Introduction

High-throughput sequencing studies have increasingly contributed to a deeper analysis of microbial diversity in aquatic systems (Debroas et al., 2017; Singer et al., 2021) and have the potential to illuminate the diversity within protist groups (Barbera et al., 2019; Bock et al., 2014). Studies confirm the wide distribution of the Chrysophyceae (Stramenopiles) in different ecosystems and habitats, be it freshwater, soil or even in the ocean through high diversity and high frequency of occurrence (Charvet et al., 2012; Kristiansen & Škaloud, 2017; Remias et al., 2013; Schiaffino et al., 2020; Triadó-Margarit & Casamayor, 2012). Despite their central role as both primary producers and grazers on bacteria, the pattern of their phylogenetic and functional diversity in European inland waters is still insufficiently resolved.

Despite the general importance and the central ecological role of Chrysophyceae, most molecular community analyses are limited to the class-level (Grossmann et al., 2016; Khomich et al., 2017) and fail to unravel such questions regarding their phylogenetic and functional diversity. The diversity within Chrysophyceae is exceptionally high with regard to phylogenetics, morphology, and functions (Beisser et al., 2017; Kraus et al., 2019; Kristiansen & Škaloud, 2017). The Chrysophyceae comprise nine different orders each comprising a substantial diversity, i.e., Paraphysomonadida Scoble and Cavalier-Smith 2014, Synurales Andersen 1987, Chromulinales Pascher 1910, Chrysosaccales Bourrelly 1954, Hydrurales Pascher 1931, Hibberdiales Andersen 1989, Segregatales Boenigk and Grossmann 2016, Ochromonadales Pascher 1910 and Apoikiida Boenigk et Grossmann 2016b (Andersen, 1987, 1989; Brodie & Lewis (2007); Grossmann et al., 2016a; Kristiansen & Škaloud, 2017). However, in a recent publication, the orders Apoikiida and Chrysosaccales and closely related lineages like Chrysastrella and relatives were combined into one clade, incertae sedis (Andersen et al., 2017). Differences and commonalities of diversity and distribution patterns between these orders are largely unresolved.

Especially, patterns related to the phylogenetic distribution of the different nutritional modes are vague: While the Synurales are phototrophic, other orders comprise only heterotrophic taxa (e.g., Paraphysomonadida) or mixotrophic and heterotrophic taxa (e.g., Ochromonadales) (Holen & Boraas, 1995; Sandgren, 1988b). Differences between the nutritional modes with respect to both diversity and distribution are also at best vague: Pigmented Chrysophyceae are characteristic specifically for slightly acidic, soft waters with low alkalinity and conductivity and low to moderate productivity (Eloranta, 1995; Sandgren, 1988a; Siver, 1995). They also seem well adapted to low nutrient conditions (e.g., arctic ponds (Charvet et al., 2012) and northern European lakes (Järvinen et al., 2013)). Heterotrophic taxa, in contrast, are an important component of bacterivores in all different types of freshwaters ranging from acidic to alkaline and from oligotrophic to eutrophic conditions (Domaizon et al., 2003; Salonen & Jokinen, 1988; Schmidtke et al., 2006). Mixotrophs are probably somewhat intermediate based on their predominant nutritional mode but again tangible and conclusive evidence for differential distribution pattern is largely missing (Domaizon et al., 2003; Rottberger et al., 2013).

On the species level, the situation is not much better and detailed knowledge is limited to very few species. About 1200 species in approximately 112 genera have been described within the Chrysophyceae (Kristiansen & Škaloud, 2017). Among the morphologically and molecularly well-studied groups are the scale-bearing representatives, e.g., Paraphysomonadida and Synurales. They traditionally belong to the best-studied groups in morphological phytoplankton analyses since their siliceous scales provide reliable characteristics for morphological species differentiation (Jo et al., 2016; Nĕmcová & Kreidlová, 2013; Škaloud et al., 2012). In contrast, such reliable characters are not available for many other lineages. For instance, the fact that the inconspicuous heterotrophic, colorless, and biflagellate Chrysophyceae evolved independently in several clades has only been deduced based on DNA sequence comparison (Grossmann et al., 2016a). For their mixotrophic counterpart, i.e., Ochromonas, extensive work based on DNA sequencing and isolation led to epitypification and characterization of new genera hardly differentiable using morphology (Andersen et al., 2017). Similarly, the Uroglena-like morphotype, consisting of colonial naked Chrysophyceae with plastid-bearing cells, evolved several times independently. Only recent studies have investigated the more precise relationships (Pusztai & Škaloud, 2019). These recent studies substantiate a large and so far mostly overlooked diversity within the unscaled Chrysophyceae. However, their environmental diversity and distribution pattern are largely unknown (but see (Nolte et al., 2010)).

