Abiotic parameters indicate ecological disturbance in the east basin of Sanabria Lake
Sanabria Lake is a warm monomictic lake with water circulation during winter and thermal stratification that begins in the spring and continues until the end of the summer (Vega et al 1992). Samples were collected at the beginning of the thermal stratification period when the eukaryotic community is expected to be homogeneously distributed in the water body following the winter mixing. This period also coincides with the time of the year with the lowest direct anthropogenic impact. Water temperature at the surface ranged from 7.1 °C to 8.4 °C and in the deepest sampling points ranged from 6 °C to 6.85 °C, with a mean range of 1.62 °C (Figure 1, Supplementary Table 1). These temperature measurements agree with the previously recorded temperatures during the homeothermic state of the lake that range between 4 to 7 °C (Vega et al 1992) and confirm the mixing state of the lake.
We assessed the trophic state of Sanabria Lake based on water turbidity and chlorophyll a values. Water turbidity is measured in FTU (Formazin Turbidity Units) and is an indicator of the trophic state of a lake as it is related to the concentration, type and size of the suspended particles in the water (Çako, Baci, and Shena 2013). During our sampling, turbidity values in Sanabria Lake were extremely low in all the sampling sites and ranged from 0.5 to 0.85 FTU (Figure 1). These values are comparable to those in ultra-oligotrophic alpine lakes (Chanudet and Filella 2007). Chlorophyll a is a reliable indicator to assess the trophic state of a lake with high values to correspond to more eutrophic ecosystems (Poikane et al. 2014). Chlorophyll a mean values in Sanabria Lake have increased in the last fifty years (Supplementary Table 5) but they have not exceeded the levels that characterise oligotrophic lacustrine ecosystems. Together, these measurements confirm the overall oligotrophic status of the lake at the time of sampling.
We observed that chlorophyll a values differed between east and west basin during our sampling (Figure 1). In Sanabria´s west basin (samples S1-S3), the mean value of chlorophyll a was below the reference value (1.5 μg/L). The reference value defines the equilibrium ecological state of the lake and confirms the absence of ecological disturbances. However, the mean values of chlorophyll a in Sanabria´s east basin (samples S4-S5) exceeded the reference values indicating the presence of ecological disturbance (Figure 1). Values of chlorophyll a above 4.2 μg/L are linked to a Good-Moderate ecological state and values above 7.1 μg/L are linked to a Moderate-Poor ecological state (BOE, 2015). Our results showed that there was some ecological disturbance that altered the values of chlorophyll a in the east basin of Sanabria Lake at the time of sampling. The altered values of chlorophyll a in the east basin may be related to higher anthropogenic impact due to the presence of three camping sites on this side of the lake. Chlorophyll a values measured in Sanabria´s east basin in March 2017 (Llorente and Seoane 2020) are lower than the ones presented in our study, implying that the stressor was temporal and that water quality has been restored.
The V4 hypervariable region captures the microeukaryotic diversity of Sanabria Lake
To characterise the diversity of the eukaryotic community in Sanabria Lake, we sequenced the V4 hypervariable region of the 18S small subunit (SSU) rRNA gene. We chose to sequence the V4 over other hypervariable regions of the 18S rRNA gene because it provides taxonomic resolution close to that of the full-length gene (Dunthorn et al. 2012; Hu et al. 2015) and it is the most suitable hypervariable region to use for phylogenetic placement (Mahé et al. 2017). A total of 15,947,744 reads from 82 samples were filtered, dereplicated and merged resulting in 31,225 Amplicon Sequencing Variants (ASVs). The study of multicellular organisms was out of the scope of the present work and thus most multicellular organisms were discarded by using physical filters of 2000 µm and 200 µm. However, some environmental DNA (eDNA) that originates from cellular material shed by multicellular organisms into the lake was sequenced together with the community DNA of unicellular eukaryotes. For our subsequent analyses, we bioinformatically filtered out all ASVs that were assigned to animals (Division/Class = Metazoa), land plants (Division = Streptophyta) and typical terrestrial fungi (Class/Order = Ascomycota, Class/Order = Basidiomycota) (Supplementary Table 3, dataset D3). After the removal of multicellular taxa, 27,790 microeukaryotic (protist) ASVs remained. We evaluated the sampling depth and the representation of microbial eukaryotes in our samples using rarefaction curves (Supplementary Figure 1). The curves reached a plateau for all samples, indicating that most of the microbial richness present in Sanabria Lake and the surrounding freshwater systems was successfully captured by our study.
