Frustular morphology and polyphyly in freshwater Denticula (Bacillariophyceae) species, and the description of Tetralunata gen. nov. (Epithemiaceae, Rhopalodiales)

, is proposed for D . vanheurckii and its allies. This new genus is restricted to Java and Bali. Conclusion – This work emphasizes the utility of evaluating morphological features within an established phylogenetic context and is part of a larger study to investigate relationships among the Bacillariales using morphological and eventually, molecular data.


INTRODUCTION
Raphid diatoms are a monophyletic lineage (e.g. Medlin et al. 1996, Medlin & Kaczmarska 2004, Sorhannus 2004, Theriot et al. 2010) that possess a slit through their siliceous cell wall called a raphe. The canal raphe system is a specialized type of raphe in which the raphe is separated from the cell interior by struts of silica referred to as fibulae. As a result of the position of the raphe and the fibulae, the raphe is essentially inside a canal with small 'windows' called portulae, connecting the raphe canal to the cell interior (Round et al. 1990). Once considered a homologous feature (e.g. Hustedt 1928, Schrader 1973, molecular phylogenies have revealed the canal raphe system has evolved in at least two independent lineages, the Bacillariales Hendey and the Rhopalodiales D.G.Mann + Surirellales D.G. Mann (e.g. Sorhannus 2004, Sims et al. 2006, Bruder & Medlin 2008, Ruck & Theriot 2011). One genus that bears a canal raphe, Denticula Kütz., is of particular interest because its systematic position has been debated.
Denticula was originally described as belonging to the Fragilariaceae Grev. (Kützing 1844), a group of diatoms with a longitudinal sternum but lacking a raphe system. Denticula was placed in this group primarily based on the presence of costae, and was recognized as separate from Diatoma Bory because of its growth as individual cells and not in the zigzag colonies typical of the latter genus (Kützing 1844). The original description of Denticula was broad and included some species that have since been transferred to a wide range of genera, including Diatoma, Rhabdonema Kütz., and Cymatopleura W.Smith (Fourtanier & Kociolek 2011). Additional authors (Schütt 1896, Van Heurck 1896, Mann 1907, Schönfeldt 1907 continued to classify Denticula as an araphid genus into the 20 th century. Van Heurck (1896) questioned the position of Denticula within the fragilarioid diatoms and suggested that it may have a closer affinity with Nitzschia Hassall based on the presence of a keel, a thickened elevated structure for the raphe on some valves, but still included it in the Tabellariineae cohort Van Heurck (most closely related to Diatoma). During its classification as an araphid diatom, Denticula tenuis Kütz. was designated the type species of the genus by Mann (1907), a status often attributed to D. elegans Kütz. by Boyer (1927), a later authority (Fourtanier & Kociolek 2011). While earlier workers included only freshwater taxa in the genus, Boyer (1927) included marine species as well.
A second approach was forwarded by Grunow (1862), who disagreed with Kützing (1844) and moved Denticula into the Bacillariaceae Ehrenb., a group that included Nitzschia, Tryblionella W.Smith, Bacillaria J.F.Gmelin, and Homeocladia C.Agardh. Grunow argued the characteristic feature uniting this group was the strongly eccentric keel. Boyer (1927), Simonsen (1979), and Round et al. (1990) agreed with Grunow's classification. However, Grunow (1862) described Bacillariaceae as a family separate from the Epithemiaceae Grunow and the Surirellaceae Kütz., whereas Boyer (1927) and Simonsen (1979) placed the nitzschioid group closer to the Surirellaceae. Boyer (1927) suggested nitzschioid diatoms were closely related to Epithemia, but closer to Surirella and that Denticula may be separate from all of these based on their presence of thickened siliceous ribs ('costae'), even mentioning Denticula had a potential affinity to Diatoma because of their similar 'striation' patterns. Based on his study of marine species (Simonsen & Kanaya 1961), Simonsen (1979) erected a new genus for the marine taxa of Denticula, Denticulopsis Simonsen, and placed this new genus and Denticula within the Bacillariaceae. Subsequently many of these marine species, recent and fossil, have been transferred into new genera within the Bacillariales (i.e., Crucidenticula, Neodenticula; Akiba & Yanagisawa 1985). Bessey (1899) and Simonsen (1979) indicate the Bacillariaceae may have evolved from the Naviculaceae Kütz., with no direct relationship to the Epithemiaceae or Surirellaceae, a theory later supported by molecular phylogenies (e.g. Sorhannus 2004, Bruder & Medlin 2008, Ruck & Theriot 2011. Round et al. (1990) agreed with Grunow (1862) and separated the Bacillariales (including Denticula) from the Rhopalodiales and the Surirellales.
