise only in Muelleria, which is probably related to the Neidiaceae and Scoliotropidaceae. We also discuss the relationship between Biremis and the recently described Labellicula and Olifantiella.

The original separation of Biremis from other naviculoid genera was based on its frustule and chloroplast morphology. The following criteria were regarded as of primary significance for Biremis by Round et al. 1990 ( [1]) and ( [2]): -the external appearance of the striae, containing large unoccluded rounded foramina arranged in two rows, one on the valve face and the other on the valve mantle -hence the genus name Biremis, referring to the fast warships of ancient Greece and Rome with their two ranks of oars. -a valve interior with transapically-elongate chambers with finely porous, sievelike internal walls, similar to those present in Scoliopleura Grunow, Scoliotropis Cleve and Progonoia Schrader (cf. [14,1]). This feature caused the genus to be included in the family Scoliotropidaceae as circumscribed by Round et al. 1990 ( [1]). -a girdle composed of several open copulae bearing one or a few rows of poroids. -usually two chloroplasts per cell, disposed one towards each pole as in Nitzschia and some Amphora (e.g. A. ostrearia Brébisson: [15], rather than the side-byside arrangement that is more common in naviculoid diatoms. Each chloroplast consists of two large plates connected by a narrow bridge containing a single compact, ¡ isodiametric pyrenoid. However, the morphology of taxa included in Biremis shows some variability in published sources. So far, three types of morphology can be distinguished. Type 1 has been found only in freshwater forms. Here the external foramina on the valve face are elongate and slit-like -either straight (e.g. B. hartzii W. Vyverman, Sabbe et R. Vyverman), curved (e.g. B. leeawuliana W. Vyverman, Sabbe et R. Vyverman) or wavy (B. clarensis W. Vyverman, Sabbe et R. Vyverman) -rather than round or elliptical. The valve face is free of areolae, apart from the foramina, so that in the light microscope (LM) valves appear to have a wide plain axial area. The second and third types of morphology are represented in marine species. Type 2 morphology conforms to the generic type and characterizes most of the Biremis species that have been described thus far. In it, the valve possesses the usual two rows of rounded foramina externally, with one of them located along the margin of the valve face and the second one along the valve mantle, and the valve face is free of areolae (see [1], [2], [10]), creating a wide plain axial area as in type 1. Type 3 morphology has been observed in a few taxa, of which only one has been formally described, i.e. B. solitaria (Cleve) Witkowski, Lange-Bertalot & Metzeltin (5 Navicula solitaria of Cleve 1896 [16]), and two are unnamed, one described from the Gulf of Gdań sk ( [11]) and the other illustrated from North Wales ([2], Fig. 3: 22,24,26). In these species, in addition to the two rows of foramina located along the margin of the valve face and along the mantle, the valves possess areolae on the valve face.
Biremis panamae, described for the first time in this paper from the tropical Pacific Ocean (coast of Panama), possesses this third type of morphology, whereas a second species, whose rbcL was sequenced earlier for us by G.R. Simpson, M. Hollingsworth and A. Clark (results included here), possesses type 2 morphology. These two species therefore provide an excellent opportunity to use morphological and molecular data to establish the phylogenetic position of Biremis. The aim of this paper is therefore to formally describe B. panamae and use morphological, gene sequence, and auxosporulation data to explore the relationship between Biremis and other raphid diatoms.

Materials and Methods
Isolation, culturing and microscopy of vegetative material A clonal culture was isolated from specimens collected on January 20 th 2011 along the drift line in the Pacific Ocean on the public beach "Playa Monagre" in Los Santos Province, Panama (7˚589 32.460 S and 80˚209 50.310 W, Fig. 1). No specific permissions were required for this location and sampling activity. The field studies did not involve endangered or protected species. The Gulf of Panama, where the sample was collected, is considered a large semi-open upwelling area. The average surface water temperature is about 26˚C but temperatures, as well as the distributions of nutrients in the Gulf, change due to seasonal wind-driven upwelling events. Thus, during the dry season, water temperatures can decrease to 16˚C while nutrients, e.g. NO 3 2 and PO 4 32 can reach relatively high concentrations compared to the rainy season ( [17]).
