A total of six containers (approx. 7 litres each) of sediment were collected on 9 June 2024 from three fishpond localities used for carp production hydrologically connected to the Blanice River in the South Bohemia Region of the Czech Republic (Fig. 1; Suppl. material 1), with two sampling sites per locality. The sites in the Vodňany area were selected as a case study to evaluate the applicability of our approach under a well-defined set of local conditions. Sampling was conducted a few metres from the bank at water depths of approximately 0.5–1 m. The sediment was transferred to the laboratory, gradually rinsed with tap water and sieved through meshes of various sizes (50–500 µm) to isolate sediment-dwelling invertebrates belonging to the Chironomidae and Oligochaeta groups. In total, 2,028 individuals were extracted from the sediment and classified into three operational sample categories/morphogroups: i) larvae of large chironomids (> 7 mm): 214 individuals (214 samples, individually processed); ii) larvae of small chironomids (< 7 mm): 718 individuals (145 samples, pooled 5 individuals per sample); and iii) oligochaetes: 1,096 individuals (220 samples, pooled 5 individuals per sample) (Suppl. material 1). Chironomids are particularly heterogeneous in larval morphology, size, colouration, habitat use and feeding biology. Field-collected chironomid larvae may, therefore, comprise several morphologically recognisable assemblage fractions, which do not necessarily correspond to formal taxonomic units, but may still reflect differences in larval composition and ecology. In the present study, we therefore distinguished two field-recognised chironomid morphogroups, based on body size, colouration and general larval appearance during sample processing, while treating this distinction as operational rather than taxonomic. The large chironomid morphogroup consisted mainly of conspicuous red “bloodworm-like” larvae, whereas the small chironomid morphogroup comprised smaller, often greenish, pale or reddish larvae. These morphogroups were retained throughout the analyses because they reflected the original sorting and sample-processing design.

The pooling strategy was based primarily on the amount of biological material available for DNA extraction. Large chironomid larvae were processed individually because a single individual provided sufficient tissue, whereas small chironomid larvae and oligochaetes were pooled (five individuals per sample) to obtain adequate DNA yield, while maintaining a practical balance between analytical resolution, laboratory workload and cost. Pooling of large chironomid larvae was avoided because it would have produced excessive amounts of tissue, reducing the efficiency of sample processing. Host taxonomic identity was not considered during pooling; individuals were grouped according to the predefined morphogroup categories described above. Each invertebrate sample pool was first mechanically homogenised using a pestle. DNA from the resulting tissue homogenates was then extracted using a modified protocol of Sellers et al. (2018), which employs bead beating (FastPrep-24™ 5G bead beating grinder and lysis system, MP Biomedicals) and silica beads (0.25–0.50 mm and 1.25–1.65 mm in diameter) to effectively disrupt resistant microsporidian spores.

For microsporidian detection, we applied a PCR protocol based on well-established and optimised primer sets targeting a ~ 400 bp fragment of the SSU rDNA spanning the V1–V3 regions. Specifically, we used the primers V1F (5’-CACCAGGTTGATTCTGCCTGAC-3’) and 530R (5’-CCGCGGCTGCTGGCAC-3’), developed by Baker et al. (1995) and Zhu et al. (1993), respectively. These primers have been successfully applied in multiple metabarcoding studies focused on microsporidian diversity (e.g. Williams et al. (2018); Chauvet et al. (2022)). Both primers were extended at their 5' ends with 10-bp multiplex identifier barcodes and a unique combination of forward and reverse MIDs was assigned to each sample, allowing sample identification after pooling. DNA extracts were amplified using barcoded primer pairs in 20 μl PCR reactions containing AccuPower PCR PreMix (Bioneer), 0.5 μl of each primer and 1 μl of template DNA. PCR consisted of an initial denaturation at 95 °C for 3 min, followed by 40 cycles (94 °C for 1 min, 58 °C for 30 s, 72 °C for 30 s) and a final extension at 72 °C for 10 min. A negative PCR control was included in each PCR run with barcoded primers to monitor potential contamination. Amplicons were purified and pooled to generate sequencing libraries. Further methodological details are provided in Lisnerová et al. (2023). Amplicon libraries were sequenced using paired-end 250 bp reads on an Illumina MiSeq platform (SEQme, Czech Republic). Forward and reverse reads were merged with FastqJoin (Aronesty 2013) and read quality was checked using the FASTX-Toolkit 0.0.13 (Gordon and Hannon 2010). Demultiplexing and barcode trimming were performed with a custom Python script (Lisnerová et al. 2023). Chimaera removal was done by UCHIME (Edgar et al. 2011). Operational taxonomic units (OTUs) were clustered at a 97% sequence similarity cutoff using USEARCH v.12 (Edgar 2013), generating the OTU table. To reduce false positives, only OTUs represented by at least 50 reads were retained for further analysis. This conservative threshold was applied to minimise the inclusion of spurious OTUs arising from sequencing artefacts, tag switching or low-level contamination, although it may also exclude some genuine low-abundance taxa. Finally, a BLASTn search was performed using the NCBI nucleotide database to identify each OTU in relation to the microsporidian group and then to a specific microsporidian group or species, if known.

