The Urotricha species were collected in three oligo-mesotrophic lakes (Lake Mondsee, Lake Piburg, and Lake Zurich) located in Austria and Switzerland (Table 1). Details on these lakes can be found elsewhere (e.g., Bossard et al., 2001; Dokulil et al., 2006; Sonntag et al., 2011; Ficker et al., 2017; Yankova et al., 2017; Niedrist et al., 2018). Sampling was conducted by boat at the deepest point of each lake (68 m in Lake Mondsee, 25 m in Lake Piburg, and 136 m in Lake Zurich) and the water samples were collected along depth gradients from the surface to above lake bottom by means of a 5-L Schindler-Patalas sampler (Uwitec, Mondsee, Austria) and a plankton net (10-μm net gauze; Uwitec, Mondsee, Austria). Samples were transported in 1-L plastic bottles at ambient lake water temperatures to the laboratory and processed right after sampling.

Mixed cultures were set up from raw water samples enriched with algal food, i.e., SAG 26.80 Cryptomonas sp. (Culture Collection of Algae at the University of Göttingen, Germany¹). For cloning, we picked individual urotrichs with finely drawn glass pipettes under a stereo microscope (Olympus SZ61, Olympus, Vienna, Austria) at up to 45 × magnification. A subsequent purification procedure following Pringsheim (1946) appeared to be crucial for both the molecular analyses and the cultivation of clonal strains. Single urotrich specimens were cultivated in a 5:1 mixture of modified Woods Hole MBL medium (WC; Guillard and Lorenzen, 1972) and Volvic^® mineral water, a previously tested cultivation medium (see below). Food algae were kept in modified Bold’s Basal Medium (3NBBM; medium 26a in Schlösser, 1997). All cultures were grown at temperatures of 15–21°C under a light:dark cycle of 12:12 h with a photon flux rate of 50 μmol m^–2s^–1. Successfully grown clones were subsequently transferred from the small volumes of 96-well plates into 24-, 12-, and 6-well plates and finally into 50-mL cultivation flasks with a ventilated cap (Biomedica).

Specifically for the strain CIL-2017/25, tentatively identified as U. castalia, we searched for optimal cultivation media and food items. We inoculated five individuals of U. castalia each in a well of a 12-well plate (in triplicates) containing one of the following media: sterile-filtered original lake water (0.2-μm filter; Minisart, Sartorius), Volvic^® mineral water, modified Bourrelly medium (Hindák, 1970), WC medium only, 1:1, 1:5, 5:1 v/v mixtures of WC:Volvic^®, and modified blue-green medium (BG11; Rippka and Herdman, 1992). For testing food preferences, diverse algae were either successfully isolated and cultivated from Lake Mondsee (indicated by “MS” with a strain number) or obtained from a public culture collection (Supplementary Figure 1). In the feeding experiment, we tested the strains MS-2017/1 Coelastrum sp., MS-2017/2 Choricystis sp., MS-2017/7 Acutodesmus obliquus, MS-2017/8 Cosmarium sp., and SAG 26.80 Cryptomonas sp. in 12-well culture plates with WC:Volvic^® medium 5:1 v/v (triplicates). The cultures of the ciliate and the respective algae were subsequently monitored by direct live counts in the plates under a stereomicroscope (Olympus SZ61) at up to 45 × magnification.

The morphology of the Urotricha strains was studied by applying standard methods for revealing the species-specific cell features and producing the drawings (Foissner, 1991, 2014). Living ciliates and algae (for the feeding experiments) were examined under differential interference contrast optics at 40–1,000 × magnification by means of an Olympus BX51 microscope (Olympus, Vienna, Austria). Two digital image analysis systems were employed for documentation and measurements (Jenoptik ProgRes SpeedXT core 5 2.9.0.1., Jenoptik PROGRES Gryphax^® Arktur). Additionally, measurements were done by a calibrated eyepiece micrometer. Regarding protargol staining, we applied Foissner’s (1991, 2014) protocol and Skibbe’s (1994) quantitative method (QPS). The specimens were identified by means of the keys published by Foissner and Pfister (1997) and Foissner et al. (1999).

For scanning electron microscopy, the specimens were fixed for 30 min in a modified Párducz’s solution (Párducz, 1967; 4 parts 2% osmium tetroxide and 1 part saturated aqueous mercuric-chloride). After three washing steps in 0.1 M sodium cacodylate buffer, the material was transferred into preparation chambers equipped with Millipore nets of 10-μm mesh size, following the method of Foissner (1991, 2014). After dehydration by an ascending ethanol series (30-50-70-90-96-3 × 100%, 5 min each), critical point drying was performed in a BAL-TEC CPD030. The material was sputter-coated with gold in an AGAR Sputter coater and viewed in a Philips ESEM XL-30.

To receive the DNA of the studied ciliates, we used the isolation method described by Pringsheim (1946). Approximately ten ciliate cells per strain were washed, one specimen each was placed into a well containing fresh medium and starved for a strain-specific period of time (Table 2). The whole genome amplification of single cells was performed, using the REPLI-g^® Single Cell method (Qiagen, Catalog no. 150343, Hilden, Germany). The used approach followed the protocol provided by the manufacturer and is briefly described as follows: a single cell was added to 4 μL of phosphate-buffered saline solution (PBS, Qiagen) in precooled 0.2-mL PCR tubes (StarLab). After adding 3 μL of D2 buffer (Qiagen) and flicking, the tubes were incubated for 10 min at 65°C in a Mastercycler (Thermal Mastercycler ^®nexus gradient, Eppendorf). Afterward, 3 μL of the Stop Solution (Qiagen) and 40 μL of the Mastermix (9 μL H₂O, 29 μL REPLI-g^® Reaction Buffer, 2 μL REPLI-g^® DNA Polymerase) were added into the PCR-tube. Finally in the Mastercycler, the following PCR program was applied: 8 h at 30°C and 3 min at 65°C. An aliquot of the 50-μL-mix (1 μL) was taken and diluted 1:10 with H₂O. Subsequently, 1 μL of the diluted mix was again diluted to obtain a 1:100 dilution. The 1:100 diluted DNA was used to amplify the targeted short DNA regions comprising the V4 and V9 regions of the SSU rDNA, ITS-1, and ITS-2 fragments in two independent PCR-reactions for obtaining the near complete SSU rDNA and ITS sequences with an overlap of approximately 180 bp of both PCR-products. The first part of the SSU rDNA (PCR1) was amplified, using the forward eukaryote-specific primer EAF3 (5′TCGACAATCTGGTTGATCCTGCCAG3′; Marin et al., 2003) and the reverse ciliate-specific primer CilR, which was actually a mix of three primers (CilR1 - 5′TCTG ATCGTCTTTGATCCCTTA, CilR2 - 5′TCTRATCGTCTTTG ATCCCCTA3′, and CilR3 - 5′TCTGATTGTCTTTGATCCCC TA3′; Lara et al., 2011). For the second part of the SSU and the ITS (PCR2), we used the ciliate-specific primer CilF (5′TGGTAGTGTATTGGACWACCA3′; Lara et al., 2007) and the eukaryote-specific primer ITS055R (5′CTCCTTGGT CCGTGTTTCAAGACGGG3′; Marin et al., 2003). Both PCR reactions were conducted in a volume of 50 μL containing 25 μL Taq-PCR-Mastermix (Qiagen, Catalog no. 201445, Hilden, Germany), 24 μL H₂O, and 0.5 μL of 10 μM of each primer, and were performed in duplicates. The following program was applied for both PCR reactions in an Eppendorf Mastercycler nexus gradient (Eppendorf, Wesseling-Berzdorf, Germany): 5 min initial denaturation at 96°C followed by 30 cycles each comprising 1 min denaturation at 96°C, 2 min primer annealing at 55°C, and 3 min elongation at 68°C, and lastly 10 min final elongation at 68°C. The resulting PCR products were purified, using the Qiagen PCR Purification Kit (Qiagen, Catalog no. 28106, Hilden, Germany) according to the manufacturer’s protocol, and sequenced (Sanger approach), using the primers N82F and 536R for PCR1, and E528F, N920F, N920R, BR, GF, and GR for PCR2 (for primer sequences see Darienko et al., 2019). The resulting sequences of these eight reactions were assembled and corrected into one contig for each Urotricha strain.