Isolates were inoculated into enriched seawater (f/2)-medium [18] and maintained at 18˚C under a 16 h light 2 8 h dark cycle, illuminated by 50 mmol m 22 s 21 of white light for about six weeks. After the cultures became visibly coloured, algal cell counts were determined using a Thoma counting chamber (Hecht-Assistent, Sondheim, Germany) and approximately 250 algal cells were streaked out on f/2-agar plates containing 1.5% (w/v) agar (AppliChem, Darmstadt, Germany). After 2-3 weeks of further incubation at the described culture conditions, single cell colonies were picked from the plates to obtain unialgal cell cultures. For all further experiments, the isolated diatoms were grown for 14 days in batch cultures in liquid f/2 medium in the same culture conditions as described above. Live cells were photographed in counting chambers using a Nikon TS300 inverted microscope (Nikon Corporation, Tokyo, Japan) equipped with a 1006 PlanApochromatic oil immersion lens (n.a.51.40) and differential interference contrast (DIC) optics.
For light microscopy (LM) and electron microscopy (EM) observations, samples were cleaned of organic material by boiling the cultured cell suspension in a few tens of millitres of 30% hydrogen peroxide for a few hours, followed by addition of ca. 10 ml of 10% HCl to remove the calcium carbonate. After oxidation, cleaned samples were repeatedly rinsed with deionized water. Diatom suspension was then pipetted onto ethanol-cleaned cover slips and left to dry in the air. Naphrax (Brunel Microscopes Ltd, Wiltshire, U.K.) was used as a mounting medium for LM observations of cleaned material, which were conducted with a Zeiss Axioscope (Carl Zeiss, Jena, Germany) using phase contrast (PhC) and DIC with a 1006 oil immersion objective (n.a.51.40). Ultrastructural analysis was made with scanning and transmission electron microscopy (SEM and TEM, respectively). For SEM examination, a drop of the cleaned sample was filtered onto Whatman Nuclepore polycarbonate membranes (Fisher Scientific, Schwerte, Germany). Filters were air-dried overnight, mounted onto aluminum stubs, coated with gold-palladium or osmium. SEM observations were made at the Goethe University in Frankfurt am Main using a Hitachi S-4500 (Hitachi, Tokyo, Japan). SEM and TEM observations were done at the Warsaw University of Technology, Faculty of Materials Science and Engineering, using a Hitachi SEM/STEM S-5500, in which the specimens were simultaneously observed in scanning and transmission mode. As the newly described species has a very small size, we have supplemented our own LM characteristics with a few images kindly taken by ing. Wulff Herwig using his advanced light photomicrography system, for which a detailed description is presented at http://www. microscopy-uk.org.uk/mag/artmar11/Advanced_Light_Photomicrography.pdf. ( [19]).

Auxosporulation
Auxosporulation was studied in a Biremis sp. obtained in January and March 1990 from the sublittoral of Loch Goil at 56˚7.79 N, 4˚54.29 W, ca. 5 km from Lochgoilhead, Argyll, W Scotland. Details of the collection site are given by ( [20]) and ( [21]). Samples were collected manually by sub aqua divers from fine silty sand at 9-12 m below mean low water, spring tides, and epipelon was harvested using 24 6 50 mm cover slips to trap diatoms moving up towards the light through lens tissue placed on sediment from which the excess water had been removed by suction (after the samples had stood for several hours). The coverslips and trapped epipelon were then incubated for up to three weeks (in the laboratory or an incubator) in enriched seawater medium, as described by ([21]). For incubations of more than a week, the medium was changed every 3-5 days.
