Fresh plant material was morphologically determined by the respective collector ( Tables S1 , S2 ) based on descriptions by various authors (Wood and Imahori, 1965; Corillion, 1975; Krause, 1997; Cirujano et al., 2008; Lambert et al., 2013; Pérez et al., 2016; Van De Weyer and Schmidt, 2018). According to those, all used T. hispanica were clearly identified by their dioecious character, whereas T. glomerata, T. salina and T. nidifica were first determined by antheridial sizes, habitat occurrences and oospore ornamentation.

Individuals that featured vegetative and antheridial characters of two taxa, e.g., T. nidifica and T. salina, were determined as T. sp.

Specimens identified as T. glomerata, T. hispanica, T. nidifica, T. normaniana and T. salina by means of vegetative characters (mostly antheridia sizes) were obtained from herbarium collections and from field collections by the authors for a total of 157 specimens. The collections span the period between 1871 and 2020 from locations in nine European countries (Austria, France, Germany, Great Britain, Greece, Ireland, Italy, the Netherlands, Norway, Portugal, Sweden) and from Chile in South America ( Table S1 ). In addition, 40 specimens (Denmark, Italy, Germany, Great Britain, Greece, France, Norway, Portugal, Sweden) could not be unambiguously assigned morphologically to any recognized taxon and are referred to as Tolypella sp. throughout the manuscript.

Specimens of T. nidifica and T. glomerata from deep water sites were collected by diving. In shallow waters, samples were gathered by snorkeling or wading. Some specimens of T. salina (France) and T. glomerata (Germany) originated from germination experiments under laboratory conditions (e.g., Holzhausen, 2016; Holzhausen et al., 2017). For all fresh material, oospores were harvested after release from cultured material in order to confidently assess oospore maturity. In addition, oospores of T. nidifica from Austria and Germany were collected from sediment samples. Herbarium specimens were sampled from collections deposited in the Herbarium Rostochiense (ROST), the New York Botanical Garden Sterre Herbarium (NY), UiO Vascular Plants Herbarium, Natural History Museum, University of Oslo (O) and private herbaria of the collectors ( Table S1 ).

Dried plant material was obtained from a total of 193 individuals initially identified as T. glomerata (60 specimens), T. hispanica (15 specimens), T. nidifica (51 specimens), T. salina (50 specimens), T. normaniana (9 specimens), and 13 morphologically ambiguous Tolypella sp. Genetic data for the atpB, psbC and rbcL plastid genes presented in this study were obtained by three different working groups: A) the University of Rostock, B) the New York Botanical Garden and C) the Norwegian Institute for Water Research by the following methods.

Method A) Genomic DNA was extracted using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany), following the manufacturer’s protocol. Amplification of the plastid genes rbcL, psbC, and atpB was performed with 10 PCR cycles with one minute each of annealing at 94°C, extension at 55°C, and denaturation at 72°C, followed by one minute each for denaturation (94°C), annealing (52°C), and polymerisation (72°C) in 25 cycles. The amplified DNA was purified using the Biometra-innuPrep Gel ExtractionKit (Analytik Jena, Jena, Germany) according to the manufacturer’s instructions. Samples were sequenced using a 3130×L GeneticAnalyzer (Applied Biosystems, NY, USA) with sequencing primers identical to the primers that were used for PCR reactions ( Table S3 ). Obtained sequences were checked visually and aligned using BioEdit v.7.0.5.2 (Hall, 1999).

Method B) Genomic DNA was extracted using the Nucleon Phytopure DNA extraction kit (GE Healthcare Gio-Sciences, Pittsburgh, PA, USA, Pérez et al., 2014). The atpB, psbC and rbcL genes were amplified by a nested PCR reaction using either a PTC-200 DNAEngine^® Thermal Cycler (Bio-Rad, Hercules, CA, USA) or a Mastercycler^® pro S (Eppendorf AG, Hamburg, Germany). Initial PCR amplicons were generated through the following cycling program: initial denaturation at 95 ˚C for 2 minutes; 35 cycles of 95 ˚C for 15 seconds; 48 ˚C for 15 seconds and 72 ˚C for 30 seconds; and followed by a final extension at 72 ˚C for five minutes. The resulting PCR product were used in a second round of PCR amplification to generate internal sequences using the same cycling program with the exception that the cycling was reduced to 30 cycles and the final extension time reduced to 30 seconds. Products from both PCR sets were sequenced at the University of Washington Genome Center (Seattle, WA, USA).