In our study, we analyzed the distribution patterns of the different Chrysophyceae lineages on a European scale. Based on an extensive phylogenetic tree, we here affiliated the 2370 environmental Chrysophyceae sequences originating from 218 European lakes to distinct phylogenetic lineages and orders as well as to distinct nutritional types. We thereby provide a basis for the linked analysis of the phylogenetic and the functional diversity within Chrysophyceae. We hypothesize based on literature data that distribution pattern systematically differs between nutritional types, as environmental factors, in particular pH and nutrient availability, differentially impact phototrophic and heterotrophic taxa. This is to be expected based on theoretical considerations and substantiated by published results on tolerance limits in laboratory studies (Bock et al., 2019; Boenigk, 2008). Further, based on the known diversity of Chrysophyceae, we hypothesize that Synurales, Paraphysomonadida, and Ochromonadales make up the majority of Chrysophyceae diversity.

Material and methods

The sequences and environmental data used in this manuscript originate from a pan-European study and occurred in August 2012 (Boenigk et al., 2018). The study included 218 freshwater lakes and ponds throughout Europe, comprising sites in Norway, Sweden, Germany, Poland, Romania, Austria, Italy, France, Spain, and Switzerland (for details on sampling sites see Supplementary Table S1). Temperature, pH, and conductivity were determined directly in the field by use of a Waterproof Tester “Combo” (Hanna Instruments, Vöhringen, Germany). Samples were taken near the shore of the respective lake from the first 50 cm of the water column and filtered on a 0.22-µm Isopore Membrane Filters (47 mm diameter, Merck Millipore, Darmstadt, Germany), air dried and frozen in liquid nitrogen (Cryoshippers). In the laboratory, the filters were stored at − 80 °C until further processing. Two technical replicates of the extracted DNA were independently amplified following the AmpliconDuo protocol (A and B variant, (Lange et al., 2015) and targeting the V9-ITS1 region of the 18S SSU rDNA, using a combination of two reverse primers with different wobble positions (5’-GCTGCGCCCTTCATCGKTG-3’ (ITS2_Dino; 10%) and 5’-GCTGCGTTCTTCATCGWTR-3’ (ITS2_broad; 90%)) and a forward primer (5’-GTACACACCGCCCGTC-3’) with an annealing temperature of 52 °C. The “rapid run” mode on the Illumina platform (paired-end HiSeq 2500 Illumina sequencing) was used for sequencing (Fasteris, Geneva, Switzerland) to generate 2 × 300 bp reads (for details see (Boenigk et al., 2018).

Bioinformatic analyses

Adapter removal, quality trimming, and demultiplexing using index sequences was performed by the sequencing company (Fasteris, Geneva, Switzerland). Base quality of raw sequence reads was checked using the FastQC software v0.11.5 (Andrews, 2010) and PRINSEQ-lite (v0.20.4; (Schmieder & Edwards, 2011)) were used for quality checks (removal of sequences with (a) average Phred quality score below 25, (b) reads with at least one base with a Phred quality score below 15). Assembled paired-end reads were again quality filtered with PANDASeq v2.10 (Masella et al., 2012). Reads with uncalled bases, an assembly quality score below 0.9, a read overlap below 20 bases, or a base with a recalculated Phred-score below 1 were discarded.

The remaining reads were clustered using the software SWARM v2.2.2 (Mahé et al., 2014) with default settings. The protist reads were further clustered by identical V9 sequences (150 basepairs, ident = 100%; R-Script “V9_Clust.R” (Jensen, 2017) and subsequently taxonomically assigned by searching the NCBI database using BLAST (BLAST + v2.7.1; NCBI nt sequences from Dec 5, 2017). The sequences were cut off at the end of V9 and further used for statistical and phylogenetic analysis. Read counts were converted to relative abundances and Hellinger transformed (Legendre & Gallagher, 2001).

Reference alignment, phylogenetic placement and trophic mode assignment

The phylogenetic reference alignment and corresponding reference tree (Fig. 1) were calculated based on sequences obtained from NCBI (http://www.ncbi.nlm.nih.gov/). Reference sequences representing the different clades of the Chrysophyceae were selected based on previous publications (e.g., (Andersen et al., 2017; Bock et al., 2017; Grossmann et al., 2016a; Kristiansen & Škaloud, 2017; Pusztai & Škaloud, 2019). Further, the NCBI database was checked for additional sequences. Priority was given to sequences for which at least partial taxonomic information was available. Only non-redundant sequences that contained the V9 region were included in the alignment. The sequences of Synchroma grande (DQ788730) and Leukarachnion sp. (FJ356265) were chosen as outgroups (Bock et al., 2017). This resulted in 167 sequences for the final reference alignment. The sequences were initially aligned using ClustalW (Thompson et al., 1994) and adjusted by eye using the SequentiX Alignment Editor (Hepperle, 2004). Ambiguous regions were excluded from the analyses, resulting in 1825 characters for the final reference tree. The phylogenetic analyses were conducted using RAxML-ng (Kozlov et al., 2019) with the GTR + FU + G4m model and 1000 bootstraps.