A third perspective was offered by Hustedt (1914Hustedt ( , 1928, who studied four Denticula taxa (i.e., D. tenuis, D. thermalis Kütz., D. valida (Pedicino) Grunow in Van Heurck, D. vanheurckii Brun) and, based on the shared features of raphe structure and presence of transapical fibulae that partition the valve, he transferred Denticula into the Epithemioideae. Karsten (1928), Patrick & Reimer (1975),  followed Hustedt's classification. Karsten (1928) based his agreement with Hustedt's classification on examination of a single taxon, D. vanheurckii.  examined a greater number of taxa (seven) in their own work and based their classification on Geitler (1977), who studied the cytoplasmic and sexual (auxosporulation) characteristics of only two Denticula species, D. tenuis and D. vanheurckii. Geitler (1977) found variability within Denticula, mainly the presence of cyanobacterial endosymbionts and one plastid in D. vanheurckii (allying this lineage with Epithemia and Rhopalodia) and the lack of endosymbionts and two plastids in D. tenuis.  followed Geitler (1977) by considering D. tenuis an aberrant taxon within the Epithemiaceae. A molecular study later revealed the presence of endosymbionts within Epithemia and Rhopalodia to be the result of a single evolutionary event (Nakayama et al. 2011).
In each of these proposals regarding the systematic position of Denticula, the genus has been regarded as monophyletic. Sims (1983), Mann (1989), and Round et al. (1990) hint at an alternative idea, where some species of Denticula may be closely related to the Bacillariales, and others (specifically D. vanheurckii) more closely related to the Rhopalodiales. Specific observations in support of this scenario have not been published.
Today, Denticula is considered a relatively small genus with only ~ 100 taxa (Fourtanier & Kociolek 2011) with few taxa (eight) reported from Europe and the United States (Patrick & Reimer 1975, but considerably higher taxon richness (nineteen) in Sumatra (Hustedt 1938) and Java (Brun 1891). Denticula species are often found in the littoral margins of lakes, ponds and streams with high conductivity, with particular species inhabiting either cold mountain streams or hot springs (e.g. Kützing 1844, Patrick & Reimer 1975. Morphologically, Denticula species are united by a linear to lanceolate valve outline, symmetry about the apical and transapical axis, and a canal raphe system subtended by large internal fibulae that either partially (e.g. D. kuetzingii Grunow) or entirely (e.g. D. tenuis) extend across the valve face from margin to margin, forming partitions within the valve (Round et al. 1990). Some authors have included taxa with either partial or entire fibulae in Denticula (e.g. , whereas others have placed taxa with only partial fibulae in different subgroups (Schönfeldt 1907, Grunow 1862 or other genera (Nitzschia, Lange-Bertalot & Krammer 1993). In addition to fibula features, raphe characteristics (i.e., position, presence/absence of a keel, and whether the raphe is continuous or interrupted) and valvocopula structure are variable and often used for species identification . Limited investigation of cytoplasmic characteristics has revealed the presence of endosymbiotic cyanobacteria in one taxon (D. vanheurckii), but not others (D. tenuis) and either one or two chloroplasts (Geitler 1977).
The goal of this study is to document with light and scanning electron microscopy the valve and cingulum structure of several Denticula taxa, including the generitype. These observations are expected to cast light on the systematic position of Denticula species as compared to Epithemia (Rhopalodiales) and Nitzschia (Bacillariales). Included in this study is an examination of Hustedt's material from Sumatra containing species that many authors (Sims 1983, Mann 1989, Round et al. 1990) have mentioned as being closely related to Epithemia.

