Diatoms are valuable bioindicators and their traditional classification and identification are mainly based on the morphological characteristics of their frustules. However, in recent years, DNA-based methods have been proposed and are rapidly growing in the scientific literature as a complementary tool to assess the ecological status of freshwaters. Diatom-based ecological status assessment uses indices calculated from sensitivity and tolerance values as well as relative abundance of species. Correct assessment requires an accurate identification of species. In the present study, diatom assemblages of an oxbow lake were investigated using light and scanning electron microscopy as well as metabarcoding using rbcL marker, and the identification results were compared, intending to match barcode sequences of species that are currently missing in the diatom reference database. The investigated oxbow is an important wetland for bird conservation, although it is impacted by land use. Taxon lists based on morphology and metabarcoding considerably differed when bioinformatics analysis involved DADA2 pipeline with Diat.barcode database. Previously unknown sequence variants of four pennate species were found with additional BLAST search. Using phylogeny and p-distance calculations sequences could be matched to three small-celled naviculoid species that were found under a microscope. One of them was found to be a new species of the genus Mayamaea and was described as a new species, Mayamaea ectorii. Additionally, spatial distribution maps for several small-celled naviculoid species are provided for the Hungarian territory.

Mayamaea, metabarcoding, new diatom species, oxbow

River floodplains are some of the most valuable ecosystems on Earth; they play an important role in maintaining biodiversity and are essential for ecosystem services, but they are vulnerable (Vörösmarty et al. 2010). They are threatened not only by global warming (faster evaporation, less water recharge, changed physico-chemical conditions in warmer waters, etc. (Prakash 2021)), but also by the increasingly long-term loss of lateral connectivity, with the causes of this loss including the significant role of bed subsidence. The loss of lateral hydrological connectivity has been identified as a major cause of biodiversity loss and ecosystem degradation in such habitats (e.g. Amoros and Bornette 2002; Wang et al. 2016).

Diatoms frequently constitute a dominant group in benthic aquatic habitats. These algae have a significant role as primary producers and also as bioindicators in ecological status assessments. For surface water monitoring, a light microscope (LM) is the primary tool available to investigators, and the EU WFD (Water Framework Directive) itself still currently requires its use (European Commission 2000). However, for accurate identification of small diatom species, a scanning electron microscope (SEM) is essential, although not mandatory.

Because traditional morphology-based identification of species is labour intensive and requires deep taxonomic knowledge (Rimet and Bouchez 2012), the application of DNA-based methods has been proposed. Metabarcoding is a process in which short, so-called barcode DNA sequences (Hebert et al. 2003) are acquired with high-throughput sequencing (HTS) and used for the identification of species (Taberlet et al. 2012). In case of diatoms, various DNA regions was suggested as barcode markers such as V4 and V9 subregions of the nuclear 18S rDNA (Zimmermann et al. 2011; Guo et al. 2016), a 312 bp region of the plastid gene, rbcL (Vasselon et al. 2017), for certain groups internal transcribed spacer (ITS) and cytochrome-c oxidase subunit 1 gene (COI, Guo et al. 2015). Vasselon et al. (2017, 2018) elaborated a diatom metabarcoding process based on a short barcode region of the plastid gene rbcL, that was tested by several authors in recent years (Mortágua et al. 2019; Borrego-Ramos et al. 2021; Duleba et al. 2021). As reference for taxonomic assignment of sequences the method by these authors uses Diat.barcode, a curated database that was developed for the taxonomic assignment of the sequences (Rimet et al. 2019) and is under continuous development and frequently updated.

Within the framework of a national scale project associated with the EU WFD, a total of 242 sampling points were selected to cover all water types in Hungary according to the typology. During this project, in an oxbow of Tisza River (Körtvélyesi Holt-Tisza near Szeged), we have found several small-celled naviculoid diatom species. Some of them were dominant; moreover one species (with a relative abundance of 10.2%) among them was unidentifiable on the basis of present literature. Small naviculoid diatoms frequently dominate in freshwaters, and when their relative abundance exceeds 5%, their ecological importance is unquestionable (Wetzel et al. 2015).