Fig. 1
figure 1

Phylogenetic reference tree. Thicker branches are those supported by bootstrap support larger than 65 (without placement of reads). The inner of the two outer circle indicates the nutritional mode of the reference sequences, the outer circle the order of Chrysophyceae. The circles at the base of the main clades correspond to the number of OTUs placed within the clade

In order to limit the large amount of OTUs (swarms) for accurate phylogenetic placement, all OTUs were selected that contained “Chrysophyceae,” “Ochromonadales,” or “Synurales” as part of the NCBI taxonomic blast hit in the first step. The corresponding sequences were automatically aligned onto the reference alignment using PaPaRa (Berger & Stamatakis, 2011). Phylogenetic placement of the OTUs was conducted using EPA-ng (Barbera et al., 2019), followed by the taxonomic assignments using Gappa (Czech et al., 2020). The taxonomic designation was performed using the “best-hit” option based on the highest likelihood weight (LWR) using Gappa. Sequences that were placed outside the Chrysophyceae were excluded from further analyses and not displayed in any figure (e.g., Fig. 1).

The nutritional mode assignment was based on the phylogenetic placement. All sequences that were phylogenetically placed within the Synurales were classified as “phototroph” following Siver (2003) and Andersen (2007). All placements within the Paraphysomonadida (including Paraphysomonas and Clathromonas) as “heterotroph” following Scoble and Cavalier-Smith (2014). The remaining sequences were classified according to their closest relative as mixotrophic if they contain a plastid, or as heterotrophic if the plastid is absent (Andersen et al., 2017; Holen & Boraas, 1995; Kristiansen & Škaloud, 2017; Sandgren, 1988b). If the trophy of the reference is not known or no exact placement was possible, the status “unknown” was assigned. The classification of the reference sequences to a trophic state is displayed in Fig. 1, that of the OTUs (SWARMs) in Table S2. The phylogenetic tree (Fig. 1) was displayed using iTOL (Version 5.6.2) (Letunic & Bork, 2019).

For each OTU, we determined the percentage of identity using the Basic Local Alignment Search Tool (BLAST) algorithm v 2.2.23 (Altschul et al., 1990) against the PR2 reference Database (Guillou et al., 2013). The best hit was determined for the best pident and an alignment length of at least 75 bases. The results were visualized using boxplot graphs in ggplot2 (Wickham, 2009). The pairwise genetic distances between the OTUs were calculated using the ‘dist.alignment’ function from the ‘seqinr’ package (Charif & Lobry, 2007).

Niche width and distribution of Chrysophyceae

For the calculation of the ecological niche width (tolerance range) of the occurrence of individual OTUs, only OTUs that occur in at least two lakes were used. For each of these OTUs, we looked at the ecological conditions under which they occur in our data set (temperature, pH, conductivity, altitude). The difference between the maximum and minimum values was then used as the tolerance range of the occurrence. Significance was examined using Wilcoxon rank test. The width of occurrence was plotted for different groups: Chrysophyceae OTUs vs. all other eukaryotic OTUs; width for the different clades within the Chrysophyceae and according to their potential nutritional mode (see above).

To assess the distribution of Chrysophyceae and their contribution to the total protist diversity, we calculated the percentage of Chrysophyceae OTUs as a percentage of the total number of OTUs. Maps were plotted using rworldMap (v.1.3–6), scatterpie (v.0.1.4), and ggplot2 (3.3.2) (Yu, 2020; South, 2011; Wickham, 2009) in which the size of the pies is proportional to the diversity contribution percentage. Ternary plots were prepared using the ggtern extension to the package ggplot2 (R software) (Hamilton & Ferry, 2018).

Results

Chrysophyceae show a high diversity in the Ochromonadales and Chrysosacca-Apoikiida clades

A phylogenetic reference tree was used to classify the occurring Chrysophyceae OTUs from our environmental DNA survey on 218 European lakes. We extracted 165 sequences from the NCBI database affiliated with Chrysophyceae as references. The main clades received medium to high bootstrap support (65–100) in the ML analyses (Fig. 1). The order of the clades among themselves found no support. After blasting and subsequent phylogenetic placement of the OTUs to the reference tree (Fig. 1), 2370 OTUs remained that can be classified as Chrysophyceae (out of 34 137 OTUs in total). Approximately 44.6% of the OTUs (1057) cluster within the Ochromonadales. Another large part clusters within the Chrysosacca-Apoikiida clade (including Chrysosaccales, Apoikiida, Chrysastrella and relatives)(576, 24.3%), followed by approx. 9.2% within the Synurales (218), approx. 7.3% in the Hydrurales (172) and approx. 6% in the Paraphysomonadida (142). The remaining groups account for a rather small proportion (sum approx. 8.6%) (details see Table 1, details on genetic distances Table S4).

