The study was conducted on a sheep farm located in Gökçeyazı Village, Gürün District, Sivas Province, Türkiye (latitude: 38.871021° N; longitude: 37.306164° E; altitude: 1790 m). The farm is situated in a mountainous steppe region characterised by extensive natural pastureland.

To comprehensively investigate the causes of infertility, a sheep flock with a history of reproductive failure in previous years was selected for the study. At the beginning of the breeding season, a total of 905 multiparous Kangal ewes (aged 3–6 years, weighing 58–66 kg, with a documented history of fertility) and 44 Kangal rams (aged 4–6 years, weighing 112–128 kg, with proven fertility) were included. Prior to mating, all ewes underwent ultrasonographic examination and evaluation of vaginal discharge. Discharge characteristics were assessed. Ewes with purulent, mucopurulent, or serosanguineous discharge were excluded from the study. Ewes with odourless discharge free of particulate content were included in the study. Following this screening, 880 ewes with normal vaginal discharge and no pathological abnormalities of the ovaries or uterus were monitored throughout the breeding season. Matings were recorded. Following each mating, pregnancy examinations were performed on the ewes via transrectal ultrasonography at intervals of 16–21 d. After the first mating, 171 ewes that failed to conceive underwent a second mating, resulting in 50 additional pregnancies. The remaining 121 non-pregnant ewes were mated for a third time, resulting in 36 further pregnancies. Blood samples were collected from the 85 ewes that remained non-pregnant after three consecutive matings. Of these, 21 tested positive for Coxiella burnetii or border disease virus (BDV) and were sent to slaughter. Vaginal swab samples were randomly collected from 25 of the remaining 64 ewes that tested negative for Brucella spp., Coxiella burnetii, and BDV. These animals were classified as repeat breeders and designated as Group 2. In addition, 25 multiparous ewes of similar age and body weight that conceived after a single estrous cycle were classified as healthy and assigned to the healthy control group (Group 1). These animals were randomly selected from the fertile ewes in the flock that conceived following a single estrous cycle. Preputial swab samples were collected from 15 of the 44 breeding rams used during the breeding season. The study population comprised 25 fertile control ewes, 25 repeat breeder ewes, and 15 breeding rams selected from the same flock according to the sampling procedures described above.

Vaginal swab samples were collected from 25 of the 64 ewes that tested negative for Coxiella burnetii, Brucella spp., and BDV. In addition, preputial swab samples were collected from 15 breeding rams. Duplicate swab samples were transported to the microbiology laboratory at approximately 4 °C immediately after collection.

Swab samples were serially diluted in Mycoplasma broth (PPLO broth without crystal violet; Becton, Dickinson and Company (BD), Franklin Lakes, NJ, USA) and incubated under microaerophilic conditions with 10 % CO₂ for 1 week. The broth cultures were examined daily for turbidity before being subcultured onto Mycoplasma agar (BD, USA), following general culture procedures described by Nicholas and Baker (1998) and WOAH (2018).

Following inoculation of each swab sample into a microcentrifuge tube containing 1 mL of sterile saline solution, DNA was extracted using a phenol:chloroform:isoamylalcohol (25:24:1; Fluka) extraction procedure, based on the general DNA extraction methodology described by Bashiruddin (1998). The extracted DNA was subsequently diluted with distilled water and stored at -20 °C until further analysis.

To identify the causative agents associated with repeat breeder cases, PCR analyses were performed using the primer sets listed in Table 1. Briefly, 50 µL PCR reaction mixtures were prepared containing 5 µL of 10× PCR buffer (500 mM KCl, 100 mM Tris-HCl (pH 9.1), and 0.1 % Triton X-100; Vivantis, ViBuffer A), 5 µL of 25 mM MgCl₂ (Vivantis), 250 µM of each dNTP (Vivantis, NP2406, 100 mM), 20 pmol of each primer pair (Macrogen, South Korea), 1.25 U of Taq DNA polymerase (Vivantis PL1202, 500 U), and 5 µL (25 ng) of template DNA. DNA amplification was performed using a T100 Thermal Cycler (Bio-Rad Laboratories, USA), following the amplification conditions reported in the corresponding methodological references listed in Table 1.

PCR products were separated by electrophoresis on a 1.5 % (w/v) agarose gel, stained with ethidium bromide for 30 min, and visualised using a Quantum ST4 gel documentation system (Vilber Lourmat, France), following the general procedure described by Lee et al. (2012).