This work aimed to compare microscopy and metabarcoding in the view of the performance in detecting diatom species in the community. We intended to identify species as precisely as possible. On one hand, during morphological investigation, scanning electron microscopy (SEM) was used in addition to light microscopy (LM), especially for small naviculoids. On the other hand, during metabarcoding, the first bioinformatics analysis (DADA2-based pipeline and taxonomic assignment using Diat.barcode) was supplemented with Basic Local Alignment Search Tool (Altschul et al. 1990) in NCBI GenBank database. Furthermore, we could identify the sequences of some naviculoid species that were found with microscopy. These included a new Mayamaea species that is described in the present paper.

Using light and scanning electron microscope, a total of 50 taxa from 15 diatom genera were identified in the samples. The Shannon diversity was 4.23 in the summer and 3.50 in the autumn sample. Dominant species (relative abundance of at least 5% or more in one sample) were Brevilinea kevei Ács & Ector in Ács, C.E. Wetzel (3.5%, 11.9%), Craticula subminuscula (Manguin) C.E. Wetzel & Ector (7.6%, 10.5%), Mayamaea permitis (14.9%, 0.2%), Mayamaea sp. (10.2%, 0.2%), Navicula cryptotenella Lange-Bertalot (11.9%, 0%), N. microrhombus (Cholnoky) Schoeman & Archibald (0.4%, 13.6%), N. veneta Kützing (5.1%, 0%), Nitzschia archibaldii Lange-Bertalot (2.0%, 10.7%), N. inconspicua Grunow (4.7%, 7.0%), N. lacuum Lange-Bertalot (1.8%, 22.0%), N. paleacea (5.5%, 8.4%), N. palea (Kützing) W. Smith (6.3%, 0%), N. supralitorea Lange-Bertalot (5.1%, 1.4%). The numbers in brackets are relative abundances, the first always refers to the sample from June, the second from September (for the whole taxon list with relative abundances see Suppl. material 1: Table S1).

In addition to the dominant species, the following small naviculoid diatom species were also found in the sample: Craticula importuna (Hustedt) Lange-Bertalot (3.3%, 0%), Sellaphora archibaldii (J.C. Taylor and Lange-Bert.) Ács, C.E. Wetzel & Ector (0%, 1.8%) and S. nigri (De Not.) C.E. Wetzel & Ector (0.6%, 2.3%). Exact identification of these small naviculoid species was only possible by electron microscopy (Fig. 1), which was important, because the accuracy of species-level identification can influence the classification results, especially if the given species is dominant in the sample.

Figure 1. 

Images of small naviculoid diatoms from the Körtvélyesi Holt-Tisza. A–C. Craticula importuna; D–F. Craticula subminuscula; G–I. Sellaphora nigri; J–L. Mayamaea permitis; M–O. Mayamaea ectorii; P–R. Navicula microrhombus; S–U. Sellaphora archibaldii; V–X. Brevilinea kevei. LM (A, D, G, J, M, P, S, V; scale bar: 10 µm) and SEM (external view: B, E, H, K, N, Q, T, W and internal view: C, F, I, L, O, R, U, X). Scale bars: 1 µm.

Based on metabarcoding in both the summer and the autumn samples, 58–58 diatom amplicon sequence variants (ASVs) were identified at species level with the DADA2 pipeline. After the final step of the bioinformatics processing of the HTS data, there were 226,469 and 124,259 reads in the June and the September sample, respectively. These constituted 88 and 110 ASVs, respectively. Of these, 19 and 26 ASVs did not belong to Bacillariophyta. Seven ASVs were additionally identified using BLAST search, these belonged to Cyclostephanos dubius (Hustedt) Round, Pantocsekiella ocellata (Pantocsek) K.T. Kiss & Ács, Nitzschia paleacea, Gomphonema gracile Ehrenberg, Amphora indistincta Levkov, and Sellaphora pupula (Kützing) Mereschkovsky.

In the case of 24 ASVs that could not be identified at species level with the DADA2 pipeline, BLAST search found 98% or lower similarity with the most similar record in the GenBank. Most of them related to the genera Nitzschia and Gomphonema. Identifying the sequences of Brevilinea kevei and Navicula microrhombus was not possible. We did not have information on the lineage of their genera. All sequences that belonged to the genus Navicula could be assigned to N. cryptotenella and N. veneta. The presence of these Navicula species in the samples was proved by microscopy.

