﻿Revision of Ardissoneaceae (Bacillariophyta, Mediophyceae) from Micronesian populations, with descriptions of two new genera, Ardissoneopsis and Grunowago, and new species in Ardissonea, Synedrosphenia and Climacosphenia

﻿Abstract Ardissonea was resurrected from Synedra in 1986 and was included as a genus by Round, Crawford and Mann (“The Diatoms”) in its own Family and Order. They commented that there might be several genera involved since the type species of the genus possesses a double-walled structure and other taxa placed in Ardissonea have only a single-walled structure. Two other genera of “big sticks,” Toxarium and Climacosphenia, were placed in their own Families and Orders but share many characters with Ardissoneaceae, especially growth from a bifacial annulus. Eighteen taxa (11 new species) from Micronesia were compared with the literature and remnant material from Grunow’s Honduras Sargassum sample to address the concepts of Ardissonea and Ardissoneaceae. Phylogenetic and morphological analyses showed three clades within Ardissonea sensu lato: Ardissonea emend. for the double-walled taxa, Synedrosphenia emend. and Ardissoneopsisgen. nov. for single-walled taxa. New species include Ardissoneadensistriatasp. nov.; Synedrospheniabikarensissp. nov., S.licmophoropsissp. nov., S.parvasp. nov., and S.rectasp. nov.; Ardissoneopsisfulgicanssp. nov., A.appressatasp. nov., and A.gracilissp. nov. Transfers include Synedrospheniacrystallinacomb. nov. and S.fulgenscomb. nov. Synedraundosa, seen for the first time in SEM in Grunow’s material, is transferred to Ardissoneopsisundosacomb. nov. Three more genera have similar structure: Toxarium, Climacosphenia and Grunowagogen. nov., erected for Synedrabacillaris and a lanceolate species, G.pacificasp. nov. Morphological characters of Toxarium in our region support separation of Toxariumhennedyanum and T.undulatum and suggest additional species here and elsewhere. Climacospheniamoniligera was not found but we clarify its characters based on the literature and distinguish C.soulonalissp. nov. from it. Climacospheniaelongata and a very long, slender C.elegantissimasp. nov., previously identified as C.elongata, were present along with C.scimiter. Morphological and molecular phylogenetics strongly suggested that all these genera belong in one family and we propose to include them in the Ardissoneacae and to reinstate the Order Ardissoneales Round.


Introduction
Bilaterally symmetrical, stick-like diatoms, 300-1300 µm long, are common and sometimes abundant epiphytes on seaweeds of coral reefs. Some have bilateral growth of the valve ("pennate" diatoms) and are in the Class Fragilariophyceae, including Stricosus Sabir & E.C. Theriot, 2018 and longer species of Hyalosynedra D.M. Williams & F.E. Round, 1986(Sabir et al. 2018). However, several common genera have valve growth from a bifacial annulus, at or near the edge of the valve face, forming striae internally and externally (Mann 1984;Kaczmarska et al. 2020) and were moved to Class Mediophyceae (bi-to multipolar diatoms) (Medlin et al. 2008) and are presently placed in four families in the Order Toxariales (Cox 2015). All are non-pennate but emerge in 7-gene phylogenetic trees in a clade sister to the pennate diatoms (Kooistra et al. 2003;Medlin et al. 2008;Theriot et al. 2015;Medlin and Desdevises 2020). Previously Round et al. (1990) had placed four genera-Ardissonea De Notaris, 1870 in De Notaris and Baglietto (1871), Toxarium J.W. Bailey, 1854, and Climacosphenia Ehrenberg, 1843, plus Synedrosphenia (H. Peragallo) Azpeitia, 1911in three Families, each in a separate Order in Fragilariophyceae. The sexual reproduction of A. crystallina (Agardh) Grunow ex De Toni, 1892 has similarities to that of Fragilariophyceae in being isogamous (Davidovich et al. 2017) and in morphology of its auxospores (Kaczmarska et al. 2018), yet the valve morphogenesis reinforces its position among Mediophyceae (Kaczmarska et al. 2020). Gliding motility is present in these diatoms, even though they lack a raphe-the structure usually associated diatom motility-and even though they are habitually sessile on mucilage stalks. Gliding was demonstrated in A. crystallina, with mucilage secretion apparently occurring via a deep groove between the valve and valvocopula at each end of the cell (Pickett-Heaps et al. 1991), and in Toxarium, which has a similar apical structure (Kooistra et al. 2003).
Each genus in the Toxariales has a confused taxonomic history that has not yet been addressed with electron microscopy and several issues need to be resolved. The genus Ardissonea, subsumed as a subgenus of Synedra Ehrenberg, 1830 and declared by Hustedt (1931Hustedt ( -1959 to be a "difficult group of … Synedra," was resurrected and emended by Poulin et al. (1986) based on SEM study, and the spelling of the name corrected (from Ardissonia) by Round et al. (1990). The generitype Ardissonea robusta (Ralfs) De Notaris, 1871 was not among the species studied by Poulin et al. (1986) but apparently was the species Round et al. (1990) used to illustrate the structure of the genus. The structure of A. formosa (Hantzsch) Grunow ex De Toni, 1892, studied by Sullivan and Wear (1995), differs significantly from the taxa studied by Poulin et al. (1986Poulin et al. ( , 1987 [A. crystallina and A. fulgens (Greville) Grunow ex De Toni, 1892] in having a double wall and unperforated girdle bands. On the other hand, the genus Synedrosphenia, as illustrated by Round et al. (1990), while having a structure very similar to Ardissonea species shown by Poulin et al. (1986), has heteropolar valves and was classified by Round et al. (1990) with the heteropolar genus Climacosphenia in the Climacospheniaceae. Round et al. (1990: 420) commented about Ardissonea that, "It may prove best to place the non-chambered species in yet another genus," but this has not yet been done. The definition of the genus has become confused because while Poulin et al. (1986) based their emendation on A. crystallina and characterized the transapical costae in these single-walled taxa as providing "chambering," Round et al. (1990) based their description on A. robusta, the generitype, and thus included the internal plate and complete chambering of the valve as a character. Kanjer et al. (2021) regarded A. fulgens as "differing only slightly" from the generitype in lacking the internal plate.
The genera Toxarium and Climacosphenia differ from the others in different ways. Climacosphenia was formerly classified alongside Licmophora, while Toxarium was in Synedra (Peragallo andPeragallo 1897-1908;Hustedt 1931Hustedt -1959. Toxarium is defined by the presence of scattered areolae within the annulus and Climacosphenia by the presence of craticular bars on the valvocopula. Both have a history of debate over whether there are one or two species of each, distinguished only by cell shape. This led to an assumption that there are only two species in each to choose from, i.e., Toxarium undulatum J.W. Bailey, 1854 (generitype) and T. hennedyanum (Gregory) Pelletan, 1889; Climacosphenia moniligera Ehrenberg, 1843 (generitype) and C. elongata J.W. Bailey, 1854. Finally, one seemingly less similar taxon that occurs among Ardissonea, Toxarium and Climacosphenia spp. in our flora has been reported as Synedra bacillaris (Grunow) Hustedt, 1932(= Synedra crystallina var. bacillaris Grunow, 1880  ; this species has a distinct central costa and apparently lacks an annulus. Even though many marine species formerly in Synedra have been removed to new genera (Williams and Round 1986), S. bacillaris has yet to be revised, despite Sullivan and Wear's (1995) observations.
The initial objective of this study was to describe several new species of Ardissonea, but in our morphological study for taxonomy and floristics we were led to address the systematics of the whole group. We here present descriptions of the taxa in Micronesia, including new genera and new species, plus some related species from Grunow's (1867, 1877) Honduras materials. We analyze sequence data and morphological characters, make the case for including Toxarium and Climacosphenia in Ardissoneacae, within a reinstated Ardissoneales Round, and, finally, make nomenclatural changes that reflect our conclusions. These include transfer of single-walled Ardissonea species into Synedrosphenia and a new genus Ardissoneopsis, amending the generic descriptions as needed, and a new genus for Synedra bacillaris.

Sample collection and imaging
Samples from our extensive collection of benthic diatoms from Micronesian coral reef habitats, preserved and prepared using the protocols standard in the U. Guam diatoms laboratory (Lobban 2021b), were surveyed for taxa of interest. Origins of the samples used in this study are listed in Table 1. Some material was observed alive, and often then transferred as a whole mount onto a filter, dried, and mounted for SEM; this enabled us to study the girdle bands in situ. In addition, we were able to observe the rich material in Grunow's Honduras sample, collected by A. Lindig ca. 1860 and published in Grunow (1867Grunow ( , 1877. They were unmounted remnants of acid cleaned material he used for preparation of slides of other species; these were listed in Lobban et al. (2021a, table 1).
Light microscopy was done on Nikon 80i microscopes with differential interference contrast, imaging with the Nikon DS-Fi1 camera system. Some SEM images were taken with a PhenomWorld G2Pro desktop SEM as in previous work, but many final images, including those using a eucentric (tilting) stage and all those of Toxarium and Climacosphenia, were made on a Thermo-Fisher Phenom XL SEM (NanoScience Instruments, Phoenix, AZ) after recoating stubs with gold using a Luxor Gold sputtercoater (NanoScience Instruments, Phoenix, AZ). The materials observed were valves or frustules on strewn slides and SEM stubs in the diatom collection of the University of Guam Herbarium (GUAM), linked to collection data by the sample number and are referenced here by that sample number. Cataloguing of selected, imaged specimens of individual specimens has begun, using the Specify™ software, but is only in the earliest stages, so that catalog numbers have not yet been assigned to the specimens studied, but the developing website is accessible via the Guam Ecosystems Collaboratorium for Corals and Oceans (GECCO) Biorepository Specify Portal at https://specifyportal.uog.edu.