Spatial biodiversity patterns
To evaluate the intra-sample diversity of Sanabria Lake and the surrounding water bodies, we calculated nine different alpha-diversity indices (Supplementary Table 4). To avoid potential biases in diversity estimates due to differences in the total number of reads, we randomly subsampled the ASVs to the minimum depth of our dataset (Supplementary Table 3, dataset D3, min sample depth = 31361 reads) before calculating the alpha-diversity indices. The number of total taxa reported was not affected by subsampling. We compared the diversity of the different water body types and we found that samples collected in the tributary stream showed significantly higher intra-sample diversity (Figure 2) and greater evenness (Supplementary Figure 10) compared to samples from Laguna (pond) and Sanabria (lake) (Wilcoxon rank sum test P value<0.01) . Previous studies have shown that small water bodies like ponds and streams can contribute significantly to regional biodiversity of macrophytes and macroinvertebrates (Williams et al. 2004). Our data support the hypothesis that the same is true for microeukaryotes. This result pinpoints the importance of small water bodies as biodiversity reservoirs and contrasts with their relative status in national monitoring and protection strategies, where they are frequently ignored. Regarding the different habitats, sediments harbour the richest microeukaryotic communities (Figure 2). Sediments have been shown to harbour richer communities than the water column for other groups of organisms like bacteria (Eckert et al. 2020) and marine diatoms (Piredda et al. 2018). However, we cannot exclude that part of the diversity recorded in the sediments can be attributed to either dormant stages of planktonic microeukaryotes or dead cells that were recently settled from the water column.
To test the effect of abiotic factors in the protist community structure we observed, we carried out permutational multivariate analysis of variance (PERMANOVA) of Bray-Curtis dissimilarities of the ASVs between communities as a function of sample spatial origin (Supplementary Table 6). All factors tested by PERMANOVA tests revealed significant differences in protist communities as a function of site (Sanabria Lake, Laguna, Stream), sampling site (S1-S10), position regarding the chlorophyll maximum (on-off), position regarding the thermocline and habitat (water column, sediments, biofilms) (Supplementary Table 6).
To visualise the compositional differences in the community structure of protists we applied nonmetric multidimensional scaling (NMDS). The communities from Sanabria Lake, the tributary stream and the Laguna were clearly separated in an ordination based on sampling site (Figure 3A). The samples were also distributed along the first axis (NMDS1) as a function of the habitat, with water column samples occupying the first three quadrants, biofilm samples in the third and fourth quadrant and sediment samples occupying only the fourth quadrant (Figure 3A). The community of microbial eukaryotes in the water column of Sanabria Lake was clearly segregated as a function of the size fraction of the filter and not the sampling depth (Figure 3B). This is what we expected given that we sampled at the beginning of the thermal stratification after the winter mixing at the point of maximum homogeneity of the community. As we observed that chlorophyll a values differed between east (S1,S2, and S3 sampling sites) and west basin (S4 and S5 sampling sites) (Figure 1) we investigated using a NMDS plot whether the microeukaryotic communities of east and west basin are grouped together but we did not observe such grouping (Supplementary Figure 2)
Our observations were statistically supported by ANOSIM tests (Supplementary Table 7), which showed significant and marked differences among communities according to habitat, sampling site and depth (Supplementary Table 1). Our results suggest that the community structure in Sanabria Lake and the surrounding freshwaters is characterised by spatial variation. The habitat is a major factor that shapes the community structure after the winter mixing period. Sediments, biofilms and water column harbour compositionally heterogeneous microbial communities that are driven by the unique environmental parameters that characterise them.