The SSU rDNA sequences were compared, using the BLASTn search approach (Altschul et al., 1990). This approach confirmed the affiliation of the new sequences with the Prostomatea. The dataset of this ciliate class was originally compiled for the genus Coleps by Pröschold et al. (2021). The Urotricha sequences were manually aligned to the secondary structure presented for CIL-2017/25 (Supplementary Figure 2) and were integrated into two datasets: (1) an SSU dataset of 35 prostomatean taxa (length 1,755 bases) to demonstrate the monophyly of the genus Urotricha and its phylogenetic position, and (2) an SSU/ITS dataset of our 18 studied Urotricha strains (length 2,343 bases) to get a better phylogenetic resolution within this genus. The automated model selection tool implemented in PAUP version 4.0a167 was employed (Swofford, 2002) to find the best fitting evolutionary model for the datasets. The best models (model settings provided in the figure legends), using the Akaike Information Criterion, were chosen (Akaike, 1974). The following phylogenetic analyses were performed with PAUP: Maximum Likelihood, using the best model estimated by PAUP, Neighbor-Joining, and Maximum Parsimony. To find molecular signatures for supporting the assigned species, the secondary structures of the hypervariable V4 and V9 regions of the SSU rDNA and the ITS-2 were reconstructed, using the program mfold, which used the thermodynamic model for folding (Zuker, 2003). The visualization of the structures was conducted by PseudoViewer3 (Byun and Han, 2006). The compensatory base changes (CBCs) were detected, using the program CBCAnalyzer version 1.1 (Wolf et al., 2005).

To investigate the global distribution of the urotrich strains, an in-house reference database was assembled consisting of V4 and V9 datasets from freshwater lakes in Austria, Chile, the Czech Republic, Ethiopia, France, Germany, Hungary, Iceland, Italy, Nepal, Norway, Poland, Romania, Slovakia, Spain, Sweden, and Switzerland. Further information on the datasets can be found in the Supplementary Table 2. The retrieved V4 and V9 sequences were then used as an exact-match query and blasted against the reference database to screen for their presence at a specific location.

Additionally, sequence and cell count data of the Urotricha species from Lake Mondsee and Lake Zurich spanning over a 1-year period were accumulated to compare the spatial and temporal distribution within both lakes. Information about the sampling strategy, sequencing, and data analyses can be retrieved from Qu et al. (2021). Based on the QPS method of Skibbe (1994), Urotricha furcata, Urotricha agilis, Urotricha castalia, and Urotricha pseudofurcata from Lake Mondsee were morphologically identified and counted and U. agilis, U. castalia, and U. pseudofurcata from Lake Zurich. Contour plots were created with the “plotly” package in R for visualizing the occurrence of the Urotricha species in the two lakes.

The available environmental data from the study of the freshwater lakes by Boenigk et al. (2018), were used to identify potential ecological preferences of the ciliate species under study (Supplementary Table 3). Spearman’s rank coefficients were calculated to relate Urotricha abundance and environmental data. A multiple logistic regression analysis was used to estimate the effect of environmental parameters on the Urotricha abundances in those lakes for which all environmental parameters were available (n = 85). All values were log-transformed to achieve additivity of their effects and to increase the homogeneity of variances in the response variable (i.e., Urotricha abundances). A stepwise procedure was then used to build the final model based on the Akaike Information Criterion (AIC) and the obtained p-values. The final model included those environmental parameters, which influenced the Urotricha occurrences. All statistical analyses were conducted with the “stats” package in R version 4.0.5.

We successfully isolated and cultivated 18 Urotricha strains, which we assigned to Urotricha furcata Schewiakoff, 1892 and tentatively to Urotricha agilis (Stokes, 1886) Kahl, 1930, Urotricha pseudofurcata Krainer, 1995, and Urotricha castalia Muñoz et al., 1987 (Table 2 and Supplementary Table 4). In the detailed descriptions of our investigated strains, we focussed on the species-specific features. Common urotrich features are thus not repeatedly mentioned, i.e., (i) the central position of the macronucleus and the micronucleus, (ii) the eccentric position of the single contractile vacuole in the unciliated posterior cell portion, (iii) the eccentric position of the one or two caudal cilia at the rear end of the small species, (iv) the pattern of the somatic ciliature with equidistantly arranged, posteriorly shortened meridional rows leaving blank the posterior cell portion, (v) the dikinetids at the anterior ends of the somatic kineties which have a cilium associated only with the posterior basal body, and (vi) the fast jumping movement (Foissner and Pfister, 1997; Foissner et al., 1999).

Urotricha agilis (Stokes, 1886) Kahl, 1930 (Figures 1A–H, 2A,B, and Supplementary Table 4).

Remarks. The strains CIL-2017/24 and CIL-2019/10 were morphologically and genetically investigated. Despite their morphological similarity, the genetic data (dissimilarity and CBCs) indicate the presence of two cryptic species; hence, their morphometric data are kept separate.

Urotricha agilis (Stokes, 1886) Kahl, 1930 - Strain CIL-2017/24 from Lake Mondsee (Figures 1A–D, 2A,B).

Remarks. The strain was studied in vivo and after protargol staining; it is genetically highly similar to the strains CIL-2019/8 and CIL-2019/9 from Lake Piburg and strain CIL-2017/23 from Lake Mondsee.