The evolutionary relationships were inferred using a dataset comprising 66 newly-obtained microsporidian SSU rDNA sequences, combined with 565 previously published sequences representing diverse microsporidian lineages. Sequence alignment was performed with the E-INS-i algorithm implemented in MAFFT (Katoh et al. 2005) within Geneious Prime 2021.0.3. The alignment was manually refined and included a total of 2,648 nucleotide positions. The alignment length reflects the inclusion of near-full-length reference sequences; positions outside the approximately 400-bp regions covered by the newly-generated amplicons were treated as missing data.

Phylogenetic trees were reconstructed using the Maximum Likelihood (ML) approach in RAxML (Stamatakis 2006), applying the GTR+G substitution model selected by jModelTest (Darriba et al. 2012). Branch support was evaluated via 1,000 bootstrap replicates. For rooting the tree, sequences of the species Chytridiopsis typographi (NCBI: MH728789), Mitosporidium daphniae (NCBI: MF278562), Nucleophaga terricolae (NCBI: KX017226), N. amoebae (NCBI: JQ288099), Paramicrosporidium saccamoebae (NCBI: JQ796369), P. vannellae (NCBI: JQ796368) and Rozella allomycis (NCBI: NG_017174) were used. Sequence divergences were identified from proportional distances (in %) calculated in Geneious Prime, based on the dataset used for the ML analysis.

Venn diagrams were generated from OTU counts, separately for ponds and for invertebrate groups. For host-category comparisons, samples were analysed according to the original operational categories used during processing: large chironomid morphogroup, small chironomid morphogroup and oligochaetes. To assess alpha diversity, we calculated Shannon and Simpson diversity indices based on relative abundance data, also at the level of ponds and invertebrate groups. All calculations were performed in R software (R Core Team 2025 (version 4.5.1), Vienna (Austria)) using the vegan package (Oksanen et al. 2022). To visualise the taxonomic composition and relative read abundance of microsporidian OTUs across phylogenetic groups, Krona charts were generated using KronaTools (Ondov et al. 2011). To visualise and summarise the host affiliations of microsporidian OTUs, we constructed a Sankey diagram using the R packages ggalluvial (Brunson 2020), Tidyverse version 1.2.0 (Wickham et al. 2019) and dplyr (Wickham et al. 2023). For each OTU, we summed the number of sequencing reads separately for each host group, based on sample metadata. These absolute read counts were then transformed into relative abundances per OTU, i.e. the proportion of reads associated with each host group out of the total reads for that OTU. The relative proportions were used as weights in a two-level static Sankey diagram (Suppl. material 7).

