The study was conducted during the 2026 breeding season at a commercial Trakehner stud farm. Twenty-one mares were enrolled, and all samples were taken during naturally occurring estrus, before the first breeding procedure of the sampled cycle in mares that were subsequently mated. At the time of sampling, no mare had ultrasonographically detectable intrauterine fluid. Because cytology and histopathology were not performed, the authors describe the mares as having no overt clinical uterine disease rather than as having a clinically healthy endometrium. Pregnancy was assessed by transrectal ultrasonography approximately 14 days after the final breeding of the studied cycle, and mares were classified as pregnant or non-pregnant through the end of the breeding season.

Four pooled sequencing libraries were constructed. A pregnant-LVL pool combined low-volume uterine lavage (LVL) samples from five naturally mated mares that established pregnancy, while a non-pregnant-LVL pool combined LVL samples from seven naturally mated mares that did not establish pregnancy and remained non-pregnant through the end of the 2026 season. A paired sampling-method comparison used LVL and double-guarded uterine swab samples from the same five mares, with swabs always collected before lavage to avoid altering the endometrial surface. In total, the four pooled libraries represented 13 unique mares, and eight enrolled mares were not represented in the sequencing libraries. Each condition was represented by a single pooled library, which meant the mare could not be retained as the experimental unit and all analyses were pre-specified as descriptive and hypothesis-generating.

Sequencing was performed on an Illumina NextSeq 1000 platform by a commercial provider. After trimming, denoising, chimera removal and feature filtering, the four libraries retained between 82,862 and 130,870 reads and between 27 and 111 final unique amplicon sequence variants (ASVs). Quantitative PCR targeting the ITS2 region, calibrated with plasmid standards, detected between 1,107 and 1,728 target copies per microliter of pooled DNA extract. The authors stress that these values quantify the amplified ITS2 target and cannot be converted to fungal genome counts or viable fungal burden, because ribosomal DNA copy number varies substantially among fungal taxa.

Taxonomic assignments were dominated by the phylum Ascomycota, which accounted for 50.7 to 72.5 percent of assignments across the four libraries, followed by Basidiomycota at 8.1 to 24.1 percent. The genus Cladosporium appeared in every pool at 13.5 to 17.6 percent relative abundance. The pregnant-LVL profile contained Microascus, Sarocladium, Wallemia and Debaryomyces among its dominant genera, whereas the non-pregnant-LVL and paired LVL libraries featured Alternaria and Malassezia, and the paired swab library contained Aspergillus, Alternaria and Sarocladium. An independent comparison of principal ASVs against the UNITE fungal reference database showed higher concordance at genus level (79.1 percent among provider-resolved assignments) than at species level (50.0 percent), leading the authors to treat species-level assignments as provisional.

Diversity analysis at a rarefaction depth of 19,970 sequences showed that the pregnant-LVL library had the lowest pool-level richness and Shannon diversity, with mean values of 26.7 ASVs and 3.502 respectively, while the non-pregnant-LVL library had the highest, at 110.8 ASVs and 5.818. Principal coordinates analysis of Bray-Curtis dissimilarity placed the pregnant-LVL library apart from the others primarily along the first principal coordinate, which explained 66.5 percent of the variation. The non-pregnant-LVL and paired LVL libraries occupied the most similar positions, with the lowest pairwise dissimilarity of 0.244, while the paired swab library differed from its LVL counterpart mainly along the second coordinate. The greatest dissimilarity, 0.610, was between the pregnant-LVL and paired LVL libraries.

Conventional microbiological culture provided clinical context but told a different story. Routine mycological culture on chloramphenicol-containing Sabouraud agar, incubated aerobically at 30 degrees Celsius for up to seven days with appropriate sterility and growth-promotion controls, yielded no fungal growth in any of the available reports. Named bacterial isolates, including Streptococcus equi subsp. zooepidemicus, were recovered from some samples. The authors emphasize that negative cultures indicate only that no fungal growth was detected under the applied conditions; they do not exclude viable fungi present at low abundance, slow-growing organisms, or fungi requiring different nutritional or atmospheric conditions. Conversely, DNA detection alone cannot establish viability, metabolic activity, endometrial adherence, inflammation or pathogenicity.

The study also confronted the persistent challenge of low-biomass microbiome work. Neither the extraction blank nor the library-preparation blank crossed the qPCR detection threshold within 40 cycles, yet both yielded sequencing reads. The dominant control ASV, assigned to Saccharomyces cerevisiae and identical to the dominant positive-control ASV, was absent from the biological libraries and was considered compatible with low-level index misassignment or sequencing cross-talk. A second ASV, assigned to Malassezia restricta, appeared in the library-preparation blank and in two biological libraries. A targeted sensitivity analysis excluding this feature produced only localized changes and did not materially alter taxonomic composition, diversity or ordination patterns. Because no field or sampling blanks were collected, the authors conclude that the uterine origin of all detected fungal DNA signals cannot be established conclusively.

Interpretation is further constrained by the design itself. Pooling eliminated individual-mare variation and prevented estimation of within-group variability, taxon prevalence, effect sizes and confounder-adjusted associations. Mares contributing to the pregnant pool were younger, had lower parity and a shorter postpartum interval than those in the non-pregnant pool, and age and parity are known influences on uterine clearance in mares, making them biologically relevant potential confounders. The fixed sampling order in the paired comparison also meant that a sampling-order effect could not be separated from the effect of sampling method. The authors therefore state that the observed differences do not establish a pregnancy-associated fungal profile, nor superiority, equivalence or interchangeability between lavage and swabbing.

The findings align with a recent ITS2-based study of pooled uterine lavage samples from postpartum beef cows, in which Ascomycota also dominated, suggesting that culture-independent fungal profiling is feasible in livestock uterine samples across species. However, differences in host, reproductive stage and study design preclude direct biological comparison. The authors call for future work that sequences individual mares, includes sufficient biological replication, sequences comprehensive field and laboratory controls, uses publicly accessible and versioned reference databases, applies standardized quantitative mycological culture, and integrates cytological and histopathological data. Until then, the detected fungal DNA signals remain an intriguing, hypothesis-generating glimpse of a poorly characterized component of the equine uterine environment rather than evidence of colonization, infection or a link to fertility.

Subject of Research: ITS2 amplicon sequencing of fungal DNA in pooled equine uterine samples from mares with and without subsequent pregnancy

Article Title: Fungal DNA in pooled equine uterine samples: an exploratory ITS2 amplicon sequencing pilot study

Article References: Pelenė, U., Nainienė, R., Šveistienė, R., Merkelytė, I., & Šiukščius, A. (2026). Fungal DNA in pooled equine uterine samples: an exploratory ITS2 amplicon sequencing pilot study. Frontiers in Veterinary Science, 13, 1978506. https://doi.org/10.3389/fvets.2026.1978506

Image Credits: AI Generated

DOI: 10.3389/fvets.2026.1978506

Keywords: equine reproduction, mare uterus, fungal DNA, ITS2 amplicon sequencing, mycobiome, low-volume uterine lavage, double-guarded swab, pregnancy outcome, endometritis, pilot study, contamination controls, veterinary science