Method C) Genomic DNA from Tolypella material was isolated after (Schneider et al., 2016). PCR for the rbcl, atpB, and psbC genes was performed on a Bio-Rad CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories, Oslo, Norway) using the iProof High-Fidelity PCR Kit (Bio-Rad Laboratories, Oslo, Norway). The following cycling protocol was used for all three genes: one cycle of 5 min at 94°C, and then 35 cycles each consisting of 10 s at 94°C, 20 s at 62°C, and 20 s at 72°C, followed by a final elongation step of 72°C for 5 min. PCR products were visualized by 1.5% agarose gel electrophoresis with GelRed staining (GelRed^® Nucleic Acid Gel Stain, Biotium, Fremont, USA) and UV illumination. Amplification of the rbcL, atpB and psbC gene region was conducted using the primers listed in Table S3 . In some cases, a nested PCR was conducted using the former PCR product as template and a second primer pair for a further PCR amplification. For sequencing the same primers and if necessary, intermediate primers were used ( Table S3 ). Sequences were analyzed and aligned using Seqassem (version 04/2008) and Align (version 03/2007) MS Windows-based manual sequence alignment editor (SequentiX – DigitalDNA Processing, Klein Raden Germany) to obtain DNA sequence alignments, which were then corrected manually. For each PCR product, both strands were sequenced on an ABI 3730 Avant genetic analyzer using the BigDye terminator V.3.1 cycle sequencing kit (Biosystems, Applied Biosystems, Thermo Fisher Scientific Oslo, Norway) according to the manufacturer’s instructions.

Complete sequences of all three plastid genes could not be generated for every sample analyzed due to the differing qualities of the specimens (age, storing conditions, drying conditions after collection, etc.). Therefore, four different datasets were used for the phylogenetic analyses in order to obtain as much information as possible for all specimens. The first dataset included the three plastid gene sequences from 88 individuals, whereas three additional data sets were compiled for each plastid gene separately; the atpB dataset with a total of 1034 positions for 94 samples, the psbC dataset with a total of 1104 positions for 125 samples, and the rbcL dataset with a total of 1265 positions for 189 samples. As outgroup, which is defined as closely related taxon or group outside of the taxon investigated, two different sequences of T. porteri were used. In order to estimate evolutionary divergence, pair-wise uncorrected p-distances and the number of substitutions were conducted using MEGA version7 (Kumar et al., 2016). To uncover phylogenetic relationships, Bayesian inference (BI) and maximum likelihood (ML) trees were constructed, with evolutionary substitution models evaluated in MEGA v.7. The method selected the same best-fitting evolutionary model (GTR+G+I) for each of the four datasets. The ML algorithm was conducted in MEGA v.7 with 1000 bootstrap replicates. BI trees were performed with MrBayes 3.2.6 (e.g. Ronquist and Huelsenbeck, 2003) with a random starting tree and two independent runs of one cold and three heated chains, each using default parameters. Each analysis was run for 2 million generations with trees sampled every 1000 generations and the first 25% generations discarded as burn-in.

Due to small genetic distances among some taxa, intraspecific data often produce a variety of possible trees when using conventional tree building methods. In such cases, the relationship among taxa is best expressed by a network that is able to show alternative potential phylogenetic relationships within a single figure (Bandelt et al., 1999). Furthermore, networks allow the identification and illustration of ancestral alleles whereas phylogenetic trees treat all sequences as terminal taxa (Posada and Crandall, 2002). For that reason, Median-Joining (MJ) network analyses were performed using the PopART software v1.7 (Leigh et al., 2015).