To further investigate potential host associations of microsporidians previously reported from Daphnia spp., we selected samples containing OTU2 and OTU8 for barcoding of the host cytochrome c oxidase subunit I gene (COI) (Suppl. material 2). OTU2 was assigned to Conglomerata obtusa, whereas OTU8 showed high sequence similarity to Microsporidium sp. MIC1. Both OTUs had previously been reported from Daphnia hosts, but were detected in chironomids and/or oligochaetes in the present study. Host COI barcoding was, therefore, used to identify the invertebrate taxa present in positive samples and to assess their potential associations with additional benthic hosts. For OTU2, these analyses were additionally intended to provide data relevant to evaluating the indirect life cycle proposed for C. obtusa by Vávra et al. (2017, 2018). For COI barcoding, we used the primers of Folmer et al. (1994), amplifying a 710 bp fragment of the gene. PCR amplifications were performed in 20 μl PCR reactions containing AccuPower PCR PreMix (Bioneer), 0.5 μl of each primer and 1 μl of template DNA. The cycling conditions consisted of an initial denaturation at 95 °C for 3 min, followed by 35 cycles of 94 °C for 1 min, 52 °C for 50 s and 72 °C for 50 s, with a final extension at 72 °C for 8 min.

As individual specimens were combined into pools, amplicon sequencing was used to resolve the composition of COI amplicons within each pooled sample. Individuals were pooled according to their sample category (large chironomid morphogroup, small chironomid morphogroup or oligochaetes), sampling site and collection event. Amplicons were sequenced using Oxford Nanopore Technologies (ONT) MinION sequencing. Sequencing libraries were prepared with the Native Barcoding Kit 96 V14 (SQK-NBD114.96; Oxford Nanopore Technologies), following the manufacturer’s protocol for ligation sequencing of amplicons with native barcoding. Briefly, amplified COI products were subjected to end-preparation, ligation of sample-specific native barcodes, pooling of barcoded amplicons, adapter ligation and purification using AMPure XP beads. The final library was loaded onto a MinION Flow Cell MIN114/R10.4.1 and sequenced on a MinION device.

Raw sequencing data were acquired, basecalled using the super high accuracy basecalling model and demultiplexed in MinKNOW software with the SQK-NBD114.96 kit configuration. Demultiplexed FASTQ files were then processed separately for each barcode/sample. Reads were quality- and length-filtered using NanoFilt v.2.8.0 (De Coster et al. 2018), retaining reads with a minimum quality score of 18 and lengths between 650 and 800 bp, while trimming low-quality terminal regions when necessary. Filtered reads were converted to FASTA format using seqtk v.1.2-r94 (Li, GitHub repository) and dereplicated with VSEARCH v.2.15.2 (Rognes et al. 2016). Singletons were removed by retaining only sequences represented by at least two reads. Potential chimeric sequences were removed using de novo chimera detection in VSEARCH. The remaining non-chimeric sequences were clustered into operational taxonomic units (OTUs) at 97% sequence identity using VSEARCH, with clustering performed on both strands. Representative centroid sequences were exported in FASTA format and annotated with sample identity.

Taxonomic identification of chironomids and oligochaetes was performed by comparing representative COI OTU sequences with sequences available in GenBank. Species-level assignments were based on the closest matches, taking into account sequence similarity, query coverage and consistency with the expected invertebrate host diversity in the analysed samples (Suppl. material 2). The COI data were used to evaluate whether the two chironomid morphogroups represented identical or different larval assemblages. Morphological, ecological and basic biological information for the chironomid taxa detected by COI barcoding is summarised in the table provided in Suppl. material 5. This table clarifies the distinction between the two field-recognised chironomid morphogroups and supports their interpretation as operational assemblage fractions rather than formal taxonomic groups.