Material
Samples examined include: recent collections from Colorado; the Southern California Coastal Water Research Project (SCCWRP) from 2004 to 2008 (www.sccwrp.org); a larger survey of the flora of Blue Lake, Utah; and the Friedrich Hustedt collection. Collection information including water chemistry data (when available) is presented in table 1. All materials discussed have been placed in the Kociolek Collection at the University of Colorado, Museum of Natural History Diatom Herbarium, Boulder, Colorado and are available upon request. Two of the taxa, D. kuetzingii and D. valida, were isolated and cultured (cultures available upon request) from the Colorado site and one of the Utah sites, respectively (see site information below) prior to examination. Morphology of these isolates was confirmed with field material to ensure that culturing conditions did not cause aberrant morphologies (both culture and field material are pictured). These two taxa were isolated and grown in WC and modified WC [same nutrients, but with artificial sea salt Instant Ocean® (Blacksburg, Virginia) to increase conductivity] media, respectively in 12:12 light:dark conditions. The Colorado site was located along the edge of Sheep Lakes within Rocky Mountain National Park at 2590 m above sea level. Sheep Lakes are a series of small ponds ~ 3 km from the Fall River entrance to the park. The sample from this site was composed of epiphytes collected from Chara spp.
The California site is located in Coldbrook Creek, a tributary off the North Fork of the San Gabriel River, in the An-geles National Forest northwest of Los Angeles, California. The site is ~ 85 km from the Pacific Ocean at 1021 m above sea level with an average depth of 9 cm and a discharge of 0.078 m 3 s -1 . Water chemistry data and a composite sample of all available habitats were collected in the summer of 2008 by SCCWRP.
The two Blue Lake, Utah sites were located on the littoral edge of Blue Lake and surrounding ponds (see table 1 for exact localities). Blue Lake is a natural warm spring located 26 km south of Wendover, Utah at 1300 m above sea level. Temperature, conductivity, and pH were measured near shore using a YSI 556 multiprobe (YSI Incorporated, Yellow Springs, Ohio) at the time of sampling. The two samples analyzed here were a composite sample from the benthos (JPK 8512) and a sample of epiphytes on grass (JPK 8535).
Samples from Sumatra were analyzed by Hustedt (1935Hustedt ( , 1938 as part of his evaluation of the freshwater diatoms of Java, Bali and Sumatra (table 1). We are indebted to Dr. Friedel Hinz for supplying four samples from the Hustedt Collection for this analysis. The Sumatran material was collected originally by Ruttner (1931) as part of a 10-month German expedition (1928)(1929) to document the biota of lakes and rivers in Bali, Java, and Sumatra. The four epiphytic samples analyzed for this study were all collected from lakes in Sumatra, including: one sample (JPK 9297/BRM AS555) of epiphytes collected from Potamogeton sp. is from the north shore of Lake Ranau; one sample (JPK 9298/BRM AS756) of epiphytes from moss growing on a rock from Lake Diatas; and two samples (JPK 9299/BRM AS869; JPK 9300/BRM AS871) of epiphytes from two sites in Lake Toba (both collected from Potamogeton sp.

Methods
Each sample was boiled in nitric acid. Cleaned materials were rinsed with filtered water and settled systematically until neutral pH was achieved. For light microscopy (LM) observations, cleaned materials were dried onto coverslips, mounted to slides with Naphrax®, and examined using an Olympus BX-51 LM (Olympus America Inc., Center Valley, Pennsylvania). Photomicrographs were captured using an Olympus DP 71 digital camera. For scanning electron microscopy (SEM) observations, cleaned material was dried on coverslips that were mounted on aluminum stubs with double-sided carbon tape. Dried material was sputter coated with 1 nm of gold using a Cressington 108 sputter coater (Cressington Scientific Instruments Ltd, Watford, UK) and examined using a JEOL JSM 6480LV low vacuum SEM at an acceleration voltage of 15 kV or a JEOL JSM 7501 field emission SEM (JEOL Ltd, Tokyo, Japan) at an acceleration voltage of 3-5 kV.