Vaginal swab PCR results were coded as binary outcomes (positive = 1, negative = 0). Culture results recorded as “antibiotic-resistant growth” or fungal contamination were considered indeterminate and excluded from the corresponding denominators. Between-group differences in pathogen prevalence were assessed using two-tailed Fisher's exact test. Because eight separate pathogen-specific comparisons were performed between repeat breeder and fertile control ewes, the Bonferroni correction was applied to control the familywise type I error rate associated with multiple hypothesis testing. Bonferroni-adjusted p values (padj) are reported throughout. Odds ratios (ORs) with 95 % confidence intervals (CIs) were estimated using the Haldane–Anscombe correction by adding 0.5 to each cell to accommodate zero-cell contingency tables. For each animal, the total number of PCR-detected pathogens was used as a co-infection burden score. Differences in this score among the three groups were evaluated using the Kruskal–Wallis test, followed by a post-hoc Mann–Whitney U test for pairwise comparison between Group 2 and Group 1. Wilson's score method was used to calculate 95 % CIs for individual proportions. All statistical analyses were performed using Python (version 3.12; Python Software Foundation, USA), with SciPy (version 1.13) and statsmodels (version 0.14).

A total of 65 genital swab samples were examined by PCR, including 25 from Group 2 (repeat breeder ewes), 25 from Group 1 (clinically healthy ewes), and 15 from breeding rams used for natural mating. The proportions of PCR-positive animals for each of the eight target pathogens are presented in Table 2 and illustrated in Fig. 1.

Among the 15 rams, Campylobacter spp. was detected most frequently (5/15; 33.3 %), followed by Mycoplasma spp. (4/15; 26.7 %), T. pyogenes (2/15; 13.3 %), and Staphylococcus spp. (2/15; 13.3 %). The detection of Mycoplasma spp. in more than one-quarter of the breeding rams is epidemiologically noteworthy. Although the difference in Mycoplasma spp. prevalence between rams and healthy ewes did not survive Bonferroni correction (padj = 0.12), the magnitude of positivity (26.7 %) and the biological plausibility of venereal transmission suggest that rams may serve as a reservoir and transmission vector for Mycoplasma spp. in this flock. Individual heatmap profiles (Fig. 2) confirm that several rams were positive for the same pathogen combinations identified in diseased ewes.

In summary, the findings provide strong statistical evidence that Mycoplasma spp., particularly M. bovigenitalium, is the pathogen most significantly associated with reproductive tract disease in ewes (Fig. 4). Direct PCR performed on vaginal swab samples is the most sensitive single diagnostic modality and should be considered the method of choice for field investigations of ovine vaginal discharge.

The prevalences of Campylobacter spp. (40.0 % vs. 32.0 %), Staphylococcus spp. (16.0 % vs 36.0 %), Streptococcus spp. (8.0 % vs 20.0 %), Salmonella spp. (4.0 % vs 24.0 %), and E. coli (16.0 % vs 40.0 %) did not differ significantly between Group 2 and Group 1 after correction (all padj > 0.78). Notably, Staphylococcus spp., Salmonella spp., and E. coli were numerically more prevalent in healthy ewes than in diseased ewes, consistent with their role as commensal members of the ovine genital tract microbiota rather than primary pathogens in this context.

Among Group 2 ewes, direct swab-based universal PCR-detected Mycoplasma spp. in 21/25 animals (84.0 %), substantially outperforming both Mycoplasma broth culture (12/25; 48.0 %) and agar culture (12/25; 48.0 %). Of the 12 broth culture-positive samples subsequently subjected to PCR, all 12 (100 %) yielded a positive universal PCR result, and 11/12 (91.7 %) were confirmed as M. bovigenitalium by species-specific PCR (Fig. 3A). These findings indicate the near-complete dominance of M. bovigenitalium among culturable mycoplasma isolates from diseased ewes, whereas direct swab-based PCR identified an additional nine cases that were missed by culture alone. These data demonstrate that conventional culture underestimates the true prevalence of mycoplasmal infection in sheep with reproductive disease.

Mycoplasma broth cultures from nine of the 15 rams (60.0 %) were overgrown by antibiotic-resistant bacteria, rendering them non-evaluable and precluding culture-based assessment in the majority of the ram cohort. Direct PCR performed on ram swab samples was therefore the only reliable diagnostic modality for this group.

The mean number of PCR-detected pathogens per animal differed significantly among the three groups (Kruskal–Wallis H=9.52, p=0.009; Fig. 3B). Group 2 ewes harboured the highest co-infection burden (mean ± SD: 2.60 ± 1.61; range 0–6), followed by Group 1 healthy ewes (1.76 ± 1.16; range 0–5) and rams (1.13 ± 1.25; range 0–4). The pairwise difference between Group 2 and Group 1 approached, but did not reach, statistical significance (Mann–Whitney U=408, p=0.059), likely reflecting the broad variability within both groups. Seven (28.0 %) diseased ewes were co-infected with three or more pathogens simultaneously, compared with three (12.0 %) healthy ewes.