A total of 579 invertebrate samples (comprising 2,028 individuals) were screened for microsporidian infections. The invertebrates included larvae of chironomids and oligochaetes. Hereafter, prevalence refers to sample-level prevalence, calculated as the proportion of analysed samples (individual specimens or pooled samples) positive for microsporidians. Amongst large chironomids, microsporidian infection prevalence was the highest at Velký Ústavní pond site 2 (57%), followed by site 1 of the same pond (46%) and Malá Outrata pond site 1 (38%). In contrast, the lowest prevalence of microsporidians was recorded at Doktorovský pond site 2 (6%). Amongst small chironomids, microsporidian infection prevalence was highest at Velký Ústavní pond site 2 (75%), while the lowest was observed at Velký Ústavní pond site 1 (0%) and then at Doktorovský pond site 2 (14%). Oligochaetes showed the highest overall infection levels, with the highest site-specific microsporidian prevalence at Velký Ústavní pond site 2 (69%) and the lowest one at Malá Outrata pond site 1 (17%). Overall, microsporidian prevalence varied considerably both amongst sampling sites and between sampling events within individual sites. Despite this within-site variability, Velký Ústavní pond generally exhibited the highest prevalence (see Table 1 for details).

The OTU composition and relative read abundance of microsporidian lineages differed substantially across habitats and host groups (Fig. 4). When comparing the three ponds, community composition was consistently dominated by Enterocytozoonida, but with notable differences in its sublineages: Velký Ústavní pond (32 OTUs) exhibited the highest proportion of Nosematida reads (approximately 45%). Malá Outrata pond (26 OTUs) was also dominated by Enterocytozoonida in terms of read abundance (approximately 70% of reads). Doktorovský pond (28 OTUs) displayed a more even distribution, with Enterocytozoonida being dominant. Regarding invertebrate host groups, read abundance patterns again varied: Large chironomids (33 OTUs) and small chironomids (31 OTUs) showed highly similar patterns, both dominated by Enterocytozoonida and to a lesser extent by Amblyosporida and Neopereziida, with Nosematida contributing only minimally. Oligochaetes (43 OTUs) presented a distinct profile, characterised by a high abundance of Nosematida reads (approximately 60%) and reduced representation of other lineages. Glugeida was not detected in any of the samples.

Eighteen OTUs were unique to Doktorovský pond, 12 to Malá Outrata pond and 20 to Velký Ústavní pond (Fig. 5A). Only four OTUs were shared amongst all three ponds. Pairwise overlaps revealed four OTUs shared between Doktorovský and Malá Outrata ponds, six between Malá Outrata and Velký Ústavní ponds and only two between Doktorovský and Velký Ústavní ponds. In panel B, showing host group comparison, 11 OTUs were exclusive to large chironomids, eight to small chironomids and 20 to oligochaetes. Fourteen OTUs were shared across all three groups. Pairwise overlaps included four OTUs shared between large and small chironomids, five between small chironomids and oligochaetes and four between large chironomids and oligochaetes.

To further explore host associations of detected microsporidian lineages, we visualised the distribution of OTUs across host morphogroups using a Sankey diagram (Fig. 6). The overall structure of the diagram reveals that most OTUs show dominant associations with a single host group, most notably oligochaete worms. Specifically, 49 out of 66 OTUs had more than 90% of their reads associated with a single host morphogroup, whereas only a few OTUs were distributed more evenly amongst multiple host groups (Fig. 6).

Host associations also differed amongst the major microsporidian taxa (Fig. 6). OTUs belonging to the Nosematida clade, particularly the Encephalitozoonidae family (10 OTUs), were predominantly associated with oligochaetes, with nearly all sequencing reads derived from this host group. For 9 out of these 10 OTUs, more than 90% of their reads were associated with a single host morphogroup. One OTU associated with the Heterovesiculidae group (Nosematida) showed host specificity for large chironomids and one was associated with the Nosematidae (Nosematida). OTUs within the Amblyosporida clade were predominantly associated with large and small chironomids. In contrast, some phylogenetic groups were more host-generalist, with OTUs showing substantial read proportions across multiple host types (Neopereziida, Enterocytozoonida).

Based on pairwise distance analysis (Suppl. material 4), we identified 12 sequences with > 99% similarity to GenBank sequences, suggesting that they represent the same species as the corresponding reference sequences. Three OTUs matched described microsporidian species, four matched previously reported environmental sequences from freshwater ecosystems and five represented undescribed microsporidian lineages.