Description. The cell is ellipsoidal to broadly ellipsoidal in vivo and in protargol-stained cells. It measures 14–23 × 9–16 μm in vivo and 10–16 × 8–14 μm in protargol preparations. The cell length:width ratio is 1.2–1.9:1 in vivo and 1.0–1.6:1 in protargol preparations. The unciliated posterior cell portion is not set off plug-like and occupies 17–37%, on average 26% of cell length, being 3–6 μm long in protargol preparations (Figures 2A,B). The globular to broadly ellipsoidal macronucleus measures 3–8 × 3–5 μm in vivo and 3–6 × 2–6 μm in protargol preparations; the micronucleus is not recognizable. Since the rod-shaped somatic extrusomes are sparse and merely about 1.5 μm long, they are hardly visible in vivo, while not recognizable in protargol preparations. The cytoplasm is colorless and includes refractive lipid droplets 1–3 μm across and often many (colorful) food vacuoles 2–7 μm, on average 5 μm across. The contractile vacuole is up to 8 μm across; its excretory pore is somewhat eccentric (Figure 2A). The somatic cilia are relatively long with 3–6 μm in vivo and 3–7 μm in protargol preparations. The somatic ciliature comprises 18–24 kineties consisting of 6–12 cilia in anteriorly unshortened rows. The length of the single caudal cilium is about half the cell length, namely, 6–12 μm in vivo and 5–10 μm in protargol preparations (Figures 1A, 2B). The 7–9 oral flaps are 1–4 μm long after protargol preparation (Figures 1C,D, 2A,B). The oral basket is 2–7 μm across distally and 2–6 μm long in vivo, while 3–5 μm across distally and 6–12 μm long in protargol preparations, in which the oral basket occupies 29–63%, on average 44% of cell width. The three adoral organelles (“brosse”) are inconspicuous, especially adoral organelles 2 and 3 (Figures 1B,D, 2A): adoral organelle 1 consists of 3-5 dikinetids and is 1.5–2.5 μm in length, adoral organelle 2 consists of two or three dikinetids and is 0.8–1.2 μm in length, and adoral organelle 3 consists of one or two dikinetids and is 0.8-1.0 μm in length. They are placed on a comparatively broad (about 2.5 μm on average) blank stripe, which separates the circumoral kinety and the somatic kineties. Occasionally, only the ciliated basal bodies of the dikinetids constituting the adoral organelles are stained.

Urotricha agilis (Stokes, 1886) Kahl, 1930 - Strain CIL-2019/10 from Lake Piburg (Figures 1E–H).

Remarks. The strain was studied in vivo and after protargol staining.

Description. The cell is ellipsoidal to broadly ellipsoidal in vivo and in protargol-stained cells. It measures 13–19 × 9–14 μm in vivo and 9–15 × 6–11 μm in protargol preparations. The cell length:width ratio is 1.1–1.9:1 in vivo and 1.0–1.8:1 in protargol preparations. The unciliated posterior cell portion is hardly set off plug-like and occupies 26–48%, on average 37% of cell length, being 3–5 μm long in protargol preparations. The globular macronucleus is 2–6 μm across in vivo and measures 2–5 × 1–5 μm in protargol preparations; the micronucleus is not recognizable. Since the rod-shaped somatic extrusomes are sparse and merely about 1.5 μm long, they are hardly visible in vivo (Figures 1E,G); in protargol preparations, they are about 1 × 0.2–0.5 μm in size. The cytoplasm is colorless and includes refractive lipid droplets and frequently many (colorful) food vacuoles of 3–6 μm, on average 4 μm across. The contractile vacuole is up to 5 μm across. The somatic cilia are relatively long with 2–6 μm in vivo and 4–7 μm in protargol preparations. The somatic ciliature comprises 14–22 kineties consisting of 5–7 cilia in anteriorly unshortened rows. The length of the single caudal cilium is about half the cell length, namely, 5–9 μm in vivo and 6–9 μm in protargol preparations (Figures 1E,F). The 8–10 oral flaps are 1.2–2.1 μm long in protargol preparations, in which the circumoral kinety is only incompletely visible (Figure 1H). In protargol preparations, the oral basket is 3–5 μm across distally, and thus comparatively wide, occupying 29–67%, on average 47% of cell width; its length was not measured because it was not visible. The three (n = 3) adoral organelles (“brosse”) are obliquely arranged in the blank stripe separating the circumoral kinety and the somatic kineties. Because they are minute, the organelles are even hardly visible in protargol preparations: adoral organelle 1 consists of four or five dikinetids and is 2–3 μm in length, adoral organelle 2 consists of two dikinetids and is about 1 μm in length (n = 5), and adoral organelle 3 consists of two dikinetids and is also about 1 μm in length (n = 3). The blank stripe is 1.8–3.0 μm broad.

Urotricha furcata Schewiakoff, 1892 - Strains CIL-2019/6 and CIL-2019/13 (Figures 2C–F, 3A–D, and Supplementary Table 4).

Remarks. Out of the six genetically identical strains, two were morphologically investigated, i.e., CIL-2019/6 was studied in protargol slides and CIL-2019/13 in vivo.

Description. The cell is ellipsoidal to broadly ellipsoidal. It measures 17–35 × 10–29 μm in vivo and 11–19 × 10–16 μm in protargol preparations. The cell length:width ratio is 1.2–1.9:1 in vivo and 0.9–1.7:1 in protargol preparations. The unciliated posterior portion is not set off plug-like and occupies 16–26%, on average 22% of cell length, being 2–4 μm long in protargol preparations (Figures 2C,D, 3A,B). The globular macronucleus measures 4–7 × 3–7 μm in vivo and 3–9 × 3–7 μm in protargol preparations; the globular micronucleus is 2–3 μm across in vivo, while it is not recognizable in stained specimens. Extrusomes are neither recognizable in vivo nor after protargol staining. The cytoplasm is colorless and includes refractive lipid droplets 1–3 μm across and often many (colorful) food vacuoles 2–8 μm, on average 5 μm across. The contractile vacuole is up to 7 μm across. The somatic cilia are relatively long with 3–7 μm in vivo and 3–6 μm in protargol preparations. The somatic ciliature comprises 18-24 kineties consisting of 10-14 kinetids in anteriorly unshortened rows. The length of the two caudal cilia is about one third to one half of cell length, namely, 6–11 μm in vivo and 4–8 μm in protargol preparations (Figure 2C). The three (n = 7) adoral organelles are hardly visible and probably arranged in a V-shaped pattern (Figure 2F): adoral organelle 1 consists of three or four dikinetids and is 1–2.5 μm in length, adoral organelle 2 consists of two or three dikinetids and is about 0.8–1.5 μm in length, and adoral organelle 3 consists of two dikinetids and is about 0.8–1.0 μm in length. The 7-12 oral flaps are 1–3 μm long in protargol preparations. In protargol preparations, the oral basket is 3–6 μm across distally and 4–10 μm long; it occupies 20–40%, on average 30% of cell width (Figures 2C,E).

Urotricha pseudofurcata Krainer, 1995 - Strain CIL-2019/3 (Figures 4A–D and Supplementary Table 4)

Remarks. Out of the two genetically identical strains, CIL-2019/3 and CIL-2019/4, the former was studied morphologically in vivo and after protargol staining.