Table 1 Taxonomic placement of the OTUs. An OTU is considered “novel” if the nearest hit has an identity of less than 97%

The genetic novelty in our dataset compared to published sequences varies among the clades. Out of our 2370 OTUs, 1159 OTUs (approx. 49%) showed at least 97% identity to known sequences from the PR2 database (Table 1). Ochromonadales (approx. 55.3%), Paraphysomonadida (approx. 84.5%), and Synurales (approx. 70.2%) comprised the highest percentage of known OTUs. In contrast, Hydrurales, Segregatales, and Chromulinales showed the highest percentage of “new sequences” (identity below 97%) (Fig. S1).

Most OTUs are affiliated with mixotrophic Chrysophyceae

Based on the phylogenetic placement, we identified the closest known relative and determined the presumable nutritional mode. A mixotrophic diet was assigned to the majority of OTUs, while only a small proportion was phototrophic (see Table 2). A considerable number of OTUs could not be assigned to a distinct nutritional mode — either because no exact placement was possible or because no diet was assigned to the next reference. This particularly concerned the separation of heterotrophic and mixotrophic taxa.

Table 2 Assigned nutritional mode of the OTUs based on next phylogenetic relative

The occurrence of Chrysophyceae within Europe

Chrysophyceae were found in nearly all sampled lakes (213 out of 218 lakes). However, their relative contribution to the diversity within the community varied widely from 0.15 to 23.67% of the OTUs detected (Table S1, Fig. 2). The relative read abundance (Hellinger transformed) varied between 0.07% and max. 50% with a median of 4.6%. The distribution of the individual clades and their OTU richness also varied greatly (Figs. S2A-H). Members of the clades Ochromonadales, Synurales, and Chrysosacca-Apoikiida were detected in almost all lakes (207, 188, and 188 lakes, respectively) whereas the number of OTUs affiliated with these clades varied considerably between the lakes (Figs. S2B, F, H). Members of the clades Segregatales (4 lakes) and Hibberdiales (72 lakes) were only found in few lakes and with low numbers of OTUs (Figs. S2D, G).

Fig. 2
figure 2

Map of the sampling localities. The size of the dots indicates the relative contribution to the OTU diversity within the lake. The colors within the pie-charts indicate the different nutritional modes

Tolerance ranges differ only slightly between members of different clades and between the different trophic modes

Based on the occurrence of individual OTUs, we evaluated whether the tolerance ranges as estimated from physicochemical parameters of the respective habitats differ systematically between clades or trophy. Only OTUs that occurred in at least two lakes were considered. The tolerance range was defined as the difference between the maximum and minimum of the respective value for the lakes in which the OTU occurred (labeled with Δ in Fig. 3, Table S2). We found that OTUs affiliated with Chrysophyceae had a significantly (Wilcoxon rank test p < 0.001) higher altitudinal range (median range 819 m) as compared to OTUs affiliated with other eukaryotes (median range 662 m; Fig. 3). The results for temperature were similar: the median temperature tolerance range was 6.28 °C for Chrysophyceae and 5.77 °C for all other eukaryotes (Wilcoxon rank test p < 0.001). No significant differences were found for conductivity (median range Chrysophyceae 317 µS, eukaryotes 296 µS; Wilcoxon rank test p = 0.431) or pH (median range Chrysophyceae 0.93, eukaryotes 0.91; Wilcoxon rank test p = 0.790).

Fig. 3
figure 3

Tolerance range of the different groups towards different environmental factors. On the left side, the Chrysophyceae were compared with the remaining eukaryotes; middle: Clades within the Chrysophyceae, right: ordered according to nutritional mode. Ranges were defined based on the difference between the maximum and minimum occurrences for the different environmental factors. Statistical significance can be found in Table S2

We found hardly any significant difference between the clades within the Chrysophyceae (see Fig. 3). The few exceptions shall be briefly mentioned: Concerning the altitude tolerance range, we found a significantly lower range for OTUs affiliated with Ochromonadales (median range = 697 m) as compared to OTUs affiliated with the Synurales (median range 1128 m; Wilcoxon rank test p = 0.03). Differences between all other clades/taxa were not significant (Wilcoxon rank test p > 0.05).

Regarding conductivity the tolerance range of OTUs affiliated with Paraphysomonadida (median = 315 µS) was significantly higher than the range of the Chrysosacca-Apoikiida clade (median = 207 µS; Wilcoxan rank test p = 0.006). The tolerance range of the Synurales (median = 387 µS) was significantly higher than the range of the Hibberdiales (median = 292 µS; Wilcoxan rank test p = 0.02). ((details Table S3).