DNA isolation, amplification and sequencing
Culture conditions for the strains sequenced and photographed in this study were dependent on where the strains were isolated. Strains isolated from Guam and Saudi Arabia were maintained in a Percival model I-36LL incubation chamber (Percival, Boone, Iowa, USA) at 27 °C under a 12:12 hr light:dark period. Strains isolated from Florida and North Carolina were maintained on a lab bench between 20-24 °C, illuminated by a north-facing window. DNA sequencing of these strains was conducted using the DNeasy Plant Minikit, adding a 45 s incubation in a Beadbeater (Biospec Products, Bartlesville, OK, USA) with 1.0 mm glass pellets for cell disruption. Three markersthe nuclear-encoded ribosomal SSU and chloroplast-encoded rbcL and psbC-were amplified by PCR, purified and sequenced following the protocols outlined in Lobban et al. (2021b). Sequence data were added to a dataset of diatoms, with Bolidomonas pacfica Guillou & Chrétiennot-Dinet, 1999 used as an outgroup (see Suppl. material 1: Table S1 for GenBank accession numbers). Data were initially partitioned by gene, by paired and unpaired sites in SSU secondary structure (determined by the SSUalign program following Lobban et al. 2019) and codon position in rbcL and psbC. Model and partition testing were performed using PartitionFinder 2 with the best model chosen using the corrected Akaike information criterion (AICc). The dataset and partitioning scheme were analyzed under maximum likelihood using RAxML ver. 8.2.7 (Stamatakis 2014) compiled as the pthread-AVX version on an Intel i7 based processor in Linux Mint 18 and IQ-TREE version 1.6.12 for Linux. For RAxML, we ran 15 replicates with 400 rapid BS replicates each with ML optimizations and 4 replicates each with 1500 rapid BS replicates with ML optimization, with bootstrap support assessed by the BS replicates from the run which produced the optimal ML score. Bootstrap support for the IQ-TREE analysis was assessed from 1000 replicates. A Bayesian Inference based phylogeny was inferred by a multi-threaded MPI hybrid variant of ExaBayes version 1.5, using four independent runs with two coupled chains where branch lengths were linked. Convergence parameters monitored included the average deviation of split frequencies of less than or equal to 5% with a minimum of 10,000,000 generations. Bayesian nodal support was assessed using posterior probabilities, with the first 25% of the trees removed as "burn-in".

Morphological analysis
Patterson (1982,1988) proposed three tests of homology: similarity, conjunction, and congruence. He further proposed that similarity in some aspect of shape, form, or development was sufficient to postulate homology, with congruence as the decisive test, which we take to mean that similarity is the first test. Cracraft (1981) even more strongly argued that similarity was the first step in testing homology, writing that we are compelled to postulate homology for observed similarity. Following that philosophy, we coded multistate characters as additive where such nested similarity was apparent, an approach simultaneously and independently argued by Wilkinson (1992), Lipscomb (1992) and Theriot (1992). Such a treatment does not rely on assumptions about character evolution but instead relies on the principle of parsimony in the same way homology was initially postulated for binary characters (Sober 1983); that is, multistate characters represent nested sets of simi-larity (and in fact, can be represented as a series of binary characters in a matrix for computer analysis.) The matrix (Suppl. material 2: Table S2) was analyzed under maximum parsimony using Winclada version 1.00.08, which serves as a tool for inputting data into a matrix, as a GUI for inputting commands into NONA version 2 for phylogenetic analysis, and as a tool for tree and character mapping visualization (Nixon 2002), with the following settings: Maximum trees to keep = 1001, Number of replications = 1000, Starting trees per hold = 1, Search strategy = Multiple TBR + TBR, and Search = unconstrained. Two characters were treated as additive: pseudoseptum (0 = absent, 1 = present, 2 = uniform rim all around valve) and valvocopula external poration (0 = absent, 1 = one row, 2 = multiple rows). Settings were heuristic search, 5000 replicates, mult*max*. Toxarium hennedeyanum was used as the outgroup.

Terminology
The frustules in the taxa treated here appear either heteropolar or isopolar, even though most are attached to the substrate at one pole by mucilage pads or stalks. Heteropolarity may be evident in the shape of the valve (cuneate or clavate), and/ or in a cuneate girdle view caused by tapering girdle bands and narrower mantle toward the basal pole. We coded polar morphology as isopolar if the shapes were essentially similar at opposite poles. We avoid the term 'apices' because it can be ambiguous, and refer instead to 'poles'; in heteropolar species we distinguish the 'basal pole', i.e., attachment end, from the 'apical pole'. In some species the valve mantle, especially at the poles, is "recurved as a lip or poorly developed pseudoseptum" (Pickett-Heaps et al. 1991: 723); for this we use the term 'pseudoseptum' without qualification (see also Kanjer et al. 2021). In these species, the valvocopula and copula are shaped to fit around the pseudoseptum leaving a groove, the copula with very long fimbriae draped over the valvocopula, as clearly shown in Pickett-Heaps et al. (1991, fig. 18); we call this 'complex polar architecture'. In some of these species, there is a broad notch in the valvocopula at the poles, producing a tunnel against the pseudoseptum (Round et al. 1990). In yet other species the rim of the valve is straight and the pars interior at the valvocopula pole is correspondingly simpler (fitting into the valve the way the sides of a cookie tin fit into the lid); we call this 'simple polar architecture'. Most species have three girdle bands in the epicingulum and mature hypocingulum, which, following von Stosch (1975) and Sullivan and Wear (1995), are designated as valvocopula, copula and pleura; the pleura is usually very small and easily hidden or lost in preparation. In Climacosphenia, the septum of the valvocopula has a costa along each side with ingrowths forming the craticular bars, i.e., having the structure of a ladder, for which the genus is named. The ingrowths usually meet in the middle and form unions that may be seamless (Round 1982 called those "complete") or complex; sometimes two outgrowths from one side will anastomose with one from the other side. Other terminology follows Anon (1975) and the revision by Ross et al. (1979).
Observations. Valves linear with broadly rounded poles, 201-300 µm long, 22-24 µm wide, striae 8-10 in 10 µm, isopolar in both valve and girdle view (Fig. 1A). Plastids numerous, lenticular, displaying peristrophy and karyostrophy depending on light intensity (Suppl. material 3: Fig. S1). Three or five longitudinal lines seen in two focal planes (Fig. 1B, C): if five, the annulus, subtending a longitudinal costa, on the valve face (not hidden at the valve margin), a central line where the offset striae meet; between these, on the inner wall, two rows of foramina ( Fig. 1B-D). If three lines, one central row of foramina with no longitudinal costa (Fig. 1F). Two additional rows of foramina along the angle of the valve-mantle junction seen in SEM but obscured in LM (Fig. 1D, F). The pattern of costae underlying the inner wall can be traced in Fig. 1D and seen directly in Fig. 1E. We have also seen specimens with one row of foramina along the midline (Fig. 1F). Commonly one or two longitudinal depressions in valve surface, giving a shallow v (Fig. 1I) or w profile. Striae comprising single row of oval areolae between each transapical costa ( Fig. 1C, E, G); loose rows of smaller, circular areolae at the poles, especially noticeable on mantle (Fig. 1I). Narrow pseudoseptum around poles (Fig. 1D, E), elsewhere the double wall forms a thick rim (Fig. 1G). Three girdle bands as follows ( Fig. 1F-J). Valvocopula with line of small pores along edge of pars interior, usually obscured in girdle view, deflecting asymmetrically inwards at pole (Fig. 1F), corresponding to location of the notch on advalvar side, which however, is nearly central (Fig. 1I). Interior edge of valvocopula with short fimbriae (Fig. 1G, J). Copula with long fringe of fimbriae at pole, overlapping the valvocopula (Fig. 1H), as described by Sullivan and Wear (1995); an irregular line of pores along abvalvar margin of copula in some specimens (Fig. 1J, arrowhead). Narrow pleura (Fig. 1I, J) with a row of small pores on pars interior.
Taxonomic comments. The biseriate striae described by Hustedt (1931Hustedt ( -1959 and the uniseriate striae in Sullivan and Wear's (1995) materials suggest that they were observing different species. Sullivan and Wear (1995) observed only two bands on the epicingulum and none on the hypocingulum but we observed the pleura. Ardissonea s.s. superficially resembles Synedra in the current restricted sense (= Catacombas) in shape and double wall but differs in lacking rimoportulae and polar pore fields.
Holotype (designated here). Specimen at 16.6 mm E and 11.5 mm S of the mark on slide 2911, deposited at ANSP accession # ANSP-GC20089. Fig. 2A.
Registration. Etymology. Densus + striatus (Latin) = densely striated, in comparison to congeners. Taxonomic comments. Given the small number of double-walled Ardissonea species, this species is easily distinguished by the small size, lanceolate shape and relatively high stria density. It is more readily confused in our flora with Grunowago pacifica sp. nov. (see below), but in LM the latter can be distinguished by the lack of an inner wall (indicated by absence of foramina) and presence of a longitudinal costa along the midline, in SEM by the extensive poration on the copula. To date this species has not been observed in Guam samples. A different lanceolate, double-walled Ardissonea was observed in samples from Chuuk and Jaluit, with stria density 9-12 µm, length 116-211 µm, but we do not have complete information from either population. These samples are similar in shape and stria density to a Red Sea species 'Ardissonea AdnU041' observed by Sabir et al. (unpublished: http://www.protistcentral.org/Taxa/get/taxa_id/586376 and Ashworth,personal communication). This unnamed species is 83-125 µm long and has no external poration on the girdle bands (as in A. formosa). The sequence data have been published as UTKSA0041, shown in our Fig. 33, sister to A. formosa. Table 3 Ardissonea crystallina (C. Agardh) Grunow, 1880 Fig. 3 References. Peragallo andPeragallo 1897-1908, p. 310, pl. 79, fig. 1;Hustedt 1931Hustedt -1959Navarro 1982, p. 260, figs 59, 60;Poulin et al. 1986, figs 28-30;Poulin et al. 1987Poulin et al. , p. 2689Pickett-Heaps et al. 1991, figs 9-19;Lobban et al. 2012, p. 259, pl. 15, figs 1-3;Davidovich et al. 2017, figs 1, 2.