Three OTUs were assigned to already described microsporidian species: i) OTU2 was assigned to Conglomerata obtusa, based on high similarity to a reference sequence in GenBank (NCBI: MH645034). This species was originally found in Daphnia (Cladocera) hosts (Vávra et al. 2018). In our study, more than 99% of OTU2 reads were associated with small and large chironomids, while OTU2 was detected only rarely in oligochaetes; ii) OTU3 was assigned to Crispospora chironomi (NCBI: GU130407), originally described from Chironomus plumosus (Diptera) (Tokarev et al. 2010). In our analysis, this OTU showed the highest affinity to chironomids, with 83% of reads. This OTU was also rarely found in oligochaetes; iii) OTU36 matched Neoperezia chironomi (NCBI: HQ396519), also originally described from C. plumosus. In our case, 99% of the reads were assigned to small chironomids.

Four other OTUs (OTU13, OTU18, OTU34, OTU53) were assigned to environmental sequences previously reported from freshwater ecosystems (MZ405816, MZ405730, MZ405717, MZ405735). Four of the five undescribed lineages matched sequences of undescribed Microsporidium spp. with known previous host associations: i) OTU8 matched Microsporidium sp. MIC1 isolate (NCBI: FJ794861), previously reported from Daphnia and, in our analysis, 99% of reads were associated with chironomids; ii) OTU24 matched Microsporidium sp. BSEI3 isolate (NCBI: FJ756159), previously reported from Asprogammarus (Amphipoda) and, in our study, it was exclusively associated with large chironomids (100% of reads); iii) OTU27 matched Microsporidium sp. BLAT8 isolate (NCBI: FJ756057), previously reported from Brandtia latissima latior (Amphipoda) and, in our dataset, 100% of reads were assigned to oligochaetes; iv) OTU43 matched Microsporidium sp. BPAR5 isolate (NCBI: FJ756102), previously found in Dorogostaiskia parasitica (Amphipoda) and, in our analysis, 100% of reads were assigned to small chironomids. Lastly, OTU39 clustered within the Nosematidae clade and showed 100% read affiliation with large chironomids.

COI barcoding of selected samples positive for OTU2 and OTU8 identified several chironomid and oligochaete taxa potentially associated with these microsporidians (Suppl. material 2). OTU2 assigned to Conglomerata obtusa was detected in both large and small chironomids representing multiple genera (Suppl. materials 2, 5) and only rarely in oligochaetes (Limnodrilus hoffmeisteri). In several PCR-positive pools, COI barcoding identified a single chironomid species, allowing confident host assignment to Chironomus plumosus, C. annularius and C. curabilis. OTU8 matched Microsporidium sp. MIC1 isolate (NCBI: FJ794861), previously reported from Daphnia cucullata. In our analysis, 99% of reads were associated with chironomids representing multiple genera (Suppl. material 2). OTU8 was most strongly associated with C. plumosus and C. curabilis, whereas only low read counts were detected in C. annularius, Glyptotendipes paripes and the oligochaete Tubifex tubifex.

Infection prevalence of microsporidians varied across host groups and sites, reflecting a combination of ecological, biological and methodological factors (Willis and Reinke 2022). The highest prevalence was observed in oligochaetes (31%), consistent with the high microsporidian diversity previously reported from this invertebrate group (Oumouna et al. 2000) and supporting the hypothesis that oligochaetes may serve as important hosts and potentially reservoirs within pond ecosystems. As sediment-dwelling detritivores, oligochaetes remain in close contact with bottom substrates (Uzunov et al. 1988), where infective spores likely accumulate and their feeding behaviour may, thus, increase exposure to contaminated material (Oumouna et al. 2000). This mechanism is consistent with patterns reported for other benthic hosts (Knudsen et al. 2004). Microsporidian prevalence also varied between sampling events within individual sites. Although our sampling design does not allow us to distinguish between temporal and small-scale spatial effects, this pattern is consistent with the patchy distribution of parasite infections and fluctuations in host population composition and infection dynamics.

