Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Does the Cosmopolitan Diatom Gomphonema parvulum (Kützing) Kützing Have a Biogeography?

  • Nelida Abarca ,

    Affiliation Botanischer Garten und Botanisches Museum Berlin-Dahlem, Freie Universität Berlin, Berlin, Germany

  • Regine Jahn,

    Affiliation Botanischer Garten und Botanisches Museum Berlin-Dahlem, Freie Universität Berlin, Berlin, Germany

  • Jonas Zimmermann,

    Affiliations Botanischer Garten und Botanisches Museum Berlin-Dahlem, Freie Universität Berlin, Berlin, Germany, AG Spezielle Botanik, Justus-Liebig-Universität, Giessen, Germany

  • Neela Enke

    Affiliation Botanischer Garten und Botanisches Museum Berlin-Dahlem, Freie Universität Berlin, Berlin, Germany

Does the Cosmopolitan Diatom Gomphonema parvulum (Kützing) Kützing Have a Biogeography?

  • Nelida Abarca, 
  • Regine Jahn, 
  • Jonas Zimmermann, 
  • Neela Enke


Diatom cultures of the G. parvulum species complex were established from seven different sites in the Faroe Islands, Sweden, Germany, Mexico and Korea, and were studied in detail. Eight morphodemes were identified which corresponded to the descriptions of the cosmopolitan taxon G. parvulum (Kützing) Kützing sensu lato: its nominate variety (var. parvulum), G. parvulum var. exilissimum Grunow and G. parvulum f. saprophilum Lange-Bertalot & Reichardt, G. [parvulum var.] lagenula Kützing plus four unidentifiable morphodemes. The concatenated analysis of the sequences of the markers 18SV4, rbcL, and ITS as well as morphological data resulted in a separation of four taxa based on their biogeography in Mexico, Korea, central Continental Europe and Northern Atlantic Europe. Mantel tests showed a significant correlation between molecular and geographical distances. The diagnoses of two taxa, G. parvulum sensu stricto, and G. lagenula, were emended, G. saprophilum elevated to species rank and epitypes designated. One species was newly described.


Diatoms are eukaryotic unicellular microalgae present in all types of water bodies and therefore often considered to be cosmopolitan and even ubiquitous. With their cell sizes ranging mainly between 10 µm and 200 µm, they seem to fit the hypothesis that microorganisms smaller than 1 mm have a cosmopolitan distribution in which the environment – here the ecology of the water – determines the occurrence of a taxon [1], [2], [3]. Recent studies on diverse protists [4], [5], [6], [7], [8], [9] and on different groups of diatoms revealed that, while some taxa are endemic for certain regions, islands or continents, others are more widely distributed [10], [11], [12], [13], [14]. For larger “flagship taxa” specific occurrences are more easily recognized than for small and inconspicuous species [15]. Although a general pattern of diatom biogeography remains to be discovered, some general ideas on the comparability of the floristic realms of diatoms and higher plants, and the cosmopolitan distribution of the most common species, to which our studied taxon Gomphonema parvulum belongs, have been discussed by Lange-Bertalot [16]. But foremost it seems that the quality and in-depth identification is key to a true and stable distribution record forming the basis for biogeography. The identification of diatoms relies mainly on micro-morphological characters which require in-depth knowledge of diatom taxonomy. If diatom diversity assessments are not done by experts of this taxon, the results are often misidentifications [17] and therefore misinterpretations of their distribution.

Molecular methods are playing an increasingly important role in the discovery and delimitation of diatom species and are revealing a hidden diversity, i.e. many taxa contain cryptic species, resulting in the description of many more species [18], [19], [20], [13]. Especially in cases where micro-morphological characters are insufficient for species delimitation, molecular data can provide additional characters and evidence for cryptic species. Several studies [21], [22], [14], [23] have shown that the molecular-phylogenetic analyses of diatom genera can support previous morphological observations. But there is also evidence that molecular sequence data sets yield significantly increased levels of taxonomic resolution compared to analyses of morphological data alone [24].

Gomphonema parvulum (Kützing) Kützing is a slender diatom of about 25 µm length and 6 µm width. It has long been recognized as a cosmopolitan, ubiquitous, and morphologically highly variable taxon for which many varieties have been described [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42] Taxonomic discrimination between morphologies is hard to disentangle because of similarities between species, supposed transitional forms, and/or high infra-specific variation. The current separation is based on micro-morphological characters of the valve, e.g., the pattern of the central area, density and branching of the striae, the form of the punctae, as well as the outline as revealed by light and scanning electron microscopy [43]. At the time of writing, Algaebase ( includes 28 taxon names with 23 varieties and 4 forms of which 12 are currently accepted taxonomically. Gomphonema parvulum s.l. has been reported worldwide as an abundant benthic species occurring over a wide range of freshwater environmental conditions, including a few saline lakes and brackish waters. In addition, cells from the Gomphonema parvulum group seem to respond sensitively to water-quality changes. Some (infra-specific) taxa in this species complex, therefore, were identified as indicators of environmental conditions in both ecological and paleoecological studies [23], [40], [44], [45], [46], [47], [48], [49]

The aim of this study is to determine whether taxa of the Gomphonema parvulum group from different ends of the Eurasian holarctic region can be differentiated by their morphological and molecular data, and if these in turn are different to taxa from the neotropics. Besides possible biogeographical distribution patterns, given the tangled taxonomy of this group mentioned above, it is necessary to clarify if morphological identification is supported by molecular data, and to assess potential cryptic species, as well as infraspecific morphological variation. These questions are tested with a number of morphometric features and three molecular markers on 21 cultures from 7 sites from the Faroe Islands, Sweden, Germany, Mexico and Korea. The recently-published paper and its available molecular and morphological data for several Gomphonema parvulum strains [50] will be compared to our data.

Materials and Methods

Field collection and cultures

Freshwater samples were collected from Northern Germany, Southern Sweden, Faroe Islands, Korea, and Mexico between 2004 and 2005 (Table 1, Fig. 1). The water samples were taken from public water bodies, no location was on protected or private land. No permits were required for the described study, which complied with all relevant regulations. The field studies did not involve endangered or protected species. 21 uni-algal cultures of Gomphonema parvulum taxa were established by micropipette isolation, for details see Romero & Jahn [51]. For morphological and genetic characterisation, cells were harvested in exponential phase and concentrated by centrifugation.

Table 1. Name, voucher number, origin, date, collector as well as Genbank Accession numbers of the studied strains.

DNA isolation

The harvested cultures were transferred to 1.5 ml tubes. DNA was isolated using either Dynal® DynaBeads (Invitrogen Corporation; Carlsbad, CA, USA) or Qiagen® Dneasy Plant Mini Kit (Qiagen Inc.; Valencia, CA) following the respective product instructions. DNA concentrations were checked using gel electrophoresis (1.5% agarose gel) and Nanodrop® (PeqLab Biotechnology LLC; Erlangen, Germany). DNA samples were stored at −20°C until further use. Long-term storage of samples will take place at −80°C within the DNA bank network [52].

PCR amplification

The samples were all amplified with the primer pair M13F-D512for 18S/M13F-D978rev 18S [53] for the V4 region of the 18S locus and the primer pair ITS1/ITS4 [54] for the ITS region. The rbcL locus was amplified in two overlapping parts using two different primer pairs; Diat-rbcL-F (5′-AAGTGACCGTKACGAATCTGG-3′) and Diat-rbcL-iR (5′-AAGAAWCGYTCTCTCCAACG-3′) as well as Diat-rbcL-iF (5′-AAGGHTTAAAAGGTGGTTTAGA-3′) and Diat-rbcL-R (5′-RTARAAACCTTTAATCA-3′). These primers were developed by Birgit Gemeinholzer at the BGBM. The polymerase chain reaction (PCR) for the V4 region was conducted following Zimmermann et al. [53]. For ITS the PCR mix (25 µl) consisted of 8.35 µl HPLC H2O, 2.5 µl 10× buffer S, 1.5 µl MgCl2, 5 µl pecGOLD dNTPs (each 1.25 mM), 2.5 µl Betaine (all products by PeqLab Biotechnology), 2 µl of each primer (20 pm/µl), 0.15 µl HotStarTaq® polymerase (Qiagen), and 1 µl DNA sample. Our ITS PCR regime followed Amato et al. [55] except for an increased annealing temperature of 48°C. For rcbL the PCR mix (25 µl) consisted of 12.9 µl HPLC H2O, 2.5 µl 10× buffer S, 1.5 µl MgCl2, 2.5 µl pecGOLD dNTPs (each 1.25 mM), 2.5 µl Betaine (all products by PeqLab Biotechnology), 1 µl of each primer (20 pm/µl), 0.15 µl HotStarTaq® polymerase (Qiagen), and 1 µl DNA sample. The PCR regime for rbcL initial denaturation 2 min at 94°C, 40 cycles of denaturation 1 min at 94°C, annealing 45 s at 52.8°C (rbcL-F/rbcL-iR) resp. 38.6°C (rcbL-iF/rbcL-R) and elongation 1 min 30 s at 72°C, followed by a final elongation of 10 min at 72°C.

PCR products were visualised in a 1.5% agarose gel and cleaned with MSB Spin PCRapace® (Invitek LLC; Berlin, Germany) following standard procedure. DNA content was measured using Nanodrop (PeqLab Biotechnology). The samples were normalised to a total DNA content >200 ng using the spectrophotometer Nanodrop® ND1000 (PeqLab Biotechnology).