Regarding the pH-range, differences between clades were slightly more pronounced. The pH range of OTUs affiliated with Paraphysomonadida (median range = 1.3) was significantly higher than for those affiliated with Chrysosacca-Apoikiida (median range = 0.91; Wilcoxon rank test p = 0.006), Chromulinales (median range = 0.51; Wilcoxon rank test p = 0.007), Ochromonadales (median range = 0.84; Wilcoxon rank test p < 0.001). Also OTUs affiliated with Ochromonadales (median range = 0.84) differed significantly from the Synurales (median range = 1.19; Wilcoxon rank test p = 0.01) concerning their pH range (details Table S3). No significant differences were found for the temperature range (details Table S3).

Even though the medians of the individual clades differ only slightly from each other, the clades contain specific OTUs that have large tolerance ranges (Table S2). The five OTUs with the largest ranges for the respective environmental factors can be found in Table 3.

Table 3 Selected OTUs with the highest tolerance ranges (defined as the difference between the maximum and minimum of the respective value for the lakes in which the OTU occurred). The taxonomic assignment was performed using the “best-hit” option based on the highest likelihood weight

For the different nutritional modes, no significant differences could be found regarding the altitudinal range (Fig. 3, Table S3). Concerning the conductivity, mixotrophic OTUs showed a significantly lower range (median range = 307 µS) thanphototrophic OTUs (median range = 387 µS) (Wilcoxon rank test p = 0.007). Regarding their pH range, mixotroph OTUs (median range = 0.9) differed significantly (Wilcoxon rank test p = 0.02) from phototrophic OTUs (median range = 1.19), even though only slightly. All other combinations were not significant (the group “unknown” was not analyzed). Mixotrophic OTUs (median range = 6.63 °C) showed a significantly larger range than heterotrophic OTUs (median range = 5.57 °C; Wilcoxon rank test p = 0.01).

Relative contribution of trophic modes to Chrysophyceae diversity

In most lakes, the richness of Chrysophyceae was dominated by mixotrophic OTUs (Fig. 2, Table S1), i.e., they accounted for more than 50% (median) of the Chrysophyceae OTUs. The proportion of heterotrophs was quite different accounting for 0 to 72% in most lakes (median = 14%) while phototrophs accounted in most lakes only for up to 10% (median). The different shares of heterotrophs, mixotrophs, and phototrophs to Chrysophyceae diversity were largely independent of the measured abiotic environmental factors (Fig. 4). Similar results were obtained by comparing the relative reads within the Chrysophyceae (results not shown).

Fig. 4
figure 4

Ternary plots showing the percentage community composition of Chrysophyceae communities in each lake (no. of OTUs per nutritional mode). Each corner of the triangle represents a nutritional mode (phototroph, mixotroph, heterotroph). The points are colored according to the values of the different environmental factors of the lakes (pH, temperature, conductivity, altitude). For results expressed as a percentage, the absolute value corresponds to 100%

Discussion

Chrysophyceae have a considerable share in the protist community composition of European lakes and belong often to the dominant representatives within the community. The diverse nutritional strategies and the polyphyletic origin of pure heterotrophy within this group hindered a comprehensive understanding of their ecological and evolutionary significance. Morphological studies mostly followed either the zoological or botanical approach and consequently focused on either bacterivorous taxa or pigmented taxa (Grossmann et al., 2016a). A comprehensive overview of overall Chrysophyceae diversity is thus hard to judge from these studies. Molecular environmental studies cover the whole diversity but are mostly restricted to class-level analyses and furthermore lack information on the nutritional mode of distinct lineages and phylogenetic clades (Grossmann et al., 2016b). By using a reference phylogenetic tree and data from 218 lakes, we link OTU diversity with phylogenetic affiliations and with that provide information on the relative importance of lineages affiliated with different nutrition modes and ecologies.

In our study, about 7% of the total OTU richness as well as total read abundances (with a median of 4.6% read abundance per lake) were associated with Chrysophyceae. This was somewhat lower than expected, as many studies report a relatively high proportion of OTU richness of up to 20% and even more (10% or even more of the read abundance) (Bock et al., 2014; Grossmann et al., 2016b; Khomich et al., 2017; Ortiz-Álvarez et al., 2018; Triadó-Margarit & Casamayor, 2012). Since a high richness of Chrysophyceae is mainly reported from Nordic areas and mountain lakes (Bock et al., 2018; Khomich et al., 2017; Ortiz-Álvarez et al., 2018), the somewhat lower total richness may result from the numerous central European lowland lakes included in our study. In fact, we found a higher richness of Chrysophyceae in samples from the Nordic countries (Norway, Sweden), as well as from the Alps, the Pyrenees, and other mountain regions (Fig. 2). Overall, our study revealed 2370 OTUs affiliated with Chrysophyceae, which is a substantial diversity (in comparison to the roughly 1200 morphologically described species worldwide (Kristiansen & Škaloud, 2017)).