Description. The cell is ellipsoidal to broadly ellipsoidal. It measures 16–24 × 10–19 μm in vivo and 9–20 × 6–14 μm in protargol preparations. The cell length:width ratio is 1.2-1.8:1 in vivo and in protargol preparations. The unciliated posterior cell portion occupies 15–55%, on average 29% of cell length, being 1–9 μm long in protargol preparations. The globular macronucleus is 3–7 μm across in vivo and measures 4–7 × 4–6 μm in protargol preparations; the globular micronucleus is 1–2 μm across in vivo, whereas it cannot be distinguished from stained food vacuoles in protargol preparations. The minute rod-shaped extrusomes are invisible in vivo, but easily recognizable in protargol preparations, where they are potentially deformed and 0.5–1.0 μm long (Figures 4B,C), and as minute stubs in scanning electron micrographs (Figure 4D). The cytoplasm is colorless and includes refractive lipid droplets 1–3 μm across and often many (colorful) food vacuoles 2–10 μm, on average 5 μm across. The contractile vacuole is up to 6 μm across (Figure 4A). The somatic cilia are relatively long with 3–5 μm in vivo and 3–6 μm in protargol preparations. The length of the two caudal cilia is about one third to one half of cell length, namely, 5–9 μm in vivo and in protargol preparations (Figures 4A–D). The somatic ciliature comprises 14–24 kineties consisting of 8–14 basal bodies in anteriorly unshortened rows. The probably three adoral organelles (n = 1) are hardly visible, especially adoral organelles 2 and 3: adoral organelle 1 consists of 3–5 dikinetids and is 1.5–2.5 μm in length, adoral organelle 2 consists of two or three dikinetids and is about 1–1.5 μm in length (n = 7), and adoral organelle 3 probably consists of three dikinetids and is about 1 μm in length (n = 1). The 8–12 oral flaps are 1–3 μm long in protargol preparations (Figures 4B,C). The oral basket is 3–5 μm across distally and 4–12 μm long in protargol preparations, occupying 28–75%, on average 45% of cell width (Figure 4C).

Urotricha castalia Muñoz et al., 1987 (Figures 2G,H, 5A–L and Supplementary Table 4).

Remarks. Out of the five genetically analyzed strains, three were morphologically investigated. The genetic data indicate a conspecificity of the strains CIL-2017/25, CIL-2017/26, and CIL-2017/27 from Lake Mondsee, whereas the strains CIL-2019/1 and CIL-2019/2 from Lake Zurich are morphologically similar but genetically distinct from those of Lake Mondsee. Since the two genetic clusters are statistically well-supported and corroborated by CBCs, they probably represent cryptic species whose morphometric data are thus kept separate.

Urotricha castalia Muñoz et al., 1987 - Strains CIL-2017/25 and CIL-2017/27 from Lake Mondsee (Figures 5I–L).

Remarks. Of the three genetically identical strains CIL-2017/25, CIL-2017/26, and CIL-2017/27, only the former and latter were morphometrically analyzed in vivo and in protargol preparations; strain CIL-2017/25 was also stained following the dry silver nitrate method.

Description. The cell is broadly ellipsoidal. It measures 26–59 × 25–44 μm in vivo and 17–29 × 13–23 μm in protargol preparations. The cell length:width ratio is 0.8-1.8:1 in vivo and 1.1-1.8:1 in protargol preparations. The unciliated posterior end is slightly set off plug-like and occupies 9–32%, on average 19% of cell length, measuring 2–6 μm in protargol preparations, whereas it is 7–15 μm long in living cells of strain CIL-2017/25. The globular to broadly ellipsoidal macronucleus measures 9–15 × 7–13 μm in vivo and is 5–12 μm across in protargol preparations; the globular micronucleus was only observed in strain CIL-2017/27 and has a size of 2–4 × 2–3 μm in vivo (n = 2) and of 2–3.5 × 1.5–3 in protargol preparations. Two types of rod-shaped somatic extrusomes occur and could only be studied in vivo, as they did not stain with protargol: the type I extrusomes are restricted to the ciliated anterior cell portion, where they are rather densely arranged and with a length of 1–3 μm shorter than the type II extrusomes of the unciliated posterior cell portion measuring 3–6 μm (Figure 5I). The cytoplasm is colorless and includes refractive lipid droplets and often many (colorful) food vacuoles 4–7 μm, on average 6 μm across. The contractile vacuole is up to 18 μm across; its excretory pore is encircled by the caudal cilia. The somatic cilia are 2–6 μm long in protargol preparations. The somatic ciliature comprises 32-42 kineties consisting of 13-27 cilia in anteriorly unshortened rows. The length of the 4-7 caudal cilia is about one fourth of the cell length, namely, 7–13 μm in vivo and 3–9 μm in protargol preparations. The three adoral organelles are difficult to observe and thus to study; they decrease in size from the anterior to the posterior brosse end; adoral organelle 1 consists of four dikinetids and is 1.2–2.5 μm long, adoral organelle 2 consists of three dikinetids and is 1.2–2.0 μm long, and adoral organelle 3 consists of 1–3 dikinetids and is 0.2–1.2 μm long (Figure 5K). The oral basket is 3–6 μm across distally and 7–18 μm long in protargol preparations, occupying 20–37%, on average 27 and 29% of cell width, respectively. The 15–21 oral flaps are 1.5–2.8 μm long in protargol preparations (Figures 5J,K). The silverline system is typical, i.e., the dry silver nitrate preparation reveals longitudinal rows of rectangular meshes each encompassing a somatic kinetid. Pairs of rows commence with a pentagonal mesh. The meshes are larger and polygonal in the unciliated posterior pole area with the caudal cilia inserting in minute circular meshes (Figure 5L).

Urotricha castalia Muñoz et al., 1987 - Strains CIL-2019/1 and CIL-2019/2 from Lake Zurich (Figures 2G,H, 5A–H).

Remarks. Two genetically identical strains were found, i.e., the strains CIL-2019/1 and CIL-2019/2. The former was studied after applying various techniques, i.e., live observation, protargol (QPS and Foissner’s method) and silver nitrate staining as well as scanning electron microscopy. Live specimens of strain CIL-2019/2 were documented (Figures 5A,B).