Morphological investigation of specimens of Denticula vanheurckii and Epithemia zebra var. denticuloides suggested these taxa belonged to the Rhopalodiales, while the remaining Denticula spp. studied here (D. kuetzingii, D. rainierensis D. tenuis, D. valida) are closely allied with the Bacillariales. To examine the phylogenetic relationships within the Rhopalodiales and determine whether D. vanheurckii and E. zebra var. denticuloides represent a monophyletic lineage from Indonesia, we performed a cladistic analysis. Other taxa included in the analysis were chosen due to their morphological similarity to these taxa (Epithemia, Rhopalodia, and Surirella). Amphora and Navicula Raphe is not in a canal (e.g., Amphora)

Raphe position
Raphe is positioned along the ventral margin of the frustule Raphe is positioned along the center of the frustule

Valvocopulae simple bands
Valvocopulae a narrow scalloped band with corrugations that clasp the 'entire' fibulae (e.g. Epithemia)  . Character states of Epithemia, Rhopalodia, Surirella, Amphora, and Navicula were determined by descriptions and images presented in Sims (1983) and Round et al. (1990). Nine unordered and equally weighted characters, described below, were utilized, two of which were multi-state characters (table 2). Coding of the character states analyzed is presented in table 3. The data matrix was analyzed using the branch and bound procedure in PAUP* 4.0b10 to determine the most parsimonious trees.

Terminology
Despite past (Anonymous 1975, Ross et al. 1979) and recent (Cox 2012) efforts to standardize descriptions and features, terminology used to describe diatom morphology generally, and Denticula in particular, is often confusing and non-uniform. In the context of Denticula specifically, this inconsistency in terminology can be attributed to the great morphological variability within the genus and the variable systematic position of the genus (and traditional usage of terms associated with each of these groups). In an effort to clarify these terms, provided here are definitions, references, synonyms, and figures illustrating terminology previously applied to Denticula and/or similar species that were used in this study. Areolar strut. Small strut of silica projecting inward from the walls of the areolae that often support volae (Sims 1983). Cribrum. A thin layer of silica across an areola and perforated by small, regularly arranged pores (Anonymous 1975 & Akiba 1990). No evidence was found that the 'deck' is a separate plate of silica from the valve in this group of species, and as such the use of this term should be discouraged for these groups. Evidence contradicting the occurrence of a deck is presented in figure 3F. Domed cap. External small cap of silica covering an areola that is supported internally by areolar struts originating from the margins of the areola (Sims 1983  Fibula. Internal strut of silica that provides structure and support for the raphe canal (Round et al. 1990 1E); or entirely across the valve, termed 'entire' (e.g. fig. 2C). Although the term costae is often used to describe fibulae that extend farther across the valve than marginally (in Epithemia and Rhopalodia by Round et al. 1990), the term 'costae' is vague and has been used to describe structures that are not homologous in disparate groups of diatoms (e.g. Diatoma, Meridion, Epithemia, Rhopalodia). Additional synonyms for fibulae include: keel puncta, pseudosepta (sensu Yanagisawa & Akiba 1990, Simonsen & Kanaya 1961, and Fibularwände (Krammer & Lange-Bertalot 1988).
Portula. The round to elliptical opening created by fibulae widening along the apical plane closer to the valve. Fig. 2C versus fig. 4F.
Vola. Flap-like areola covering that originates from the edge of areolae (Sims 1983). Fig. 5H, arrowheads. (1880); see Schönfeldt (1907) and Lange-Bertalot & Krammer (1993) who suggested these synonymies. LM: description -Frustules with nitzschioid symmetry. In valve view, valves narrowly lanceolate to elliptical to linear with convex to parallel sides and narrowly rounded to acute apices, 13-33 µm long and 4-5 µm wide ( fig. 1A, appendix 1A-G). Striae parallel in the center to convergent to the apices, but evenly spaced throughout the valve, 16-18 striae in 10 µm. Canal raphe continuous and eccentric with no obvious keel, fibulae number 5-8 in 10 µm. In LM, fibulae appearing well defined near the canal raphe, merging with the valve face across the apical axis (appendix 1C-F), but possibly visible across the entire valve if focused on their base (appendix 1B). If focusing on the top of the valve, a small round nodule visible along the fibulae close to the canal raphe (appendix 1A & G, arrows). In girdle view, cells rectangular and the small round nodules on each fibula more obvious (appendix 1H, arrow). SEM: description -Canal raphe raised slightly above the valve face lying at the junction of the valve face and the mantle ( 1). Patrick & Reimer (1975) Fig. 3, appendix 1P-U LM: description -Frustules with nitzschioid symmetry. In valve view, valves lanceolate to linear with convex to parallel sides with the raphe side often having a slight constriction indicating an intermissio, and rounded to acute apices, 18-37 µm long and 5-7 µm wide ( fig. 3A, appendix 1P-T).

