Microspora

Microsporidia MB specific primers (MB18SF: CGCCGGCCGTGAAAAATTTA and MB18SR: CCTTGGACGTGGGAGCTATC) were designed to target the Microsporidia MB 18S rRNA gene region and tested for specificity on a variety of Microsporidia-infected mosquito controls (including Hazardia, Parathelohania and Takaokaspora). For detection, the PCR reaction volume was 10 µl, consisting of 2 µl HOTFirepol^® Blend Master mix Ready-To-Load (Solis Biodyne, Estonia, mix composition: 7.5 mM Magnesium chloride, 2 mM of each dNTPs, HOT FIREPol^® DNA polymerase), 0.5 µl of 5 pmol µl⁻¹ of both forward and reverse primers, 2 µl of the template and 5 µl nuclease-free PCR water. The PCR cyclic conditions used were; initial denaturation at 95 °C for 15 min, further denaturation at 95 °C for 1 min, followed by annealing at 62 °C for 90 s and extension at 72 °C for a further 60 s, all done for 35 cycles. Final elongation was done at 72 °C for 5 min. Microsporidia MB was also quantified by qPCR using MB18SF/ MB18SR primers, with normalization against the Anopheles ribosomal S7 host gene^{56 (link)} (primers, S7F: TCCTGGAGCTGGAGATGAAC and S7R: GACGGGTCTGTACCTTCTGG). The primer efficiencies were determined using published methods^{57 (link)}. Samples were considered negative if the cycle threshold (Ct) value was greater than 35 or if the melt curve did not align with the positive control. These conditions were met for all negative control samples, which was DNA extracted from An. arabiensis from the insectary at icipe, Nairobi. The qPCR reaction volume was 10 µl, consisting of 2 µl HOT FIREPol^® EvaGreen^® HRM no ROX Mix (Solis Biodyne, Estonia, mix composition: 12.5 mM Magnesium chloride, EvaGreen^® dye, BSA, dNTPs, HOT FIREPol^® DNA Polymerase and 5× EvaGreen^® HRM buffer), 0.5 µl of 5 pmol µl⁻¹ of both forward and reverse primers, 2 µl of the template and 5 µl nuclease-free PCR water. Finally, melt curves were generated including temperature ranges from 65 °C to 95 °C. Standard curves were also generated to determine amplification efficiency. An. arabiensis were considered to be infected if Microsporidia MB was detected in either head and thorax or abdomen compartments. All reactions were carried out on a MIC qPCR cycler (Bio Molecular Systems, Australia).

Whole-genome sequences of 10 O. colligata samples, each from a different D. magna clone collected from a different population, were obtained from the illumina sequencing (Table 1). Sequences from clones FI-SK-17-1, NO-V-7, and GB-EP-1 were reused from Haag et al. (2020) (link) (NCBI database; SRA accession: SRP211974, Bioproject ID: PRJNA419750), sequences from clone RU-BAYA1-1 were reused from Angst et al. (2022) (link) (NCBI database; SRA accession: SRP346323, Bioproject ID: PRJNA780787), and for clone FI-SK-17-1 additional sequencing was done. We used the D. longispina-specific Ordospora pajunii for analyses that needed an outgroup (NCBI database; Bioproject ID: PRJNA630072; de Albuquerque et al. 2022 (link)). Some samples with observed microsporidia gut infection (i.e. based on published criteria for identifying O. colligata infection) showed <1× average whole-genome coverage for O. colligata. To verify the presumed parasite at species level, we performed polymerase chain reactions (PCRs) of the small subunit of microsporidian ribosomal DNA followed by Sanger sequencing (Supplementary Table 1 and Methods).

We used parasites derived from material collected within the framework of a large-scale biogeographic study of the host species, D. magna (Fields et al. 2015 (link), 2018 (link), 2022 (link); Seefeldt and Ebert 2019 (link)). From each population, animals were brought to the laboratory and one iso-female line, i.e. clone, was created. We checked these clones for infections with microsporidia by phase-contrast microscopy, using squash-preparations or samples of the gut. The panel includes whole-genome sequencing of D. magna clones with illumina paired-end reads using HiSeq 2500 and NovaSeq 6000 sequencers.

For calculating the average number of nucleotide differences (π) between O. colligata genomes, we used pixy v.0.95 (Korunes and Samuk 2021 (link)). Beforehand, we filtered out alleles with less than half or more than double the average sample coverage. Finally, we averaged the estimated π-values over one Kilobase pair (kb) windows. Similarly, we calculated the non-/synonymous per-site nucleotide diversity (π_N and π_S) for the coding sequences of O. colligata using the script selectionStats.py (https://github.com/tatumdmortimer/popgen-stats). Both methods are consistent with theoretical expectations and comparable among species as they take invariant sites into account for their calculations (Korunes and Samuk 2021 (link)). The ratio of π_N and π_S was separately calculated for BUSCO genes (Benchmarking Universal Single-Copy Orthologs) identified using BUSCO v.4.0.1 (Seppey et al. 2019 ) and its microsporidia_odb10 database (Creation date: August 05, 2020) and compared with published π_N/π_S values of H. tvaerminnensis (Angst et al. 2022 (link)).