Auxosporulation occurred after a week or more of incubation and was examined in situ on cover-slips incubated as above, after removing and mounting the slips on enriched seawater medium on a microscope slide; each cover-slip was ringed with Vaseline to prevent evaporation during a few hours of observation. Vegetative and auxosporulating cells were photographed on Kodak Technical Pan film (Estar base), using a Reichert Polyvar photomicroscope (1006 objective, n.a. 1.32), or drawn with the aid of the Polyvar's drawing attachment. Selected line drawings were used by Mann ([22], Fig. 13-15) in a publication reviewing patterns of sexual reproduction in diatoms and so are not included here. The terminology of sexual reproduction and auxospores follows ( [23]). and two chloroplast-encoded gene (rbcL, psbC) were amplified from genomic DNA using the proof reading polymerase Phusion (Finnzymes, Thermo Scientific, Schwerte, Germany) following the PCR protocol as described in ( [24,25]). The primers used for amplification are listed in Table 1. PCR products were visualized in a 1% agarose gel and then purified using Exo/Sap enzyme mixture (Thermo Scientific Fermentas) and sent to oligo.pl DNA Sequencing Laboratory IBB PAS, Warsaw, Poland for Sanger sequencing with use of BigDye Terminator v. 3.1 chemistry and ABI3730 xl sequencer. For rbcL, purified PCR product has been sent to MWG Operon (now Eurofins Genomics Ebersberg, Germany) for sequencing on ABI 3730 xl machine.

Phylogenetic analysis
Analysis of the three-gene (SSU, rbcL and psbC) dataset was performed using 73 diatom taxa (Table 2), using Ctenophora pulchella (Ralfs ex Kützing) D.M. Williams & F.E. Round and Tabularia cf. tabulata (C. Agardh) Snoeijs as the outgroups. GenBank accession numbers for the sequences used can be found in Table 2. Sequences were aligned by eye using Mesquite version 2.75; alignments are available in Appendix S1. Prior to phylogenetic analysis, the SSU rDNA inserts of B. panamae were excluded and the terminal ends of all sequences were trimmed to have the same length (aligments with B. panamae inserts are included in Appendix S2). The data were partitioned by gene and by codon position (in the case of the chloroplast markers) with a GTR+G+I model. Maximum Likelihood (ML) analyses were run using RAxML v. 7.2.6 ( [26]). The analysis consisted of multiple runs (100), each with 1000 bootstrap replicates and the tree with the best log likelihood score was chosen as our maximum likelihood estimate (see results section).
Two Bayesian Inference analyses, each with 3 chains, were run with MrBayes v.3.2 ([27]), using a 6-substitution model, partitioning the dataset by gene and by codon. Thirty million generations were run per analysis and all but the final million generations were discarded as ''burn-in''; the remaining trees were compared in Tracer v.1.5 ( [28]) for evidence of convergence. These final 10,000    trees were used to generate a majority rule consensus tree and obtain posterior probabilities for nodes (see results ection).

Nomenclature
The electronic version of this article in Portable Document Format (PDF) in a work with an ISSN or ISBN will represent a published work according to the International Code of Nomenclature for algae, fungi, and plants, and hence the new names contained in the electronic publication of a PLOS ONE article are effectively published under that Code from the electronic edition alone, so there is no longer any need to provide printed copies. The online version of this work is archived and available from the following digital repositories: PubMed Central and LOCKSS.

Taxonomic treatment
Order: Naviculales Suborder: Neidiineae Family: Scoliotropidaceae Biremis panamae Barka, Witkowski & Weisenborn, sp. nov.  doi:10.1371/journal.pone.0114508.g002 (Fig. 2H). Plastids plate-like, two per cell (sometimes only one?), arranged diagonally, or with one towards each apex ( Fig. 2A-G). Valves elliptical to linear elliptical with broadly rounded apices, 4.0-7.0 mm long and 2.0-2.6 mm wide. Axial area barely distinguishable, linear; central area distinguishable. Raphe straight, filiform, with external central endings very close to each other; apical endings not observable in LM (Fig. 2I-O). Transapical striae distinct, 17-20 in 10 mm, in the middle parallel, becoming slightly radiate towards apices. With bright-field optics, valves appear like a tiny Navicula; however, with phase contrast optics, the chambered nature of the striae is recognizable (Fig. 2K). The complex nature of the valve structure is also detectable with advanced light microphotography, where the rows of marginal areolae can be observed in valve view (Fig. 2I) and especially in girdle view (Fig. 2H).