The Sanger sequencing was conducted by Starseq® (GENterprise LLC; Mainz, Germany). As sequencing primers M13 tails were used for the V4 region, following Ivanova et al. [56]. The sequencing for ITS was conducted with the same primers used for the amplification.

The sequences were edited in ChromasPro (Technelysium Pty. Ltd.; Tewantin, Australia) aligned using ClustalW [57], and manually improved in BioEdit [58]. For part of the sequences the alignment was done in PhyDe® [59].

Acquisition of morphometric data

Diatom frustules were cleaned in the laboratory at 80°C in H2O2. The hydrogen peroxide was removed by repeated centrifugation and washing events using distilled water. Cleaned subsamples were dispersed on cover glasses, dried, and embedded in the diatom resin Naphrax® for further microscopic observation. Photomicrographs of the cleaned valves were taken, using a Zeiss Axioscope-microscope with Differential Interference Contrast (DIC) and an AXIOAM MRc camera. The morphology of cleaned frustules was examined and photographed using both LM and scanning electron microscopy (SEM Philips 515).

Identification of the cleaned valves was based on Krammer & Lange-Bertalot [60], Lange-Bertalot [41], Lange-Bertalot & Metzeltin [61]. Voucher slides (see Table 1) of all cultures are kept in the BGBM-Berlin and are documented in the AlgaTerra Information System.

Valve length, width, density of striae and density and number of areolae of at least 10 valves per strain were measured using ImageJ 1.37v software (Table 2). The valves were screened for areolae openings, central raphe endings, alveoli and stigma openings. Table 3 gives an overview of the 11 frustule characters. The different characters are described with micrographs in Figs. 2, 3, 4; terminology follows Ross et al. [62] and Round et al. [43].

Figure 2. LM photos of strains.

Figs. 2.1–22. Gomphonema parvulum (Kützing) Kützing var. parvulum. Figs. 2.1–3. Strain D16_042, Figs. 2.4–6. Strain D16_044, Fig. 2.7. Strain D16_005, Fig. 2.8. Strain D16_009, Figs. 2.9–11. Strain D16_008, Figs. 2.12–15. Strain D16_004, Figs. 2.16–18. EPITYPE Strain D16_045. Figs. 2.19–20. Strain D13_034, Fig. 2.21. Strain D16_030, Fig. 2.22. Strain D16_027. Figs. 2.23. Strain D16_026, Gomphonema parvulum var. nov.? Figs. 2.24–25. Strain D12_022, Gomphonema parvulum var. parvulum [morphodeme exilissimum]. Figs. 2.26–28. Gomphonema lagenula Kützing. Figs. 2.26–27. Strain D33_006. Fig. 2.28. EPITYPE Strain D33_024. Figs. 2.29–34. Gomphonema parvulum var. nov.? Figs. 2.29. Strain D16_028. Figs. 2.30–34. Strain D16_011. Figs. 2.35–43. Gomphonema saprophilum (Lange-Bertalot & Reichardt) Abarca et al. comb. nov. Figs. 2.35–38. EPITYPE Strain D36_003. Figs. 2.39–41. Strain D20_027, Figs. 2.42–43. Strain D03_167. Figs. 2.44–49. Gomphonema narodoense R. Jahn et al. sp. nov. Figs. 2.44–46. HOLOTYPE Strain D23_012. Figs. 2.46–49. Strain D23_009. Scale bars represent 10 µm.

Figure 3. SEM morphology of studied strains; external valve view.

Frustule outline. Apices: rostrate (Fig. 3.1), capitate (Fig. 3.4) and subrostrate to subcapitate (Fig. 3.2). External raphe fissure: head pole: hooked (Fig. 3.1), strongly hooked (Fig. 3.2), undulate (Fig. 3.3), slightly deflected (Fig. 3.4) and foot pole: hooked (Fig. 3.5), strongly hooked (Fig. 3.6), undulate (Fig. 3.7), slightly deflected (Fig. 3.8, 3.9). Fig. 3.1. Strain D16_026, Gomphonema parvulum var. nov.?. Fig. 3.2. Strain D16_008, Gomphonema parvulum var. parvulum. Fig. 3.3. Strain D13_034, Gomphonema parvulum var. parvulum. Fig. 3.4. EPITYPE Strain D33_024, Gomphonema lagenula. Fig. 3.5. Strain D20_027, Gomphonema saprophilum. Fig. 3.6. Strain D16_008, Gomphonema parvulum var. parvulum. Fig. 3.7. Strain D16_042, Gomphonema parvulum var. parvulum. Fig. 3.8. Strain D12_022, Gomphonema parvulum var. parvulum [morphodeme exilissimum]. Fig. 3.9 Strain D23_009, Gomphonema narodoense R. Jahn et al. sp. Nov. Scale bars usually represent 2 µm; the bar for Fig. 3.4 represents 5 µm.

Figure 4. SEM morphology of studied strains, external valve view.

Central raphe ending: droplike (Fig. 4.1), straight (Fig. 4.2), droplike deflected (Fig. 4.3), droplike spatulate (Fig. 4.4), hooked (Fig. 4.5). Areolae open: small rounded pores (Fig. 4.9), C shape (Fig. 4.3), E, 3 and anchor-like (Fig. 4.6), C, E, 3 (Fig. 4.8), kidney-shaped (Fig. 4.7). Alternating striae in the middle: short striae (Fig. 4.1), 2 short striae (Fig. 4.2), absent (Fig. 4.3). Fig. 4.1. Strain D16_030, Gomphonema parvulum var. parvulum. Fig. 4.2. Strain D16_042, Gomphonema parvulum var. parvulum. Fig. 4.3. Strain D16_011, Gomphonema parvulum var. nov.? Fig. 4.4. Strain D16_026, Gomphonema parvulum var. nov.? Fig. 4.5. Strain D16_008, Gomphonema parvulum var. parvulum. Fig. 4.6. HOLOTYPE Strain D23_012, Gomphonema narodoense. Fig. 4.7. Strain D16_044, Gomphonema parvulum var. parvulum. Fig. 4.8. Strain D20_027, Gomphonema saprophilum. Fig. 4.9. Strain D23_009, Gomphonema narodoense. Scale bars usually represent 2 µm; only in Figs. 4.6, 4.7 and 4.8 they represent 1 µm.

Phylogenetic analyses

The morphological data were coded in a 0/1 matrix including the characters given in Table 3 plus valve length and breadth as well as the number of striae per 10 µm given in Appendix S1/A. Categories in 0.5 µm spans were established for valve length and breadth. The range for each strain was then coded as absence/presence (0/1). The number of striae found in each strain was also coded comparably, the categories being the numbers of striae per 10 µm in 0.5 steps. The binary 0/1 coding was chosen to avoid assumptions concerning the homology of characters. Five different data sets were used for the phylogenetic analysis: morphological matrix, ITS, 18SV4 and rbcL individual datasets (trees given in Appendix S1/B). After establishing that the individual tree topologies were not conflicting (considering only nodes with bootstrap support >75) a concatenated dataset including the morphological matrix and all three molecular markers was formed. Each dataset was analysed using three different approaches (Maximum Parsimony MP, Maximum Likelihood ML, Bayesian Likelihood BL) except for the morphological matrix which was analysed using two different approaches (MP, BL).

To test the stability of the recovered clades in a broader context, all sequences for Gomphonema parvulum available from Genbank were downloaded (Table 4) [50]. Again, the individual datasets for each marker were first analysed and then an analysis on a concatenated dataset was run that included those strains from Genbank for which all three markers were available (Table 4) as well as all the 21 strains from the present study.

Table 4. Gomphonema parvulum strains from Genbank with accession number and country of origin.

Prior to the phylogenetic analysis, the datasets were tested for their phylogenetic information content using SAMS [63] with standard options. First, the MP analysis was run on PAUP 4.0b10 [64] with equal weights, 10,000 closest sequence additions and tree bisection-reconnection (TBR) branch swapping, permitting 1000 trees to be held at each step. A strict consensus tree was computed. The trees were evaluated by a bootstrap analysis [65] with 10,000 replicates (using the same search strategy as the MP analysis) and MulTrees option in effect (but limiting the maximum tree number to 10,000). For the ML and BL analysis of the molecular datasets, the optimal model of sequence evolution that best fitted the sequence data was calculated under the hierarchical likelihood ratio test (hLRT) and selected according to the Akaike information criterion (AIC) using Modeltest 3.7 [66]. This resulted in the model GTR+G+I [67]. Second, an ML analysis was conducted using RAxML 7.0.4 [68], [69], ML search option (GTR+G+I) and 10,000 bootstrap replicates (model GTRCAT as implemented in RAxML for the rapid bootstrap algorithm). The RAxML run was not conducted on the morphological matrix. A third analysis was run on MrBayes 3.1.2 [70] using gamma distribution rate variation among sites and 10,000,000 generations of the MCMC chains in two independent runs of four chains apiece; otherwise the default parameters were used. Convergence of MCMC chains was assumed when the standard deviation of split frequencies was lower than 0.01. Every hundredth tree was sampled. The first 25,000 trees were discarded as burn-in; the rest was used to calculate a 50% majority rule consensus tree. The strict consensus tree of the MP analysis was compared to the 50% majority rule tree of the BL analysis, and the best ML tree found by RAxML. Trees were drawn using FigTree v1.2.2 [71] and Adobe Illustrator (Adobe Systems, San Jose, CA).


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

Genetic distances

Uncorrected p-distances were calculated for each marker individually as well as a concatenated data set using PAUP 4.0b10 [64] Those parts of the alignment including missing data were excluded from the analysis as PAUP cannot discriminate between gaps and missing data when calculating uncorrected p-distances. Results are given in Appendix S1/C.