Furthermore, we used the taxon lists obtained with microscopy to help identify more sequences that could not be assigned based on the reference DNA databases (Diat.barcode, GenBank).

One ASV showed 95% similarity with Mayamaea permitis and it was dominant in the June sample but occurred only in a few read numbers in the September sample. This abundance distribution in the samples was similar to that shown by a presumably undescribed Mayamaea species (listed as Mayamaea sp.) under microscope. A variant of this ASV (1 nucleotide difference) was also detected, however, in low abundance (94 reads) and only in the June sample. Based on the Mayamaea sequences in Diat.barcode, the mean p-distance of this region of rbcL in this genus is 0.04, the minimum distance between two taxons [M. atomus (Kützing Lange-Bertalot and a M. permitis variant] was 0.008, the maximum distance [between M. alcimonica (E. Reichardt) C.E. Wetzel, Barragán & Ector and M. permitis] was 0.13. This ASV showed 0.053–0.068 p-distance with variants of M. permitis (Suppl. material 2: Table S2). Therefore, these ASVs were accepted as the sequences of the undescribed Mayamaea species. Using this sequence in phylogenetic analysis, this species was positioned between Mayamaea species but on distinct lineage as a sister to a group containing lineages of M. permitis, M. atomus, Mayamaea terrestris N. Abarca & R. Jahn, Mayamaea pseudoterrestris Tuji et H.Yamaguchi and Mayamaea vietnamica Glushchenko, Kezlya, Kulikovskiy & Kociolek (Fig. 2, the new species referred as Mayamaea ectorii KHT; see description and naming of this species below; the abbreviation KHT refers to Körtvélyesi Holt-Tisza). This separation was supported by a relatively high bootstrap value. Various M. permitis sequences were identified from Körtvélyesi Holt-Tisza (M. permitis KHT1-6) by DADA2-based analysis and these were located in a group of M. permitis that supported their identification. Mayamaea alcimonica represented a distinct lineage formed by a sequence from the database and an ASV from Körtvélyesi Holt-Tisza. This environmental sequence occurred in low abundance. It was identified by DADA2-based analysis that was confirmed by phylogenetic analysis. Mayamaea sequences grouped separately from Rossia, Fallacia and Sellaphora.

Figure 2. 

Maximum likelihood phylogenetic tree of Mayamaea species using the rbcL fragment. Rossia, Fallacia and Sellaphora taxa were used as outgroups. Bootstrap values are indicated at nodes. Sequences acquired in present study are in bold and indicated with “KHT” (referring to Körtvélyesi Holt-Tisza). Sequences from database are provided with NCBI GenBank accession number (if available) or culture ID of Thonon Culture Collection.

Based on the morphological investigations, a dominant Mayamaea species showed a distinct morphology, different from all other known Mayamaea species, and this was also confirmed by the metabarcoding results of the rbcL amplicon. So we described it as new as follows:

Mayamaea ectorii Ács, Kiss & C.E. Wetzel, sp. nov.

Fig. 3A–M

Morphology. Valves are small, oval with slightly pointed apices. The observed range of valve dimensions (n = 35): length 6.2–7.9 μm, width 4.2–5.3 μm, 32–36 striae in 10 μm.

In light microscope the “three dots” can be detected as spots, which is characteristic for the genus Mayamaea (Fig. 3A–I). Striae are uniseriate, radiate throughout; two to three shorter striae present in the very middle of the valve (Fig. 3J–M). Striae composed of 8 to 11 areolae located on the valve face, extending down onto an extremely shallow mantle, bearing a single row of areolae (Fig. 3K). Small silica depositions can be observed in some specimens on the valve mantle/face junction (Fig. 3J).

Figure 3. 

Mayamaea ectorii sp. nov. LM (A–I; scale bar: 10 µm) and SEM (external view: J, K and internal view: L, M). Scale bars: 1 µm.