Group 2: Single-walled taxa with pseudosepta
Description from literature. Valves 200-700 µm long, 8-20 µm wide, linear but slightly wider at the middle and poles; 11 striae in 10 µm but much closer in "transitional forms such as var. dalmatica" (Hustedt 1931(Hustedt -1959. Internally with prominent costae developed on every virga except at poles, and with two longitudinal costae corresponding to location of annular ring; spines completely absent. Edge of valve recurved into a pseudoseptum; an interpretive diagram in Pickett-Heaps et al. (1991) shows the relationship between the valve, valvocopula, copula and a small, plain pleura at the pole. Valvocopula and copula have interior comb-like fringe, exterior part with four or more rows of pores arranged decussately; an apical groove between the pseudoseptum and a fold in the valvocopula provides an exit for attachment mucilage (Poulin et al. 1987;Pickett Heaps et al. 1991) but Pickett-Heaps et al. (1991 argued that this groove was a very different structure from the apical pore in A. formosa as shown by Round et al. (1990). Shorenko et al. (2016) obtained auxospores of this species and determined a maximum length of 678 µm.
Materials examined. Guam: GU44Y-13!, GU44AV-8!, GU44BJ-4!, GU56A-2!. Federated States of Micronesia: Chuuk: TK4, TK28. Marshall Islands: M1! Observations. Specimens from Guam and Chuuk 220-350 µm long, 11-16 µm wide, isopolar in both valve and girdle views, striae 16-19 in 10 µm, slightly offset on internal and external sides of annulus (Fig. 3A, B). Annulus well separated from valve-margin junction and clearly visible in LM and SEM, including around the poles (Fig. 3A, B).   (Fig. 3F). Pores on pars exterior discontinuous around pole in front of the groove (Fig. 3F-H). Copula (Fig. 3D, I, J) broader and flat, even at pole, with up to 10 rows of circular pores in a decussate pattern; pars interior fimbriate with free fimbriae highly developed at pole, where they overlap the valvocopula, elsewhere shorter fimbriae arising from fused base (arrows, Fig. 3I, J). Small difference in shape, number of rows of pores and length of fimbriae between Fig. 3I and J may be due to origin of material ( Fig. 3D and 3I are from Chuuk wild samples, Fig. 3J from Guam culture) but copulae from wild material from Guam (not shown) also had >6 rows of pores. Pleura ( Fig. 3D) fimbriate, an apical cap extending along the valve as a very narrow ribbon.
Taxonomic comments. Ardissonea crystallina cannot remain in Ardissonea sensu stricto and will be transferred in the Taxonomic Revisions section at the end of the paper. Álvarez-Blanco and Blanco (2014) show Mediterranean specimens of Ardissonea dalmatica (Kützing) De Toni, restored to species status without comment, but the length ranges only up to 151 µm. There was a row of papillae/spines around the pole, not present in our material and the length range in the literature is 100-151 µm, but the stria density matches our specimens more closely. We leave our identification at A. crystallina pending broader study. Although Hustedt (1931Hustedt ( -1959 noted A. crystallina as distributed on coasts of warmer waters, Poulin et al. (1987) reported it as the only Ardissonea on the coast of Québec, an area that is far from tropical. Likewise, Shorenko et al. (2016) and Davidovich et al. (2017) collected it from the Black Sea. We do not agree with Pickett-Heaps et al. (1991) that there is a fundamental difference in the architecture at the pole of the valvocopulae of A. formosa and A. crystallina as both involve irregularities in the surface of the flange of the valvocopula where it lies below the pseudoseptum. Hustedt, 1931-1959 References. Janisch and Rabenhorst 1863, p. 13, pl. 2, fig. 6;Grunow 1877, p. 169, pl. 194, fig. 3a, b, c, d ;Peragallo andPeragallo 1897-1908, p. 313, pl. 78, figs 1, 2 (as S. cuneata Grunow); Hustedt 1914in Schmidt et al. (1874-1959Hustedt 1931Hustedt -1959Cheng and Gao 2011, figs 147, 148;Hein et al. 2008, pl. 14, fig. 3, pl. 15, fig. 1;Hernández-Almeida et al. 2013, figs 39-42 (as Synedrosphenia cuneata).

Synedrosphenia gomphonema (Janisch & Rabenhorst) Hustedt, 1932 in
Description from literature. The valve shape as shown by both Grunow (1877) and Hustedt (1914, has a basal part with nearly parallel sides that then tapers wider, flaring more abruptly near the apical pole. Hustedt gave length as 250-475 µm, maximum width 27-36 µm, basal and apical stria densities 12 and 14 in  10 µm, respectively. Striae change abruptly from parallel to radiate where the apical pole narrows. Transapical costae on the virgae often continuous across the valve and obscuring the midline; a longitudinal rib near each margin. Round et al. (1990: 444) noted that in Synedrosphenia stria density is higher on the girdle bands than on the valve (Fig. 10A), whereas in Climacosphenia it is lower on the girdle bands (Fig. 28A). This holds true for all our species.
Observations. Frustules heteropolar, cuneate in both valve and girdle view, numerous small, oblong plastids ( Fig. 4A-D). Valve margins parallel near basal pole, then gradually widening with slight flare below subrostrate apical pole (Fig. 4E, F). Length 255->414 µm, greatest width 27-29 µm, stria densities in 10 µm 13-14 basal, 14-16 apical. Striae parallel except changing abruptly to radial at apical pole with several interpolated short striae (Fig. 4B A frustule of Sceptroneis cuneata in remnants of Grunow 639 ( Fig. 6), showed the valve and copula. Valve stria densities were lower than in our material: basal 12, apical 13 in 10 µm. The copula differed in having a hyaline abvalvar border with no tendency to form fimbriae (Fig. 6D, E) and a longer hyaline area at the basal pole ( Fig. 6E vs Fig. 4F). The apical cap of the pleura may be smaller than in our specimens. A valve apical pole with part of the valvocopula (Fig. 6F, G) had 11 striae in 10 µm and there were no pores in the exterior, in contrast to Guam specimens.
Taxonomic comments. There are too many differences from Grunow's specimens to identify our species as Sceptroneis cuneata. We also cannot confirm whether our specimens match Synedrosphenia gomphonema without SEM of authentic specimens. The taxonomy of Synedrosphenia is convoluted. As explained by Fourtanier and Kociolek (1999): "Originally described (and validly published) as a subgenus of Synedra: Synedra subgenus Synedrosphenia H. Peragallo in H. and M. Peragallo 1900[Peragallo and Peragallo 1897-1908, p. 308, 312. Three species were included, Sceptroneis [Synedra] cuneata Grun., S. clavata Grev. and S. dubia Grun. and no type was designated. Although Azpeitia 1911, p. 220 introduces the name as "El genero ó subgenero Synedrosphenia" (the genus or subgenus Synedrosphenia), he treats it as a genus when describing his new species Synedrosphenia giennensis. Although S. giennensis Azpeitia is the only species formed as binomial with Synedrosphenia in Azpeitia 1911, Azpeitia, in his treatment of Synedrosphenia also clearly included the 3 species that constituted Peragallo's subgenus (but did not make transfers to genus Synedrosphenia). No type was designated by Azpeitia. The generic name cannot be attributed directly to Azpeitia because he did not provide a generic description. The authorship of Synedrosphenia is therefore "(H. Peragallo in H. Perag. and M. Perag.) Azpeitia," and the type has to be chosen among the species originally included in the subgenus by Peragallo. The choice of S. giennensis as the type species by Round, Crawford and Mann (1990) p. 444 and 704 is therefore untenable." Figure 6. Specimens of Sceptroneis cuneata from Grunow's Honduras sample A-C whole frustule at 0° tilt and details of basal pole and apical poles, respectively, showing internal costae and the copula; arrows point to long fimbriae overlapping valvocopula at both poles D, E same specimen, tilt = 60°, details of apical and basal poles, respectively, showing copula, including shallow cut-away matching smaller pleura (arrow on D), hyaline abvalvar margin of copula (black arrow on D), and reduced poration near basal pole (arrow on E) F apical part of valve in external view G valvocopula associated with same specimen. Scale bars: 50 µm (A); 10 µm (B-F); 5 µm (G). Hustedt (1931Hustedt ( -1959 very clearly synonymized Sceptroneis cuneata into Synedrosphenia gomphonema (1931( -1959( , fig. 723, pp. 220-221 in Jensen's (1985 translation, pp. 241-242 in orig.). The drawing in Janisch and Rabenhorst (1863) is, as Hustedt (1931Hustedt ( -1959 said, "very schematic and erroneous to be sure," showing a prominent sternum, but Hustedt based his synonymy on examination of many specimens from both Janisch and Grunow, both of whom worked on collections from Honduras. The other two species, S. clavata and S. dubia, are poorly known, perhaps not seen since their original descriptions, both characterized as very rare, and perhaps not even attributable to this genus. We thus propose Sceptroneis cuneata Grunow as the lectotype, below, recognizing the synonymy with S. gomphonema as the current name for the Honduras species, but pending evidence from Janisch and Rabehnorst's materials to test whether our species is the same. Synedrosphenia giennensis Azpeitia (1911, p. 219, pl. 6, figs 1, 2) was based on LM of a single clavate valve, maximum width ¼ of the way from one end, costae between striae not always continuous across the valve. Azpeitia made no mention of longitudinal costae. The dimensions were 195 × 35 µm, 12 striae in 10 µm, areolae 19-20 in 10 µm. This does appear to belong in Synedrosphenia but certainly needs further study.
Description. Cells heteropolar, probably attached by basal pole to seaweeds. Valves spathulate, 214-277 µm long with linear, wider apical portion 12-14 µm wide, narrowing abruptly about halfway down the valve and then more gradually to 6 µm near the basal pole, which is weakly inflated to 7 µm (  had a comb along inner edge of pars interior and a row of pores just under the overlapping valvocopula (Fig. 8F, G). Copula internal comb continued around the apical pole; pars exterior dominated by a series of long slits (rimae) that often extended to abvalvar margin (Figs 7I, J, 8G), resulting in a fringe along the edge of the cingulum except at poles. Both valvocopula and copula narrowed at the poles, leaving a space filled by the pleura, which is possibly a continuous band (Fig. 7I, J). We did not see the pars interior of the pleura but it is likely to be fimbriate (see S. parva below).
Holotype (designated here). Specimen at 14.9 mm E and 9.6 mm S of the mark on slide 2920, deposited at ANSP, accession # ANSP-GC20090.  Etymology. Named for Bikar Atoll, Republic of the Marshall Islands, where it was collected.
Taxonomic comments. The apparent absence of a bifacial annulus is unique among the Synedrosphenia species described here, where there is usually at least a hyaline line visible on the valve face or at the face-mantle margin. There is, however, a misalignment of striae evident in places along the valve-mantle junction (Fig. 7D, E), as observed in the other bifacial-annulus taxa described here, and hence no reason to suppose the growth of the valve proceeds differently. We hypothesize that there is an effective annulus present.