Mantel tests

A matrix of direct line geographic distances (km) and genetic distances (uncorrected p distances; calculated using PAUP 4.0b10 [64]) between populations was created (data available from the corresponding author upon request). Mantel tests [72] were used to examine associations between geographic and genetic distances using R 2.7.1 [73] with 9999 permutations.


Morphological characters (Figs. 2, 3, 4, 5)

Figure 5. SEM morphology of studied strains, internal valve view.

Central raphe endings: hooked (Fig. 5.2), deflected (Fig. 5.1), strongly hooked (Fig. 5.4) and parrot head-shape (Fig. 5.5). Alveoli: rectangular separated by silica struts (Fig. 5.6) and oval elongated separated by opened silica struts (Fig. 5.7). Stigma opening in internal view: short slit (Fig. 5.3), slit (Fig. 5.1), sperm-shape (Fig. 5.4) and round (Fig. 5.2). Scale bars represent 2 µm; only in Figs. 5.3 and 5.7 it represents 1 µm. Fig. 5.1. HOLOTYPE Strain D36_003, Gomphonema saprophilum. Fig. 5.2. Strain D23_009, Gomphonema narodoense. Fig. 5.3. Strain D20_027, Gomphonema saprophilum. Fig. 5.4. Strain D16_026, Gomphonema parvulum var. nov.? Fig. 5.5. Strain D12_022, Gomphonema parvulum var. parvulum [morphodeme exilissimum]. Fig. 5.6. Strain D33_006, Gomphonema lagenula. Fig. 5.7. D16_028, Gomphonema parvulum var.nov.?

The morphometric data (Table 2) and morphological features recorded for each strain (Table 3, Figs. 2, 3, 4, 5) showed the typical characteristics of Gomphonema parvulum: lanceolate or linear-lanceolate heteropolar valves with more or less subrostrate or rostrate heads. The central area is asymmetrical with a single long stria on one side terminating in a distinct stigma, and a single short stria on the other side. Striae are almost perpendicular to the raphe. Such characteristics, associated with the length (17–29 µm), breadth (5–8 µm), length/breadth ratio (3–5) and 12–21striae in 10 µm, allowed us to group all studied specimens into Gomphonema parvulum s.l.

In LM observations (Fig. 2), four of the eight different morphodemes that could be differentiated on the basis of their outline corresponded to named varieties and/or forms forms. These were: parvulum sensu stricto (D13_034, D16_004, D16_005, D16_008, D16_009, D16_027, D16_030, D16_042, D16_044, D16_045), exilissimum (D12_022), lagenula (D33_024, D33_006), saprophilum (D03_167, D20_027, D36_003). The remaining four morphodemes, as yet unnamed, did not correspond to named taxa. These morphodemes were (D16_028, D16_011, D16_026, D23_009 and D23_012).

The four unnamed morphodemes had a slightly different valve outline, making it difficult to assign them to any known group; the first two cannot be delimited convincingly because there seem to be transitions (Übergangsformen) between the outlines. Some of the proposed differentiating micro-morphological features are present within the same strain such as raphe fissures (Fig. 3), central raphe endings (Figs. 4 and 5), areolae (Fig. 4), alveoli (Fig. 5), stigma openings (Fig. 4), alternating striae in the center (Fig. 4),. For example, the areolae of the morphodeme exilissimum (Figs. 3.8 and 5.5) are supposed to be on average significantly larger than G. parvulum and have the shape of a horseshoe or ‘C’ as shown in Fig. 4.3 but the strain D12_22 shows large and small areolae with shapes including not only ‘C’ forms, but also ‘3’ and ‘E’. A further example is the morphodeme saprophilum which on average should have smaller areolae mostly comma to kidney-shaped but the strains D03_167, D20_027, D36_003 have a variety of forms and sizes such as ‘C’, ‘3’, ‘E’ and kidney-shapes (Fig. 4.8).

Cladistic analysis of morphological data (Fig. 6)

Figure 6. Strict consensus tree of the ML analysis of the morphological matrix with the results of the SAMS analysis (arrows with split names), bootstrap values (>75) of MP (PB) and ML (LB) as well as Bayesian posterior probabilities (PP) (>0.75).

Fig. 6 shows the strict consensus tree of the ML analysis of the morphological matrix (Table 2 and Table 3) with the results of the SAMS analysis, bootstrap values (>75) of MP (PB) and Bayesian posterior probabilities (>0.75, PP). Only 2 clades are well resolved; these were the two Mexican strains (Clade 3, Fig. 6, (PP 0.92) and the two Korean strains (Clade 4, Fig. 6, PP 0.99) which separated well from the rest of the strains. These two clades were the only ones that, following the SAMS analysis, were supported unambiguously by the data (shown by arrows in Fig. 6). In addition the sample pair D16_008 and D16_009 was also supported by BL (PP 0.97) as well as MP analysis (PB 89).

Since the morphometric data and the few differentiating morphological features overlap even within one strain, molecular data were used in order to identify and substantiate further differences between the strains.

Phylogenetic analysis of molecular and combined data (Fig. 7, Fig. 8)

Figure 7. Strict consensus tree of the ML analysis of the combined dataset (molecular markers 18SV4, internal transcribed spacer ITS, rbcL and morphological matrix) with the results of the SAMS analysis (arrows with split names), bootstrap values (>75) of MP (PB) and ML (LB) as well as Bayesian posterior probabilities (PP) (>0.75).

Figure 8. Strict consensus tree of the ML analysis of the combined dataset (molecular markers 18SV4, internal transcribed spacer ITS, and rbcL) including Genbank sequences with the results of the SAMS analysis (arrows), bootstrap values (>75) of MP (PB) and ML (LB) as well as Bayesian posterior probabilities (PP) (>0.75).

Split names indicate the splits from Fig. 7.

The results of the phylogenetic analyses are summarised in Figs. 7 and 8. First the individual markers were analysed (trees are given in Appendix S1/B). In both approaches (excluding and including additional Genbank sequences) ITS and rbcL recovered non contradicting tree topologies (Appendix S1/B2, S1/B3, S1/B7, S1/B8). The analysis of 18SV4 recovered contradicting topologies (Appendix S1/B1, S1/B4, S1/B5, S1/B6); the main difference to the ITS and rbcL datasets being the position of strains D16_026 and D16_028 that cluster with D33_006 and D33_024 (e.g. Appendix S1/B1). However, the contradicting nodes are supported only by values below the threshold (>75 bootstrap support or >0.95 posterior probability), so 18SV4 could be included in the concatenated dataset.

Fig. 7 shows the strict consensus tree of the ML analysis of the combined dataset including only our own strains from the present study with the results of the SAMS analysis, bootstrap values (>75) of MP (PB) and ML (LB) as well as Bayesian posterior probabilities (>0.75, PP). As there are no outgroup taxa included in our analysis we chose an unrooted tree design. The SAMS analysis showed that the combined dataset comprising morphological characters and morphometric data (Tables 2 and 3) and the molecular markers (18SV4, rbcL and ITS) is the phylogenetically most informative dataset. The splits supported by the combined dataset are the same as in the individual subsets: however not all subsets support all splits. The splits that are unambiguously supported by the combined dataset following SAMS are indicated by arrows in Fig. 7. The supported splits are (a) the split between the strains from the Faroe Islands and the central Continental European, Mexican and Korean clones, (b) the split between the central Continental European clade and the rest of the clones, (c) the split between the Korean and Mexican clones and the rest, (d) the split between Korean clones and the rest and (e) the split between Mexican clones and the rest (Fig. 7).

The combined dataset shows the best resolved tree in comparison with the individual datasets. However, no contradictions are found in the other trees. In the following, all bootstrap values and posterior probabilities are given for the concatenated dataset if not stated otherwise. ML bootstraps (LB) are strong (>75) for the splits (a), (b), (c) and (e) (Fig. 7). BL posterior probabilities (PP) are high (>0.95) for splits (b), (c), (d) and (e); PP for split (a) is somewhat lower at 0.87 (Fig. 7). The splits (d) and (e) are also well (PB >75) supported by MP bootstraps (PB) (Fig. 7). The BL analysis also recovered three more sample pairs: D16_005 and D13_034 (PP 0.85), D16_008 and D16_009 (PP 0.90, also supported by MP analysis: PB 91) as well as D16_004 and D12_022 (PP 0.82) (Fig. 7). The long branch of D16_011 is due to missing data in the rbcL gene (Fig. 7).

Fig. 8 shows the outcome of the analysis of the concatenated dataset including additional Genbank sequences (Table 4). Clades 1–4 are recovered. Clade 1 is supported by ML (LB100) and MP (PB100, Fig. 8). Clade 2 is supported by BL (PP0.81), MP and ML recover a corresponding clade that includes D03_164, D20_027 and D03_036 as well as three further strains from France (TCC545, TCC612) and Luxembourg (TCC664, Fig. 8). The Mexican Clade 3 was supported by BL, ML and MP (PP1.00, LB100, PB98, Fig. 8) and was the sister group of a clade of strains from Mayotte (Appendix S1/B7, S1/B8; represented in Fig. 8 by strain TCC429). A strain from Portugal (TCC736) is sister to the strains from Mexico and Mayotte, however this is only weakly supported by BL (PP0.82). The Korean clade is supported by BL, ML and MP; however, not very strongly (PP0.86, LB76, PB75). The ITS and rbcL analysis show it is sister to another clade from Mayotte (Appendix S1/B7, S1/B8; represented in Fig. 8 by strain TCC492). In Fig. 8 there are also two strains from Portugal (TCC725, TCC734) clustering in this clade.