Alpha diversity (Shannon and Simpson indices) also varied amongst ponds and host morphogroups, with chironomids showing higher within-sample diversity than other examined groups. Oligochaetes appeared comparatively less diverse and were dominated by a smaller number of OTUs. This suggests that both pond-specific environmental conditions and host traits (habitat use, feeding ecology and exposure pathways) shape microsporidian assemblages and within-sample diversity.

COI barcoding revealed partial overlap between the two chironomid morphogroups, but also showed that they represented non-identical operational assemblages (Suppl. materials 2, 5). The large chironomid morphogroup consisted exclusively of Chironomus spp., i.e. large, haemoglobin-rich larvae typically associated with soft, organic and often oxygen-poor sediments. In contrast, the small chironomid morphogroup comprised a broader range of non-Chironomus taxa, including tube-building Tanytarsini, vegetation- and detritus-associated larvae and predatory Tanypodinae. The occurrence of Chironomus spp. in both morphogroups indicates that the small fraction also included earlier instars of larger Chironomus species. The two morphogroups should, therefore, be regarded as partly overlapping operational assemblage fractions rather than strictly distinct taxonomic or biological groups.

This broader ecological composition may account for the higher alpha diversity and more even OTU community observed in small chironomids, because this fraction included larvae occupying different microhabitats and feeding niches. Oligochaetes, by contrast, are sediment-dwelling detritivores that remain in close contact with organic bottom sediments, where resistant microsporidian spores may accumulate. This exposure route may explain their high infection prevalence and distinct read-abundance profile, particularly the strong contribution of Nosematida. Overall, differences amongst the host categories likely reflect a combination of host identity, developmental stage, larval ecology, feeding behaviour, exposure route and microhabitat use rather than taxonomic identity alone. As metabarcoding does not itself demonstrate active infection, these patterns should be interpreted as host-association and exposure signals that require confirmation through parasite localisation or transmission experiments.

Our results reveal a high diversity of freshwater microsporidians, with 66 OTUs detected across three fish ponds, of which 51 represent OTUs not previously reported in reference databases/previous studies. Notably, this diversity was recovered mainly from only two screened benthic invertebrate groups, chironomids and oligochaetes, highlighting how much microsporidian diversity can be uncovered even by targeted sampling of selected host taxa. This finding strongly supports growing evidence from recent tissue-based metabarcoding studies and freshwater eDNA surveys that a substantial portion of microsporidian diversity remains undocumented across ecosystems (Trzebny et al. 2020, 2022, 2023; Dubuffet et al. 2021; Chauvet et al. 2022). Traditional morphological or molecular single-marker approaches often fail to capture the full extent of diversity, highlighting the advantage of metabarcoding methods in detecting cryptic and novel lineages (Dubuffet et al. 2021; Chauvet et al. 2022; Lisnerová et al. 2023; Grabner and Fiala 2025). The diversity observed here underscores the ecological complexity of microsporidians and their potential roles in aquatic host communities, as well as the need for further integrative studies combining molecular data with ecological and life cycle investigations (Bojko et al. 2022; Grabner and Fiala 2025). Characterising this undocumented diversity is essential for clarifying host-parasite interactions and the evolutionary dynamics (e.g. phylogenetic relationships, co-evolutionary patterns) shaping freshwater systems.