Molecular diversity and differential pattern between distinct orders of Chrysophyceae

The Chrysophyceae comprise nine different orders with numerous phylogenetic lineages (Kristiansen & Škaloud, 2017), which were all well supported as monophyletic lineages in our analysis (Fig. 1). Exceptions are Paraphysomonadia where Clathromonas and Paraphysomonas do not cluster in a well-supported clade as it is known from several other studies (Andersen et al., 2017; Kristiansen & Škaloud, 2017). The addition of Neotessella sequences in the Synurales reduced the support for this order as well (Fig. 1). In this context, the sampling of reference sequences for the tree has a major impact not only on the support of individual phylogenetic lineages, but also on the basal arrangement of orders.

Using our phylogenetic reference tree, we could assign the majority of the OTUs to the already established clades (2304 OTUs out of 2370), whereby the selection of the reference strains has a great influence on the placement of the OTUs. Especially for OTUs without a direct close relative, phylogenetically distant sequences also seem to influence the placement.

The (phototrophic) Synurales and the (colorless) Paraphysomonadida comprise silica-scaled taxa. Based on the morphological richness of these scales, these orders belong to the morphologically best-investigated groups with numerous species described (Scoble & Cavalier-Smith, 2014; Škaloud et al., 2012, 2020). The Ochromonadales do not comprise scaled taxa but comprise a high variety of morphological forms including lorica-bearing and naked taxa as well as single-celled and colonial taxa (Kristiansen & Škaloud, 2017). The morphological diversity and consequently (morpho-)species diversity within these three orders is therefore considerably higher than in the other orders (Kristiansen & Škaloud, 2017). Consequently, we expected a high share of these orders in molecular diversity surveys. However, our results only partly confirmed this assumption.

Most OTUs (1057) were placed within the clades Ochromonadales while Synurales and Paraphysomonadida were only represented by 218 and 142 OTUs, respectively (Table 1). Surprisingly the Chrysosacca-Apoikiida clade, comprising the orders Chrysosaccales and Apoikiida and related lineages (e.g., Chrysastrella) was represented by 576 OTUs. This high diversity within this morphologically less diverse order was completely unexpected. Within the morphologically well-investigated Synurales and Paraphysomonadida, only approx. 30% and 16% of the OTUs in our study were considered novel. For the morphologically diverse but less intensively investigated Ochromonadales, the fraction of novel OTUs was higher, i.e., 45%. For all other orders/clades, the share of novel OTUs was considerably higher and ranged from 63% (Chrysosacca-Apoikiida) to 100% (Segregatales). This study therefore highlights the need for enhancing the research effort in particular for these morphologically less conspicuous taxa. The majority of environmental sequences within these clades cannot be affiliated with species and/or a morphological cellular identity and the ecology and distribution pattern of these taxa therefore remain vague. Since some of the most abundant freshwater eukaryotes on planet Earth are affiliated with these orders (see discussion) and these taxa play, therefore, a presumably dominant role in planktonic food webs we consider a deeper understanding of their ecology and evolution as crucial for further progress in understanding microbial eukaryotic ecology and diversity.

Diversity within the order Ochromonadales

The Ochromonadales include diverse morphological forms and no clear synapomorphic character. Apart from inconspicuous heterotrophic or mixotrophic colorless biflagellate forms, colonial plastid-bearing cells or lorica bearing forms (single or colonial) also evolved in different phylogenetic lineages within this order. Heterotrophy has manifested itself in different lineages and mixotrophic and heterotrophic taxa are closely related (Andersen et al., 2017; Bock et al., 2017; Kristiansen & Škaloud, 2017). Based on known high morphological and functional diversity, a high diversity of OTUs within this clade was to be expected and was confirmed by our investigations (Fig. 1). However, the proportion of new OTUs to the total number of OTUs within the Ochromondales is somewhat lower than expected but still considerably high (45%). This probably reflects the high research interest in recent times that lead to a large number of phylogenetic studies (Andersen et al., 2017; Bock et al., 2017; Pusztai & Škaloud, 2019).

Diversity within the Chrysosacca-Apoikiida clade

As mentioned above, the high OTU richness within the clade Chrysosacca-Apoikiida (including Chrysosaccales, Apoikiida, Chrysastrella) was a surprise. The representatives of this clade show a reduction of the plastid (Kristiansen & Škaloud, 2017) and comprise substrate-bound taxa with lorica, e.g., Lagynion, coccoid forms like Chrysophaera, as well as naked single-celled taxa similar to Ochromonas, Chromulina, or Spumella. Given this morphological diversity and no distinctive characters, the fraction of novel OTUs was within the expected range, i.e., higher than in Ochromonadales but lower compared to the morphologically less diverse orders. But even though a considerable number of species are morphologically described within the Chrysosaccales, their molecular identity is largely unknown (and vice versa). For instance, about 20 species are described within the genus Lagynion (Guiry & Guiry, 2021), but there is only limited molecular data published or isolates available in public culture collections, which would allow for linking morphospecies to sequence data. Further, a high number of OTUs is affiliated with an environmental clone (NCBI accession number JF730796). Given this high and largely novel OTU richness, the genetic diversity (genetic distance) within this clade is surprisingly low (Table S4). This may have various causes: Besides a presumably high speciation rate and a high intraspecific genetic variation in this group, methodological approaches may result in a certain overestimation of diversity (e.g., PCR artifacts, sequencing errors, cluster algorithm) (Forster et al., 2019; Kunin et al., 2010; Patin et al., 2013). However, methodical artifacts are expected to be of similar importance in other clades as well and, since evidence for such bias is low, we consider methodical reasons as unlikely to explain this pattern. However, an extended molecular taxonomic analysis is needed in future studies and will likely affiliate a higher genetic diversity within this clade. Given the high numbers of OTUs, the Chrysosacca-Apoikiida are certainly an interesting group waiting for intensive treatment.