Clade 1 is composed of strains from the Faroe Islands (Fig. 7). The strains showed a wide phenotypic plasticity, with specimens varying from isopolar to clearly heteropolar in shape, and with subrostrate, rostrate or subcapitate apical endings (Figs. 2.1–22). All of the strains present a mixture of valve characters such as dimensions, striae density, raphe characters, areolae, alveolae, stigma openings and even alternating striae in the center. This clade also includes the strain D12_022 which corresponds to the species description of Gomphonema exilissimum (Grunow) Lange-Bertalot & Reichardt 1996 in Lange-Bertalot & Metzeltin [61] and Jüttner et al. [23]; specimens have long, slender valves, are weakly clavate to linear-lanceolate with shortly protracted, rounded head poles and sharply narrowed foot poles (length 24–26.5 µm, width 5–6 µm, 13–15 striae in 10 µm) (Figs. 2.24, 2.25). There is no difference within a group of 7 strains across all three molecular markers (D16_030, D16_027, D16_044, D16_042, D16_045, D16_009, D16_008, Appendix S1/C4) and within a group of two strains (D16_005, D13_034, Appendix S1/C3, only available for rbcL, other markers too much data missing) and only very little (<0.2%, Appendix S1/C4) between all the strains of these two groups. The strains D16_004 and D12_022 show a little more difference to the above mentioned strains but stay below one percent (Appendix S1/C4). The three further strains of the Faroe Clade are difficult to compare since their data are not complete: The morphologically unidentified strain D16_028 (Fig. 2.29) shows a distance of about 1.3% in its 18SV4 (Appendix S1/C1) towards all the other strains from the Faroe Islands but it clusters closely with the strain D16_026 (Fig. 2.23); morphologically unidentified strain D16_011 (Figs. 2.30–34) shows clear differences in its ITS data of up to 4.6% to all the other strains of this clade (Appendix S1/C2); data for rbcL are incomplete and data for 18SV4 are missing.

As can be seen in Fig. 8, none of the additional Genbank sequences clustered in or close to our Clade 1.

Clade 2 is composed of the two strains from Northern Germany (Berlin; D03_167, D36_003) and southern Sweden (Skepparkroken; D20_027) (Fig. 7); there are no differences in the 18SV4 data (Appendix S1/C1) and only 0.2% in the rbcL (Appendix S1/C3) and about 0.5% in the ITS (Appendix S1/C2). Morphologically, these three strains fit the concept of Gomphonema parvulum var. parvulum f. saprophilum [41].

In the analysis including additional Genbank sequences (Fig. 8) three more strains clustered in this clade originating from France (TCC545, TCC612) and Luxembourg (TCC664); in the corresponding paper by Kermarrec et al. [50], they belong to clade C which includes only strains from Europe (in that paper [50] the names clade C and clade D have been interchanged in the concatenated tree, we refer to the clade-naming in the single marker trees). Including these, the genetic distance over all three markers within this clade ranges between 0.2% and 0.3% (Appendix S1/C5). The highest variation is found in ITS where D36_003 differs by about 1% from TCC545, TCC612 and TCC664 (Appendix S1/C6).

Clade 3 is composed of the two Mexican strains (Fig. 6, Fig. 7); the morphological data – apical ends subcapitate to capitate – fit the concept of Gomphonema lagenula (Figs. 2.26–28). In addition, the molecular data show very little variation: 0.3% for 18SV4 (Appendix S1/C1), none for rbcL (Appendix S1/C3).

Including the additional Genbank sequences (Fig. 8), there is a clade of strains from the tropical island Mayotte (Appendix S1/B7, S1/B8) that is represented in Fig. 8 by TCC429; in the corresponding paper [50], they have been identified as belonging to clade A which also includes only Mayotte strains. The genetic distance between the Mexican strains and TCC429 is 0.5% over all three markers (Appendix S1/C5), for rbcL there is no genetic difference between the Mexican strains and TCC429 (Mayotte), Another strain from Mayotte (TCC500), being identical with TCC429 in ITS and rbcL (Appendix S1/B7, S1/B8), has been identified as G. cf. lagenula by Kermarrec et al. [50].

The strain TCC736 from Portugal is neighbouring the Mexican and Mayotte strains (Fig. 8); the uncorrected p-distance between the Mexican clade and TCC736 is 1.3% over all three markers (Appendix S1/C5); for rbcL there is a genetic difference of 0.6% between these three and TCC736 (Portugal, Appendix S1/C7). In the corresponding paper [50], this strain is a representative of clade D (single marker trees in [50]) which is found in Europe only.

Clade 4 is composed of the two Korean strains (Fig. 6, Fig. 7); the morphological data do not fit any of the described taxa; their outline is only slightly different from the other demes (Figs. 2.44–47) but is distinct enough to form a well supported clade even in the purely morphological analysis (Fig. 6). The molecular data show only little variation: none for 18SV4 and rbcL (Appendix S1/C1, S1/C3) and only 1.2% for ITS (Appendix S1/C2).

Including the additional Genbank sequences (Fig. 8), our Clade 4 forms the sister group to clade B and B′ from the corresponding paper [50], including strains from Mayotte (TCC492) and Portugal (TCC725, TCC734). The genetic differences over all three markers between the Korean clade and these strains range between 0.2% and 0.3% (Appendix S1/C5). However, if the markers are considered individually, there are some interesting patterns: there is no variation in 18S between all five strains (Appendix S1/C7). ITS shows no difference between TCC725, TCC734 (both Portugal) and TCC492 from Mayotte, with other differences ranging from 0.9% to 1.7%, the largest being from D23_012 to either TCC725 or TCC792 (Appendix S1/C7). Both Korean strains have an insert of 6 bp in ITS that is missing in the other three strains. For rbcL only TCC492 from Mayotte shows any difference (0.7%) from the other strains (Appendix S1/C7).

Nomenclatural and taxonomical consequences

Because the molecular and morphological data of Kermarrec et al. [50] do not overlap with our study, the nomenclatural findings and taxonomical consequences of this section can only be based on our strains; but similarities will be pointed out in the discussion. The morphodeme parvulum (D16_004, D16_005, D16_008, D16_009, D16_027, D16_042, D16_044, D16_045) is restricted in our study to the Faroe Islands as can be seen in Fig. 6 and Fig. 7 as well as Fig. 8. The morphodeme exilissimum is found in only one culture on the Island Vagar (D12_022) on the Faroe Islands (Figs. 2.24–2.25, Fig. 7); it clusters within a subclade of the Faroe Islands clade together with a typical parvulum morphodeme (D16_004, Fig. 7), but this clade is not statistically supported. Three of the unidentified morphodemes (D16_028, D16_026, and D16_011) originate from three strains from the island Viöoy; they appear together in the same Faroe Islands subclade which is not supported (Fig. 7); in addition, some of their data are not complete (Table 1). These might represent varieties but we cannot solve their taxonomy. The fourth unidentified morphodeme is represented only in the Korean strains (D23_009, D23_012; Fig. 2.44–2.47, Fig. 7) which cluster together even morphologically (Fig. 6) and should be separated taxonomically with a name. The morphodeme saprophilum (D03_167, D20_027, D36_003) is found only in the northern central Continental European clade (Fig. 2.35–2.43, Fig. 7). The morphodeme lagenula (D33_024, D33_006) is observed only within the Mexican clade (Fig. 2.26–2.28, Fig. 6, Fig. 7). The above data show that the four unambiguous clades (Fig. 7) should be separated and named.

Gomphonema parvulum (Kützing) Kützing var. parvulum (Figs. 2.1–22; 2.24–25; 3.2–3; 3.6–8; 4.1–2; 4.5; 4.7; 5.5)

Kützing, Species Algarum, p.65. 1849 [26].

Basionym: Sphenella parvula Kützing, Die kieselschaligen Bacillarien oder Diatomeen, p.83; taf. 30, fig. 63. 1844 [25].

Heterotypic synonyms: G. parvulum var. exilissimum GrunowG. exilissimum (Grunow) Lange-Bertalot & Reichardt 1996 [61], Figs. 22–27.

Lectotype designated here from the syntypes designated by Krammer & Lange-Bertalot [40] 1986: 358ff; fig. 154:2.

Type locality: Falaise, France.

Epitype (designated here): B 40 0040914 (strain D16_045); see Fig. 2.16.

Type locality of the epitype (designated here): Faroe Islands, Viöoy, Waterfall below church, Lat 62.360335°, Lon −6.542833° ±20 m, 04.08.2004, leg. J. Bansemer.

Emended species description:

This species is highly variable in its valve shape: lanceolate, linear-lanceolate or oval heteropolar valves with more or less subrostrate or rostrate head. The central area is asymmetrical with a single long stria on one side terminating in a distinct stigma, and a single short stria (sometimes two) on the other side. Striae slightly radiate or parallel, not distinctly punctuate. Length 22–29 µm, breadth 5–7.5 µm, striae 12–20 in 10 µm.

Gomphonema saprophilum (Lange-Bertalot & Reichardt) Abarca, R. Jahn, J. Zimmermann & Enke comb. nov. (Figs. 2.42–43; 1.39–41; 2.35–38; 3.5; 4.8; 5.1; 5.3)

Basionym: G. parvulum var. parvulum f. saprophilum Lange-Bertalot & Reichardt 1993, Bibliotheca Diamologica 27, p. 69 [41]; figs. in Krammer & Lange-Bertalot [40] 1991, figs. 76: 8–13; not figs. 77: 5–9.