Taxonomic composition of the protist community
To gain an overview of the microeukaryotic taxonomic composition in the Sanabria Lake and the surrounding freshwater systems, we plotted the relative abundance of ASVs at division level (based on the PR2 taxonomy) across sampling sites (Figure 4). The phylogenetic diversity of ASVs covered all currently recognized eukaryotic supergroups (Adl et al. 2019). The group of Stramenopiles was the most abundant supergroup in all sampling sites, accounting for the 33% of total reads in Sanabria Lake, 34% in the nearby pond (Laguna) and 40% in the tributary stream respectively (Supplementary Figures 3, 4, and 5). In addition to being abundant, Stramenopiles were diverse, encompassing 22% of total ASV richness (6,988 ASVs) (Supplementary Table 3). Among Stramenopiles, Ochrophyta was the most abundant group in all sampling sites (Supplementary Figure 6). Most Ochrophyta in the tributary stream (85%) and Laguna (81%) were affiliated with Chrysophyceae (Supplementary Figure 6), a group that is generally common in low-nutrient lakes (Nicholls and Wujek 2003). In Sanabria Lake, together with the Chrysophyceae (36%), we report a high relative abundance of Bacillariophyta (37%) and Synurophyceae (24%) within Ochrophyta, two phototrophic lineages that produce silica skeletons or scales (Supplementary Figure 6). Alveolata was the second most abundant and diverse supergroup, accounting for 26%-28% of the total eukaryotic reads in each site (Supplementary Figures 3, 4, and 5) and a total of 4,609 ASVs in the study (Supplementary Table 3).
The plankton community of Sanabria Lake (excluding the surrounding freshwater systems) was dominated by Ochrophyta (in the Stramenopiles supergroup; 26%), Ciliophora (Alveolata; 14%), Dinoflagellata (Alveolata; 10%), Cercozoa (Rhizaria; 10%), Cryptophyta (10%) and unicellular Opisthokonta (7%) (Supplementary Figures 3). The presence of all these groups except for unicellular Opisthokonta was previously documented by light microscopy in Sanabria Lake (Vega et al. 1992). We further explored the taxonomic composition of Sanabria Lake by separately examining benthic and pelagic samples. The taxonomic composition of the benthic protist community, as represented by ASVs in the sediments, was dominated by Stramenopiles (36%), Alveolata (29%), Rhizaria (13%), Opisthokonta (7%), Amoebozoa (5%), Archaeplastida (3%), Hacrobia (3%) and Excavata (3%). In contrast, the planktonic microbial community was characterised by the prevalence of Hacrobia (16%) as the third most abundant eukaryotic supergroup after Stramenopiles (31%) and Alveolata (28%). The planktonic Hacrobia (Cavalier-Smith, Chao, and Lewis 2015) in Sanabria Lake included Cryptophyceae (84%), Katablepharidophyta (13%), Centroheliozoa (1.5%), Telonemia (1%) and Haptophyta (0.5%). Excluding Cryptophyceae, this is the first record for these taxonomic groups in Sanabria Lake. (Katablepharidophyta were previously classified inside Cryptophyceae until electron microscopy and molecular phylogenies provided evidence to consider them as a separate taxonomic group (Okamoto and Inouye 2005).)
Protist parasites in a temperate oligomesotrophic lake
Here we provide the first description of the taxonomic composition of the unicellular eukaryotic parasites (Supplememtary Table 3, dataset D6) present in Sanabria Lake, the biggest natural lake in the Iberian Peninsula. Parasitic unicellular eukaryotes modulate large scale ecological processes by regulating the abundance and dynamics of their hosts population (Bråte et al. 2010). As their study by microscopy is tedious, little was known about their prevalence in freshwater systems until the advent of metabarcoding (Frenken et al. 2017).