Mycoplasma spp. was the pathogen most strongly associated with vaginal discharge. It was detected by universal PCR in 21 of 25 (84.0 %) diseased ewes (Group 2), whereas none of the 25 healthy ewes (Group 1) tested positive (padj < 0.001; Fig. 4). M. bovigenitalium was identified in 10/25 (40.0 %) diseased ewes and, likewise, was not detected in any of the healthy ewes (padj = 0.005). Among the rams, Mycoplasma spp. was detected by PCR in 4/15 (26.7 %) animals, of which one (6.7 %) tested positive specifically for M. bovigenitalium.

T. pyogenes was detected in 13/25 (52.0 %) Group 2 ewes compared with 6/25 (24.0 %) Group 1 ewes (OR = 3.43, 95 % CI 1.00–10.5; padj = 0.634; Fig. 4). Although this difference did not reach statistical significance after Bonferroni correction, reflecting the limited statistical power associated with the sample size, the direction and effect size were consistent with a pathological role for this organism in the reproductive tract of diseased animals.

The 84.0 % detection rate of Mycoplasma spp. in Group 2 ewes exceeds the 30 %–60 % prevalence reported in earlier culture-based studies of ewes with vulvovaginitis or vaginal discharge (Cottew, 1984; Nicholas et al., 2008). This discrepancy is most likely attributable to the diagnostic limitations of conventional culture rather than to a true difference in pathogen burden, as direct-swab PCR identified nine additional positive cases that were not detected by culture. Mycoplasma bovigenitalium was first isolated from the ovine genital tract in association with mild vulvovaginitis and endometritis by Jones (Jones et al., 1985), and subsequent molecular surveys have identified it as one of the most prevalent mycoplasmal pathogens of the sheep reproductive tract across geographically diverse populations (Halium et al., 2019; Stipkovits and Kempf, 1996). Its complete absence from all 25 healthy control ewes, yielding an undefined odds ratio, argues against a commensal or incidental carrier status and supports a pathogenic role in ovine reproductive tract disease.

Two features distinguish the present findings from those of previous descriptive surveys. First, the pathogen was detected in both ewes and rams within a single infertile flock. Second, it consistently co-occurred with Trueperella pyogenes in the diseased phenotype. Mycoplasmal species have been reported to be entirely absent from microbiome profiles obtained throughout healthy ovine pregnancies (Cassas et al., 2024), reinforcing the interpretation that their detection in non-pregnant ewes reflects a pathological rather than a normal colonisation state. Collectively, these observations provide, to the best of our knowledge, the first systematic characterisation of M. bovigenitalium as a contributor to reproductive failure in a naturally managed sheep flock under field conditions.

The reservoir for this pathogen appears to extend to rams. Mycoplasma spp. were detected in 26.7 % of rams, with one-quarter of the positive animals confirmed as M. bovigenitalium, consistent with the well-established ability of rams to harbour mycoplasmas within the prepuce and urogenital tract as long-term asymptomatic carriers (DaMassa et al., 1992; Nicholas, 2002). The concordance between the pathogen profiles of rams and ewes (Fig. 2), together with the exclusive use of natural mating in this flock, suggests that venereal transmission is a plausible route of infection. This interpretation is further supported by an Australian study involving 629 cattle, in which mycoplasmal colonisation, predominantly M. bovigenitalium, increased following natural mating, consistent with venereal acquisition (Hazelton et al., 2020). The same transmission dynamic is biologically plausible in sheep and warrants further investigation in prospective studies.

The co-occurrence of T. pyogenes and M. bovigenitalium emerged as a consistent feature of the diseased phenotype. T. pyogenes (formerly Arcanobacterium pyogenes) was detected in 52.0 % of diseased ewes compared with 24.0 % of healthy ewes (OR = 3.43, 95 % CI 1.00–10.5). It is a well-characterised opportunistic pathogen of the ruminant reproductive tract and is routinely isolated from cases of metritis, endometritis, and pyometra in cattle, sheep, and goats, where it typically acts synergistically with other bacteria following primary mucosal disruption (Jost and Billington, 2005; Rzewuska et al., 2019). In cattle, its synergistic interaction with M. bovigenitalium following a complicated puerperium has been shown to sustain subclinical endometritis (Osawa, 2021). The present study identified a comparable pattern in sheep, with the two organisms co-detected in the vaginal swabs of most infertile ewes and jointly isolated from the preputial swab of one ram. To the best of our knowledge, this co-infection pattern has not previously been described in ovine populations. The moderate prevalence of T. pyogenes in healthy ewes (24.0 %) is consistent with commensal carriage within the normal genital microbiota, with pathogenicity likely emerging under conditions of immunosuppression or concurrent mycoplasmal infection that compromises mucosal barrier integrity, a model supported by the co-infection data presented in the present study.