The terminology of oospore characters in this study is based on Soulié-Märsche and García (2015). Descriptions of membrane ornamentation follow those of Frame (1977) and Urbaniak et al. (2012). Altogether, 712 mature oospores were harvested from herbarium specimens, fresh plant material and sediment samples. Oospores were collected from 12 specimens of T. glomerata, 19 of T. nidifica, 18 of T. salina and 10 of T. sp. The individual numbers of oospores examined and the appropriate pre-treatments are given in Table S2 . Oospores were stored in well plates for eventual re-examination and are part of the Rostocker oospore database (Holzhausen et al., 2015).

For stereomicroscopic analysis, oospores were photographed in lateral, apical, and basal views with a mounted camera. Qualitative oospore characteristics that were examined included colour, shape and membrane ornamentation. To differentiate among the various brown hues of oospores, colour terms used in this study are clay brown, fawn brown, nut brown, chestnut brown, dark brown wine red and black brown (RAL COLOUR SYSTEM). Quantitative characteristics included: number of striae, expression of striae (prominence of striae), angle of striae with respect to the longitudinal axis, oospore length and width, fossa width (average of 4 fossae), and length of the outer lines of the pentagonal basal impression. Length measurements were calculated using ImageJ 1.50i. The ISI (isopolarity index; 100*(length/width)) was also calculated.

Scanning electron microscope (SEM) analyses of oospores were performed at the Senckenberg Forschungsinstitut und Naturmuseum Frankfurt. Prior to SEM observations, few oospores were pre-cleaned (HoAc 5%) and all were dried by lyophilisation. Dry specimens were later sputter-coated with gold. SEM images of the surface of oospores and fossa walls (magnification 200-2500X) were produced with a JEOL JSM-6490 LV in high-vacuum mode by using secondary electrons and routinely applying an acceleration voltage of 20kV.

Oospore characters were tested for normality using the Shapiro-Wilk Test. Pairwise tests were performed for different levels of analyses (taxon, country, region, type of location and plants) by the Kruskal-Wallis Test (SPSS). P ≤ 0.05 was used as statistical significance for oospore analyses.

To identify (a) the correlation between oospore characteristics and regionality and (b) parameter combinations that might provide reliable discrimination, combined analyses of all quantitative and qualitative oospore features as well as their ratios, with the exception of the membrane ornamentation, were performed by nonmetric multivariate techniques using the Primer7 software package (Clarke and Gorley, 2015). Principal component analysis (PCA) was based on standardised, square root transformed data and Euclidean distance matrices. Multiplot-Analysis based on non-treated raw data.

The phylogenetic analyses of the plastid gene sequences in each dataset recovered the 196 Tolypella individuals into five general clades that were denoted as ‘T. glomerata’, ‘T. nidifica/salina’, ‘Tolypella sp.’, and two distinct ‘T. hispanica’ clades (‘T. hispanica I’ and ‘T. hispanica II’; Figure 1 ; Table S4 ). Supporting values for each cluster were given in the sections below. The gene sequence similarities of Tolypella individuals within each group were generally over 99.8% ( Table 1 ). However, support for their phylogenetic placements were unresolved or weakly to moderately supported in the single gene analyses. Phylogenetic resolution and support were greatest in the three-gene analyses. The results of the network analyses were comparable to the phylogenetic trees, with the same clusters recovered in both approaches ( Figures 1 , 2 ; Table S1 ; Figures S1 – S3 ). Results for the ML tree and the MJ network of concatenated gene sequences are shown in Figures 1 , 2 . Complete trees and networks for single gene analyses are shown in the supplement ( Figures S1 – S3 ). The labels used to identify genetic groups correspond to those in Table S1 .