Denticula valida sensu
Striae parallel in the center to slightly convergent to the apices, but evenly spaced throughout the valve, 19-22 striae in 10 µm. Canal raphe eccentric with no obvious keel, fibulae number 3-4 in 10 µm. Fibulae extending entirely across the transapical axis and parallel throughout the valve with a slightly larger gap between fibulae towards the valve constriction. If focusing on the fibulae, two humps visible on the fibulae towards the valve margins with a slightly more silicified area between them, giving the fibulae the appearance of having a small bow tie (appendix 1T). In girdle view, cells rectangular with capitate fibulae (appendix 1U, arrow). SEM: description -Raphe situated between the valve and the mantle ( fig. 3B) with simple proximal raphe ends ( fig. 3C, arrow) and hooked distal raphe ends ( fig. 3D). In the external view, striae consisting of elliptical to rectangular areolae increasing in size and density towards the margin opposite the raphe ( fig. 3B & C). Striae extending transapically across the valve and onto the mantle ( fig. 3B). From the interior,     table 1). Remarks -Aside from symmetry differences, this taxon resembles D. vanheurckii with regard to valve and cingulum morphology. Cladistic analysis identified one most parsimonious cladogram with 14 steps, a consistency index of 0.73, and retention index of 0.64 ( fig. 8). Synapomorphies uniting the ingroup included the presence of a canal raphe and frustules with asymmetry along the apical axis. Surirella was diagnosed by the continuous raphe around the entire periphery of the valve and the presence of a keel. Synapomorphies uniting the Rhopalodiales included lunate areolae and 'entire' fibulae extending across the internal valve face. Rhopalodia was defined by the presence of a keel. An alternative interpretation, equally parsimonious, of the evolution of the keel (character 4) is that it diagnoses all of the canal raphe forms considered in this lineage, and was subsequently lost in Epithemia spp. + D. vanheurckii clade. The Epithemia spp. + D. vanheurckii clade was defined by the raphe shape and valvocopulae structure, two features that were subsequently modified in Epithemia. Epithemia zebra var. denticuloides was most closely related to Epithemia, not D. vanheurckii. The slightly arched raphe and unique valvocopulae structure expected to unite E. zebra var. denticuloides and D. vanheurckii were shown to be symplesiomorphic characters, and most likely ancestral to Epithemia. The synapomorphy distinguishing D. vanheurckii and its allies from Epithemia (including E. zebra var. denticuloides) is its symmetry along the apical axis.

DISCUSSION
As presently circumscribed, the genus Denticula includes a wide range of morphological diversity. Aside from their more or less symmetry about both the apical and transapical axes, and presence of fibulae extending partially or entirely across the valve face, few additional features serve to unite the species within Denticula that would indicate they represent a natural group. In fact, not only may this assemblage of species be allied with different genera, but also their morphological differences are great enough to ally them with different orders within the diatoms.