In contrast, Campylobacter spp., Staphylococcus spp., Streptococcus spp., Salmonella spp., and Escherichia coli showed no significant differences in prevalence between the groups, suggesting that these organisms were not associated with repeat breeding in the present flock. Campylobacter spp. were detected in approximately one-third of animals across all groups, a prevalence comparable to that reported in apparently healthy sheep populations. Although certain species, particularly C. fetus subsp. fetus, remain important causes of ovine abortion under specific flock conditions (Mshelia et al., 2010; Sahin et al., 2017), species-specific PCR did not detect C. fetus subsp. fetus, C. fetus subsp. venerealis, or C. jejuni in the present study. These findings suggest that the Campylobacter spp. detected were unlikely to have contributed to the reproductive failure observed, although broader species-level characterisation would be valuable in future investigations.

Two of the principal associations – the difference in Mycoplasma spp. prevalence between rams and ewes (padj = 0.12) and the increased prevalence of T. pyogenes in diseased ewes (padj = 0.634) – did not remain statistically significant after Bonferroni correction. This outcome most likely reflects the conservative nature of the correction in combination with the moderate sample size, both of which reduce the statistical power to detect effects of the observed magnitude, rather than indicating a lack of biological relevance. The effect size for T. pyogenes (OR ≈ 3.4) is comparable to associations considered aetiologically relevant in the bovine endometritis literature, while the concordance between rams and ewes is further supported by the observed co-infection patterns and the proposed venereal transmission framework described above. These associations should therefore be regarded as biologically and clinically meaningful findings that warrant confirmation in adequately powered cohorts and are interpreted throughout the present study accordingly, rather than as definitive evidence of causality.

The 36 percentage-point difference between the PCR (84.0 %) and culture (48.0 %) detection rates for Mycoplasma spp. represents the most clinically relevant diagnostic finding of the present study. Mycoplasmas present inherent challenges for culture, including fastidious growth requirements, prolonged generation times, and susceptibility to overgrowth by contaminating flora, all of which render conventional culture unreliable under field conditions. This limitation was reflected in the present study by the 60 % non-evaluable rate among ram cultures due to overgrowth by antibiotic-resistant bacteria. This pattern is consistent with findings in the bovine and caprine literature, where PCR-based methods have repeatedly demonstrated superior sensitivity compared with culture (Nicholas and Ayling, 2003; Wisselink et al., 2019). Species-specific PCR confirmed M. bovigenitalium in 91.7 % of the culture-confirmed broth samples, establishing this species as the predominant culturable mycoplasmal isolate in clinically affected ewes and a clear diagnostic target for future flock surveillance. These findings indicate that direct-swab PCR should replace culture as the first-line diagnostic method for field investigations of ovine vaginal discharge and that negative culture results should not be used to exclude a mycoplasmal aetiology.

Beyond detection sensitivity, the co-infection burden itself distinguished the diseased animals. Diseased ewes harboured a higher mean pathogen load than healthy ewes (2.60 vs 1.76 pathogens per animal; Kruskal–Wallis, p=0.009), and 28 % carried three or more pathogens simultaneously, more than twice the proportion observed in the healthy controls. These findings suggest that the most severe clinical presentations are characterised by polymicrobial rather than single-pathogen infections, consistent with a cascade model in which primary mycoplasmal infection compromises mucosal immune defences and facilitates secondary colonisation by opportunistic organisms such as T. pyogenes and E. coli – analogous to the mechanisms described in bovine metritis (Sheldon et al., 2009). Clinically, this profile suggests that antimicrobial regimens directed against a single pathogen are unlikely to be sufficient; instead, combination treatment strategies targeting both the mycoplasmal and opportunistic bacterial components should be evaluated in future studies.

Several limitations should be considered when interpreting these findings. First, the cross-sectional study design precludes temporal inference regarding the sequence of pathogen acquisition and the onset of clinical disease and does not permit definitive conclusions about causality. Longitudinal studies incorporating pre-mating and serial post-mating sampling are therefore required to determine whether mycoplasmal colonisation precedes reproductive failure or develops as a secondary consequence of impaired uterine immunity. Second, the single-flock design limits the generalisability of the prevalence estimates to other geographic regions and production systems. Third, the absence of sequencing-based strain characterisation or multilocus sequence typing (MLST) limited our ability to provide direct molecular evidence of venereal transmission between rams and ewes, and these approaches should be prioritised in future investigations. Finally, the moderate sample size (n=65) reduced statistical power, contributing to wide confidence intervals and preventing marginal associations from reaching statistical significance after correction. Larger cohorts combined with whole-metagenome sequencing of the reproductive tract microbiome throughout the mating cycle would substantially enhance our mechanistic understanding of host–pathogen interactions underlying ovine reproductive failure.