Valve exterior in EM. Fig. 3A-D, Fig. 4A-D, Fig. 5A-D The valve face is slightly but distinctly domed. The transition between valve mantle and valve face is fairly abrupt and is marked by a narrow hyaline strip,  (Fig. 3B, D, Fig. 4A, B). The axial area is narrow and linear and the central area very small and asymmetrical, comprising only a slight shortening of a central stria on the valve's secondary side. The raphe-sternum is slightly elevated above the general valve surface. The raphe is filiform, straight to slightly bent, with the external central endings very close to each other, slightly expanded, and straight or bent slightly towards the secondary side of the valve. The external apical raphe endings are strongly hooked in the same direction and terminate on the valve mantle in a small groove in line with the row of mantle foramina (Fig. 3A, B, 4A-C). The girdle is composed of several copulae, each finely perforated (Fig. 4A). TEM observations show that the transapically elongate areolae of the proximal part of the valve face are covered by a delicate membrane of porous silica with tiny round holes resembling the hymenate occlusions of other raphid diatoms ( [29]). This fine membrane, which lies slightly below the remainder of the valve face Description Biremis panamae and Phylogenetic Position of Biremis externally, is lost in eroded specimens, revealing the shape of the areolae and also that the areolae are wider than the virgae (transapical ribs, Fig. 3B, Fig. 4A, B, Fig. 5A-D).  Fig. 6A and the complete specimen in Fig. 6B) and eroded specimens (Fig. 6A) show that it extends from the edge of one valve face areola, beneath the hyaline strip at the margin of valve face and mantle, to the valve margin; it opens to the exterior by two foramina (compare Fig. 4A-C). In other Biremis species, the interior wall of the chambers is porous (e.g. [1]) but we did not detect pores in our material. The raphe-sternum is somewhat elevated internally and the raphe slit is straight and simple. The internal central endings are simple and coaxial and terminate in a 'double helictoglossa', i.e. a slightly elongate, almost beak-like mass of silica capping the raphe slits. Likewise, the apical internal endings terminate in a very small and simple helictoglossa (Fig. 6A-D).

Comparison with established taxa
Most of the Biremis species described from marine or brackish habitats are much larger than B. panamae and could not be confused with it, likewise the freshwater taxa. The only similar species in terms of size is B. lucens, though its valves are always longer and broader (7.6-25.5 mm long and 2.5-5.6 mm wide) and the striae Auxosporulation in Biremis sp. Fig. 7A-J In order to help place Biremis phylogenetically and taxonomically, we studied auxosporulation in an unnamed Biremis species (Fig. 7). Cells almost always lay in girdle view (e.g. Fig. 7A, B) because the valves were much narrower than the girdle. Valves had marginal chambers and two rows of external foramina (visible in Fig. 7G, I, as in B. panamae). Living cells contained two chloroplasts, one towards each end of the cell (Fig.7A, B). Each chloroplast comprised two X-or butterfly-shaped plates, one lying against each side of the girdle (Fig. 7A), connected by a narrow bridge containing a compact, ¡isodiametric pyrenoid (Fig. 7B). In valve view each chloroplasts resembled a narrow 'H' (not illustrated), with the pyrenoid occupying the cross-bar.
Prior to sexual reproduction, cells became paired, usually lying with their girdles adjacent, and entered meiosis. Fig. 7C and D are two focuses (C above, D below) of a single pair of cells (gametangia) that had completed meiosis I and produced two gametes per cell. In the pair shown in Fig. 7E, on the other hand, the cells lay displaced from each other, with only the valves touching. The gametes were formed by cleavage of the gametangium parallel to the valves after meiosis I (not shown: see [22], Fig 13). Subsequently, the gametes rounded up and swiveled around to lie one towards each end of the gametangium (Fig. 7C, D). During fertilization, one gamete moved out from its gametangium and into the other gametangium, so that finally each gametangium contained a single subspherical zygote, which contained two nuclei (Fig. 7E, left), the other two nuclei from meiosis having previously degenerated, and also four chloroplasts (see also [22], Fig. 13-15). There was then a phase during which the zygote matured, becoming elliptical and acquiring a robust wall, and transformed into an auxospore. The first stage of auxospore expansion could be detected through the appearance of the primary transverse perizonial band around the equator of the cell (Fig. 7F,  arrow; cf. Fig. 7H). Auxospore