The first taxon examined, Denticula kuetzingii, is more similar to other Denticula species along what appears to be a morphological gradient between Nitzschia and Denticula based on the degree the fibulae extend across the valve face and the extensions on the valvocopulae (Lange-Bertalot & Krammer 1993). Lange-Bertalot & Krammer (1993) suggest other taxa in this category include: N. amphibia Grunow, N. denticuloides Hust., N. robusta Hust., N. semirobusta Lange-Bert., and N. subdenticula Hust. This was a change of opinion by Lange-Bertalot & Krammer (1993), since their earlier treatment of the Bacillariales allied Denticula with Epithemiales . In fact, Lange-Bertalot & Krammer (1993) referred to D. kuetzingii as Nitzschia denticula and described a subgenus, Denticuloidea (Lange-Bertalot 1993), which includes these 'borderline' taxa. This debate is not new; many authors have placed D. kuetzingii in Denticula (Grunow 1862, Schönfeldt 1907 or Nitzschia (Cleve & Grunow 1880, Boyer 1927. A recent molecular study by Ruck & Theriot (2011) places D. kuetzingii firmly within the Bacillariales. Although the morphology discussed here agrees with the conclusions of Ruck & Theriot (2011) (i.e. Denticula sensu stricto is closely allied with Nitzschia), placement of Denticula within the Bacillariales requires further examination because of the absence of the generitype of Denticula in current molecular analyses. Round et al. (1990), reviewing the morphology of D. tenuis, suggested its affinity within the Bacillariales. This placement of D. tenuis, the generitype of Denticula, is suggested by the chloroplast arrangement, presence of a keel, raphe structure, areola structure, and girdle structure (Mann 1989, Round et al. 1990). Denticula tenuis may be most closely related to N. sinuata (Thwaites in W. Smith) Grunow in Cleve & Grunow and its relatives (Mann 1989, Round et al. 1990), a group that had been included previously in the genus Grunowia Rabenhorst (1864). Should these species be shown to form a natural group (sensu Kociolek 1997) and be recognized within the Bacillariales at the level of genus, Denticula Kütz. (Kützing 1844) would have nomeclatural priority over Grunowia. Also included here would be Denticula valida and its allies (e.g. Denticula elegans Kütz., Denticula kittoniana Grunow in Van Heurck). This group includes species that possess a raphe with or without an intermissio. The ecological breadth of this putative group is great, with species that are found in cold mountain lakes and streams (e.g. D. tenuis; Schönfeldt 1907, Patrick & Reimer 1975, to high conductivity environments (e.g. N. sinuata var. tabellaria; , to those able to tolerate elevated temperatures (e.g. D. valida ;Patrick & Reimer 1975). Denticula as circumscribed here is firmly within the Bacillariales. Further work is warranted to determine whether the species discussed here form a monophyletic group.
Denticula rainierensis, by virtue of lacking a keel, the organization of its striae, and position of the raphe on the mantle rather than the valve face, differs significantly from the previous two groups discussed above. Denticula rainierensis is very similar to D. subtilis Grunow, differing primarily in valve metrics and the presence/absence of portulae (Johansen et al. 1994, Lange-Bertalot & Krammer 1993. Denticula rainierensis and D. subtilis are most commonly reported from inland hot or warm springs and estuaries, respectively (Sovereign 1963, Wojtal 2013. In terms of morphology, these species appear to be related to Fragilariopsis and Neodenticula, being comparable to species in these genera by their striae and raphe structure (Round et al. 1990, Poulin et al. 2010. In phylogenies of the Bacillariales based on partial nuclear-encoded large subunit (LSU) rDNA, Fragilariopsis and Neodenticula form a natural group, part of a lineage that also includes Pseudonitzschia, with Nitzschia frustulum (Kütz.) Grunow in Cleve & Grunow being a sister taxon to this lineage (Lundholm et al. 2002, Poulin et al. 2010. Denticula rainierensis and D. subtilis differ from Fragilariopsis by having parallel striae at the apices and internal protrusions on the valvocopulae (Johansen et al. 1994, Round et al. 1990. In contrast to Neodenticula, these two species have a shallower mantle and an interrupted raphe (Poulin et al. 2010). A more robust phylogenetic analysis of this group of species is required before proposing their placement in the classification system of the Bacillariales.
Denticula vanheurckii was described originally from Java (Brun 1891), and Hustedt (1935, 1938 described eighteen taxa with similar morphology from Sumatra. This species has a raphe system quite similar to naviculoid diatoms, with two branches positioned in the middle of the valve face. A keel is lacking. In terms of valve morphology, the complex arrangement and structure of the areolae firmly place this species group within the Rhopalodiales. Similar complicated areolar structure can be seen in Epithemia species (Sims 1983, pers. obs.) and a modified version in Rhopalodia (Round et al. 1990). In addition, D. vanheurckii has a complex series of closed and open girdle bands, additional support for its placement within the Rhopalodiales (Sims 1983, Round et al. 1990). In contrast, the Bacillariales have either open or closed girdle bands, but both states do not occur within the same species (Round et al. 1990).