The phylogenetic analysis of the detected microsporidian OTUs revealed several patterns. Most OTUs belonged to Enterocytozoonida, a group of parasite species infecting a broad range of protists, invertebrate and vertebrate hosts and occurring in diverse environments, including aquatic ones (Dubuffet et al. 2021; Bojko et al. 2022; Chauvet et al. 2022). In our dataset, some Enterocytozoonida OTUs were detected across multiple host groups, whereas others showed a predominant association with a single host group, suggesting that host specificity may vary considerably within this lineage (Bojko et al. 2022; Trzebny et al. 2023). Several OTUs clustered within Encephalitozoonidae, a few in Nosematidae and Heterovesiculidae (all Nosematida). Nosematida is a clade containing economically and medically important species, often characterised by complex life cycles, primarily involving terrestrial hosts (hymenopterans, butterflies and other insect groups), less frequently aquatic hosts (Keohane et al. 1999; Pilarska et al. 2015; Bojko et al. 2022). Their detection in our samples suggests that these lineages can also occur in freshwater environments and may represent previously uncharacterised aquatic diversity within Nosematida, potentially forming novel clades within this group (e.g. Dubuffet et al. (2021); Chauvet et al. (2022); Grabner and Fiala (2025)). The group Amblyosporida, which included 11 OTUs, is predominantly known from terrestrial and aquatic invertebrates, particularly insects, such as dipterans (Bojko et al. 2022). The group Neopereziida, which encompassed the least number of OTUs (5), includes lineages from terrestrial and/or freshwater environments (Bojko et al. 2022).

In general, a high number of OTUs within a clade may indicate broad host associations and/or high infection prevalence amongst the sampled taxa, suggesting that members of a lineage are common and ecologically significant in the studied ponds. However, OTU richness within a clade can also be influenced by sampling intensity and marker resolution and, thus, should be interpreted cautiously. In contrast, clades represented by few OTUs may reflect narrower host ranges, lower prevalence or incidental presence of spores. Low-OTU clades may also represent potentially poorly-characterised lineages, emphasising the utility of the metabarcoding approach for uncovering hidden parasite diversity. The detection in aquatic host tissues of OTUs closely related to GenBank sequences originating from terrestrial hosts (e.g. bees, lepidopterans and nosematid parasites) may reflect terrestrial–aquatic connectivity in microsporidian transmission, for example, via infection of aquatic stages of insects with terrestrial relatives, spillover from terrestrial hosts or introduction of spores into ponds through decomposing infected insects and other environmental inputs.

Our findings provide novel evidence supporting the hypothesis of Vávra et al. (2017, 2018) that Conglomerata and related taxa (Berwaldia) have an indirect life cycle involving additional non-crustacean hosts. OTU2 assigned to C. obtusa and OTU8 assigned to Microsporidium sp. MIC1 isolate have previously been reported from Daphnia hosts. In our study, both OTUs were predominantly detected in chironomids, suggesting that chironomids may act as alternative or additional hosts in these microsporidian life cycles. These results indicate that C. obtusa and Microsporidium sp. MIC1 isolate may be associated with a broader range of hosts than previously reported, including multiple chironomid species and, to a lesser extent, oligochaetes. Although low read counts in some species may suggest occasional infections or ingestion of spores, repeated detection across multiple individuals/samples within the same host and consistently higher read abundances in a subset of species (C. plumosus, C. curabilis and C. annularius for OTU2; C. plumosus and C. curabilis for OTU8) increase confidence that these represent true host associations rather than sporadic contamination. Both C. obtusa (OTU2) and Microsporidium sp. MIC1 isolate (OTU8) cluster phylogenetically within the Amblyosporida clade, a group which often involves species with alternating hosts, typically crustaceans and dipterans, such as mosquitoes (larval and adult stages) (Andreadis 1985; Sweeney et al. 1985). Further research, including transmission experiments and parasite localisation studies within host tissues, is needed to verify whether chironomids serve as active hosts and to resolve the specific transmission routes.

Two OTUs (OTU3 and OTU36) matched microsporidian species originally described from C. plumosus and both exhibited a strong association with chironomid hosts in our dataset. OTU3 was assigned to Crispospora chironomi (NCBI: GU130407) and was detected predominantly (and repeatedly across multiple individuals/samples) in large and small chironomid morphogroups. OTU36 matched Neoperezia chironomi (NCBI: HQ396519) and occurred almost exclusively in small chironomids. These findings are consistent with the original species descriptions (Issi and Voronin 1979; Tokarev et al. 2010) and support a strong chironomid affinity. Crispospora and Neoperezia represent two of the few described microsporidian lineages known to infect chironomids and both appear to have monoxenous life cycles (Tokarev et al. 2010; Issi et al. 2012). Our data reinforce the notion that these genera exhibit strong host affinities to chironomids, although detection of OTU3 in both chironomid morphogroups may reflect differences in host range, host developmental stage preference or ecological niche amongst parasite strains or closely-related lineages.