The axial area is narrow, linear over most of its length. Central area is relatively wide, symmetrical, marked by the “three dots” corresponding to a thickened siliceous area. Proximal raphe endings are slightly curved towards the same side, with teardrop proximal pores (Fig. 3J). Distal external raphe fissures with different lengths, bending strongly to the same side of the valve, continuing onto the mantle (Fig. 3K).

Internally, proximal raphe endings slightly deflected to the same side and distal raphe endings terminating on helictoglossae (Fig. 3L). Areolae showing a wide diameter and rounded shape (Fig. 3M) are covered by hymens located towards the outer opening of the areolae (Fig. 3J, K). The species was found in hypertrophyic conditions (Suppl. material 3: Table S3).

Holotype. (Fig. 3K, M) designated here. Type material: CER-14/13 and CER-14/14, deposited at the Centre for Ecological Research, Institute of Aquatic Ecology, Budapest, Hungary, epiphyton, collected by B. Csányi 22/06/2019. HNHM-ALG-D002301 and HNHM-ALG-D002302 deposited at the Hungarian Natural History Museum, Budapest, Hungary.

Type locality. Körtvélyesi Holt-Tisza, Hungary (46.423980°N, 20.230290°E).

Isotypes. Fig. 5J, L designed here.

Etymology. The name is chosen in honour of our dear friend, the famous Belgian diatomologist Luc Ector (1962–2022†).

Similar logic to that presented for Mayamaea species led to finding the sequence of two Craticula species. One ASV was dominant in samples and showed 95% similarity with Craticula species [Craticula cuspidata (Kützing) D.G. Mann, C. ambigua (Ehrenberg) D.G. Mann]. Within this genus, the mean p-distance was 0.08 in the studied region of rbcL. The minimum pairwise distance was 0.004 between C. ambigua and C. cuspidata, the maximum was 0.14 between C. accomoda (Hustedt) D.G. Mann and C. importuna (Hustedt) K. Bruder & Hinz. Thus, this ASV belonged to the Craticula genus. Moreover, C. subminuscula was dominant in both samples based on microscopy. The ASV showed the lowest p-distance (0.046) with C. subminuscula (Suppl. material 4: Table S4). Therefore, it was accepted as the sequence of C. subminuscula. Another ASV occurred only in the June sample and showed 0.015 p-distance (4 nt differences) with C. importuna AT-70Gel14a (accession number AM710444) that was regarded as C. molestiformis in its original publication (Bruder and Medlin 2007) and later was transferred to C. importuna by Bruder and Medlin (2008). High similarity (99%) between sequences may suggest conspecifity of our taxon with C. importuna AT-70Gel14a (accession number AM710444).

Phylogenetic analysis (Fig. 4) showed that the two Craticula species that occurred in our samples (marked with “KHT”) were located within Craticula species that formed several lineages. Craticula subminuscula from Körtvélyesi Holt-Tisza (KHT) was located most closely to two C. subminuscula strains. Our C. importuna was the closest relative of C. importuna AT-70Gel14a (AM710444), forming a lineage distinct from C. molestiformis sequences (Fig. 4). Separation of C. subminuscula as well as C. importuna from its closest relatives was supported by medium bootsrap values.

Figure 4. 

Maximum likelihood phylogenetic tree of Craticula species using the rbcL fragment. Stauroneis taxa were used as outgroups. Bootstrap values are indicated at nodes. Sequences acquired in present study are in bold and indicated with “KHT” (referring to Körtvélyesi Holt-Tisza). Sequences from database are provided with NCBI GenBank accession number (if available) or culture ID of Thonon Culture Collection.

Sequences of M. ectorii, C. subminuscula and C. importuna have been deposited in the NCBI GenBank database under accession numbers OP354489–OP354492.

The sequence of Sellaphora archibaldii could not be identified either. Sellaphora nigri and S. saugerresii sequences were identified by DADA2-based pipeline, BLAST search the sequence of S. pupula. Almost all Sellaphora sequences were S. nigri, one of them was S. saugerresii (Desm.) C.E. Wetzel & D.G. Mann in Wetzel et al., and the other one was S. pupula. On the phylogenetic tree (Fig. 5) S. nigri sequences (KHT1–4) grouped with S. nigri sequences from the database, S. saugerresii KHT formed a group with other S. saugerresii sequences. This confirmed their original identification. Sellaphora pupula KHT located in the group of taxa of S. pupula complex. The presence of S. nigri was verified under microscope in both samples. The other two Sellaphora sequences were present only in small numbers (S. saugerresii had merely 18, S. pupula had 15 reads).