‘T. glomerata’

A first group comprising mostly T. glomerata contained 49 individuals that represented T. glomerata (39 specimens), T. normaniana (7 specimens), and T. sp. (1 specimen). Specimens originated from nine European countries (France, Germany, Great Britain, Greece, Ireland, Italy, Netherlands, Norway, Sweden), and from the United States, Canada, and Chile. The analysis of concatenated sequences of ‘T. glomerata’ identified a difference of ≥ 18 nucleotides with respect to other clusters of sect. Tolypella ( Table 1 ). With an average of about 3.85 bp substitutions, the genetic variability within the group was relatively high compared to the intragroup variability shown by the other taxa. However, differences in their sequence data were not regionally correlated; European and North American specimens showed identical sequences. In contrast, 8 substitutions were observed between samples collected in Greece (GR02-UW22) and Italy (RB13_16-1 and RB12_14-1).

‘T. nidifica/salina’

A second cluster (labelled ‘T. nidifica/salina’) consisted of 110 individuals which have traditionally been assigned mainly to the taxa T. nidifica (44 specimens) and T. salina (50 specimens). Additionally, two T. normaniana, and 14 morphologically ambiguous T. sp. were found in this cluster ( Table S4 ). They were collected in Austria, France, Germany, Greece, Italy, Norway and Sweden. Interestingly, the specimen originally determined as T. glomerata f. littorea from France (KGK4867) and two of nine sequenced T. normaniana (T70/T71) clustered within T. nidifica/salina. Sequence data for each of the atpB, psbC and rbcL genes could not be obtained for every specimen in this cluster. rbcL sequence data was obtained from 188 specimens, whereas sequences for atpB and psbC were obtained from 38 specimens ( Table S4 ). Overall, however, there was little genetic variation within the ‘T. nidifica/salina’ cluster when comparing each of the gene sequences. Depending on the dataset, between 95.2 and 97.6% of the analysed specimens had identical sequences. Minor genetic differences were observed in this group; two T. sp. collected from Italy (rbcL, Figure S1 ), and a T. normaniana from Norway (psbC, Figure S3 ) differed by a single nucleotide substitution each. Regional differences were not reflected in the sequence data with identical haplotypes throughout Europe. Consistent nucleotide differences were found only among two T. sp. collected in Austria (AU03-Tg01, AU03-Tg03).

‘Tolypella sp.’

A third cluster (labelled ‘Tolypella sp.’) consisted of 21 individuals which have been classified as morphologically ambiguous specimens due to the presence of vegetative characters of more than one Tolypella taxon mentioned above. The specimens were partly originally determined as T. glomerata f. humilor. They were collected in Denmark, France, Germany, Great Britain, Greece, Italy, Norway, Portugal, and Sweden. ‘Tolypella sp.’ revealed unique sequence data and differed from ‘T. nidifica/salina’ and ‘T. glomerata’ by averaging 19.4 and 27.9 bp substitutions respectively ( Table 1 ). Nucleotide differences within ‘Tolypella sp.’ ranged from 0 to 4 bp substitutions for concatenated sequences (mean 0.06%, Table 1 ; Figure 2 ).

‘T. hispanica’

Two clusters labelled as ‘T. hispanica I’ and ‘T. hispanica II’ consisted of ten and five individuals, respectively, which have traditionally been assigned to the dioecious T. hispanica. ‘T. hispanica I’ included ten individuals collected in France, Greece, Italy, and Algeria. The samples collected in France had two unique nucleotide substitutions for the combined sequences (0.06%, Table 1 ). ‘T. hispanica I’ formed a strongly supported clade together with ‘T. glomerata’ ( Figure 1 ). ‘T. hispanica II’ contained five individuals from four field collections in Italy, Sardinia that shared identical rbcL sequences whereas three individuals had identical sequences for all three genes ( Table 1 ). In the ML analysis, however, ‘T. hispanica I’ was sister to ‘Tolypella sp.’ in a weakly supported relationship ( Figure 1 ).

Differences in quantitative and qualitative oospore characters were considered with respect to taxa determined by either vegetative morphology or genetically determined cluster. The results of oospores grouped in taxa determined by vegetative morphology are summarized in Table 2 .