Synedrosphenia licmophoropsis Lobban, sp. nov. Figs 9, 10
Diagnosis. Extremely long heteropolar cells, differing from other species in the cuneate shape in valve and girdle views.
Description. Frustules heteropolar, tapering uniformly in both valve and girdle view (Figs 9D, 10A, B), attached to seaweed substrata by mucilage stalks; plastids numerous, lenticular to oval (Fig. 9A, B). Valve margins parallel near basal pole, 14 µm wide, widening evenly to 27 µm wide just before the cuneate obtuse or slightly rostrate apical pole, length 600-735 µm ( Fig. 9A-D). Striae parallel except radiate near apical pole, 19 in 10 µm near the apical pole, 16-17 near the basal pole ( Fig.  9E-H). Annulus, evident even in living specimens (Fig. 9A), nearly halfway between centerline and valve margin ( Fig. 9B-H), underlain by a longitudinal costa except at apical pole (Fig. 9G, H). Valve border with a rim along its length and shallow pseudosepta at each pole (Fig. 9G, H), the latter matched by extensive development of the valvocopula. In girdle view, three bands evident, all fimbriate ( Fig. 10A-C, F). Valvocopula bearing a row of pores at the edge of the pars interior and five rows of pores on the exterior near the apical pole, reduced to two near the basal pole, the pores in a decussate pattern, 27 in 10 µm, the edge of the pars interior bearing short fimbriae near the apical pole, longer elsewhere ( Fig. 10D-F). The pars interior of the valvocopula at the poles formed a deep shelf with an irregular surface (Fig. 10D, E), creating a notch on one side and a shallower, longer gap on the other (Fig. 10A, B); notches were on opposite sides of both the valvocopula itself and when comparing epitheca and hypotheca (Fig. 10A, B arrows). The line of pores along the edge of pars exterior moved to inner edge of the gap and became slitlike (Fig. 10G, arrow). Copula (Fig. 10A, B, F, H) wider, again with pores in a decussate pattern, about 10 rows near apical pole diminishing to five at the basal pole, a wide delicate fringe along the edge of the pars interior. The relationship between the first two girdle bands at the poles is shown in Fig. 10I. The pleura ( Fig. 10A-C) had a broad apical cap, quickly tapering to a narrow external line with fimbriate pars interior, broadening again somewhat at the basal pole; this band lacked pores.
Holotype (designated here). Specimen at 13.0 mm E and 5.0 mm S of the mark on slide 455, deposited at ANSP accession # ANSP-GC20086. Etymology. Named for its resemblance to certain large Licmophora spp., especially L. attenuata Lobban, Tharngan and Ashworth. Taxonomic comments. Longer and more evenly tapered, apical pole broadly rounded, striae denser compared to S. gomphonema; valves appearing narrower than S. gomphonema because of the greater length, but the apical poles are equally wide.

Synedrosphenia parva Lobban, sp. nov. Figs 11, 12
Diagnosis. Shorter and narrower than S. gomphonema and more finely striated. Girdle bands distinctive: copula with very long internal fimbriae at poles, pleura a large apical cap.
Description. Frustules heteropolar in valve and girdle views (Fig. 11A, F, G) with numerous small plastids (not shown). Valves 196-250 µm long, 18 µm wide near the apical pole, clavate with acutely rounded to weakly rostrate apical poles (Fig. 11A-E). The annulus evident in LM but close to edge of valve face (Fig. 11A, C). Valve surface with weak longitudinal furrows, mantle deeper toward apical pole (ca. 11 vs. 5 areolae deep) (Fig. 11 B, C, F, G). Striae 16 in 10 µm near basal pole, 18-20 in 10 µm near apical pole, parallel except radiate at apical pole ( Fig. 11A-E). Pseudosepta present at both poles (Fig. 11D, E); longitudinal costae subtending annulus and transverse costae on vimenes except at apical pole (Fig. 11D, E). Three girdle bands are present (Figs 11F, G, 12E-G). Valvocopula with a row of pores along outer edge of pars interior plus three rows decreasing to one on pars exterior; pars interior with a notched flange at each pole and a fimbriate edge (Fig. 12B). Copula with up to nine rows of pores at apical pole in decussate pattern, decreasing to three near basal pole (Figs 11F, G, 12C, D). Pleurae a pair of polar caps (connecting band not seen), the apical cap relatively massive with small pores and fimbriate (Figs 11F, 12E-G), basal cap smaller, with fewer pores (Fig. 11G arrowhead).
Holotype (designated here). Specimen at 15.3 mm E and 12.7 mm S of the mark on slide 2326, deposited at ANSP accession # ANSP-GC20088. Fig. 11A.
Synedrosphenia recta Lobban, sp. nov. Fig. 13 Diagnosis. Long, straight valves with isopolar, bluntly rounded poles, differing from Synedrosphenia (Ardissonea) crystallina in the marginal annulus, shape of pole, and poration on girdle bands, and from Synedrosphenia (Ardissonea) fulgens in the notch and poration of the valvocopula. Description. Valves linear, slightly wider in the middle, isopolar with bluntly rounded poles, length 406-430 µm, width 13-16 µm, striae parallel except at pole, 16-18 in 10 µm, areolae slightly elongated in apical axis ( Fig. 13A-D). Valve face longitudinally depressed along midline (Fig. 13F, I, K). Annulus detectable externally as a narrow space between areolae around pole and along edge of valve face (Fig. 13D  arrows); internally, the end is seen between the areolae at the pole and can be traced around the same as externally (Fig. 13H, L arrows). There are no longitudinal costae (Fig. 13E, J arrow) and transverse costae underlie the vimines as continuous costae from one marginal rim to the other (Fig. 13J). Valvocopula (Fig. 13E-J) lacking external areolae, pores along edge of pars interior have costae between them on the pars interior, appearing to be slits (compare Fig. 13H, J) and the line deflects around the asymmetric sculpting at pole (Fig. 13H). A discrete notch at pole of valvocopula ( Fig. 13F, G, I); fimbriae not observed. Copula (Fig. 13E, F) broader with a line along interior-exterior boundary dividing internal bifurcated fimbriae from external part perforated by long slits (rimae) that sometimes continued to the abvalvar edge; perforations discontinued around pole. Pleura glimpsed in Fig. 13K, apparently apical caps joined by very narrow perforated band lacking fimbriae.
Holotype (designated here). Specimen at 14.3 mm E and 5.7 mm S of the larger mark on slide 154, deposited at ANSP accession # ANSP-GC20113. Fig. 13A Additional materials examined. Guam: GU76B-2!. Marshall Islands: Jaluit, J5!. Etymology. L. rectus, straight. Named for its parallel sides and isopolarity. Taxonomic comments. This species resembles two former Ardissonea species which we are separating into different genera. The complex polar architecture places it in Synedrosphenia, within which it is differentiated from S. (Ardissonea) crystallina by the marginal annulus, bluntly rounded poles, plain valvocopula and slits on the copula. The shape of the poles and the marginal annulus resemble Ardissoneopsis fulgicans, and in LM of Chuuk samples S. recta was distinguishable only by length, the A. fulgicans specimens being only 300 µm long. We first noticed it when we found a specimen, apparently of A. fulgicans, with a valvocopula pole evidently of the Synedrosphenia type (Fig. 13H). It was necessary to tilt specimens to see girdle band characters.
Specimens from Bikar Atoll (not illustrated) with longer (to 583 µm) but narrower (7 µm) valves, weakly inflated at the poles and in the middle, were structurally indistinguishable from the Chuuk specimens on which the species is defined, including the lack of longitudinal costae. Without further evidence, we cannot decide whether to assign this to a formal varietal status, describe it as a separate species, or leave it within the scope of S. recta. We mention it here to underline the point that the present study does not exhaust the biodiversity of these genera in Micronesia. Table 3 References. Peragallo andPeragallo 1897-1908, p. 311, pl. 79, fig. 5;Hustedt 1931Hustedt -1959Navarro 1982, p. 260, figs 64, 65;Poulin et al. 1986, figs 31, 32;Witkowski et al. 2000, p. 44, pl. 31, figs 9-11;Hein et al. 2008, p. 23, pl. 8: 1, 2).