The relationship of these Sumatran species with members of the Rhopalodiales is underscored by Geitler's (1977) report of the presence of blue-green endosymbionts in D. vanheurckii. Symbionts in other members of the Rhopalodiales have been well documented (Lowe et al. 1984, DeYoe et al. 1992, and a recent phylogeny of the Rhopalodiales and the symbionts themselves, suggest the endosymbionts were acquired by a common ancestor of these diatoms and have been retained throughout host speciation (Nakayama et al. 2011). Denticula vanheurckii differs from other members of the Rhopalodiales by its valve symmetry and position and structure of its raphe. In terms of the raphe structure, D. vanheurckii is similar to Nagumoea Kociolek & Witkowski , a canal-bearing genus that has two distinct raphe branches per valve positioned in the middle of the valve and lacking a keel . Structure of the valvocopula (enveloping the fibulae versus simply undulated) and areolae (complex alveolate versus simple pores) distinguish the two. Witkowski et al. (2011) suggested Nagumoea to be more closely related to the Bacillariales, but further research is needed to verify the systematic position of this enigmatic genus.
The suite of features related to position and structure of the raphe as well as areolar morphology, diagnose a lineage with D. vanheurckii to the exclusion of Epithemia and Rhopalodia. To include D. vanheurckii in Epithemia would require the broadening of the concept of Epithemia to include valves that are either symmetrical or asymmetrical to the apical axis, placement of the raphe as ventral or central, and the arched or biarcuate shape of the raphe system. Instead of dramatically proposing this new, broad concept of Epithemia, we propose here a new genus for Denticula vanheurckii and its apically symmetrical allies:  fig. 8; pl. 14, fig. 10. 1891 (Brun 1891). Description -Valves symmetrical, slightly asymmetrical, or strongly asymmetrical about the apical axis. Raphe medial, straight to barely biarcuate, composed of two distinct branches. Externally, raphe in a distinct, wide, straight axial area with proximal raphe ends dilated and deflected in the same direction and distal ends extending onto the mantle; keel lacking. Internally, raphe within a canal containing a large, ovoid central portula and a series of smaller, round portulae extending to the apices. Portulae across a wide hyaline strip that runs the length of the valve. Central nodule small, linear in outline. Fibulae distinct, extending entirely transapically across the valve. Striae distinctly areolate, 4-8 rows of striae between the fibulae. Areolae clover-shaped, with 4-5 cshaped occlusions. Valvocopula closed with extensions over the fibulae and meeting at the center with small, interdigitating teeth. Additional copulae a complex series of closed and open bands. One large chloroplast present, in additional to spherical bodies that are endosymbiotic blue-green algae. Etymology -The generic name refers to the 'four crescents' shape of the areolae. Note -Based on the detailed light microscope images provided in Simonsen (1987: plates 345-349), on the taxa described by Hustedt from Lake Toba and environs (Hustedt 1935(Hustedt , 1938, which indicate similarities in valve structure, the following transfers to the new genus are presented below:  fig. 8; pl. 14, fig. 10. 1891 (Brun 1891 Basionym -Denticula vanheurckii var. angusta Hust., Archiv für Hydrobiologie, Supplement 15: 450;pl. 38, fig. 17, 18. 1938pl. 38, fig. 17, 18. (Hustedt 1938 fig. 31. 1935 (Hustedt 1935).
Epithemia zebra var. denticuloides, which is morphologically similar to D. vanheurckii, but is asymmetrical to the apical axis with the raphe positioned towards the ventral margin, was initially expected to be closely related to D. vanheurckii and its allies. However, cladistic analysis revealed these characters (raphe shape and valvocopulae structure) to be symplesiomorphies. Therefore, E. zebra var. denticuloides was not transferred to Tetralunata.