Description. The cell is broadly ellipsoidal. It measures 25–53 × 26–46 μm in vivo and 24–42 × 21–38 μm in protargol preparations. The cell length:width ratio is 0.8-1.4:1 in vivo and 1.1-1.4:1 in protargol preparations. The unciliated posterior end is slightly set off plug-like and occupies 11–25%, on average 21% of cell length, measuring 3–9 μm in protargol preparations. The globular macronucleus measures 7–15 × 7–12 μm in vivo and 7–14 × 5–13 μm after protargol impregnation; the globular micronucleus was observed in living cells and was 1–5 μm across and only once after protargol impregnation where it measured 3 × 2 μm. Two types of rod-shaped somatic extrusomes occur and could only be studied in vivo, as they do not stain with protargol: the type I extrusomes are restricted to the ciliated anterior cell portion, where they are rather densely arranged and with a length of 1–2 μm shorter than the type II extrusomes of the unciliated posterior portion measuring 3–5 μm (Figure 5A). The cytoplasm is colorless and includes refractive lipid droplets 1–3 μm across and often many (colorful) food vacuoles 3–10 μm, on average 6 μm across. The contractile vacuole is up to 17 μm across; its excretory pore is encircled by the caudal cilia as also recognizable in scanning electron micrographs (Figures 5B,H). The somatic cilia are 4–9 μm long in vivo and 2–5 μm in protargol preparations. The somatic ciliature comprises 34-46 kineties consisting of 18-30 cilia in anteriorly unshortened rows. Each of the 4-7 caudal cilia inserts in a pit surrounded by an elevated ring and is about one fifth of the cell length, namely, 4–11 μm both in vivo and in protargol preparations (Figures 2H, 5A,B,H). The three (n = 14) adoral organelles (“brosse”) are difficult to observe and thus to study; they are separated by distinct ridges and decrease in size from the anterior to the posterior brosse end: adoral organelle 1 consists of four or five dikinetids and is 2–3 μm long, adoral organelle 2 consists of three or four dikinetids and is 1–2 μm long, and adoral organelle 3 consists of two or three dikinetids and is 0.5–1.5 μm long (Figures 2G, 5C,E,G). The oral basket is in vivo 6–19 μm across distally and 6–23 μm long, occupying 20–60%, on average 40% of cell width, while it is 4–10 μm across and 8–16 μm long in protargol preparations, occupying 20–30%, on average 30% of cell width (Figure 2H). The 19-29, on average 23 oral flaps are 2–4 μm long in protargol preparations (Figures 5A,C–E,G). The silverline system is typical, i.e., the dry silver nitrate preparation reveals longitudinal rows of rectangular meshes each encompassing a somatic kinetid. The meshes are larger and irregular in the unciliated posterior pole area with the caudal cilia inserting in minute circular meshes (Figure 5F).

Of all media tested, a mixture of WC:Volvic^® 5:1 with SAG 26.80 Cryptomonas sp. as food resource was most successful for the growth of U. castalia; hence, it was also used for all other Urotricha strains (Table 2). The medium WC:Volvic^® 1:1 was also suitable, but not employed for this study. Overall, the results revealed that U. castalia was feeding only on the mobile alga SAG 26.80 Cryptomonas sp., while not on immobile food items (MS-2017/1 Coelastrum sp., MS-2017/2 Choricystis sp., MS-2017/8 Cosmarium sp., and MS-2017/7 Acutodesmus obliquus; Supplementary Figures 1, 3).

Phylogenetic analyses of SSU rDNA sequences revealed that all investigated strains of Urotricha formed a highly supported monophyletic group together with Plagiocampa sp. (KY980324; Zhang et al., 2014) and Halodinium verrucatum (LC424401; Gurdebeke et al., 2018), which is placed in a distinct clade. Hence, the identification of the latter two sequences is incorrect or the genus Urotricha is not monophyletic. The sister group of Urotricha were the Colepidae including the genus Coleps (Figure 6).

The analyses of the concatenated dataset comprising SSU rDNA and ITS sequences provided a better resolution among the Urotricha strains than the analyses of the SSU rDNA dataset alone. The genus Urotricha formed four highly supported lineages, i.e., U. agilis, U. furcata, U. pseudofurcata, and U. castalia in all bootstrap analyses (Figure 7). The strains from Lake Mondsee tentatively identified as U. castalia differed slightly in their SSU rDNA and ITS sequences from the strains isolated from Lake Zurich; both form two well-supported clades, probably representing cryptic species.

The tree topology regarding the strains tentatively identified as U. agilis was incongruent with the geography, displaying a grouping of two sequences from Lake Mondsee with two from Lake Piburg. This cluster represented again a statistically supported adelphotaxon to another sequence from Lake Piburg. Hence, at least two genetically distinct strains of U. agilis occurred in Lake Piburg (Figure 7). All strains of U. furcata and those tentatively identified as U. pseudofurcata came from Lake Zurich; they form highly statistically supported adelphotaxa; within each clade, the sequences were identical.

The secondary structures of the ITS-1 and ITS-2 sequences were analyzed for detecting compensatory base changes (CBCs) in the four species. Both ITS regions of the urotrichs studied here consisted of only three helices (ITS-1: Helix 1-3; ITS-2: Helix II-IV) as Helix 4 of the ITS-1 and Helix I of the ITS-2 are missing. Using the ITS-2/CBC approach, all species had specific loops and were supported by CBCs and hemi-CBCs (one-sided base changes; HCBCs) (Figure 8 and Table 3). Not only the phylogeny of the concatenated sequence data, but also the detection of CBCs suggested genetically distinct clades within the specimens tentatively identified as U. agilis and U. castalia. Unfortunately, no ITS sequences of Urotricha species beyond those of the present study were available in GenBank. Therefore, we analyzed the variable regions of the SSU, V4, and V9, which were used in high throughput sequencing approaches for the discovery of species distribution. The V9 cannot always discriminate the different Urotricha species, i.e., the strains identified as U. furcata and tentatively as U. pseudofurcata had identical V9 sequences (Supplementary Figure 4). However, the V4 region appeared promising for species-delimitation of the Urotricha strains tested and can be considered as a diagnostic barcode region (Figure 9).

In total, 252 freshwater lakes in 17 different countries were screened for the presence of urotrich V4 or V9 barcode markers (Figure 10 and Supplementary Tables 2, 3). The barcode markers occurred in all investigated countries, except for the Czech Republic, in which, however, only a single lake was sampled. The strains from Lake Mondsee tentatively identified as U. castalia and the strains of U. furcata/U. pseudofurcata were found in 15 different countries and in most lakes, making them globally and regionally the most widespread strains. The strain from Lake Zurich again tentatively identified as U. castalia showed a similar global distribution as the sister clade, but was less common. The strains tentatively identified as U. agilis could also be found worldwide with a high occurrence in Austria and Germany (Figure 10). In contrast, LC424401_Halodinium appeared to be restricted to Europe, being present in only seven out of 17 countries.