expansion took place parallel to the long (apical) axis of the gametangium (Fig. 7I, J) and was accompanied by formation of further transverse perizonial bands, clearly visible as periodic thickenings of the auxospore wall (e.g. white arrows, Fig. 7J). During expansion it became obvious that the original wall of the zygote (comprising the incunabula sensu Kaczmarska et al. 2013 ([23]) had been composed of two smooth, evenly curved hemispheres (e.g. black arrows, Fig. 7I, J). These survived oxidation (by burning) to remove  Fig. 7A, the other being out of focus beneath, on the opposite side of the cell) connected by a narrow bridge containing the pyrenoid (e.g. p). C, D. Two paired gametangia, each containing two rounded, rearranged gametes. The gametangia were paired with their girdles adjacent, the cell shown in Fig. 7D lying immediately below that in Fig. 7C. E. Two paired gametangia, unusual in being in contact only via their valves. Each gametangium contains a single subspherical zygote. Two nuclei are visible in the left-hand cell (arrows) and two of the four chloroplasts in the right-hand cell. F, G. Two focuses of a gametangium containing a zygote on the point of transformation into an auxospore. Note the slight central inflection of the zygote's outline, marking the deposition of the primary transverse perizonial band (cf. Fig. 7H, arrowhead). The two rows of foramina on the valves can be seen in Fig. 2G (arrowheads). H. Expanded auxospore containing the initial epivalve (in section at arrow). The auxospore is encased in a well developed perizonium, containing a primary transverse band flanked by several secondary bands (see in section: see also Figs 7I, J). I. Peripheral focus of an expanded auxospore containing the initial epivalve. The two rows of foramina on one of the gametangium valves can be seen (arrowheads). The end of the auxospore is covered by a siliceous cap (arrow). J. Expanded auxospore containing a completed initial cell. The initial hypovalve (in section at h) lies at a distance from the perizonium, as a result of a strong contraction of the protoplast immediately before its formation; the initial epivalve lies opposite, directly moulded by the interior of the perizonium. The auxospore casing can be seen to consist of a perizonium of transverse bands (e.g. at white arrows) and two silicified hemispherical caps (e.g. at black arrow). [Scale bar 10 mm].
Once expansion was complete, the initial epivalve was formed within the perizonium, beneath one of the valves of the gametangium. The initial epivalve was closely appressed to the perizonium over its whole length, except at the centre, where there was a slight contraction (Fig. 7H). In contrast, the initial hypovalve was formed after a marked contraction of the cell away, so that there was a wide space between it and the perizonium (Fig. 7J). Expanded auxospores were ca. 30.0-32.0 mm long and gametangia 14.0-18.0 mm.

Phylogeny
The size of the amplified B. panamae SSU rDNA gene was found to be ca. 3300 bp, whereas we expected its size to be ca. 1700 bp, based on the published sequences of other raphid diatoms. Through alignment of our SSU rDNA sequences to existing SSU sequences of diatoms, we discovered the presence of 12 inserts with a total length of ca. 1500 bp (Appendix S2). In a three-gene maximum likelihood (ML) and Bayesian Inference (BI) phylogenetic analysis (Fig. 8, 9 respectively) Biremis appeared most closely related to Neidium and Scoliopleura. The two Biremis spp. formed a grade, with Biremis sp. sister to the Scoliopleura-Neidium clade (with low support: 70% ML bootstrap, 0.73 BI posterior probability) and B. panamae sister to the Biremis sp.-Scoliopleura-Neidium assemblage (again with rather low support: 50% bootstrap, 0.99 posterior probability). As for the relationship of the Biremis-Scoliopleura-Neidium clade to the rest of the raphid pennate diatoms, not much can be said at this point. In both ML and BI analyses, the Biremis clade was in an unresolved trichotomy sister to the Navicula clade (ML; 95% bootstrap) or Diploneis (BI; 1.00 posterior probability). It should be noted that Biremis sp. is represented by a single rbcL sequence, which likely affects the support values in the Biremis clade and may explain why the two Biremis strains do not form a clade. However, when we performed a RAxML analysis based only on the rbcL alignment the topology of the Biremis-Scoliopleura-Neidium clade was almost identical to that in the three gene tree (the rbcL tree is included in the supplementary data), with Biremis panamae sister to the Scoliopleura-Neidium clade and Biremis sp. sister to that clade. The bootstrap support was lower than that in the three-gene tree (less than 50, Appendix S3).