Ardissonea fulgens (Greville) Grunow ex De Toni, 1892
Description from literature. Kanjer et al. (2021) reported the ultrastructure of Greville's type material of this species, collected at Mull, Scotland, and clearly showed a pseudoseptum and sculpted apices on the valvocopula; no apical notch was observed. That architecture is not visible in LM and so we can only assume that collections reported from Atlantic and Mediterranean Europe are the same until they can be examined. The species as traditionally understood is described as follows: Valves narrow, linear except slightly wider in the middle, 170-450 µm long, 10-15 µm wide, striae 12-14 in 10 µm, offset slightly where they meet such that a central line ("pseudoraphe") is distinct but very narrow; longitudinal costae very near to the valve edge and normally difficult to see (Hustedt 1931(Hustedt -1959. Kanjer et al. (2021) show an undulation in the annulus near the middle of the valves in holotype material, a feature not reported elsewhere; they reported a stria density of 15-16 in 10 µm. Poulin et al. (1986, figs 31, 32) showed a valve without pseudoseptum, i.e., of the Ardissoneopsis type, with transapical costae on the virgae except at the poles, and the annulus near the valve-mantle junction marked externally by a gap in the areolae and internally by a longitudinal costa on each side. SEMs of var. mediterranea, with 17 striae in 10 µm, in Güttinger (1989), include an oblique view suggesting that the annulus is not internally thickened. Navarro (1982) gave dimensions of 554 µm long by 13.5 µm wide, 14-15 striae in 10 µm. Hein et al. (2008) showed "Ardissonea cf. fulgens" with dimensions of 123-186 µm long by 6.5 µm wide, striae 20 in 10 µm-i.e., smaller and more finely striated than classical descriptions, and they noted that their counts of striae from images in Witkowski et al. (2000) showed stria densities of 19-22 in 10 µm.
Taxonomic comments. For the reasons given below for separating single-walled Ardissonea taxa, it cannot remain in Ardissonea sensu stricto. If this species is confirmed to lack the apical notch, its pseudoseptum and sculpted valvocopula are sufficiently strong characters to place it in Synedrosphenia; it certainly does not belong in Ardissoneopsis, proposed below, which lacks those characters, and we therefore must describe our fulgens-like taxa as new species and propose a new combination of Synedrosphenia fulgens for Greville's species.
Group 3: Single-walled taxa without pseudosepta Diagnosis. Differing from congeners in the straight outline, only weakly inflated at the center. Description. Valves 205-320 µm long, 9-15 µm wide, usually slightly widened in the middle and bluntly rounded at poles (Fig. 14A-F); 17-19 parallel striae in 10 µm. Cells attached by one pole to sturdy mucilage stalks (not shown). Annulus along valve-margin junction and not visible in LM, underlain by longitudinal costae (Fig. 14C, D, F, G); sternum absent but striae often out of alignment at the midline (Fig. 14B, H); often spines on poles (Fig. 14D, I). Valve rims straight (i.e., lacking a pseudoseptum) and valvocopula simple, therefore no apical groove (Fig. 14E, F); transverse costae on the mantle thinning toward the margin ( Fig. 14G; contrast Fig.  13J). Transverse costae on most vimenes but absent from a relatively large area at the poles (compared to Synedrosphenia recta, Fig. 13H) (Fig. 14F, H). Three girdle bands were present (Fig. 14E, I, J). Valvocopula (Fig. 14E, H, J) with generally two rows of pores on pars exterior; pars interior with a single row along outer edge and a fimbriate inner edge. Copula (Fig. 14D, E, I) with multiple rows of pores forming a decussate pattern, and a fimbriate advalvar margin. Pleura barely visible as a narrow cap at each end, possibly connected, apparently fimbriate at the poles (Fig. 14E, I).
Taxonomic comments. In our region, this species is hard to positively identify in LM from the new species Synedrosphenia recta described above, and some other species so far only glimpsed but differs in (1) having simple apical structure unlike Synedrosphenia spp.; (2) the presence of two or more rows of pores on the valvocopula exterior. Lobban et al. (2012) previously identified this species as Ardissonea fulgens on the basis of literature but Kanjer et al.'s (2021) surprising finding about the type material (see above) forced the reevaluation.
Diagnosis. Differentiated from congeners in the straight valve with inflated center and apices, and valve-appressed colony formation. Description. Cells valve-appressed to form flat colonies from sturdy mucilage pads, cells toward the outside increasingly curved in the apical axis ( Fig. 15A-D). Nucleus in widest part of the cell and appressed to one side, numerous lenticular plastids (Fig.  15B). Valves very long and narrow, inflated at the middle and poles (Fig. 15E-G); frustule also inflated in the middle in girdle view because of deeper mantle and wider girdle bands (Fig. 15C vs D). Length 560-900 µm, width at center 9-12 µm, poles 7-10 µm, 5 µm in between; striae 18 in 10 µm (Figs 15F, G, 16A-C). Annulus along junction of valve face and mantle except at poles, where it is further back (Fig. 16B, C); some thickening under annulus but not as thick as transverse costae (Fig. 16E). Striae parallel except radiate at poles (Fig. 16B, C). Internally, costae on most virgae except near poles (Fig. 16C, E, F). Spines absent. No pseudosepta (Fig. 16C); transverse costae on the mantle diminishing toward margin (Fig. 16E arrowhead; contrast  Etymology. Adjective (L.) appressus, appressed, with reference to the colony formation.
Taxonomic comments. We previously ) identified this species as Ardissonea fulgens var. gigantea (Lobarzewsky) Rabenhorst, based on literature. Starting from Peragallo andPeragallo (1897-1908), that taxon has been described as extremely long (400 µm to 1 mm), with a broad central portion joined to capitate poles by very narrow linear portions, the striae 13-14 in 10 µm, but not forming distinctive colonies. Cells from Guam matched this description, except for somewhat higher stria density, but adhered valve to valve while still attached to massive mucilaginous stalks, giving colonies a unique form, a characteristic not mentioned in the literature. The variety is distinctive enough by the shape and colony formation to differentiate it from the nominate variety but has been even less studied than A. fulgens. However, drawings of Synedra gigantea Lobarzewsky (1840, p. 276, pl. 6, fig. a-c) show a highly elongated diatom occurring in valve-appressed clusters, somewhat similar to those seen in Guam, but not showing any bulge in the middle and being apparently very flexible; as far as one can tell, the frustule is uniformly linear in both views, but he stated that the side (i.e., valve) view is 3-4 times narrower and with a wavy outline. This does not well accord with valve views of Ardissonea fulgens var. gigantea, which has never been depicted as wavy but has been shown with a strong central inflation. Lobarzewsky's drawings look more like Licmophora flucticulata Lobban, Schefter and Ruck (2011), which is wavy and very narrow and occurs in valve-appressed clusters, except that the clusters of that species in girdle view are very obviously tapered. The descriptions of this taxon were carried forward into Rabenhorst (1864: 140) and De Toni (1892: 674) but there is no mention of the valve outline. The comparison of its shape to Toxarium hennedyanum (W. Gregory) Pelletan seems to have begun with Peragallo andPeragallo (1897-1908), who clearly describe the valve shape and stria density for the first time, but show a smooth valve outline. This taxon was said to be frequent in the Mediterranean (Peragallo andPeragallo 1897-1908), yet neither they nor Hustedt (1931Hustedt ( -1959 refer to valves sticking together in distinctive fan-shaped clusters. The extreme length is the sole character that could link the acid cleaned frustules so accurately described by the Peragallos to the taxon drawn by Lobarzewski (1840), and that is insufficient, so that the identity of Synedra gigantea is a mystery, while the lack of colony information and the lower stria density leave doubt about how the Guam population is related to any European taxa. Given findings of Kanjer et al. (2021) on the nominate variety, we reject our earlier identification and propose a new species for our material. Observations. We have not encountered this in the Guam flora but found several fragments in Grunow's Honduras gathering that showed internal and external structure, confirming the similarity of structure to the two aforementioned Ardissoneopsis spp. A fragment including the central inflation and one pole suggests a total length of ca. 690 µm (Fig. 17A). Widths were 10-11 µm across the center, 9 µm across the poles and 4-5 µm between; stria density 19-21 in 10 µm. The annulus along valve face-mantle junction at least partially subtended by longitudinal costae (Fig. 17B-E).
Taxonomic comments. According to AlgaeBase (Miranda in Guiry and Guiry 2020), S. undosa is currently referred to as Toxarium undosum (Grunow) De Toni, 1892, which De Toni recognized as distinct from T. undulatum. We have not found more recent mention of this species, but it clearly cannot remain in Toxarium. Hustedt (1931Hustedt ( -1959 stated that, while Grunow had made clear the abundance of Synedra undosa in his Honduras material, "one must assume he also included S. (Toxarium) undulata," because Hustedt had not observed specimens matching Grunow's drawings in Honduras material he examined. Hustedt concluded that this species requires further examination. Grunow, however, was clear that S. undosa was different from S. undulata and the SEM images here clearly prove him right. On the evidence that it is a species in its own right, we propose below its transfer to Ardissoneopsis.
Ardissoneopsis gracilis Lobban, sp. nov. Fig. 18 Diagnosis. Differing from A. fulgicans in valves being more slender, slightly wavy outline, lower stria density, and markedly heteropolar frustule in girdle view and from A. undosa by lack of central inflation.
Description. Frustules cuneate in girdle view (Fig. 18C, D, N, O), both valve mantle and width of girdle bands decreasing from apical pole to basal pole; hypothesized to be attached to mucilage pads at the basal pole, but not observed in live or raw material. Valves slender, weakly undulating, slightly inflated in the middle and at poles, slightly heteropolar but poles not consistently distinguishable except for mantle depth (Fig. 18A, B, G, E). Dimensions: 206-431 µm long, 5 µm wide except 7 µm across the middle and poles (Fig. 18A-G). Striae parallel, 10-12 in 10 µm, except radiating at poles, areolae small, circular. The annulus near valve-mantle junction, indicated only in places where striae on mantle are offset from those on valve face (oval on Fig. 18I). There is often also an offset across the midline but no sternum. Several spines occur on both poles. (Fig. 18H, K, N, O). Internally, interstriae are barely thickened, not forming costae, and there are no longitudinal costae on the annulus (Fig. 18L, M). Three girdle bands were present (Fig. 18N-R): valvocopula with a row of pores along the edge of the pars interior and a fimbriate internal margin (Fig. 18Q); two rows of pores occur around basal pole, but become one within 10 µm from base, further out becoming two and then three or more rows (Fig. 18N, O). In the copula, also bearing a fimbriate flange (Fig. 18P, R), pars exterior has numerous pores arranged in a generally decussate pattern but becoming looser near basal pole (Fig. 18N, O). The pleura is very narrow even at the poles (Fig. 18N, O, P).