Figure 5. 

Maximum likelihood phylogenetic tree of Sellaphora species using the rbcL fragment. Mayamaea, Rossia and Fallacia taxa were used as outgroups. Bootstrap values are indicated at nodes. Sequences acquired in present study are in bold and indicated with “KHT” (referring to Körtvélyesi Holt-Tisza). Sequences from database are provided with NCBI GenBank accession number (if available) or culture ID of Thonon Culture Collection.

In the summer sample (June 2019), Mayamaea permitis, Navicula veneta, Nitzschia palea and Fistulifera saprophila (Lange-Bertalot & Bonik) Lange-Bertalot reached relative abundance of more than 5% based on metabarcoding using the DADA2 pipeline and supplemented with a BLAST search. Based on microscopy, all of these species were dominant in the sample except for Fistulifera saprophila that was not detected. After comparison with morphology, the sequences of Craticula subminuscula and the presumably new Mayamaea species were identified and seemed to be dominant instead of Navicula veneta.

In the autumn sample (September 2019), Nitzschia filiformis (W.M. Smith) Van Heurck, N. inconspicua, N. palea and N. supralitorea were dominant. After BLAST search, N. paleacea also became dominant. Under microscope, N. filiformis and N. palea were not detected, N. supralitorea was present in 1.36%, N. inconspicua and N. paleacea were dominant. After comparison with morphology, Craticula subminuscula was also recognised as a dominant species based on DNA (Suppl. material 1: Table S1).

In the summer sample, eight dominant species were found based on light microscopy, of which six were detected also based on metabarcoding (using both DADA2 and BLAST), however, the identification of the sequence of two other species (Craticula subminuscula and Mayamaea sp.) became possible after the comparison with microscopy results. In the autumn sample, microscopy showed seven dominant species, however, only two (Nitzschia inconspicua and N. paleacea) were detected based on metabarcoding (with DADA2 and BLAST). The comparison of the morphology- and the DNA-based method added one further identification (Craticula subminuscula).

The MIL index calculated from the results of microscopy and metabarcoding are presented in Table 2. The MIL index calculated based on metabarcoding results (using any of the analyses) was lower than the morphology-based ones.

When microscopy results were also used for the taxonomic assignment of the sequences, Craticula subminuscula and C. importuna could influence the value of the index. Mayamaea ectorii could be added as Mayamaea sp. (specified only at genus level) to OMNIDIA. The genus has species with different index values, so in such form sensitivity and tolerance values were not available for M. ectorii and it could not be involved in index calculation. Further research on samples from various environments will be required to assign index values to this species. Comparison with morphology did not help reduce the difference between microscopy- and metabarcoding-based indices. In the case of the June sample, this may be due to the lack of ecological parameters of the newly described species. In the case of the September sample, the reason may be that sequences of dominant species e.g. B. kevei and N. microrhombus could not be identified.

The dynamics of the main branch-sub-branch relationship play a fundamental role in the life of rivers. The diversity and abundance of biota in tributaries fundamentally depend on the dynamics of water flow (Bayley 1991; Nilsson and Svedmark 2002). The issue of the vulnerability of river floodplains caused by changes in water flow due to river regulation is a priority and this is particularly true for sanctuary-like oxbows, which can maintain high and unique biodiversity. Comparing the benthic diatom species richness of different-sized lakes in Hungary, Bolgovics et al. (2016, 2019) found that lakes in the 10⁵ size range were the most diverse. This size range includes Körtvélyesi Holt-Tisza. TN and TP concentrations and Chl a values indicate the high nutrient content of the Körtvélyesi Holt-Tisza: this is a common phenomenon in wetlands which are important for bird conservation (Boros et al. 2008), moreover the land use also has significant effects on the water quality of this oxbow.

R-selected, small-celled species often develop in these oxbows, which are highly exposed to land use or natural eutrophication (Ács et al. 2016, 2017).