Several OTUs in our dataset (OTU24, OTU27, OTU43) were putatively assigned to Microsporidium spp. originally reported from amphipod hosts, yet they were found here in different taxa, including chironomids and oligochaetes. OTU24 matched a microsporidian sequence from Asprogammarus (NCBI: FJ756159), yet was exclusively detected in large chironomids. Similarly, OTU27, originally reported from the amphipod Brandtia latissima latior (NCBI: FJ756057), was exclusively associated with oligochaetes in our dataset. OTU43, previously found in Dorogostaiskia parasitica (NCBI: FJ756102), appeared only in small chironomids. These findings can be interpreted in several ways. First, detections in non-original hosts could result from environmental exposure, contamination or ingested material (e.g. passive transmission of spores via feeding) (Grabner 2017), particularly because chironomids and oligochaetes feed on detritus within sediments where spores may accumulate (Uzunov et al. 1988; Nogaro et al. 2008). Alternatively, these results may indicate broader host ranges or a more complex life cycle than previously recognised. Some microsporidian lineages exhibit heteroxenous life cycles involving crustaceans and insect larvae and adults (Andreadis 1985; Sweeney et al. 1985), raising the possibility of host alternation between amphipods and chironomids or oligochaetes. The exclusive detection of these OTUs within a single host group argues against sporadic carryover and is more consistent with stable host association. However, without parasite localisation (e.g. histological confirmation, in situ hybridisation) or experimental transmission data, it remains difficult to distinguish true host-parasite associations from incidental detections. This represents a general limitation of metabarcoding-based approaches for parasite detection. Therefore, metabarcoding should be regarded as a powerful screening tool that identifies candidate host–parasite associations, which subsequently require confirmation using complementary methods.

The four additional OTUs (OTU13, OTU18, OTU34, OTU53) matched environmental microsporidian sequences previously reported from freshwater ecosystems (Dubuffet et al. 2021). While these reference sequences lack confirmed host associations, their repeated detection in host tissues in our dataset suggests they represent undescribed freshwater microsporidian lineages that are consistently associated with pond biota. Given that the positive samples were predominantly from sediment-associated taxa, these OTUs may be linked to benthic habitats and/or benthic hosts, but confirming host identity and infection status will require parasite localisation and life cycle studies.

In this study, we used the primers V1F and 530R, originally developed by Baker et al. (1995) and Zhu et al. (1993) targeting the SSU rDNA gene. These primers are widely applied for microsporidian detection and generally provide reliable genus-level identification. They have been successfully used in multiple metabarcoding studies focusing on microsporidian diversity (e.g. Williams et al. (2018); Chauvet et al. (2022)). However, our results also indicate a known limitation of SSU rDNA metabarcoding, which lies in a restricted resolution amongst closely-related species within certain genera, notably Nosema. Consequently, OTU39, affiliated with Nosema (clustering near N. imperfecta and N. ceraces), could not be confidently assigned to a described species due to insufficient sequence variation within the relatively short amplified fragment (~400 bp of the complete 1,200–1,300-bp SSU rDNA). This limitation highlights the value of incorporating more variable markers, such as the ITS regions (Bojko et al. 2022) or sequencing longer SSU rDNA fragments (Bojko et al. 2022; Stratton et al. 2024), to increase taxonomic resolution and improve discrimination amongst Nosema species. Despite this limitation, the SSU rDNA still enabled robust ecological inference by linking OTU39 to large chironomid hosts, indicating that even genus-level identifications can provide meaningful host-association patterns. In contrast, for other genera in our dataset, this marker was sufficient to discriminate species-level differences, underscoring that its resolving power varies across taxonomic groups.