Diversity within the silica-scaled orders Synurales

While the Ochromonadales and Chrysosaccales accounted for a surprisingly high fraction of OTUs in our study, the number of OTUs was surprisingly low for the Synurales. Synurales include the three photo-autotrophic genera Mallomonas, Synura, and Neotessella (Kristiansen & Škaloud, 2017; Pusztai et al., 2016). Especially Mallomonas and Synura are relatively well studied and have been the focus of detailed morphological and molecular analyses. However, species differentiations within Synurales are mostly based on gene regions with higher variation than SSU, like ITS or rbcL (Gusev et al., 2018; Jo et al., 2016). Therefore, it is possible and even likely that some OTUs encompass several species and much of the diversity within the clade remained undiscovered by our approach. We do, however, not expect a systematic bias of species resolution between different clades, i.e., we consider the relative contribution of distinct clades to overall Chrysophyceae richness as robust.

Tolerance range of Chrysophyceae

A previous study suggested that ecophysiological tolerances and the associated distribution of organisms are strongly related to different nutritional strategies rather than phylogenetic and systematic classification (Bock et al., 2019). Therefore, Chrysophyceae are the ideal test organisms as they comprise flagellates with contrasting survival strategies, which typically co-exist in different habitats (Fig. 2). The calculated tolerance ranges of the individual clades depend strongly on the placed OTUs. Since the placement depends in many cases on the reference sequences, as already explained above, the tolerence range of the clades is subject to variation due to reference sequence sampling.

Differences and commonalities between nutritional types

We expected significant differences in tolerance ranges between different nutritional modes: First, due to the lack of carbon concentrating mechanisms in Chrysophyceae (Bhatti & Colman, 2005; Raven et al., 2005; Saxby-Rouen et al., 1998), the phototrophic taxa are particularly expected to rely on sufficiently high ambient pCO2 levels and, thus, to be less competitive in alkaline environments. Heterotrophic taxa in contrast do not have such limitations. In accordance with these expectations and in contrast to the heterotrophic taxa, the phototrophic taxa are mostly reported from low pH environments and can hardly grow above pH 8.5 with many taxa having even lower maximal pH tolerances (Eloranta, 1995; Sandgren, 1988b; Siver, 1995). A higher potential to tolerate a wide range of physicochemical conditions was shown for several colorless Chrysophyceae, in particular within Ochromonadales and Paraphysomonadida (Bock et al., 2014; Boenigk et al., 2006; Nolte et al., 2010; Pfandl et al., 2009; Scoble & Cavalier-Smith, 2014). Field data indicate that in particular mixotrophic Chrysophyceae seem to prefer more narrow temperature ranges (e.g., Dinobryon (Heinze et al., 2013; Jost, 2010)) as compared to heterotrophic taxa. Third, pigmented Chrysophyceae are used as bioindicator species in fossil as well as extant phytoplankton assemblages (e.g., (Järvinen et al., 2013; Smol, 2010)), which also supports the assumption of narrow ecological niches of these taxa.

In contrast to these expectations, our analyses cannot confirm generally deviating tolerance ranges for the different nutrition modes (Fig. 3). In each of the investigated groups, some representatives occurred in a wide range of environmental factors while others were restricted to a narrow range of the analyzed environmental factors (Fig. 3). Chrysophyceae species richness seems rather independent of lake chemistry (Fig. 4). Potentially factors such as biotic interactions strongly modulate the realized niche and override potential signatures of the fundamental niche width in the distribution pattern (Bock et al., 2020). In particular, food availability and quality (Bock et al., 2019; Grujčić et al., 2015; Šimek et al., 2013), as well as predation influence abundance and distribution pattern (Haraguchi et al., 2018; Samuelsson & Andersson, 2003; Sanders & Porter, 1990; Segovia et al., 2014). The influence of these biotic interactions has probably been underestimated in the past (Bock et al., 2020) and may explain a similar distribution pattern despite potentially different fundamental niche widths. The realized niche as estimated from distribution data can differ considerably from the fundamental niche as inferred from laboratory studies. Nevertheless, the lack of differential tolerance ranges was unexpected and surprising.