Synedra bacillaris (Grunow) Hustedt, 1932
Description from literature. As described from LM (Hustedt 1931(Hustedt -1959, this taxon is easily distinguished from Ardissonea spp. by the prominent longitudinal costa on the midline and the absence of lateral longitudinal costae (i.e., no observable annulus). There is only one wall layer. Valves linear, usually slightly wider in the middle, 320-700 µm long, 16-20 µm wide, 8-9 striae in 10 µm. Robust transapical costae on each of the virgae. As seen in SEM (Sullivan and Wear 1995), the valve has a single wall, areolae are circular, those at the poles slightly smaller and not organized into striae, elsewhere the striae are framed by the transapical costae. Sullivan and Wear stated that the costae extend from the margin to the median costa, but their fig. 12 shows some costae extending from the median costa and the costae are not consistently thick. Sullivan and Wear reported an absence of spines, polar pore fields and rimoportulae. Girdle bands closed, not more than five (three on the epicingulum): the valvocopula broad with a single row of small, circular pores near the advalvar edge of the pars exterior and a large flange that fits inside the valve mantle; there was said to be short fimbriae along its inner edge but they are not evident in the micrographs. There are additional rows of pores on the valvocopula just around the poles. The valvocopula has a flange that fits around the valve mantle but there is no apical notch. The copula is also broad but with several rows of pores (their fig. 11 shows these loosely organized in straight striae of at least three pores), and there is a narrow pleura (Sullivan and Wear 1995); none of their images show the inner edge of the girdle bands. Although not specifically stated in these descriptions, the frustules are implicitly isopolar. Sullivan and Wear (1995) particularly noted the circular areolae, in contrast to the large oval areolae they saw in Ardissonea formosa. Grunow's (1877) published drawings (as Synedra crystallina var. bacillaris) do not include details of areolae but unpublished drawings archived in the Grunow drawing collection at W (W1901-5465) include several sketches where the areolae appear to be oval. Grunow's drawings, like those of later authors, show this species as linear and isopolar, with bluntly rounded poles, but his unpublished girdle view shows a slightly tapered frustule.
Observations. We were able to find a few fragments of Synedra crystallina var. bacillaris Grunow in an SEM preparation from Grunow's Honduras gathering (Fig. 19). The external view (Fig. 19A-C), of a fragment 172 µm long that appeared to reach the slightly inflated middle, was distinguished from Ardissonea formosa, also present on the stub, by the row of pores on the exterior of the valvocopula. The pole was blunt, with small spines (Fig. 19B, arrowheads), and the valve linear, although a little wider in the middle. Both valves had 11 striae in 10 µm, comprising oval areolae, and the small fragment, width 10 µm, showing the interior had the characteristic central longitudinal costa and prominent transapical costae (Fig. 19D). A second frustule fragment (Fig. 19E-G), 174 µm long, showed the single line of pores on the valvocopula were 19 in 10 µm, striae on the copula 18 in 10 µm. The broken tip of this valve showed the transverse costae (Fig. 19G).
Taxonomic comments. Specimens provisionally identified as S. bacillaris were found in samples from across Micronesia but there was a range of stria densities and dimensions. All had oval areolae and many had spines. However, there were few fitting the literature description, even overlooking the shape of the areolae, i.e., those with linear valves, blunt poles and 8-9 striae in 10 µm. Those in Guam were mostly among populations of lanceolate specimens and probably part of them (see next taxon). The small, circular areolae shown by Sullivan and Wear (1995) are certainly different from the oval areolae in our specimens but determining the identities of these specimens relative to one another and to the literature will require much further study.
Diagnosis. Differing from Grunowago (Synedra) bacillaris in width and lanceolate outline of valve.
Description. Valves lanceolate, often slightly rostrate, 194-341 µm long, 11-16 µm wide, striae 9-10 in 10 µm (Fig. 20A-D), the striae comprising large oval pores; poles with smaller, circular pores, but not organized into polar pore fields (Figs 20E,21A). Isopolar but wider in the middle in girdle view (Fig. 18F-H). Two small spines usually present on each pole (Figs 20G-I, 21B). Rim present on the valve border but not extended into a pseudoseptum (Fig. 21A). Valve single-walled (Figs 20J, 21A), internally costate with a distinctive central costa the entire length of valve and thickened vimines (Figs 20J, 21A), no annulus detected either as a break in the striae, or any other longitudinal costae (Fig. 21A). Three girdle bands were present (Fig. 20F-H). Valvocopula (Figs 20I, J, 21A, C) without extended lip but having groove on pars interior that fits rim of valve. A single row of pores on advalvar side of pars exterior, circular except elongated near poles (Figs 20G, H, 21A, C), a short line of pores around poles on pars interior (Figs 20I, 21C). Copula has a row of pores on outer edge of pars interior (Fig. 21A, B, D), often hidden under valvocopula and a well-developed fringe on the pars interior; up to 5 rows of pores on pars exterior (Figs 20G, H, 21B, D). Pleura a very thin fimbriate band, continuous along cell length (Figs 20G, H, 21B, D).
Etymology. Adjective (Latin) with reference to its presence in the Western Pacific Ocean.
Taxonomic comments. Specimens that we ) referred to Synedra bacillaris were 180-400 µm long, 14 µm wide, with 9-14 striae in 10 µm. This range of stria density is suspiciously large for Ardissoneaceae and suggests that two or more species are being confounded. Some specimens in GU44BJ-4 population were nearly linear but most likely represent an extreme within G. pacifica rather than an occasional G. bacillaris. Both Hustedt (1931Hustedt ( -1959 and Sullivan and Wear (1995) remarked on the ease of identifying this species by the strong central costa, compared to other Ardissonea spp., but as with other taxa where a strong identifying character has masked species diversity (e.g., Cyclophora tenuis - Ashworth et al. 2012, Bleakeleya notata -Lobban and Perez 2016), we suspect that there is hidden diversity within "Synedra bacillaris". The genetic branches are long between lanceolate samples from Guam (GU44AI-5) and Florida (HK291) (see, below, Fig. 33). For now, we can only describe Grunowago pacifica and deal with the generic placement of that species cluster.
Taxonomic comments. Kooistra et al. (2003) show a single row of pores on the mantle and a copula that matches those shown here for T. hennedyanum. They do not state a stria density, but we count 11 in 10 µm, measured from their fig. 1E. There has been a tendency to ignore stria density, reflecting the marginal striae and the domination of the valve face by scattered areolae. Kooistra et al. (2003) had monoclonal cultures in which both hennedyanum-like and undulatum-like cells were present, and they proposed a merger of the two species. The specimens they show had a rippled edge, but not a strongly wavy outline (contrast Poulin et al. 1986;Hein et al. 2008, and our Fig. 23); i.e., a midline along the apical axis is straight, whereas that in T. undulatum is sinusoidal. If the Micronesian specimens shown here are indeed T. hennedyanum and T. undulatum (by no means certain), our results would suggest that the rippled edge seen in the cultures of Kooistra et al. (2003) was a phenotypic expression of T. hennedyanum and was not T. undulatum. We previously noted ) the significant genetic variation in Guam isolates of T. hennedyanum and T. undulatum reported by Theriot et al. (2010Theriot et al. ( , 2011 and this is even more evident now; however, in reviewing the vouchers, we find no data from specimens attributable to T. undulatum; all the variation shown in Fig. 33 appears to be part of a T. hennedyanum complex.
We also documented T. undulatum in Grunow's Honduras sample (Fig. 24). A whole valve was 700 µm long ( Fig. 24A) with long-wavelength undulations. The deep mantle sometimes exhibited longitudinal waves in the areolae (not solely an artefact of the wave in the valve, shown by short additional longitudinal lines of areolae- Fig. 24D, arrows). The striae on the valve face were two areolae long on each side bordering scattered areolae in the inflated center and poles (Fig. 24C, E), internally, the virgae were thickened into costae (Fig. 24C). Valvocopula (Fig. 24C, E-G) with a single line of areolae on the exterior, short fimbriae along the interior margin except at the sculpted poles, and a series of arched slits (Fig. 24F, G; compare Fig. 23K). Copula (Fig. 24C-E, H) wide with coarse fimbriae on the pars interior, forked at the tips, exterior with striae of pores tending to merge into slits, especially on the advalvar side. Fimbriate pleura clearly seen in Fig. 24H. Taxonomic comments. The specimens from Grunow's Honduras material are near the length ("about 35m" = 890 µm; see Itaki and Bjørklund 2006) and shape described by Bailey (1854) for specimens from Sargassum in Narragansett Bay. As noted by Hustedt (1931Hustedt ( -1959, Grunow (1877) did not mention this species (in the context of presenting new species), but Hustedt says he observed it on Grunow's slides.
In the Micronesian samples, depths of the mantle, and structure of the copula were different in the two species, so we can assert that the specimens shown here for T. hennedyanum and T. undulatum are different species. The large genetic variation in Toxarium materials we have sequenced (Fig. 33) casts doubt on whether either of them might be identified to authentic Atlantic types, a question highlighted by the Grunow specimens. It similarly raises questions about the number of species there might be in Micronesia. Both questions require much more work to define the range of character states and their combinations, but the oft-repeated claim that the two species differ only in outline is already not supported by morphological evidence. The more important point for the present study is to establish generic characters for comparison to those of the genera in Ardissoneaceae and Climacospheniaceae.