The parasites accounted for 21.3% (5,925) of the total protist ASVs identified in our study. The parasitic community composition was dominated in all sampling sites by Chytridiomycota, whose relative abundance within parasitic taxa was 29% in the tributary stream, 32% in Sanabria Lake and 42% in Laguna. The prevalence of Chytridiomycota in the pelagic zone of lakes has been previously reported (Lefèvre et al. 2007; Sime-Ngando, Lefevre, and Gleason 2011). Chytridiomycota, which includes more than 1,000 described species (Powell 1993; Shearer et al. 2007), was also the most diverse group of parasites in our study, including more than 2,200 of the 5,925 total parasite ASVs, distributed among more than 50 genera (Supplementary Figure 7). Almost half of the chytrids in terms of abundance identified in our study belonged to the order Rhizophydiales, that are host‐specific chytrids that infect various phytoplankton species, mostly diatoms (Jobard, Rasconi, and Sime-Ngando 2010; Rasconi, Jobard, and Sime-Ngando 2011). The prevalence of Rhizophydiales in the Sanabria Lake ecosystem was not surprising given that they are the most common planktonic chytrids in lacustrine ecosystems (Monchy et al. 2011). A species of Rhizophydiales was probably the causative agent of a chytrid infection in Sanabria Lake in 2014 that diminished the population of the diatom Tabellaria fenestrata and controlled an algal bloom caused by eutrophication (Llorente and Seoane 2020). The relative abundance of Perkinsea, a group of parasitic alveolates, ranged from 13% to 18% of total parasite abundance across the different sampling sites. Perkinsea were previously considered as strictly marine parasites (Norén, Moestrup, and Rehnstam-Holm 1999; Erard-Le Denn, Chrétiennot-Dinet, and Probert 2000; Villalba et al. 2004; Figueroa et al. 2008; Leander and Hoppenrath 2008) until molecular environmental surveys revealed an unprecedented diversity of these organisms in freshwaters (Richards et al. 2005; Lefèvre et al. 2008; Lepère, Domaizon, and Debroas 2008; Bråte et al. 2010).
The parasitic community of each sampling site carried a unique taxonomic signature. The parasitic community of the tributary stream was characterised by a higher proportion of Apicomplexa (17%) and Labyrinthulomycetes (12%) in comparison to the other sampling sites. Most of the apicomplexan ASVs in the tributary stream fell into eugregarines, the most abundant apicomplexan group in environmental surveys (del Campo et al. 2019). Parasitic Stramenopiles (Pseudofungi), a significant component of freshwater ecosystems (Cooper, Pillinger, and Ridge 1997), constituted the second most abundant group in Laguna and represented 20% of the Stramenopiles and 7% (76,070 reads) of all eukaryotes in this small pond (Supplementary Figure 6). Within the group of parasitic Stramenopiles (Supplementary Figure 6), there was observed a higher prevalence of Oomycetes that are common fish pathogens (van West 2006; Phillips et al. 2008) in Laguna in comparison to the other sampling sites. Finally, Sanabria Lake harboured a higher relative abundance of Ichthyosporea (12%, 96,491 reads) in comparison to the other sampling sites (Stream: 2%, Laguna: 1%). The majority of the Ichthyosporea in Sanabria were associated with the marine genera Abeoforma (69%), Sphaeroforma (17%) and Pseudoperkinsus (10%), none of them previously identified in a freshwater environment.
To confirm the identity of the ichthyosporean ASVs in Sanabria Lake we analysed them by phylogenetic placement. We compiled a dataset that encompassed all the extant diversity of unicellular Holozoa (n=234). Half of the complete 18S rDNA gene sequences used to build the reference tree belonged to uncultured environmental taxa. A total of 132 ASVs identified as Ichthyosporea by the Ribosomal Database Project (RDP) classifier were placed into the 465 branches of the reference tree (Supplementary Figure 8). Most of the queries were placed in a clade formed by the freshwater anuran parasite Anurofeca richardsi, the marine Creolimax fragrantissima, Pseudoperkinsus tapetis and Sphaeroforma arctica, and some uncultured environmental taxa (Supplementary Figure 8). The 132 ichthyosporean queries were clustered into 17 clades in the best-hit placement tree (Supplementary Figure 9). Most of the clades were associated with freshwater sequences. Clade 4, the one formed by the larger number of sequences, was assigned to the FRESHIP2 group (del Campo and Ruiz-Trillo, 2013), expanding the known molecular diversity of these environmental taxa. Clades 13, 14 and 15 were assigned to Anurofeca richardsi and clade 9 to Caullerya mensii, another freshwater parasite that infects Daphnia pulex (Lu et al. 2020). We identified two clades that were directly associated with marine Ichthyosporea, clade 6 that branched as sister to Abeoforma whisleri and clade 16 that branched as sister to Sphaeroforma arctica (Supplementary Figure 9). The genera Abeoforma and Sphaeroforma were previously considered exclusively marine and this is the first record that connects these taxa to freshwater habitats. As freshwater habitats are increasingly explored by molecular means, the number of taxa that have been previously reported as exclusively marine and later were found in freshwater surveys continues to expand (Bråte et al. 2010; Simon et al. 2015; Richards and Bass 2005; Annenkova, Giner, and Logares 2020; Massana et al. 2006; Simon et al. 2015; Yi et al. 2017; Mukherjee et al. 2019).