Heterotypic synonym?: G. parvulum var. parvulum f. neosaprophilum. Kobayasi et al. [42]

Lectotype (designated here): fig.76: 11 in Krammer & Lange-Bertalot 1991 [40]

Type locality: Kommunale Kläranlage der Stadt Frankfurt a.M.; “voll” gereinigtes Abwasser mit einer Restbelastung von BSB5-Werten um 20 mg O2/L (15.1.1975).

Epitype (designated here): B 40 0040919 (strain D36_003), see Fig. 2.35

Epitype locality: Germany, Berlin, Tiergarten, Landwehrkanal, Lat 52.511°, Lon 13.339°, ±20 m, 11.06.2005, leg. W.-H. Kusber.

Emended Diagnosis: This taxon is separable from G. parvulum based mostly on the valve-shape; the taxon G. saprophilum has a rather rhomboidal shape with slightly rostrate ends, the head pole and the valve are on average also wider than those of G. parvulum (Figs. 2.35–2.43). Valves are clavate-lanceolate, 22–27 µm long, and 6–8 µm wide; head poles shortly rostrate and foot poles obtuse or subcapitate. The axial area is linear and the central area is narrow. The central area is asymmetric due to the presence of shorter striae opposite to the stigma. Stigma opening is circular in valve exterior, internally it shows a variety of transversely elongated short, long and sperm shaped openings (Figs. 5.1 and 5.3). The outer fissure of the raphe undulates slightly, but the inner fissure extends straight. The external central endings of the raphe branch are slightly swollen and terminate straight, slightly expanded as well as pore-like deflected or spatulate. The internal central endings are hooked, deflected or have a strongly hooked “parrot head” toward the side of the stigma (Figs. 5.1, 5.3). In both poles, polar fissures are abruptly curved toward the side of the stigma and then extended to the opposite side to terminate. In one side of the valve, the central striae extend long towards the stigma, while on the other side the central striae is very short or even two short striae are observed.

Comment: In order to have micro-morphological as well as molecular data available for this taxon, we have decided to describe this taxon as new and not elevate it from the level of forma to species rank.

Gomphonema lagenula Kützing (Figs. 2.26–28; 3.4; 5.6)

Kützing, Die kieselschaligen Bacillarien oder Diatomeen. p.85, pl. 30, fig.60. [25] 1844.

Homotypic synonym: Gomphonema parvulum var. lagenula (Kützing) Frenguelly 1923.

Lectotype designated by Krammer & Lange-Bertalot [40] 1986: 358ff; fig. 154:8.

Type locality: Trinidad, Tobago.

Epitype (designated here): B 40 0040918 (strain D33_024), see Fig. 2.28

Epitype locality (designated here): Mexico, Ixtlán de los Hervores, 06.12.2004, leg. N. Abarca.

Emended species description:

This taxon is separable from G. parvulum based on the shape of the head pole, which is consistently more rostrate to capitate. This taxon is differentiated from G. parvulum by its rostrate to capitate valve shape with strongly protruded valve ends and areolae divided by struts in the alveolus. This taxon has a broadly lanceolate shape with markedly rostrate to slightly subcapitate polar ends (Figs. 2.28 and 3.4). Striae vary from parallel in the central area to parallel or slightly convergent at the poles. Within the central area, there is a short stria located opposite to the stigma. This also results in an asymmetrical central area.

Gomphonema narodoense R. Jahn, Abarca, J. Zimmermann & Enke nov. sp. (Figs. 2.44–49; 3.9; 4.6; 4.9; 5.2)

Holotype (designated here): B 40 0040917 (strain D23_012); see Fig. 2.45. Type locality: South Korea, DukheungRi, NaeNarodo, spring at rice paddy, 14.10.2004, leg. R. Jahn.

Diagnosis: This taxon is separable from G. parvulum based on the shape. Frustule in girdle view narrow rectangular, hardly perceptible cuneate. Lanceolate to linear-lanceolate, almost naviculoid with only moderate rostrate head pole in comparison to the foot-pole (Figs. 2.44–2.47). Length 17–24 µm, breadth 5–6 µm, striae 16–18 in 10 µm, with transverse orientation. Areolae open and extremely variable from small rounded pores to C-, E-, 3- and anchorlike-shapes (Figs. 4.6 and 4.9). The alveoli are oval, elongated and separated by opened silica struts (Fig. 5.2). Stigma has a round opening (Fig. 5.2). Axial area is linear and the central area is narrow. The central area is asymmetric due to the presence of shorter striae opposite the stigma which open circular in valve exterior, internally it shows round openings. The external central endings of the raphe branch terminate straight, slightly expanded as well as pore-like deflected. The internal central endings are hooked toward the side of the stigma. In both poles, polar fissures are abruptly curved toward to the side of the stigma and then extended and terminate towards the opposite side.

Mantel tests

The Mantel test revealed a positive correlation between genetic and geographic distances separating the Gomphonema strains (rm = 0.88; p = 0.001). Also for the central Continental European Gomphonema strains there was a positive correlation (rm = 0.84; p = 0.001). No association between geographic and genetic distance was found for the strains within the Faroe Islands (rm = 0.40; p = 0.101).


Phylogeny and taxonomy

G. parvulum s.l. has long been recognized as a highly variable species complex with broad ecological tolerances and many varieties. Many references suggest there is great richness of morphologies that seem to inter-grade with each other and form a morphologically diverse but single taxon [32], [31], [36], [39], [40], [41], [50]. The separation of these morphologies using the valve structure is very difficult. Even though some separation has been done based on slight shape differences, there is great overlap in characters such as dimensions and striae density. Wallace and Patrick [36] considered that the morphological variability in G. parvulum is due to the varying conditions of the environment in which these organisms are found, and synonymized many varieties under the single taxon G. parvulum. In recent years, studies on the taxonomy of this group have resulted in the splitting off of many new varieties and forms, and old names synonymized in the past have been restored in order to accommodate some of the morphological variations [41]. In addition, many of these varieties have been recognized as sufficiently morphologically different from the nominate variety to be raised to species status [40], [41], [74]. However, other varieties have yet to be examined concerning their micromorphology with SEM or morphometrically assessed to see how morphologically different they are from nominate and other conspecific varieties [75].

The current taxonomic separation of strains within G. parvulum s.l. is based on morphological criteria alone. But this current taxonomic concept is only partly supported by our study on the morphological variability of strains (Table 3). Some strains exhibited overlapping morphological features (shape, striation, and stigma). Morphological features such as “attached by a stalk” or “free living” have already been recognized as unreliable [26], [27], [28], [29]. The features shape and striation density of the closely related Gomphonema augur Ehrenberg have been recognized to be influenced by environmental conditions [76]. Our results and those of Kermarrec et al. [50] also show that the molecular separation of the taxa does not completely coincide with the morphological taxa, we have therefore called them here morphodemes.

Our combined morphological and molecular results separated at least four taxa/clades clustering geographically (Fig. 7) according to our investigated sites in Korea, Mexico, Central Europe and Northern Atlantic Europe. This is partly supported by Kermarrec et al. [50] who discuss the existence of four or five semi-cryptic taxa in their paper which includes only sites from Europe or the tropical island of Mayotte.

Gomphonema parvulum sensu stricto (Clade 1)

In our study, we found Clade 1 (Fig. 7) to be morphologically and molecularly the most heterogenous clade for which we also had the most strains available. But a main G. parvulum–group can be defined because of no or little difference (below one percent) between 11 strains (see details in results). We have chosen D16_045 as the epitype of Gomphonema parvulum var. parvulum. In this main group though, we found one strain with apparently typical specimens of G. exilissimum (D12_022) with characters hardly separable using the light microscope (Fig. 2.24–2.25). But in SEM studies of the cultured material, the main differentiating feature, that the foramina of the areolae are supposed to be significantly larger [61], could not be confirmed. In our molecular results the morphodeme G. exilissium was resolved within the main group of G. parvulum (Fig. 7); interestingly however, this strain is the only one from the Island Vagar (D12). Kermarrec et al. [50] also show a strain (TCC439) with exilissimum features which molecular analyses resolve within their clade A which is restricted to their tropical Island Mayotte. This points at infraspecific variability of morphological features; we believe that G. exilissimum should be considered a morphodeme or perhaps an ecodeme if data shows that it only appears in specific oligotrophic waters. G. exilissimum is reported as a characteristic species in slightly acid waters with low electrolyte content and low nutrients on silica bedrock for Central Europe [23], [49]. By contrast, Hofmann et al. [49] report the occurrence of G. parvulum (var. parvulum) in oligosaprobic and mesosaprobic waters independent of their trophic situation. The wide variability of G. parvulum Kützing has also been recorded as producing a morphological oddity, the Janus cell [77]. This occurs when two morphologically distinct valves are produced on either side in a single diatom frustule. This phenomenon is considered rare but inducible, and indicates that various phenotypes can be produced by the same genotype. Mann [78] suggests that because the two valves of a diatom frustule form at different times, each may be influenced by different environmental conditions and hence exhibit two different morphological forms.