Among the small-sized naviculoid diatoms from the oxbow, Craticula importuna, C. subminuscula, Mayamaea permitis and Sellaphora nigri were also common in the other waters we studied (Suppl. material 6: Fig. S1). These diatoms are cosmopolitan species, and were also very common in studies including those from e.g. France (Bey and Ector 2013; Peeters and Ector 2019), or Reunion (Gassiole et al. 2019). Brevilinea kevei, Navicula microrhombus and Sellaphora archibaldii were rarely found in our surveys (their frequency was 2.9%, 1.4% and 4.5% respectively). Brevilinea kevei and Navicula microrhombus were not found, e.g. in the above-mentioned surveys in France or Reunion.

Brevilinea kevei is a relatively recently described species (Ács et al. 2016), but since then it has been found in Sweden (Sundberg 2022). Brevilinea genus contains only two species and so far, DNA sequence information is not available for either of them. Navicula microrhombus was described in 1970 by Cholnoky (Cholnoky 1970) from a fishpond in South Africa. Its first occurrence outside South Africa was reported from Slovakia in 2000 (Hindáková 2000), and it was first recorded in Hungary during a sampling campaign in 2001 (Szabó et al. 2004). The uncertain taxonomic position of this species has been reported by Hindáková (Hindáková 2001), and has not been clarified so far. Morphology (e.g. the raphe characteristics) of N. microrhombus suggests that it belongs to another genus other than Navicula sensu stricto. Though we could not identify the sequence of N. microrhombus, all sequences belonging to the Navicula genus were assigned to other species whose presence was proved under microscope, suggesting that N. microrhombus is not a Navicula species.

The applicability and comparison of HTS data and microscopy observations for finding barcode sequences of diatom species were first shown by Rimet et al. (2018). One of the identified sequences was the sequence of Gomphonema clavatuloides Rimet, D.G. Mann, Trobajo & N. Abarca that they described as a new species. This method has the advantage that the reference database can be enriched both with taxa and sequence variants of a given taxon complementing the labour intensive and often unsuccessful culturing (Rimet et al. 2018; Mora et al. 2019). This method helped us to identify sequence variants of species already recorded in the database and the sequence corresponding to a new species.

Our new species has the main characteristics of the genus Mayamaea, but no exact match could be found with any described Mayamaea species.

The genus Mayamaea was described by Lange-Bertalot in 1997 (Lange-Bertalot 1997); main characteristics: cells are small, with a maximum length of 16 µm and a maximum width of 7 µm, but mostly less than 10 µm long and 5 µm wide. Valves are always elliptical with broadly rounded poles. The axial area is thickened and contains a straight, filiform raphe. Distal raphe ends are curved in the same direction. Proximal ends are also slightly curved to the same side, but to the opposite side to the distal ends, and terminating in well-developed pores on the outside. Internally, the proximal ends of the raphe terminate in a central nodule and the distal ends in a medium-sized helictoglossa. These three nodules are clearly visible under a light microscope. Each stria is formed by a single row of areolae. Mayamaea species are found in ephemeral habitats, waters with high nutrient content and soils (Barragán et al. 2017).

Currently, 30 freshwater or terrestrial valid Mayamaea taxa are listed in the AlgaBase (Guiry and Guiry 2022, https://www.algaebase.org/) and/or the DiatomBase (Kociolek et al. 2022, https://www.diatombase.org/aphia.php?p=searh). Of these, four have overlapping striae number with Mayamaea ectorii, and these are M. arida, M. lacunolaciniata, M. nolensoides and M. permitis (Table 1). Mayamaea arida has wide, well-developed axial area and the striae are parallel in the central part, while M. ectorii has a narrow, linear axial area and striae strongly radiate along the whole length of the valve. Mayamaea lacunolaciniata is narrower (width 3.5–4 µm) and its axial area is different (its sternum expanded to broadly rhombic-lanceolate, structureless or with some irregularly dispersed areolae). Mayamaea nolensoides differs from M. ectorii in its distal raphe ends and completely hyaline area with short and higher striae density (impossible to solve even with DIC). The terminal nodules are in more proximal position, whereas in M. ectorii the distal raphe end continues on the mantle. Mayamaea permitis is narrower (width 3–4 µm) and it has broadly rounded apices, while M. ectorii has slightly pointed ones. In appearance, Mayamaea ectorii most resembles M. alcimonica, but it is wider and has a higher number of striae.