Group 6: Genus Climacosphenia
Description from literature. C. elongata has been described (e.g., by Peragallo and Peragallo 1897-1908and Hustedt 1931-1959 as identical to C. moniligera Ehrenberg, 1843 except for the valve tapering more abruptly to a stem with parallel sides. Yet Bailey (1854) had specifically distinguished his species from C. moniligera by the finer stria density (without giving numbers) and described a species 330 µm long, with very wide craticular bars (as shown also in Peragallo andPeragallo 1897-1908, pl. 86, fig. 5; Hustedt 1914, pl. 307, figs 1-4;and Round 1982, fig. 3) (Table 5). Peragallo andPeragallo (1897-1908) gave stria densities of 18-21 near the basal pole, 27->30 near the apical pole for C. elongata versus 16-17 in 10 µm basal and 19-20 in 10 µm apical for C. moniligera but expressed the opinion that there was intergradation between the two species. Hustedt (1931Hustedt ( -1959 said that Peragallo and Peragallo had confused their European specimens of C. elongata with C. moniligera and he attempted to resolve the confusion with an original drawing of a cell with a very long, very narrow stem. The drawing (his fig. 626) was based on Samoan specimens he had published in "Schmidt's Atlas" (Hustedt 1914). Round (1982) attempted to clarify the difference between the two species, not considering the variation within C. elongata, but his drawings of C. elongata show specimens no more than 320 µm long. In recent literature, Álvarez Blanco and Blanco (2014), show short cells similar those in Peragallo andPeragallo (1897-1908), while Navarro and Lobban (2009) reported cells 750-1200 µm long from a Guam sample, similar in shape to those in Hustedt (1931Hustedt ( -1959, a range that otherwise occurs only in Hustedt's report (table in Álvarez Blanco and Blanco 2014: 103). Ultrastructural characters: Apical spines present; costae generally present; frustules have only two girdle bands, both with fimbriate inner margins (Ricard 1987, fig. 677;Montgomery 1978, pl. 114, fig. G;Round 1982).
Taxonomic comments. The shape and morphometrics accord adequately with the literature on C. elongata, the narrow craticular bars, especially in the stem, differing from the sturdy bars and oval spaces of C. moniligera, and from the very longstemmed specimens depicted by Hustedt (1931Hustedt ( -1959). Based on the available evidence, we accept the hypothesis that the specimens shown in Fig. 25 are Bailey's species.
Diagnosis. Differing from C. elongata in the greater length, especially in the stem, and narrower, linear apical portion, and from C. truncata Hustedt ex Simonsen in the shape of the wide part of the valve.
Description. Cells attached in small groups to stout, branched mucilage stalks (Fig. 26A, E), at least 110 µm wide at the base and up to 2 mm long. Frustule generally deeper than wide, especially where stem first narrows below the apical pole, ratio there about 15 µm deep to 5 µm wide (Fig. 27A). Nucleus situated between two more-widely spaced craticular bars about 120 µm from apical pole (Fig. 26A, D); plastids lenticular. Valves 750-1305 µm long, linear toward apical pole, 22-24 µm wide, then tapering to a long linear stem, 5 µm wide, expanding to 10 µm at the pole (Figs 26A-C, 27A). Stria density 27-28 in 10 µm through most of the valve but 18-21 in 10 µm near basal pole, sparser and more scattered inside annulus near basal pole (Fig. 27D, E, G). There was a crescent of scattered areolae at top of mantle at apical pole (Figs 27D, G, 28A) (noticed in Climacosphenia spp. by Round 1982: figs 15, 16, 25). Annulus apparently open at basal pole, sides ±parallel in the wide area, 8.3 µm apart, in the stem 2.3 µm apart, expanded to 5 µm at the basal pole ( Fig. 27C-G). Two or more stout spines on external apical pole (Figs 27D, 28A), shallow transapical costae throughout except near basal pole (Fig. 27F, G). Pseudoseptum absent (Fig. 27F, G). Two girdle bands (   ( Fig. 28B), stria density 19 in 10 µm (Fig. 28A, B) but increasing around foot pole, where there was also a row of pores on pars interior (Fig. 27I). Craticular bars widely spaced in wide part of the cell, with widest space corresponding to the location of the nucleus (Figs 26B, 27B); in the stem becoming much closer together (Figs 26B, 28C), often paired with the two bars apparently growing from opposite sides (Fig. 28C). In one specimen from culture, we observed porous bands of silica around several craticular bars near the apical pole (Fig. 28D). Copula fimbriate, more densely striated than the valvocopula but less than the valve, 22 in 10 µm (Fig. 28A, E). A search of abundant girdle views in whole mounts showed no sign of a pleura. Holotype (designated here). Specimen at 16.1 mm E, 12.8 mm S of the mark on slide 1861; deposited at Academy of Natural Sciences, Drexel University, Philadelphia, accession number ANSP GC20108. Fig. 27B.
Etymology. Superlative of elegans, with reference to the slender appearance of the extremely long cells.
Taxonomic comments. The width of the apical pole and especially the distance between the annular lines are narrower than in C. elongata (Table 5) and the stem is also narrower, though the basal pole is inflated and slightly wider than C. elongata. Hustedt's (1914) drawings of C. elongata have the greatly extended stems that are present in our new species, but the apical portions of his cells are generally elliptical rather than linear. However, the valvocopula (Hustedt 1914, pl. 308, fig. 6) is notable in having parallel sides as far as the large space (i.e., where the nucleus would have been), so it is possible they represent C. elegantissima. Climacosphenia truncata from Borneo is also illustrated there (pl. 308, figs 1-4); it also has a very long stem but possesses an essentially triangular blade with a bluntly rounded apical pole. C. elegantissima is common as an epiphyte on filamentous seaweeds in Guam farmer fish territories and such habitats around the western Pacific/ IndoPacific may yield new material of C. truncata and other Climacosphenia spp. Round (1982) showed porous bands of silica on craticular bars near the basal pole of both species only found in Hawaiian material. We saw them only once on apical bars in C. elegantissima. Their distribution and function are unknown.
Climacosphenia scimiter A. Mann, 1925 Figs 29, 30 Reference. Mann 1925, p. 59, pl. 12, fig. 4. Description from literature. Cells tapering uniformly and strongly curved, the septa "delicate," very narrow, "and showing at the middle either obscurely or not at all the break or sutural division common to specimens of this genus." Length 414 µm, width at apical pole 28 µm, stria density not specified.
Observations. Specimens fitting this description have been found in Guam and Chuuk, often together with C. elegantissima. Colonies on branched mucilaginous stalks, the nucleus near the apical pole between two of the widely spaced craticular bars (Fig.  29A). Length 390-612 µm, maximum width 25-26 µm near apical pole, 8-10 µm across basal pole (Fig. 29A, B). Stria densities outside the annulus 21 in 10 µm in lower part, 29 in 10 µm at apical pole; striae within annulus at basal pole about 16 in 10 µm, areolae more, or less aligned (Fig. 29C-E vs Fig. 30A). Maximum distance across annulus 10 µm (Figs 29B,D,30A). While annulus appears discontinuous in SEM, phase contrast suggests continuity (Fig. 29C). Valves delicate except at basal pole, apical striae hard to resolve in LM even with phase contrast (Fig. 29C, E), costae not developed on either vimines or annulus, except near basal pole (Fig. 29D). Two spines on apical pole of valve (Fig. 30A, D). Two girdle bands present, both broad and with regular striae, but at different densities, valvocopula 20 in 10 µm, copula 24 in 10 µm (Fig. 30D). Valvocopulae fimbriate with narrow, widely spaced craticular bars, unions all seamless (Fig. 30E-G). Taxonomic comments. Observations are completely in accord with Mann's (1925) description of specimens from the Philippines (Mann noted that data on the locations of the dredge samples were lost). Given also that this species was discovered not far from Micronesia, we are confident in the identification. We cannot accept that this is merely a curved C. moniligera, as suggested by Round (1982), because of the great difference in septum characters and differences in stria densities. Whether there exist straight versions of this species, or curved C. moniligera populations, remain open questions.  Fig. 31 Diagnosis. Differing from C. moniligera in apical stria density and delicate craticular bars throughout, seamless except for complex unions in the most basal 1-4 bars and from C. scimiter in straight outline, presence of complex unions, and cell width.
Taxonomic comments. Valves resemble C. moniligera in size and stria densities (Table 5) but differ in the valve shape and the narrow and mostly seamless craticular bars are unlike the massive septum with complex unions in the bars reported for C. moniligera (Peragallo andPeragallo 1897-1908;Hustedt 1931Hustedt -1959Round 1982).

Morphological cladistics
The cladistic analysis of the character set (Suppl. material 2: Table S2) resolved 18 equally parsimonious trees, which we have summarized as a strict consensus tree (Fig. 32). The genera with striae perpendicular to the bifacial annulus across the entirety of the valve (Ardissonea, Synedrosphenia, Ardissoneopsis, Climacosphenia and Grunowago) formed a monophyletic group (Toxarium hennedeyanum was used as the outgroup). Three genera appear to have morphological synapomorphies: Climacosphenia (craticular bars on the valvocopula), Ardissonea (the internal wall) and Grunowago (the central longitudinal costa). Morphological synapomorphies for Ardissoneopsis and Synedrosphenia are less clear but each genus has a unique combination of the main characters ( Table 6). The species of Ardissoneopsis form a polytomy with Climacosphenia, sharing the heteropolar shape of the frustule in girdle view (though this character is lost in A. fulgicans). The taxa in Synedrosphenia have a range of character states, and the only characters shared by all the species (internal transverse costae, pseudosepta, apical notch on the valvocopula and fimbriate girdle bands) are also shared by species in other genera.
The genera Climacosphenia and Toxarium were also monophyletic in all three analyses (bs > 99%, pp = 1.0), though their position relative to other taxa changed between the analyses. The Climacosphenia sequences were sister to the rest of the clade in the ExaBayes analysis (pp = 1.0). In the Maximum Likelihood analyses, Climacosphenia was either sister to the Ardissonea clades (IQTREE, bs = 63%) or sister to the Ardissoneopsis-morphotype strains, but with extremely low support (RAxML, bs = 18%). Strains with the Ardissonea sensu stricto morphology were either sister to the Synedrosphenia morph strains (IQTREE bs = 75%, ExaBayes pp = 0.73) or sister to a clade containing the Grunowago and Synedrosphenia morphs (RAxML bs = 77%). The Ardissoneopsis morph clade was sister to the rest of the Ardissonea sensu lato clades in the IQTREE and ExaBayes analyses (bs = 85%, pp = 0.98, respectively), but in a clade with Climacosphenia which was sister to Toxarium in the RAxML analysis (bs = 30%).