T. glomerata

Oospores analysed in this study were usually fawn brown (97.4%), occasionally chestnut brown (1.7%) or dark brown wine red (0.9%), with an elongated rounded shape with 7-8 striae. The expression of striae is flat to prominent (0.0–17.9µm). Oospores showed lengths of 252.2 to 487.4 µm ( ± 49.8-50µm), widths from 197.6 to 351.2 µm ( ± 33.5-34.1µm), a mean fossa width of 31.0 to 57.9 µm, mean lengths of the outer basal impression from 29.0 to 61.8 µm and an ISI of 110–160. All oospores exhibited a reticulate ornamentation in varying expression and size ( Figure 3 ; Table S2 ).

T. nidifica/salina

Oospores of T. nidifica/salina were mainly dark brown wine red (48.3%), chestnut brown (32.6%) or black brown (16.7%) with a terete or broad rounded/globose shape and a flattened base. Oospores showed (4-)5-7(- 8) striae that were flat and prominent (0.0–17.7 µm), oospore lengths of 194.3–457.5 µm ( ± 45.2µm), oospore widths of 149.9–404.7 µm ( ± 40.4µm), mean fossa width of 27.2–69.7 µm and an outer mean basal impression length of 27.1 to 89.9 µm. The calculated ISIs ranges between 100 and 190. Ornamentation patterns of T. nidifica/salina were highly variable, from smooth to smooth with some pustules or with fine linear structures ( Figure 3 ; Table S2 ).

Tolypella sp.

The oospores of the morphologically ambiguous specimens were dark brown wine red (78.9%), occasionally fawn brown (8.4%), chestnut brown (7.4%) or black brown (5.3%) with a broad range of shape variations (ellipsoid, elongate rounded, broad rounded/globose, or peanut- shaped). Oospores showed (5) 6–7 (-8) striae with widths between 3.0–11.2 µm, oospore lengths of 266.7–472.0 µm ( ± 38.7µm) and widths between 195.6–346.7 µm ( ± 27.9 µm). Fossae ranged from 37.2 µm to 62.2 µm and outer basal impression lengths from 33.1 up to 57.9 µm. Calculated ISIs ranges from 80 to 190 (-220). The membrane of T. sp. Oospores was smooth to smooth with some pustules or with fine linear structures ( Figure 3 ; Table S2 ).

The analysis of taxon-related oospore characteristics shows that variations, especially with regard to the features colour, shape and length exist within each taxon ( Figure 3 ; Tables 2 , S2 ). Especially for T. glomerata, large discrepancies between oospores along a geographical gradient and between populations could be observed ( Figure 4 ). The PCA shows that the two axes explain 64.8% of the cumulative variation of oospores (eigenvalue 1 = 6.72, eigenvalue 2 = 3; Table S5 ). The first component is determined by the characters length and width and the ratios of length/angle and width/angle, whereas the second component is defined by the striae and width/fossa ratio. Depending on the level of analysis, significant intraspecific differences between countries, regions, type of locations and plants can be detected. Oospores from Germany (length: 335µm – 487µm, width: 255µm – 351µm) and Austria (length: 378µm – 438µm, width: 241µm – 313µm) were significantly larger and wider than those from Greece (length: 294µm – 326µm, width: 197µm – 221µm; p ≤.001) and Italy (length: 252µm – 362µm, width: 203-µm – 274µm; p ≤.001). No significant differences could be detected between oospores from Germany and Austria or between those from Italy and Greece.

Given the vegetative determination in T. nidifica, T. salina and T. sp., significant differences could be obtained for the oospore characters shape (p ≤.006), length (p ≤.005), fossa (p ≤.005) and width (p ≤.005) between T. nidifica and T. salina. Tolypella sp. could be separated from T. nidifica by the width, fossa (p ≤.006) or shape (p ≤.017) whereas T. sp. could be differentiated by oospore length (p ≤.005) and shape (p ≤.006) from T. salina although overlapping areas exist. However, these differences can only be obtained by means of statistical tests. Multiplot Analysis ( Figure S4 ) on the other hand show the overlapping areas, which were partly caused by single plants or individual oospores. Due to the lack of statistical power, they should be seen only as trend.

Interestingly, depending on the type of location, oospores of permanent water bodies (lagoons, harbours and lakes) and pools are wider than oospores from temporary water bodies ( Figure 4 ).