Molecular information on the genus Mayamaea is scarce (Kezlya et al. 2020); sequence data are available only for a few (seven) Mayamaea species. We used only six species because the sequence of M. fossalis was renamed to M. permitis in the Diat.barcode database. Moreover, in these data, M. permitis is overrepresented, and most of the other species are represented by one sequence. We also found several M. permitis sequence variants. A large diversity within M. permitis was observed by Kezlya et al. (2020) and Bagmet et al. (2021), raising the possibility of a cryptic species. Mayamaea ectorii was close to M. permitis variants, however, it was located on a distinct lineage suggesting that it can be accepted as an independent species.

The ecological requirements of Mayamaea ectorii and also the codominant species are nutrient rich freshwaters.

Our phylogenetic analysis showed the genus Craticula was not monophyletic which could be an artefact maybe due to short sequences. But deciding this is beyond the scope of this investigation. Nevertheless, phylogeny along with p-distance calculations confirmed that these two environmental sequences belonged to C. subminuscula and C. importuna detected under microscope.

Craticula importuna was identified for the first time in Hungary. It was probably previously misidentified as either C. minusculoides or C. molestiformis. The main distinguishing characteristics of C. minusculoides are the following: the areolae are not occluded with hymens externally and the distal raphe fissures are curved and do not continue on the valve mantle, while the areolae of C. molestiformis are covered by hymens externally and the distal raphe endings are long, strongly hooked and continuing onto the valve mantle (Levkov et al. 2016). We have not any information about the hymens in the case of C. importuna, but its distal raphe fissures are curved and do not continue on the valve mantle (Bruder and Medlin 2008). Based on the p-distance and the known morphological characteristics, we accepted our ASV as the sequence of C. importuna. However, as the studied gene sequence is short, the question requires further investigation.

Among the Sellaphora species, we identified S. archibaldii, S. nigri, S. pupula and S. saugerresii. Sellaphora archibaldii was described as an endemic species of South Africa for a long time (Taylor and Lange-Bertalot 2006). However, Ács et al. (2017) showed that it has been found in Hungary, France and Germany and more recently found on the island of Reunion as well (Gassiole et al. 2019). This species was not dominant in our studied samples and various explanations can be given for failing to find its sequence: e.g. its cell wall was more resistant, dead cells were observed under microscope or the primers did not match its sequence (Duleba et al. 2021). Identification of S. nigri and S. saugeresii sequences with DADA2 pipeline was confirmed by phylogeny. Sellaphora pupula is a species complex involving several cryptic species (Evans et al. 2008), our environmental sequence (identified after BLAST search) grouped with the taxa of this complex so it could be accepted as a member of the complex. Sellaphora nigri was also detected by microscopy. Sellaphora saugerresii and S. pupula occurred in very low read numbers so it was not surprising that they could not be found among 500 valves counted under microscope.

In this study, we investigated the diatom assemblages of an oxbow using two approaches, metabarcoding as well as microscopy (LM and SEM) in comparison. Both methods revealed high diversity, however, significant differences were found in the taxon lists. These discrepancies were mitigated when some sequences being unassigned by DADA2 pipeline were assigned using BLAST. Identifying sequences of three species (including two dominant species) was possible only after the comparison with the results of morphological analyses. Moreover, one of these taxa could be described as a new species of Mayamaea genus. Our results also demonstrate that metabarcoding comparing with microscopy can contribute to the complementing of the reference database.

The research presented in the article was carried out within the framework of the Széchenyi Plan Plus program with the support of the RRF 2.3.1 21 2022 00008 project and was supported by KEHOP-1.1.0-15-2016-00002 project as well.

With thanks to Dr Krisztina Buczkó for helping our taxonomic works in the project KEHOP and to Biomi Ltd for the DNA work.

We would also like to express our unusual thanks to our dear colleague, Luc Ector, who died before this article was completed, but he was here with us in our minds all the time, shaping our sentences as if we had just written the article together, as so many times before.