The three cultures (D16_011, D16_026, D16_028) which do not show up in the main G. parvulum-group (Fig. 7) show much morphological and molecular diversity; since some data are missing, they are not completely comparable to the other 11 cultures (Fig. 7). Nevertheless, it is interesting to note how much molecular diversity is present in such a small place since 12 strains (D16) were established from the same sample from Viöoy of the Faröer Islands; the other two are from adjacent islands Sandoy (D13) and Vagar (D12). If this is normal molecular diversity of a taxon, or if this could hint at concealed diversity representing different varieties, or an early stage of an island radiation process, needs to be left for later studies. In the analysis including the additional Genbank sequences (Fig. 8), none of the additional strains but only the 14 strains from the Faroe Islands belong to Clade 1 which in our study represents Gomphonema parvulum sensu stricto.

Reported occurrences for Gomphonema parvulum are hard to unravel since it has been used as a collective name for the entire species complex for two centuries; being characteristic of organically polluted waters, but also found in samples from a wide range of water qualities [79], [39], [60]. Further studies are needed to determine if this taxon sensu stricto is really ubiquitous and cosmopolitan or if it is restricted to specific environments or geographies.

Gomphonema saprophilum (Clade 2)

The three cultures which we have described as G. saprophilum form a well-supported clade. Although the three cultures were sampled from three different sites, they are morphologically and molecularly almost homogeneous. This clade has a combined distance of 1.0–1.5% (Appendix S1/C4) from the Faroe clade and might be representative of a northern-central, Continental European clade, since it has been found in waters in Berlin (northern Germany) as well as Skepparkroken, Ängelholm (southern Sweden) which are less than 600 km apart; both waters are highly anthropogenically influenced. In the analysis including the additional Genbank sequences (Fig. 8) two strains originating from France and one from Luxembourg clustered next to our Clade 2 and represent clade C in the corresponding paper [50]. Clade C was identified for Europe only but few other localities were examined in [50] and supports our findings of a (Central) European distribution of this entire clade. Morphologically however, these three strains of clade C [50] do not belong to our concept of G. saprophilum; they might represent a different semi-cryptic species.

Since this taxon has only recently been differentiated as G. parvulum var. parvulum f. saprophilum [41], little is known about its occurrence. Hofmann et al. [49] who differentiate it from G. parvulum var. parvulum f. parvulum, report this species for Central Europe as occurring in organically loaded, alpha-mesosaprobe to polysaprobe waters. Kobayasi et al. [42] describe a Gomphonema parvulum var. neosaprophilum Kobayasi ex K.Osada from a river in Tokyo; whether this is the same taxon or one with a similar morphology needs to be checked.

Gomphonema lagenula (Clade 3)

Although Kützing [26] had described this species from the tropical nation Trinidad and Tobago, the conspecificity of G. parvulum and G. lagenula has been discussed for a long time. Frenguelli [32], for instance, understood that there were differences between G. parvulum and G. lagenula, but these differences were considered by him too subtle to distinguish them as different species. Thus, this author regarded G. lagenula as a variety of G. parvulum. Since then, most authors have accepted Frenguelli's [32] proposal, differentiating both taxa at infraspecific level [60]. Our results show a well-supported clade (Clade 3, Fig. 6, Fig. 7), separating the Mexican cultures from all the others with 1.2–2.0% difference across all three markers (Appendix S1/C4).

In this study, this taxon has been found in an anthropogenically influenced river. Besides this, it is regularly reported mainly from tropical Central and Southern America [80], [81]; but it has also been reported as occurring in central Europe [60], in Japan [42] in Korean waters [82], and in Burundi (Africa) [83]. Two GenBank sequences from Mayotte Island (Indian Ocean) appear close to this clade (Appendix S1/B7, S1/B8); strain TCC429 is shown in Fig. 8 and belongs to clade A in Kermarrec et al. [50]. The genetic distance between our Mexican strains and TCC429 is 0.5% across all three markers (Appendix S1/C5). The strain TCC500 from Mayotte (not shown in Fig. 8) is molecularly identical with TCC429 in ITS and rbcL (Appendix S1/B7, S1/B8) and has been identified as G. cf. lagenula in Genbank. Since in the corresponding paper [50] clade A is restricted to Mayotte strains, this points at a tropical distribution of this taxon; which is also supported by the morphology of the strains in clade A [50], except for TCC 500 which has very small valves and is therefore difficult to identify. Although Kermarrec et al. [50] did identify one strain as G. cf. lagenula, their morphological concept as well as their restricted geographical data did not allow them to separate this taxon from their broad Gomphonema parvulum complex.

The strain TCC736 from Portugal, sitting on the branch of Clade 3 (Fig. 8) with a distance of 1.3% to the Mexican strains across all three markers (Appendix S1/C5) and representing clade D, is probably a different species. This is also supported by the morphology; we would not have identified this taxon as belonging to Gomphonema parvulum sensu lato. Zimmermann et al. [53] have shown that in the well studied Sellaphora pupula group a distance between taxa of 0.3% to 0.8% for 18SV4 usually indicates that they are different species. In addition, the ITS marker of TCC736 has an unique insertion of 12 bp; a second insertion of 6 bp found in TCC736 has been found in all members of Clade 1 and Clade 2 plus TCC595, TCC612 and TCC664 and the Mexican strains of Clade 3.

Gomphonema narodoense (Clade 4)

Although this taxon is very hard to differentiate morphologically; it is sufficiently distinct to be supported on these characters alone (Fig. 6). Molecularly, it is a well supported clade clearly separated from the Mexican, the central Continental European and the Faroe Clade with 1.2–2.3% difference across all three markers (Appendix S1/C4). This new taxon has been discovered in a spring on one of the many islands in Southern Korea. It has never been reported before; the recently compiled Algal Flora of Korea [82] lists Gomphonema parvulum as very common in Korea, it might be that G. narodoense is not restricted to this island but its distribution is hidden within the broad species-complex concept of G. parvulum.

The three additional Genbank sequences from Mayotte (TCC492) and Portugal (TCC725, TCC734), named clade B and B′ respectively in the corresponding paper [50], form a sister clade to our Clade 4 because the genetic difference across all three markers between our Korean strains and these strains range between 0.2% and 0.3% (Appendix S1/C5). But when the markers are considered individually, there is no variation in 18SV4 between all five strains (Appendix S1/C7) and for rbcL only TCC492 from Mayotte shows 0.7% difference to all the other strains in this clade (Appendix S1/C7). But the ITS morphology of the two sub-clades shows a decisive difference as our Korean strains have an insert of 6 bp which the three TCC-strains from the sister sub-clade do not have. This insert of 6 bp is found in all of our own strains except for Clade 3. Morphologically, only strains TCC492, TCC428 and TCC436 from the clade B, Mayotte, [50] show some resemblance to our concept of Gomphonema narodoense; all the other strains of clade B are very different.

Interestingly, this clade, the combined data of our findings (our Clade 4) plus the data of Kermarrec et al [51] (their Clade B & B′) (see Fig. 8), is the most cosmopolitan in this study as it clusters strains from such distances as a South Korean Island in the Western Pacific Ocean with strains from Portugal in southern Europe at the Atlantic Ocean and a tropical strain from Mayotte Island in the Indian Ocean. But this might be due to arbitrary statistics because, although the 18SV4 is the same, the rbcL data show differences; the biggest differences can be found in the ITS data where our Korean strains (plus our other strains except for Mexico) have a 6 bp insert which is missing in the other strains of Kermarrec et al [51], Many more strains from many different localities are needed and have to be studied carefully to determine if we are dealing here with mistaken or missing data simulating a specific distribution pattern.


Our own findings of four clearly separated taxa (see Fig. 7), partly supported by recent data (see Fig. 8) [50] within the G. parvulum complex contradict the hypothesis of cosmopolitanism of microorganisms [1], [3]. Our four well defined clades and taxa, G. parvulum sensu stricto, G. saprophilum, G. lagenula, and G. narodoense, which are on average only 25 µm long, are not distributed randomly but hint at a biogeographical or phylogeographical pattern (see also Boo et al. [4]). Although we cannot back it up with detailed water chemistry data, we think that they also do not support the hypothesis that only the environments selects [1], [2]. The tropical Mexican G. lagenula and the strains of G. saprophilum from the northern-central Continental European clade are from anthropogenically influenced waters; if water quality alone would decide on their occurrence, both should be the same species; this we had originally expected since the polluted Mexican River Lerma had been sampled for comparable European taxa to be used in bio-monitoring. The same argument applies to the Korean G. narodoense and the Faroese G. parvulum sensu strictu which are both from very clean islands' fresh waters. The difference in climate such as water temperature between tropical Mexico and temperate Europe as well as between islands in the North Atlantic and Western Pacific might also have an influence on clade occurrences [84] but our data - supported by the significant correlation of genetic and geographical distances between the four geographic sites as shown by Mantel tests - implies that in this case geographic distances resulting from historical distributions might be one of the reasons which determined which species is living where. This supports recent findings that diatoms have a biogeography [10], [11], [12], [13]; which is best documented by large “flagship taxa” [15]. Our findings go beyond this, because we are showing that even those species which are much smaller and inconspicuous, and which used to belong to the “250 well-known cosmopolitan species” [16], [84] seem to have a biogeography.

The most important research resources for taxonomic and biodiversity studies on bio geography are well documented strains from different parts of the world, but these are hard to come by as single, living cells need to be isolated in order to establish a culture. However these cultures are essential in order to study the strains micro-morphologically and molecularly in detail, because only in this way can DNA sequences be linked unambiguously to a certain morphology. Our data and those recently collected by Kermarrec et al. [51] show, that there is a great deal more diversity in Gomphonema parvulum sensu lato to be discovered around the world. But the question is, if there will ever be enough data available to infer the absence of particular genetic lineages from any geographic area. Nevertheless, further studies on more strains of many more taxa from different parts of the earth will show if our findings are an isolated phenomenon or if they are part of a pattern in diatom biogeography [85].