Discussion
The results support a conclusion that there is one monophyletic group containing six genera (Table 6), i.e., Ardissonea sensu stricto with double walls and pseudosepta; (2) Synedrosphenia, including Ardissonea crystallina and A. fulgens, with single walls and pseudosepta; (3) Ardissoneopsis gen. nov., including Synedra undosa Grunow, with single walls but without pseudosepta; (4) Grunowago gen. nov. including Synedra bacillaris, with a central costa; (5) Toxarium with scattered areolae on the valve face; and (6) Climacosphenia with craticular bars on the valvocopula. The following paragraphs further discuss the new and emended genera, before considering the limitations of the study and what we can conclude about the Family/Ordinal level.
There is a clear distinction within Ardissonea sensu lato, between species with a double wall and those with a single wall, as noted by Round et al. (1990), i.e. the double wall is a synapomorphy for Ardissonea sensu stricto. Even without Ardissonea s.s., single-walled taxa under the name Ardissonea were not monophyletic and resolved into two clades. In Micronesia, several species differ subtly but perhaps significantly from the known taxa, and there are species sufficiently different to be recognized as undescribed even without sequencing. The two clades of single-walled Ardissonea spp. can be separated on the structure of the poles, one group, including Synedrosphenia gomphonema and therefore taking that genus name, with complex polar architecture (also seen in Ardissonea s.s.) including a narrow, often asymmetrical extension (pseudoseptum) of the rim of the valve at the apical pole, a corresponding folding of the valvocopula to form a shelf with a groove or notch, often overlapped internally by very long fimbriae of the copula. The other group has simple polar architecture, commonly seen in diatoms, with a straight edge to the mantle and the pars interior of the valvocopula; we propose Ardissoneopsis for this group, including three new species, and typified by Grunow's obscure species Synedra undosa. Toxarium is within a clade with the other genera discussed here in two of three molecular analyses and does not clearly separate in morphological cladistics, suggesting that all belong in one family.
Synedrosphenia was kept separate from Synedra by Hustedt (1931Hustedt ( -1959, even though he rejected Ardissonea as a genus. His argument was that, despite the many similarities, the heteropolar shape of the valve was more important, and he cited pairs of genera, e.g., Gomphonema and Navicula, in which heteropolar species were separated from isopolar. Synedrosphenia was said by Round et al. (1990) to be in "urgent need of re-investigation," but, based on the ultrastructure and cell shape, they recognized a relationship with Climacosphenia. With more recent EM studies on single-walled Ardissonea species (Poulin et al. 1987;Kaczmarska et al. 2020), especially the recognition of complex polar architecture in Ardissonea fulgens type material (Kanjer et al. 2021), and our observations on Synedrosphenia from Micronesia and Grunow's historical material from Honduras, it is now clear that the ultrastructural similarities must outweigh the heteropolarity. (We reserve judgement on specimens identified in the literature as A. fulgens var. gigantea for reasons detailed in Results.) As seen in our new species, the shape presents more of a gradient and strongly suggests that all the single-wall Ardissonea and new species with a pseudoseptum should be placed in Synedrosphenia rather than erecting a new genus. A dearth of sequenced taxa means that this conclusion is not yet supported by molecular data.
The new genus Ardissoneopsis is necessary to accommodate Ardissonea-like species with single walls but simple polar architecture. These include not only two species in Micronesia that we had previously identified with Ardissonea fulgens and its var. gigantea, but also Grunow's Synedra undosa, which we were able to observe for the first time in SEM, and a new species. While it is better to have a positive synapomorphy for a genus, the complexity of the apical structure of valve and valvocopula in some other genera in this group is a strong character, comparable to the presence of an inner wall layer, and warrants separating those that have it from those that do not. This genus is presently small but we anticipate it will grow as Ardissoneaceae are more fully explored in other regions.
The new genus Grunowago is required to accommodate a new species and Synedra bacillaris. The latter cannot remain in Synedra, which, although defined by Round et al. (1990) as a freshwater genus and now clarified as a small marine genus, with Catacombas as a junior synonym (Williams and Karthick 2021), has a double wall, rimoportulae, and polar pore fields. Valve structure in Grunowago has some important differences that seem to set it apart from the Ardissoneaceae, including a strong central costa and lack of lateral costae that implies bilateral growth rather than annular growth. Nevertheless, sequence results show that it is part of this Family. Not all the other Ardissoneaceae have evident annular rings, however. As noted above, the ring is not visible in Synedrosphenia bikarensis; this species has not been sequenced yet. Our hypothesis to explain this is that there is annular growth in these species, since that is the defining character of the group. Regardless of its eventual systematic position, this genus helps resolve Synedra.
Our study was largely restricted to observations on materials from Western Pacific Islands, those from Guam usually examined fresh, others only after acid cleaning. Having stubs with remnants from Grunow's Honduras materials, we made use of them, but the large specimens were very broken up and we had much more success with species with distinctive shapes (Synedra undosa,Toxarium undulatum,Figs 17,24) than with the linear species with several similar species that we could not identify. We managed to identify putative Synedra crystallina var. bacillaris (Fig.  19) but it was hard to be sure that the specimens did not have a double wall when Ardissonea was also present. We restricted descriptions of new species to those that could be confidently differentiated from known species, leaving others, such as the two Toxarium spp. with the widely used names as working hypotheses. The other major limitation in this study is low taxon sampling for sequencing. Low taxon sampling may not change quickly, as Medlin et al. (2008) commented that Ardissonea and Climacosphenia are difficult to culture, and that has been our experience as well. Many of the new species were found only after acid cleaning of samples that had been preserved in the herbarium and therefore could provide no information on organic components, including DNA. Low taxon sampling is also the cause of the apparently sister relationship with the centric diatom Lampriscus (Fig. 33); it is closest but not close.
Despite the limitations of the study, we can draw some conclusions about the Families and Order. Molecular and morphological evidence support a hypothesis that the six genera in our flora are in one monophyletic family but there is no known common ancestor, from which to derive a synapomorphy for the Family, which now appears to stand alone in an Order with no other Mediophyceae nearby. In the past they were placed in several Families and usually more than one Order; the most recent summary of diatom taxonomy has them in four families under one Order, Toxariales (Cox 2015). Round et al. (1990) described Ardissonea based on the type species, A. robusta, without reference to the single walled forms that Poulin et al. (1986) had used as the basis for their emendation. However, they also defined a Family and an Order with one and the same paragraph, which, because it was based entirely on the characters of A. robusta, even excludes the single-walled species from these higher taxa. Cox (2015) partially resolved these issues by eliminating the Ardissoneales and placing the family in Toxariales. However, her description of Ardissoneaceae mirrors Round's and she lists only one genus-Ardissonea, with only 2 species (presumably A. robusta and A. formosa), still leaving the single-walled taxa in limbo. Moreover, she also follows Round et al. (1990) in placing Synedrosphenia in Climacospheniaceae, which is defined as having valvocopulae with septa that meet in the middle, true only for Climacosphenia. We therefore need to emend the description of Ardissoneaceae. First, Ardissonea, Ardissoneopsis and Synedrosphenia (Groups 1-3 as defined above) clearly all belong in Ardissoneaceae as presently conceived. Second, there seems to be little justification for maintaining separate Family+Order for each of Toxarium and Climacosphenia, given the similarities to the Ardissoneaceae. Toxarium has characters of both Ardissoneopsis (weakly developed costae) and Synedrosphenia (pseudoseptum + valvocopular notch) and is distinguished only by the scattered areolae in the wider areas (when there are any inside the annulus). While the craticular bars of Climacosphenia are its defining character, unique within the Order, we do not see this as warranting a separate family, especially considering the distribution of such valvocopulae within the biraphid genus Climaconeis, where craticular bars distinguish only 2 of 20 species (Lobban 2021a). We therefore propose to transfer Toxarium and Climacosphenia to the Ardissoneaceae and to restore that family to the Ardissoneales, emended.
Much work remains to be done even in our region to tease out the biodiversity of these genera, and that work depends on progress in other regions and especially on type materials. Some species, e.g., Ardissonea formosa and A. pulcherrima, were described from the East Indies, others from Atlantic/Caribbean waters, e.g. A. robusta from Kattagat Strait, Europe, and several species from Honduras. In many cases precise type locations are unknown -Hantzsch received samples labeled only from the East Indies Archipelago; Lindig did not inform Grunow whether his Honduras samples were from its east or west coast. It is tempting but counterproductive to assume that everything is everywhere. Moreover, Williams (2011) and Williams and Kociolek (2017), among others, have found evidence of regional endemicity in freshwater diatoms, and our studies of Licmophora (Lobban and Santos in review) indicate the same in that marine genus. Williams (2011) even proposed that we should start from the assumption that everything is endemic, rather than that everything is everywhere. At least we should not assume that everything is everywhere, nor that there is anywhere that the full biodiversity of Ardissoneaceae has been discovered. Likely more species will be resolved with further study both in our region and in other places where these large but understudied species occur.

ArdissoneAles Round 1990, emend. Lobban & Ashworth
Description. Highly elongate valves with development from a bifacial annulus (not always visible) and lacking rimoportulae and apical pore fields (ocelli/ocellulimbi). Frustules often morphologically heteropolar in valve view, girdle view, or both, attached to substrata by mucilage stalks. Valves with simple areolae in uniseriate striae, internal transapical costae more or less developed. Valvocopula and copula closed, fimbriate internal margins. Pleura (when present) narrow, fimbriate inner margin, with larger apical caps.