Representatives of this new genus appear restricted to Java and Sumatra, a region with many endemic plants and animals (Whitten et al. 1987). It is interesting to consider the biogeographic context of Tetralunata, as very few genera of freshwater diatoms have restricted distributions. Examples of other freshwater diatom genera with restricted biogeographic distributions include Eunophora Vyverman et al. (restricted to Tasmania and New Zealand, Vyverman et al. 1998), Gomphocymbella O.Müller (restricted to East African Rift Valley lakes region, Kociolek & Stoermer 1993), as well as much smaller genera Bicudoa (Wetzel et al. 2012) and Eunotioforma (Burliga et al. 2013) both reported only from Brazil, and Tibetiella (Li et al. 2010) known only from China. A large number of diatom genera have been described from Lake Baikal, Russia (Kulikovskiy et al. 2012), a region that has harbored endemic fossil genera and families of freshwater diatoms (e.g. Khursevich & Chernyaeva 1989. In In-donesia, where there are high levels of endemic species in terrestrial ecosystems, there are also well known instances of fish and crustacean adaptive radiations in freshwater lakes of Sulawesi (Herder et al. 2006, von Rintelen & Cai 2009). These Tetralunata species from Sumatra may form a species flock, i.e., many species of a monophyletic group that are geographically constrained and have evolved quickly (e.g. Sullivan et al. 2002). Most other putative examples of species flocks in diatoms include only a few species (Mann 1999, Edlund & Soninkhishig 2009, Seddon et al. 2011, with the exception of the genus Gomphoneis in Lake Baikal ). While we do not have divergence rates of this species flock from Sumatra, this is an area for future research.
Although Tetralunata and other members of the Rhopalodiales share a number of features including fibula and areola structure; their ability to utilize endosymbiontic cyanobacteria; and chloroplast characteristics (Geitler 1977), Tetralunata also differs from both Epithemia and Rhopalodia. Valve morphology of Tetralunata differs from Epithemia in the following: the ventral to apical instead of biarcuate position of the raphe; a large hyaline area above the canal raphe internally (compare to Epithemia turgida (Ehrenb.) Kütz., the generi type; Sims 1983); the lack of thickened 'secondary' fibulae that are thicker than the virgae but do not extend across the raphe, the valvocopulae suture along the apical axis, not the dorsal side of the valve; and the valvocopulae extend across the entire valve with adjacent copulae barely grasping the valvocopulae (Epithemia is vice versa sensu Sims 1983). The differences between Tetralunata and Rhopalodia are more striking, including: the ventral to apical instead of dorsal position of the raphe; the lack of a keel (some Rhopalodia spp.); raphe fissures not bordered by siliceous flanges or ridges; a more elaborate interlock between the valve and valvocopulae; and the absence of any difference in girdle band width between the dorsal and ventral margins (Round et al. 1990).
Tetralunata also differs from the recently described subgenus, Denticuloidea (Lange-Bertalot 1993). Denticuloidea includes the 'border-line' taxa between Nitzschia and Denticula (e.g. D. kuetzingii, N. robusta, N. semirobusta, etc.) and therefore is within the order Bacillariales. In contrast, Tetralunata is within the Rhopalodiales. Whereas both Tetralunata and Denticuloidea share the presence of fibulae, they differ in areolae and valvocopulae structure, raphe position and shape, and the presence/absence of endosymbionts.
The previous debate about the systematic position of the genus Denticula was hindered by acceptance of the group as monophyletic, and the assumed close relationship between canal raphe bearing groups, stated explicitly or implied. Whereas some focused on the raphe and keel structure of certain species, such as D. tenuis (e.g. Round et al. 1990), others noted the similarity of D. vanheurcki with Epithemia (Hustedt 1928, Mann 1989. Although previous authors such as Bessey (1899) and Simonsen (1979) had suggested separate origins of the Bacillariales and the Rhopalodiales + Surirellales, not until this was suggested by molecular data (Sorhannus 2004, Sims et al. 2006, Ruck & Theriot 2011 and the implications of these results detailed by Ruck & Theriot (2011), did diatom sys-tematists fully grasp the significance. Previously, in the system where canal raphe diatoms were thought closely related or indeed monophyletic, it may have been difficult to reconcile the possibilities of multiple origins of endosym biotic blue-green algae, or the diversity of forms within a genus such as Denticula. In the context of the morphological data presented here, despite the presence of a canal, Denticula as previously understood is polyphyletic. This study helps underscore the ability of morphology, in the context of an established phylogeny, to resolve issues of systematic position across what were once thought to be closely related species.