Supporting Information

Appendix S1.

S1/A 0/1-Matrix for valve length and breadth as well as number of striae; S1/B Trees of Individual Gene Markers (18SV4, ITS, RBCL); S1/B1 18SV4: MP strict consensus tree, LB, PB and PP given, black arrows indicate SAMS support, clades named according to Fig. 7; S1/B2 ITS: MP strict consensus tree, LB, PB and PP given, black arrows indicate SAMS support, clades named according to Fig. 7; S1/B3 rbcL: MP strict consensus tree, LB, PB and PP given, black arrows indicate SAMS support, clades named according to Fig. 7; S1/B4 18SV4 with Genbank sequences: Bootstrapped (10 000 replicates) MP strict consensus tree, PB given, black arrows indicate SAMS support, clades named according to Fig. 7; S1/B5 18SV4 with Genbank sequences: Bootstrapped (10 000 replicates) RAxML best scoring ML tree, LB given, black arrow indicate SAMS support,clades named according to Fig. 7; S1/B6 18SV4 with Genbank sequences: Majority rule consensus tree from MrBayes, PP given, black arrows indicate SAMS support, clades named according to Fig. 7; S1/B7 ITS with Genbank: MP strict consensus tree, LB, PB and PP given, black arrows indicate SAMS support, clades named according to Fig. 7; S1/B8 rbcL with Genbank: MP strict consensus tree, LB, PB and PP given, black arrows indicate SAMS support, clades named according to Fig. 7; S1/C Uncorrected P-Distances; S1/C1 18SV4; S1/C2 ITS; S1/C3 rbcL; S1/C4 18SV, ITS and rbcL combined; S1/C5 18SV, ITS and rbcL combined, compared to Genbank sequences; S1/C6 Clade 2: Individual p-Distances for 18SV, ITS and rbcL compared to Genbank sequences; S1/C7 Clade 3: Individual p-Distances for 18SV, ITS and rbcL compared to Genbank sequences; S1/C8 Clade 4: Individual p-Distances for 18SV, ITS and rbcL compared to Genbank sequences.



The authors wish to thank Dr. Daniel Lauterbach for his support with R. We are grateful to Jana Bansemer and Dr. Oliver Skibbe for sampling and cultivating. We thank Dr. Harry Enke for providing support with the calculations.

Author Contributions

Conceived and designed the experiments: NA RJ NE. Performed the experiments: NA JZ. Analyzed the data: NA NE RJ JZ. Contributed reagents/materials/analysis tools: RJ NA. Wrote the paper: NA RJ JZ NE.