Using Lake Mondsee and Lake Zurich as case studies, the spatial and temporal distribution of Urotricha species was studied in greater detail. For this purpose, the cell count data obtained from morphological investigations and the HTS data were compared. Overall, the morphology-based data and HTS data seem to be more divergent for Lake Mondsee than for Lake Zurich (Supplementary Figures 5, 6). In Lake Mondsee, both datasets for U. agilis revealed a broad spatial distribution ranging from 0 to 40 m depth. Differences, however, occurred on a temporal scale, as, based on the cell count data, U. agilis was mainly present in June and in the winter months, whereas the HTS data showed a presence throughout the year, except for the winter months. A more congruent picture is obtained from the U. castalia datasets. Both indicated a broad spatial distribution with high abundances in the upper water layers (0–12 m depth) and a confined seasonal occurrence from June to November/December. Since U. furcata/U. pseudofurcata could not be differentiated based on their V9 region (Supplementary Figure 4), the cell count data of U. furcata and U. pseudofurcata were compared against the single HTS dataset. The latter indicate the presence of U. furcata/U. pseudofurcata over all depths and the year. The cell count data, however, revealed morphotype-specific differences, as U. furcata preferred upper water layers (0–12 m depth) and warmer months (May to September), while U. pseudofurcata was distributed throughout the entire water body, preferably in the winter-spring months (November to May).

In Lake Zurich, the abundance of U. agilis demonstrated a more restricted spatial distribution than the HTS data (20 m and 40 m depth, respectively, Supplementary Figure 6), however, apart from the coinciding temporal distribution in both approaches from April to November, cell counts revealed individuals occurring also between June and October, respectively. Equally congruent are the data for U. castalia, which revealed an occurrence from 0 to 20 m depth and, in low abundance, also at 120 m depth, however, with a temporal restriction to the warmer months (April to October). The greatest deviations were observed for the cell count and HTS data of U. furcata/U. pseudofurcata. Whereas U. pseudofurcata showed its presence in 0–40 m water depths and an absence over the winter months, i.e., December to March, the HTS data for U. furcata/U. pseudofurcata indicated a distribution all over the entire water column restricted to June through October.

Statistical analyses were conducted to reveal whether urotrich ciliate abundances (read numbers) were a function of environmental variables and altitude. In general, hardly any significant correlations between urotrich ciliate abundance and environmental factors could be detected. Only the strains of U. castalia from Lake Mondsee correlated, albeit weakly, negatively with K⁺ (cor = –0.22, p = 0.04), and U. furcata/U. pseudofurcata also only weakly negatively with altitude (cor = – 0.18, p = 0.01) and weakly positive with SO₄^2– (cor = 0.23, p = 0.03) (Supplementary Figure 7).

A multiple logistic regression analysis was used to model the effect of a set of environmental factors on the abundance of individual urotrich species. The fitted regression models are summarized in Table 4 and Supplementary Table 5. Significant positive correlations were only found for two out of the five tested species. The final model for U. agilis included Na²⁺, DOC, conductivity, and SO₄^2– (p = 0.005) and the final model for U. furcata/U. pseudofurcata DP, H⁺, DOC, Ca²⁺, altitude, DRSi, and TP (p < 0.001).

Urotricha agilis (Stokes, 1886) Kahl, 1930. In the original description, Stokes (1886) mentioned and illustrated a conical cell shape, a single caudal cilium of roughly cell length, kineties occupying the anterior two thirds of the about 14 μm long cell, an eccentric contractile vacuole in the posterior cell portion, and a globular and eccentrically located macronucleus near mid-body. The unique conical cell shape was used by Penard (1922) and Kahl (1930) for identifying the species, which has been redescribed from life and after silver staining by Foissner (1979); further, it had been included in two taxonomic and ecological monographs on freshwater ciliates (Foissner et al., 1994, 1999). Again, the conical shape of the 10–20 μm long ciliate was mentioned as the main distinguishing feature. According to the redescription, the species has 12-14 somatic kineties and about seven cilia/kinetids per kinety (as inferred from the illustrations). Data on the adoral organelles and the circumoral kinety are not available.

In the present study, five strains were genetically analyzed and four of them form a statistically well-supported and by CBCs corroborated clade being sister to strain CIL-2019/10. The latter strain and CIL-2017/24 were also morphometrically studied. Both are highly similar, merely differing in the average number of somatic kineties (14-22, on average 18 vs. 18-24, on average 22) and cilia in an anteriorly unshortened kinety (5-7, on average 6 vs. 6-12, on average 9) as well as in the length of the unciliated posterior cell portion occupying 26–48%, on average 37% vs. 17–37%, on average 26% of cell length. The three tiny adoral organelles were recognizable in few specimens in vivo and after protargol staining; however, the number of dikinetids per organelle slightly varied among the strains CIL-2017/24 and CIL-2019/10 (three to five vs. four or five, both on average four in adoral organelle 1, two or three vs. two, both on average two in adoral organelle 2, one or two vs. two, both on average two in adoral organelle 3) (Figures 1B,C, 2A).

Our strains match U. agilis in cell size (in vivo 14–23 μm and 13–19 μm vs. 10–20 μm; Foissner, 1979) and the blank stripe separating the somatic kineties from the circumoral kinety, not present in other congeners of this size class with one caudal cilium (Figures 1C,D). However, the strains are not conical (Figures 1A,B,E–G) and have more somatic kineties (18-24, on average 22 and 14-22, on average 18 vs. 12-14). Strain CIL-2017/24 might also differ by a higher number of kinetids in an unshortened kinety (6-12, on average 9 vs. about 7). Despite these differences, we tentatively assign our strains as two cryptic species to the Urotricha agilis-complex. A far-reaching revision requires the analyses of further marker sequences, particularly, of U. agilis collected at the type locality or in its surroundings in the United States, which is beyond the focus of this study.

Two species have been considered being synonyms of U. agilis, i.e., Urotricha gyrans Stokes, 1888 (Foissner, 1979) and Urotricha nais Muñoz et al., 1987 (Foissner et al., 1999). Foissner (1979) attributed the slightly different (ellipsoidal) shape of U. gyrans to a slight squeezing of the cell by the cover glass pressure. Additionally, Foissner et al. (1999) proposed U. nais Muñoz et al., 1987 as a junior synonym of U. agilis; yet, both species differ in the numbers of kineties (12-14 vs. 18-21 in U. nais). Since U. nais also has minute extrusomes, a distinct blank stripe between the somatic kineties and the circumoral kinety, an unciliated posterior portion that is not set off plug-like, and 18-21 somatic kineties, strain CIL-2017/24 and the genetically similar strains CIL-2017/23, CIL-2019/9, and CIL-2019/8 might be conspecific.

Urotricha furcata Schewiakoff, 1892. Three freshwater urotrichs with two caudal cilia are known: Urotricha macrostoma Foissner, 1983 (cell length in vivo 30-40 μm, three adoral membranelles, and minute extrusomes); U. pseudofurcata Krainer, 1995 (distinct extrusomes); and U. furcata Schewiakoff, 1892 (no extrusomes).