Abundant and potentially novel freshwater microbial eukaryotes
Metabarcoding biodiversity studies have shown that great part of the extant microbial diversity remains undocumented (Pawlowski et al. 2012; del Campo et al. 2014). In a metabarcoding survey, a species can be described as novel either because it is completely unknown to science or because the particular molecular marker database used in the study does not include available information on the species. In this study, we used the term molecular novelty to define any organism whose V4 hypervariable region of the 18S rRNA gene is not present in our reference database without discriminating between the two aforementioned cases.
To check whether Sanabria Lake and its surrounding freshwater systems could be a potential sampling site to isolate new organisms, we investigated the molecular novelty by first selecting potentially novel ASVs. We used the Ribosomal Database Project (RDP) classifier (Wang et al. 2007) to assign taxonomy to the ASVs. The RDP classifier provides for each ASV an assignment of the best matching taxa together with a bootstrap confidence score at each taxonomic rank. This score represents the level of confidence of the taxonomic assignment at each rank, from supergroup to species. Here, we define as poorly assigned, thus potentially novel, all ASVs with bootstrap confidence score value <97 at the supergroup level. We were interested in identifying the most abundant and novel microbial eukaryotes in our study site, so we selected all ASVs with more than 1,000 reads and bootstrap confidence score value lower than the aforementioned established novelty threshold.
To assign taxonomy to the queries of our dataset, we analysed them using phylogenetic placement (Figure 5). We first constructed a comprehensive eukaryotic reference tree with 618 eukaryotic taxa that encompassed all the extant eukaryotic diversity according to the latest classification of eukaryotes (Adl et al. 2019). We designed the reference tree with two criteria. First, to be inclusive in order to minimise phylogenetic placement artefacts related to taxonomic sampling and second to be non-redundant in order to be smaller and thus easier to handle in the post placement analyses. The amplicon short sequences were aligned to the reference alignment and the amplicon sequences that were not aligned in the V4 region were removed as artefacts after manual inspection. We placed a total of 113 ASV V4 queries into 1,233 branches of the reference tree.
Most of the ASV placements in the tree were found in the leaf nodes of Rhodophyta (Archaeplastida), Bigyra (Stramenopiles), early-branching Nucletmycea (also known as Holomycota) and Apicomplexa (Alveolata) pinpointing these clades as parts of the tree with potential novel undescribed molecular diversity (Figure 5). An elevated number of placements was spotted in the internal nodes of Dinophyta and the divergence between Opisthokonta and Apusomonadida (Figure 5). Apusomonadida is a recently defined phylum with a key phylogenetic position to understand the origin of the eukaryotic cell. Apusomonads are rarely detected in environmental studies (Purificación López-García et al. 2003; Not et al. 2008; Takishita et al. 2007; W. Orsi et al. 2012; Lesaulnier et al. 2008) and can be considerably more diverse than is currently perceived (Torruella, Moreira, and López-García 2017). We report previously undocumented diversity associated with the genera Cryptomonas and Chilomonas inside Cryptista, the naked filose amoebae of the genus Vampyrella (Endomyxa) and the frequently detected by 18S rRNA gene sequencing eukaryovorous biflagellate Aquavolon (Bass et al. 2018). No placement was recorded inside Excavata.