  1. 1. Becking LGMB (1934) Geobiologie, of Inleiding Tot de Milieukunde: Met Literatuurlijst en Ind. Van Stockum
  2. 2. De Wit R, Bouvier T (2006) Everything is everywhere, but, the environment selects; what did Baas Becking and Beijerinck really say? Environ Microbiol 8: 755–758
  3. 3. Finlay BL, Monaghan EB, Maberly SC (2002) Hypothesis: the rate and scale of dispersal of freshwater diatom species is a function of their global abundance. Protist 153: 261–273.
  4. 4. Boo SM, Kim HS, Shin W, Boo GH, Cho SM, et al. (2010) Complex phylogeographic patterns in the freshwater alga Synura provide new insights into ubiquity vs endemism in microbial eukaryotes. Mol Ecol 19: 4328–4338.
  5. 5. Cermeño PVC, Abrantes F, Falkowski PG (2010) Phytoplankton Biogeography and Community Stability in the Ocean. PLOS ONE 5(4): e10037
  6. 6. Chao A, Li PC, Agatha S, Foissner W (2006) A statistical approach to estimate soil ciliate diversity and distribution based on data from Wve continents. Oikos 114: 479–493.
  7. 7. Foissner W (2006) Biogeography and dispersal of micro-organisms: a review emphasizing protists. Acta Protozool 45: 111–136.
  8. 8. Padisák J (2009) The phycogeography of freshwater algae. In: Likens, G, editors. Encyclopedia of Inland Waters, Vol. I. Elsevier Inc. Oxford, pp 219–223.
  9. 9. Krienitz L, Bock C, Kotut K, Pröschold T (2012) Genotypic diversity of Dictyosphaerium–morphospecies (Chlorellaceae, Trebouxiophyceae) in African inland waters, including the description of four new genera. Fottea, Olomouc 12(2): 231–253.
  10. 10. Vyverman W, Verleyen E, Sabbe K, Vanhoutte K, Sterken M, et al. (2007) Historical processes constrain patterns in global diatom diversity. Ecology 88: 1924–1931
  11. 11. Moser G, Lange-Bertalot H, Metzeltin D (1998) Insel der Endemiten: Geobotanisches Phänomen Neukaledonien. Bibliotheca Diatomologica 38. Berlin. Stuttgart: Strauss Offsetdruck press. 464 p.
  12. 12. Cocquyt C, Jahn R (2007) Endemic taxa related to Surirella ovalis from Tanzania, East Africa: Typification of O. Müller's taxa and description of a new species. Cryptogamie Algol 28: 107–116.
  13. 13. Poulícková A, Vesela J, Neustupa J, Skaloud P (2010) Pseudocryptic diversity versus cosmopolitanism in diatoms: a case study on Navicula cryptocephala Kütz. (Bacillariophyceae) and morphological similar taxa. Protist 161: 353–369.
  14. 14. Souffreau C, Verbruggen H, Wolfe PA, Vanormelingen P, Siver AP, et al. (2011) A time-calibrated multi-gene phylogeny of the diatom genus Pinnularia. Mol PhylogenetEvol 61: 866–879.
  15. 15. Vanormelingen P, Verleyen E, Vyverman EW (2009) The diversity and distribution of diatoms: from cosmopolitanism to narrow endemism. Biodivers Conserv 17: 393–405.
  16. 16. Lange-Brtalot H (2007) Kosmopoliten versus Endemiten: Biogeographische Implikationen steigender Taxazahlen. In: Kusber WH, Jahn R, editors. Proceedings of the 1st Central European Diatom Meeting 2007. pp. 107–112.
  17. 17. Besse-Lototskaya AA, Verdonschot PFM, Coste M, Van de Vijver B (2011) Evaluation of European diatom trophic indices. Ecol Indic 11: 456–467.
  18. 18. Mann DG, McDonald SM, Bayer MM, Droop SJM, Chepurnov VA, et al. (2004) The Sellaphora pupula species complex (Bacillariophyceae): morphometric analysis, ultrastructure and mating data provide evidence for five new species. Phycologia 43: 459–482.
  19. 19. Alverson AJ (2008) Molecular systematics and the diatom species. Protist 159: 339–353.
  20. 20. Kaczmarska I, Lovejoy C, Potvin M, MacGillivary ML (2009) Morphological and molecular characteristics of selected species of Minidiscus. Eur J Phycol 44: 461–475.
  21. 21. Chen GF, Wang GC, Zhang BY, Fan XL (2007) Morphological and phylogenetic analysis of Skeletonema costatum-like diatoms (Bacillariophyta) from the China Sea. Eur J Phycol 42: 163–175.
  22. 22. Moniz MB, Kaczmarska I (2010) Barcoding of diatoms: nuclear encoded ITS revisited. Protist 161: 7–34.
  23. 23. Jütter I, Ector L, Reichardt E, Van de Vijver B, Jarlman A, et al. (2013) Gomphonema varioreduncum sp. nov., a new species from northern and western Europe and a re-examination of Gomphonema exilissimum. Diatom Res 303–31.
  24. 24. Wortley AH, Scotland RW (2006) The effect of combining molecular and morphological data in published phylogenetic analyses. Systematic Biol 55: 677–685.
  25. 25. Kützing FT (1844) Die kieselschaligen Bacillarien oder Diatomeen. Nordhausen. 152 p.
  26. 26. Kützing FT (1849) Species Algarum. Lipsiae. F.A. Brockhaus. 922 p.
  27. 27. Heiberg PAC (1863) Conspectus criticus Diatomacearum Danicarum.,Forlag, Copenhagen. 135 p.
  28. 28. Rabenhorst L (1864–1868) Flora Europaca Algarum aquae dulcis et submarinae, 1(3). apud E. Kummerum.
  29. 29. Grunow A (1878) Algen und Diatomaceen aus dem Kaspischen Meere. Seperat aus den Sitzungsberichten der natur. Abh zooL-bot Ges. Wien, 10.
  30. 30. Cleve PT (1894) Synopsis of the naviculoid diatoms. Kongl Svenska Vetenskaps Akad Handl. 26: : 194 p.
  31. 31. Van Heurck H (1896) A treatise on the Diatomaceae. William Wesley & Son. 558 p.
  32. 32. Frenguelli J (1923) Contribuciones para la sinopsis de las Diatomeas Argentinas. Boln Acad nac Cienc. Cordoba 32: 11–119.
  33. 33. Mayer (1928) Die bayerischen Gomphonema. Denkschr K bayer bot Ges Regensb XVII NF Bd XI: 83–127.
  34. 34. Hustedt F (1930) Bacillariophyta (Diatomeae). Heft 10. Die Süsswasser-Flora Mitteleuropas. G. Fisher, Jena.
  35. 35. Geitler L (1932) Der Formwechsel der pennaten Diatomeen (Kieselalgen). Arch Protistenk 78: 1–22.
  36. 36. Wallace JH, Patrick R (1950) A consideration of Gomphonema parvulum Kiitz. Butler Univ bot stud 9: 227–234.
  37. 37. Hohn M, Patrick R (1959) Variability in three species of Gomphonema (Bacillariophyceae) undergoing auxospore formation. Notul Nat 316: 1–7.
  38. 38. Carter JR (1960) Diatom notes: British freshwater forms of the genus Gomphonema. J Queckett microsc Club 12: 255–265.
  39. 39. Patrick R, Reimer CW (1975) The Diatoms of the United States, Exclusive of Alaska and Hawaii, Volume 2, Part 1 – Entomoneidaceae, Cymbellaceae, Gomphonemaceae, Ephitemaceae. Academy of Natural Sciences of Philadelphia Monograph.13: : 213 p.
  40. 40. Krammer K, Lange-Bertalot H (1986–1991) Süsswasserflora von Mitteleuropa.–Bacillariophyceae 1–4. Gustav FischerVerlag, Stuttgart, Jena, 876: : 576 p.
  41. 41. Lange-Bertalot H (1993) 85 neue Taxa und uber 100 weitere neu definierte Taxa ergänzend zur Süßwasserflora von Mitteleuropa 2/1–4: 85. Biblioth Diatomol 27. J. Cramer. 454 p.
  42. 42. Kobayasi H, Idei M, Mayama S, Nagumo T, Osada K (2006) H. Kobayasis's Atlas of Japanese Diatoms based on electron microscopy. - Tokyo, Uchida Rokakuho Publishing Co. Ltd. v. 1: 531 p.
  43. 43. Round FE, Crawford RM, Mann DG (1990) The diatoms: biology and morphology of the genera: Cambridge University Press. 747 p.
  44. 44. Van Dam H, Mertens A, Sinkeldam J (1994) A coded checklist and ecological indicator values of freshwater diatoms from the Netherlands. Neth J Aquat Ecol 28: 117–133.
  45. 45. Lange-Bertalot H, Genkal SI (1999) Diatomeen aus Sibirien, I: Inseln im Arktischen Ozean (Yugorsky-Shar Strait). Iconogr Diatomol 6: 304.
  46. 46. Walley WJ, Grbović J, Džeroski S (2001) A reappraisal of saprobic values and indicator weights based on Slovenian river quality data. Water res 35: 4285–4292.
  47. 47. Bellinger BJ, Cocquyt C, O'Reilly CM (2006) Benthic diatoms as indicators of eutrophication in tropical streams. Hydrobiologia 573: 75–87.
  48. 48. Krivograd KA, Toman MJ (2010) Influence of environmental variables on benthic algal associations from selected extreme environments in Slovenia in relation to the species identification. Periodicum biologorum 112: 179–191.
  49. 49. Hofmann G, Werum M, Lange-Bertalot H (2011) Diatomeen im Süsswasserbenthos von Mitteleuropa: A. r. G. Gantner Verlag K. G. Ruggell. 908 p.
  50. 50. Kermarrec L, Bouchez A, Rimet F, Humbert J-F (2013) First evidence of the existence of semi-cryptic species and of a phylogeographic structure in the Gomphonema parvulum (Kützing) Kützing complex. Protist 164: 686–705.
  51. 51. Romero OE, Jahn R (2013) Typification of Cocconeis lineata and Cocconeis euglypta. Diatom Res 28: 175–184.
  52. 52. Gemeinholzer B, Droege G, Zetzsche H, Knebelsberger T, Raupach M, et al.. (2009) - (continuously updated) DNA Bank Network Webportal. Available:
  53. 53. Zimmermann J, Jahn R, Gemeinholzer B (2011) Barcoding diatoms: evaluation of the V4 subregion on the 18S rRNA gene, including new primers and protocols. Org Divers Evol 11: 173–192.
  54. 54. White TJ, Bruns T, Lee S, Taylor J (1990) Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In: PCR Protocols. A Guide to Methods and Aplications. In Innis MA, Gelfand D H, Sninsky JJ, White TJ. San Diego. Academic Press. pp. 315–322.
  55. 55. Amato A, Kooistra WH, Levialdi Ghiron JH, Mann DG, Pröschold T, et al. (2007) Reproductive isolation among sympatric cryptic species in marine diatoms. Protist 158: 193–207.
  56. 56. Ivanova NV, Zemlak TS, Hanner RH, Hebert PD (2007) Universal primer cocktails for fish DNA barcoding. Mol Ecol Notes 7: 544–548.
  57. 57. Larkin MA, Blackshields G, Brown NP, Chenna R, McGettigan PA, et al. (2007) Clustal W and Clustal X version 2.0. Bioinformatics 23: 2947–2948.
  58. 58. Hall TA (1999) BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl Acid S 41: 95–98.
  59. 59. Müller K, Müller J, Neinhuis C, Quandt D (2005) PhyDe® - Phylogenetic data editor. Available:
  60. 60. Krammer K, Lange-Bertalot H (1997) Bacillariophyceae, 2 (1–4). Süsswasserflora von Mitteleuropa. Stuttgart: Fischer press. 1991.
  61. 61. Lange-Bertalot H, Metzeltin D (1996) Oligothrophie-Indikatoren: 800 Taxa repräsentativ für drei diverse Seen-Typen Kalkreich-Oligodystroph-Schwach gepuffertes Weichwasser. Iconogr Diatomol 2. Koenigstein: Koeltz Scientific Books. 390 p.
  62. 62. Ross R, Cox EJ, Karayeva NI, Mann DG, Paddock TBB, et al. (1979) An amended terminology for the siliceous components of the diatom cell. Nova Hedwigia Beih 64: 513–533.
  63. 63. Wägele JW, Mayer C (2007) Visualizing differences in phylogenetic information content of alignments and distinction of three classes of long-branch effects. BMC Evol Biol 7: 147
  64. 64. Swofford DL (2002) PAUP*: Phylogenetic analyses using parsimony (* and other methods), version 4. 0beta. Sunderland: Sinauer.
  65. 65. Felsenstein J (1985) Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39: 783–791.
  66. 66. Posada D, Crandall KA (1998) Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817–818.
  67. 67. Tavaré S (1986) Some probabilistic and statistical problems in the analysis of DNA sequences. Lectures Math Life Sci 17: 57–86.
  68. 68. Stamatakis A (2006) RAXML-VI-HPC: Maximum-Likelihoodbased phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22: 2688–2690.
  69. 69. Stamatakis A, Hoover P, Rougemont J (2008) A rapid bootstrap algorithm for the RAxML web-servers. Systematic Biol 75: 758–771.
  70. 70. Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574.
  71. 71. Rambaut A (2008) Figtree v1.2.2. Insitute of Evolutionary Biology, University of Edinburgh. Available: software/figtree.
  72. 72. Mantel N (1967) The detection of disease clustering and a generalized regression approach. Cancer Res 27: 209–212.
  73. 73. R Development Core Team (2008) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available: Accessed June 2008.
  74. 74. Reichardt E (1999) Zur Revision der Gattung Gomphonema: Die Arten um G. affine/insigne, G. angustatum/micropus, G. acuminatum sowie gomphonemoide Diatomeen aus dem Oberoligozän in Böhmen. Iconogr Diatomol 9. ARG Gantner. 203 p.
  75. 75. Dawson PA (1972) Observations on the structure of some forms of Gomphonema parvulum Kütz,. Brit Phycol J 7: 255–271.
  76. 76. Jahn R (1986) A study of Gomphonema augur Ehrenberg: the structure of the frustule and its variability in clones and populations. – In MRicard (ed.): Proceedings 8th International Diatom Symposium, Paris 1984. pp. 191–204.
  77. 77. McBride SA, Edgar RK (1998) Janus cells unveiled: frustular morphometric variability in Gomphonema angustatum. Diatom Res 13: 293–310.
  78. 78. Mann DG (1999) The species concept in diatoms. Phycologia 38: 437–495.
  79. 79. Hustedt F (1985) Die Kieselalgen Deutschlands, Österreichs und der Schweiz Bd. 7, Teil 2. Leipzig.
  80. 80. Metzeltin D, Lange-Bertalot H, Garcia-Rodriguez F (2005) Diatoms of Uruguay compared with other taxa from South America and elsewhere. Iconogr Diatomol 15. - Königstein, Germany: A.R.G. Gantner Verlag K.G. 736 p.
  81. 81. Michels-Estrada A (2003) Ökologie und Verbreitung von Kieselalgen in Fliessgewässern Costa Ricas als Grundlage für eine biologische Gewässergütebeurteilung in den Tropen. Department für Ökologie Fachgebiet Limnologie. Technischen Universität München. Dissertation: 244 p.
  82. 82. Algal Flora of Korea (2011) Freshwater Diatoms IV. Vol. 3, Nr.4. compiled by Lee, Jung Ho, National Institute of Biological Resources, Ministry of Environment.
  83. 83. Mpawenayo B, Cocquyt C, Nindorera A (2005) Les diatomées (Bacillariophyta) et autres algues des sources thermales du Burundi (Afrique centrale) en relation avec les caractéristiques physiques et chimiques des eaux. Belg J Bot 138: 152–164.
  84. 84. Souffreau C, Vanormelingen P, Van de Vijver B, Isheva T, Verleyen E, et al. (2013) Molecular Evidence for Distinct Antarctic Lineages in the Cosmopolitan Terrestrial Diatoms Pinnularia borealis and Hantzschia amphioxys. Protist 164: 101–115.
  85. 85. Mann DG, Vanormelingen P (2013) An inordinate fondness? The number, distributions and origins of diatom species. Eukaryot Microbiol 60: 414–420.