The latter species, originally described from Hawaii by Schewiakoff (1892), has authoritatively been redescribed from life and after silver staining by Foissner and O’Donoghue (1990) from Australia and included in two taxonomic and ecological monographs on freshwater ciliates (Foissner et al., 1994, 1999). Since the Australian specimens matched the original description and the later illustrations by Schewiakoff (1892, 1893), Foissner and O’Donoghue (1990) merely described the stained cell and provided morphometrics. Urotricha furcata matches our genetically identical strains in the apparent lack of extrusomes, the cell size in vivo (20–30 × 15–20 μm vs. 17–35 × 10–29 μm; Foissner et al., 1999) and after protargol staining (14–20 × 11–15 μm vs. 11–19 × 10–16 μm; Foissner and O’Donoghue, 1990), the cell shape, the number of somatic kineties (22–24 vs. 18–24) and kinetids in unshortended rows (13–21 vs. 10–14), however, not in the number of adoral organelles (two vs. three; Foissner and O’Donoghue, 1990). Despite the latter difference, the strains CIL-2019/6 and CIL-2019/13 are identified as U. furcata.

Urotricha pseudofurcata Krainer, 1995. As mentioned above, there are three urotrichs with two caudal cilia: Urotricha macrostoma Foissner, 1983 (minute extrusomes); U. furcata Schewiakoff, 1892 (no extrusomes); and U. pseudofurcata Krainer, 1995 (distinct extrusomes). Our strain deviates from U. macrostoma in the cell length (in vivo 16–24 μm vs. 30–40 μm) and the diameter of the oral opening (4 μm vs. 8 μm; Foissner, 1983) and from U. furcata in the presence of extrusomes (vs. absent; see above).

The species description of U. pseudofurcata by Krainer (1995) is based on live observations and protargol-stained material from Austria. The strain CIL-2019/3 is most similar to U. pseudofurcata, but differs in the number of somatic kineties (15–24 vs. 25–27; Krainer, 1995). Due to the different preparation methods, the length of the rod-shaped extrusomes (0.5–1.5 μm in protargol preparations vs. 3–4 μm in vivo; Figures 4B–D) is hardly comparable. Although our strain thus probably represents a distinct species, it is tentatively assigned to the U. pseudofurcata-complex until all taxonomically relevant data on our strain and the marker gene sequences for U. pseudofurcata collected at the type locality are available.

Urotricha castalia Muñoz et al., 1987 (Table 5). The original description by Muñoz et al. (1987) is based on silver carbonate-stained and probably squashed and swollen specimens collected in an artificial pond in the “Parque de Berlin,” Madrid, Spain; their size and shape are thus hardly comparable with specimens prepared with different methods. Later, Foissner and Pfister (1997) investigated supposedly conspecific specimens from an Austrian and a German population and synonymized U. rotunda Fernandéz-Leborans and Novillo, 1994 with U. castalia.

The results of our study obtained from the genetic data, particularly the CBCs, suggest two species in our material, i.e., one species that we found exclusively in Lake Zurich and the other one in Lake Mondsee (Figure 8). They are morphologically very similar, merely differing in the average number of circumoral dikinetids (19-29, on average 23 vs. 15-21, on average 17 and 18, respectively).

Our strains morphologically match the specimens from Austria (Salzburg) and Germany (Lake Constance), especially in the possession of two different types of extrusomes (Figure 5I), but deviate slightly in the number of caudal cilia (4-7, on average 5 in our strains vs. 8-10, on average 9 in the Salzburg population and 4-9, on average 7 in the Lake Constance population; Figures 5A,B,H; Foissner and Pfister, 1997). In the light of the present genetic data, the minute morphological discrepancies might indicate further cryptic species; however, DNA barcodes are not available for the specimens from Salzburg and Lake Constance. Likewise, data on marker sequences and extrusomes are missing for the Spanish type population (Muñoz et al., 1987) and the suggested synonym, U. rotunda (Fernandéz-Leborans and Novillo, 1994). These two populations have more somatic kineties than the previously mentioned ones (Table 5). This difference and another discrepancy, i.e., in the number of cilia in a somatic kinety (16-18 vs. 30-36) in the type population and in U. rotunda suggest another two species in the U. castalia-complex. Eventually, future morphological and genetic studies are required to verify that the minute deviations in the ciliature observed should not be attributed to intraspecific variability, but rather indicate different species. At the current state of knowledge, these species share 4-10 caudal cilia arranged in a circle, usually three adoral organelles, and an ellipsoidal to broadly ellipsoidal cell shape.

Ciliates form a statistically well-supported and specious mono-phylum (Lynn, 2008; Gao et al., 2016, 2017; Adl et al., 2019). Their high diversity, heterogeneous morphology, biogeographic distribution patterns, and phylogenetic relationships are put more and more into a comprehensive perspective. Ciliate (re-) descriptions and taxonomic assignments should be based on thorough studies of the species-specific diagnostic features, the relevant literature, and biogeographic aspects (Agatha et al., 2021). Furthermore, microscopic and molecular approaches need to be linked, which is still a major challenge in ciliate ecology (Warren et al., 2017).

Identification is difficult, particularly in ciliates less than 50 μm in length and in complexes of cryptic species, e.g., those of the genera Paramecium and Coleps (Sonneborn, 1975; Krenek et al., 2015; Pröschold et al., 2021). In such cases, molecular markers with an appropriate phylogenetic resolution (V4 and V9 of the SSU rDNA, ITS, D1-D2 of the LSU, mitochondrial COI) and their secondary structure might provide clues for species delimitations (Coleman, 2005; Dunthorn et al., 2012; Stoeck et al., 2014a; Shazib et al., 2016; Pitsch et al., 2019; Pröschold et al., 2021). Molecular barcode markers are needed to place a taxon into ecological guilds as closely related taxa might differ distinctly in their ecological functions (Foissner et al., 2003). This becomes crucial when the ecological processes are inferred from HTS samples (Pawlowski et al., 2014; Stoeck et al., 2014b; de Vargas et al., 2015; Massana et al., 2015; Gimmler et al., 2016; Piredda et al., 2018; Grattepanche and Katz, 2020; Qu et al., 2021). The correct taxonomic annotation of OTUs (Operational Taxonomic Units) or ASVs (Amplicon Sequence Variants) relies on publicly accessible databases; the EukRef platform is a curated database, while others, e.g., NCBI, still tend to be incomplete for ciliate species and the data are not curated (Boscaro et al., 2018; del Campo et al., 2018).

The present study combines morphological, ecological, and genetic approaches and offers a valuable basis for future ciliate research (Spanner et al., 2020; Forster et al., 2021; Pröschold et al., 2021; Qu et al., 2021). It provides the first single cell barcodes of morphologically investigated urotrich ciliates. The findings confirmed the resolution of the SSU rDNA gene at the generic level (Figures 6, 7). The hypervariable V4 region of the SSU rDNA had a higher resolution, allowing even the discrimination of the various Urotricha strains, whereas the V9 failed in distinguishing the morphospecies assigned to U. furcata and U. pseudofurcata, respectively (Figure 9 and Supplementary Figure 4). Interestingly, the closely related Coleps group could be separated by means of the V9 sequences, while not using the V4, and the phenotypic plasticity among the 19 clones analyzed was considerable (Pröschold et al., 2021). In the SSU rDNA tree of the prostomatids, two Urotricha-related sequences differ morphologically, i.e., Plagiocampa sp. (KY980324; Zhang et al., 2014) and H. verrucatum (LC424401; Gurdebeke et al., 2018). They probably represent misidentifications or cross-contaminations, especially as both the phylogenetic placement and the signatures of the V4 and the V9 gene regions indicate that they are urotrichs (Figures 6, 9). However, a general genetic marker gene region for ciliates requires more taxon-related studies in the future.