Discussion
Superficially, B. panamae looks in LM like one of the numerous tiny Navicula species found in fresh and marine waters. Only the high magnification LM images (Fig. 2H-N) taken with a phase contrast or examination with EM reveal the true structure of the valve and identify the species as belonging to Biremis. The most characteristic feature is the stria structure, with aligned chambers forming a  of apically elongate foramina near the margin of the valve face (which open into the marginal chambers) and the raphe-sternum. Until now, two types of valve face have been observed in marine Biremis: in one the valve face is plain (B. ambigua and B. lucens: [1], [2], [10]), while in the second the valve face is striated (e.g. [2], [11]). The Biremis species in which we observed auxosporulation belongs to the first type (our unpublished observations), whereas B. panamae clearly belongs to the second type, which was previously known only in an unnamed Biremis species ([2], Fig. 3: 22, 24, 26), B. solitaria, and Biremis spec. 155/2 from the Gulf of Gdań sk (the Baltic Sea: [11], Fig. 154: 17, 18 and 155: 8 respectively). Because of the striated valve face, B. panamae, B. solitaria and the two unnamed Biremis species studied by Cox (1990 [2]) and Witkowski et al. (2000, [11]) seem to provide a link between Biremis and the freshwater genus Pulchella, which was erected by Krammer (2000, [30]) for two species previous assigned to Biremis,  Fig. 9-17) and thus resembles the freshwater Biremis species described by Vyverman et al. (1997, [9]). The distinction, if any, between these two genera needs further study.
The marginal chambers of B. panamae are very similar in their general features (chambers closed internally by a delicate membrane that bulges into the cell interior; opening externally via two large foramina, which are aligned to form two longitudinal rows on each side of the valve) to the chambers seen in other Biremis and also to the more elaborate and longer chambers present in Scoliopleura, Scoliotropis and Progonoia ( [1]). In all too, the raphe terminates internally in a double helictoglossa and there are either two H-shaped chloroplasts as in Biremis (Fig. 7A, B), or four chloroplasts, like the four plate-like halves of the two chloroplasts of Biremis, but without the pyrenoid connection between them.
Biremis was established as a genus quite recently ( [1]) and currently contains only ca. 17 species (see Introduction), some described as new but most transferred from Pinnularia, Amphora, or Navicula [where they were described and illustrated by authors such as Cleve, Heiden, Hustedt and Giffen: ( [3,16,4,5,6,7]). The three gene based phylogeny resulting from our research shows, however, that Biremis is not closely related to Pinnularia or Navicula, nor to Amphora (which is not included in our tree but belongs close to the Rhopalodiales and Surirellales: [31], [32]). Instead, Biremis belongs in the same well supported clade as Neidium and Scoliopleura. The classification of Biremis and Scoliopleura in the same family (Scoliotropidaeae) was proposed without explanation by ( [1]), but was based on the marginal chambering of the valve, chloroplast structure and double helictoglossa internally at the centre (D.G. Mann, unpublished information). In these respects, morphological observations and molecular data are consistent, supporting classification of both in the same family. However, in our tree, Scoliopleura is sister, not to Biremis but to Neidium, which argues against placing this Neidium in a separate family Neidiaceae (as done by [1]), if at the same time Biremis and Scoliopleura are placed together in the Scoliotropidaceae, because it would make the Scoliotropidaceae paraphyletic.
Although the link between Biremis and Neidium may initially look surprising, there are in fact several non-molecular (cytological, raphe and auxospore) characters that are consistent with a close relationship between them: this has been the subject of several studies in the last decades (e.g. [22], D.G. Mann and G.E. Simpson, unpublished). Morphological observations reveal that the frustule structure and habitat of Neidium and Biremis differ in numerous aspects. The major differences are chambered striae, curved polar raphe endings and simple external central raphe endings in Biremis, whereas in Neidium there is a longitudinal canal at the valve face periphery, forked apical raphe endings, and oppositely curved central raphe endings (e.g. [33], [1]). In addition, Neidium is an obligately freshwater genus, whereas Biremis spans marine and freshwaters. However, despite these contrasting characteristics, Biremis and Neidium are linked by the mode of auxosporulation (discussed further below) and a cytological character, namely bilineal transmission of chloroplasts through the mitotic cell cycle: because of the way the chloroplasts divide and are inherited (documented in B. ambigua by Cox 1990 [2] and in Neidium by Mann 1984 [34], 1996 [35]), each Neidium and Biremis cell possesses one or two individual organelles of each of two 'clones' of chloroplasts. Furthermore, both genera have a double helictoglossa internally (between the central raphe endings), and in them and also in Scoliotropis, Scoliopleura and Progonoia the valves have a two-layered structure over at least part of the valve, with fine internal pore occlusions internally and larger open pores or foramina externally ( [1]). Neither bilineal transmission of chloroplasts nor the presence of a double helictoglossa is unique to Neidium, Biremis and related taxa, but they are unusual in raphid diatoms ( [35], D.G. Mann unpublished observations) and tend to corroborate the molecular phylogeny.