Tolerance range of individual OTUs

In contrast to the lack of differentiation on the level of orders and of nutritional types, we found OTUs associated with an exceptionally broad tolerance range width in all dominant orders and all nutrition types. These OTUs often were also rather abundant and belong to the most abundant and widespread eukaryotes in freshwater ecosystems. For instance, within Ochromonadales, the OTU N15 (assigned to the heterotrophic genus Poteriospumella) showed a considerable ecophysiological tolerance. Not surprisingly this OTU had a wide distribution (occurrence in 160 sampled lake out of 218). A species of this genus, Poteriospumella lacustris, has been reported (and isolated) from different geographical regions (e.g., China, New Zealand, Europe) and high relative abundances have been reported from distinct lakes in particular during the warmer season (Nolte et al., 2010). The tolerance range inferred from our study corresponds to high tolerances for abiotic factors such as temperature, pH, and salinity (Boenigk, 2008; Boenigk et al., 2007).

Regarding the Chrysosacca-Apoikiida clade, several OTUs could not be placed accurately, except for some OTUs closely related to Chrysosaccus and Lagynion (see Table S2). Most of the sequences affiliated with this clade have a high similarity to an environmental clone (published under the acc. number JF730796 in NCBI). This clone was originally reported from the arctic and represents a phylogenetic lineage with slight sequence variation and without close relatives (Charvet et al., 2012). OTUs assigned to this clone show a wide tolerance range and widespread occurrence (Tables 3, S2), similar to Poteriospumella OTU N15 (discussed above). In particular, OTU N21 occurred in 149 sampled lakes (out of 218) and showed large tolerance ranges in terms of altitude, pH, and temperature. However, in contrast to P. lacustris, ecophysiological laboratory data are not available for N21 and also the morphological identity cannot be judged. This again highlights the need for ecological research on these inconspicuous but seemingly dominant and widespread taxa.

Also within order Synurales, we detected OTUs with an exceptionally high tolerance range. In particular, OTU N117 assigned to Mallomonas showed a wide distribution with respect to pH, conductivity, and temperature. With more than 200 species, the genus Mallomonas is one of the most species-rich genera of the scaled Chrysophyceae (Guiry & Guiry, 2021; Kristiansen & Billard 2001; Siver, 1991). OTU N117 could not be assigned to species level in our analyses. Similar to the two formerly discussed OTUs (N15 and N21) the OTU N117 proves a wide distribution of a distinct phylogenetic lineage within the Synurales. Again the wide distribution (151 of 218 lakes sampled) was related to a high tolerance range.

Novel and unassigned OTUs

Only 66 OTUs could not be assigned to any of the reference orders (Table 1). Evidence for potentially new main clades (e.g., orders) within the Chrysophyceae without cultured representatives has been reported in several studies (Del Campo & Massana, 2011; Ortiz-Álvarez et al., 2018; Shi et al., 2009). However, detailed information on where the potential new lineages are phylogenetically placed is only available from studies using cloning or classical isolation and cultivation techniques (e.g., (Grossmann et al., 2016; Škaloud et al., 2020). In a study on Arctic lakes, clone libraries were used to demonstrate the high diversity of new phylogenetic lineages, potentially representing new clades/orders (Charvet et al., 2012). Similar results were found in high mountain lakes in Spain (Triadó-Margarit & Casamayor, 2012) or Lake George (Richards et al., 2005). However, in high-throughput sequencing studies, the diversity within large taxonomic groups is usually not subjected to detailed phylogenetic or taxonomic analyses (e.g., (Filker et al., 2015; Shi et al., 2009). However, the fraction of new formerly unknown lineages is comparably low in our study (approx. 3%). Even though some of these OTUs may belong to potentially new clades, these clades are presumably not very diverse and ecologically of minor importance. However, further studies are necessary to illuminate this unresolved diversity within the Chrysophyceae.

Conclusions

Our study confirmed that the Chrysophyceae are one of the most common groups in freshwater lakes but the current perception of Chrysophyceae diversity is strongly biased towards the morphologically diverse taxa, in particular towards Synurales and Paraphysomonadida followed by Ochromonadales. In particular, and contrary to our original hypotheses based on literature data, the Chrysosacca-Apoikiida clade comprise numerous OTUs, which can currently not be assigned to known species. Regarding the phylogeny, by far the most Chrysophyceae OTUs were assigned to Ochromonadales and with respect to the nutritional mode to mixotrophs (covering the gradient from predominantly heterotrophic to predominantly phototrophic). Ecophysiological tolerance range differed neither systematically with phylogeny nor with nutritional mode. However, some taxa with an exceptionally high tolerance range are among the most abundant and widespread eukaryotes in freshwater ecosystems. The inability to link (some of) these important taxa to their cellular identity highlights the need for enhanced research on the ecology and diversity of these morphologically inconspicuous nanoflagellates.