Morphology-based species identification of urotrichs alone is difficult, particularly in the small congeners. Apart from their small cell size, their rapid locomotion hampers live observations, especially at high magnifications; thus, silver-staining methods are indispensable but require some experience to be successfully applied (Foissner and Pfister, 1997; Foissner, 2014). These issues might be responsible for the low number of species descriptions since the monograph on the taxonomy and ecology of planktonic freshwater ciliates by Foissner et al. (1999), i.e., only Urotricha spetai and Urotricha psenneri have been established (Sonntag and Foissner, 2004; Foissner, 2012). Moreover, genetic datasets are missing or could not be correctly assigned. The present data, especially the genetic clues, propose the existence of small (pseudo-cryptic) urotrichs in freshwater habitats, which also await their description (Sonntag et al., unpublished).

Recently, DNA metabarcoding studies in oceans and lakes have opened up new perspectives on the high diversity and geographic distribution of protists and ciliates in particular (de Vargas et al., 2015; Gimmler et al., 2016; Qu et al., 2021). Such studies are commonly dedicated to the detection of diversity and its correlation with environmental parameters but rarely focus on individual taxa. However, ciliate abundance can hardly be estimated in environmental barcoding approaches based on OTU or ASV counts, because the number of SSU rDNA gene copies varies considerably within and among species (Zhu et al., 2005; Stoeck et al., 2014b). Therefore, morphospecies counts and OTU abundance rarely coincide (Pitsch et al., 2019; Pröschold et al., 2021). However, this aspect has previously been exhaustively discussed (e.g., Stoeck et al., 2014b; Stern et al., 2018). Nevertheless, the comparison of the taxon-specific spatial and temporal distribution patterns inferred from both HTS datasets and morphological surveys (e.g., Pröschold et al., 2021 and Supplementary Figures 5, 6) might provide hints for niche separation and speciation in cryptic species.

Linking taxa/species with more or less known ecological preferences to DNA metabarcoding datasets is essential for inferring food web structures and processes from HTS datasets (e.g., Forster et al., 2021; Pröschold et al., 2021; Qu et al., 2021). Therefore, we performed an experiment for at least one of our barcoded urotrichs, i.e., strain CIL-2017/25 to gain insights into the ciliate’s autecology. These findings can then be implemented into analyses of the spatial and temporal distribution of the population in natural communities. Accordingly, key players in aquatic food webs can be identified and future research questions and experimental approaches on climate change scenarios, for example, directly addressed. Moreover, we aim tagging OTUs with a name and fill up public databases with sequences to identify and record their specific distribution patterns. In environmental studies, individual taxa can serve as indicators for water or soil quality (e.g., Foissner et al., 1999; Stoeck et al., 2018; Dias et al., 2021). Actually, only some ciliates can exclusively be related to specific environmental conditions, namely, the obligate anaerobic Cariacotrichea or the Armophorea (Lynn, 2004; Orsi et al., 2012).

Autecological studies further form the basis for generating knowledge on the functional ecology and diversity by elucidating a protists’ individual placement in an aquatic or soil food web (e.g., Geisen, 2016; Weisse et al., 2016; Qu et al., 2021). In contrast to the marine environment, for many freshwater ciliates at least some ecological information on food items and preferred environmental conditions (temperature, pH, oxygen concentration etc.) is available (see compilation on the autecology of planktonic species in Foissner et al., 1999 and references therein). For example, the U. castalia-complex has previously been tested for its food and temperature response and cultures were kept in modified WC medium (Guillard and Lorenzen, 1972; Weisse et al., 2001). Obviously, members of the U. castalia-complex grow well on cryptomonad food. However, we assumed that the ciliate preferred motile prey that triggered a kind of hunting behavior; though, this has not been directly tested here. This finding supports the results of Qu et al. (2021) who suggested diverse flagellates as potential food source for two U. castalia populations. Apart from studying morphological and molecular characters for species description as emphasized by Warren et al. (2017), we strongly encourage researchers to also include autecological datasets to experimentally address specific research questions on predator-prey relationships or food preferences.

Despite some anecdotal notes on food items and environmental conditions, the successful collection of the desired species in the field and its cultivation under laboratory conditions prior to an experiment are problematic. Particularly, the development of cultivation media with appropriate food items for establishing long-term cultures is tricky. Consequently, only those ciliates that are relatively easily kept under laboratory conditions, such as Paramecium, Tetrahymena, or Euplotes species, are intensively studied. Yet, long-term cultures of planktonic species are more difficult to manage and we aim depositing our strains in public culture collections for future research.

Classifying planktonic organisms, particularly ciliates, according to their sizes was a common procedure in lake ecology in the 20^th century, owing to methodological difficulties caused by the application of certain fixatives (e.g., Lugol’s solution) that obscure the taxonomically relevant features. Admittedly, live observations by means of interference contrast optics and the staining methods compiled by Foissner (1991, 2014) are time-consuming and necessitate some experience. Occasionally, scanning electron microscopy might be useful for displaying surface structures, e.g., the adoral organelles and oral flaps in urotrichs. Yet, in modern taxonomy (species identification and description), these methods are indispensable because they distinctly increase the taxonomic resolution and reliability. The quantitative protargol stain (QPS) does not only reveal species-specific cytologic characters (e.g., ciliary pattern and nuclear apparatus) but allows for estimating taxon abundance (Skibbe, 1994). The application of the QPS therefore provides a useful tool for ecological plankton studies (e.g., Pfister et al., 2002; Sonntag et al., 2006, 2011, 2017; Van Wichelen et al., 2013; Kammerlander et al., 2016).

While the usage of ciliate size classes is frequently misleading in ecological studies, urotrichs represent an exception as their two classes obtain somewhat different placements in the food web. Species more than 70 μm long not only feed on picoplankton and nanoplankton as almost all urotrichs, but are also predators of rotifers (see compilation of species and their food preferences in Foissner et al., 1999). Within the size classes, urotrichs might show species-specific food preferences, necessitating the identification to species level. The about 20 μm small urotrich species have one or two caudal cilia (Foissner et al., 1999; Sonntag and Foissner, 2004), whereas the larger urotrichs have three or more caudal cilia. Additional distinguishing features are the presence (and types) of extrusomes and the ciliary pattern.

The example of the urotrichs, demonstrates the need of solid knowledge on the autecology of the taxa under consideration for deciding whether a size-dependent approach for inferring the food web position is justified. Particularly, in genera comprising mixotrophic and heterotrophic species (e.g., in Coleps, Vorticella, and Monodinium), an identification down to species level is indispensable for estimating carbon flux in the habitat.