With respect to auxosporulation, ( [22]) has already pointed several similarities between Biremis and Neidium, especially the presence of ¡ hemispherical silicified caps over the ends of the auxospores, which are produced by the zygote before it begins to expand. The nature of the caps has since been investigated in detail by Mann & Poulíčková (2009 [36]; see also Fig. 7H-J). Although caps (derived from the incunabula of the zygote) occur on the poles of many pennate diatoms, they are usually either wholly organic (e.g. [37]) or contain small silica scales (e.g. [38], [39]): fully silicified unitary caps are known so far with certainty only in Biremis (this paper), Neidium ( [36]), probably Scoliopleura, and Muelleria ( [39]), and their restriction to these genera suggests they comprise a synapomorphy for a group at family or subordinal level. Muelleria was not considered by Round et al. (1990, [1]) but shares several characteristics of valve structure with Neidium (two-layered valves with striae of small round pores in each layer; presence of a longitudinal canal within the thickness of the valve; development in many species of a flap over the external polar raphe endings, making these appear forked; central internal raphe endings accompanied by a double helictoglossa, though this is elongate in Muelleria and almost forms two separate helictoglossae) and Spaulding and Stoermer ([41]) therefore tentatively placed the genus in the Neidiaceae (see also Van de Vijver et al. 2010 [42] for descriptions of Muelleria valve morphology). The Neidium-Biremis-Scoliopleura-Muelleria group also exhibits similarities with respect to sexual reproduction and auxosporulation, in addition to the possession of incunabular caps discussed above. These similarities include rounding off and rearrangement of the gametes before plasmogamy (e.g. Fig. 7C, D); formation of the zygotes wholly within the gametangia; expansion of the auxospores parallel to the apical axes of the gametangia; linear auxospores; a narrow primary transverse perizonial band; and, in allogamous species, trans behavioural anisogamy ( [34], [43], [36], [40]), allowing the auxosporulation pattern to be classified as type IA1a in Geitler's scheme ( [44]). However, several of these features are widespread among biraphid diatoms. For example, type IA1a allogamous auxosporulation is also found in Frustulia, Amphipleura, Gomphonema, Cymbella, Placoneis and Nitzschia ( [44], [45]). So, although the wider aspects of auxosporulation are consistent with the relationships suggested by the concatenated gene tree, they do not give it strong support.
Recently two further diatom genera showing a certain degree of similarity with  [47]). Whereas Labellicula is a monotypic genus with one species described from Subantarctic ( [46]), Olifantiella seems to be very diverse, especially in the tropical regions and coral reef habitats in particular ( [48]). Interestingly, although Labellicula and Olifantiella bear stigmata (not found in any of the Scoliotropidaceae and Neidiacaeae genera mentioned above), they nevertheless share some characters with Neidium (Labellicula, cf. [46]) and Biremis (Olifantiella, cf. [47] and [48]). The similarity between Labellicula and Neidium lies in the valve internal raphe structure, with double helictoglossae, whereas with Biremis, Labellicula shares aspects of girdle structure, composed of numerous punctate copulae. Olifantiella also has a girdle composed of numerous punctate copulae and double helictoglossae internally, but more interesting is that the chambered transapical striae resemble very much those of Biremis. However, since both Labellicula and Olifantiella show the presence of stigmata, which are missing in Biremis and Neidiaceae, further analysis is necessary before any formal conclusion is made about the relationships between them.