The ecological and evolutionary outcomes of plant–fungal interactions are strongly influenced by genome size and ploidy, yet the ploidy level of both partners is rarely assessed simultaneously. Epichloë symbioses with Pooideae grasses are established model systems for exploring these dynamics, but associations between polyploid hosts and haploid endophytes remain poorly documented. In this study, the association of the Iberian endemic Festuca rothmaleri—which includes tetraploid, hexaploid, and octoploid cytotypes—with Epichloë fungal endophytes is documented for the first time. An integrative, method-rich framework combining cytogenetics, morphometrics, and multilocus phylogenetics revealed a strikingly asymmetric interaction, with all cytotypes harboring a single haploid strain of Epichloë festucae. Two methodological innovations were developed: (i) an image-based tool for automated measurement of asexual structures, including the novel metric “conidial area,” and (ii) a flow cytometry protocol for estimating fungal genome size. Despite morphological variability, all fungal isolates shared similar genome sizes and formed a well-supported monophyletic lineage in a coalescent species tree based on nuclear loci sequences (actG, CalM, ITS, tefA, tubB). This work provides the first comprehensive characterization of a haploid Epichloë endophyte spanning multiple naturally distributed host polyploid levels and highlights a rare but promising system for future evolutionary, physiological, and ecological studies of plant–fungal interactions.

Sampling and characterization of host plants

Sampling and taxonomic identification. Individuals of Festuca rothmaleri representing three putative ploidy levels (tetraploid, hexaploid, and octoploid) were collected from three mountainous sites in northwestern Spain, Montemayor del Río (Mon), Candelario (Can), and El Cabaco (Cab) (Salamanca province), located 20–50 km apart and separated by mountain ranges (Fig. S1). Between 20 and 35 individuals were sampled per site during two consecutive flowering seasons (June 2022 and June 2023). To reduce the likelihood of clonal sampling due to the rhizomatous growth habit of the host species, individuals were collected at a minimum distance of 3 meters from one another. Plants were transplanted into pots with universal substrate (Blumenerde, Gramoflor) and maintained in a greenhouse under relatively constant temperature conditions (22–27 °C) and watering regimes (adjusted according to seasonal needs, on average three times per week) throughout the study.

Taxonomic identification of the host plants was performed based on morphoanatomical examination and measurements of vegetative and reproductive structures, following the diagnostic criteria and identification keys provided by De la Fuente and Sánchez (1987), Al-Bermani et al. (1992), De la Fuente et al. (2001), Loureiro et al. (2007), and Devesa et al. (2020).

Cytogenetic characterization. The host plant genome size was determined by flow cytometry (Ploidy Analyzer, Sysmex) in fresh mature leaf tissue, following the protocol and reagents (Otto I and Otto II) described by Doležel et al. (2007). A preliminary analysis confirmed the existence of three genome size categories corresponding to different ploidy levels. Based on these results, up to 15 individuals were selected per population of origin and ploidy level (Mon4x, Can4x, Cab6x, and Can8x), analyzing two technical replicates in each case. Measurements were made on more than 5,000 nuclei with a coefficient of variation (CV) < 3%. Primary standards included Solanum lycopersicum ‘Stupické polní rané’ (1.96 pg/2C) for tetraploids, Pisum sativum ‘Ctirad’ (9.09 pg/2C) for octoploids, and Secale cereale ‘Daňkovské’ (16.19 pg/2C) for hexaploids according to Doležel et al. (2007).

Chromosome number was determined from meristematic cells of the root tip using the protocol of Jenkins and Hasterok (2007). For each ploidy level, five independent counts were performed for each of three randomly selected individuals from the genome-size dataset, resulting in a total of 15 counts per detected ploidy level. Images were obtained at 400× magnification using a Zeiss Axio Lab.A1 phase-contrast microscope equipped with a Canon EOS 2000D digital camera.

Comprehensive multi-method characterization of Epichloë endophytes

Detection, isolation and incidence. The presence of the endophyte in the host's aerial tissues was initially detected by aniline blue staining (Florea et al. 2015). In parallel, fragments of floral stems or the bases of vegetative tillers from all collected specimens, with the surface disinfected, were cultured on potato dextrose agar (PDA, Potato Dextrose Agar EP/USP/BAM, Condalab) plates containing chloramphenicol (Chloramphenicol BioChemica, PanReac AmpliChem; 25 µg/ml) to inhibit bacterial growth. When an Epichloë-like endophyte emerged from the host tissue, it was isolated on PDA plates and cultured at room temperature (22–25°C) in the dark. The incidence of *Epichloë *endophytes in Festuca hosts varies by holobiont and locality (Zabalgogeazcoa et al. 1999; Saikkonen et al. 2000; Clement et al. 2001; Bazely et al. 2007; Gundel et al. 2014). The proportion of infected individuals was estimated by sampling site (Mon, Can, Cab), host ploidy level (4x, 6x, 8x), and their combination (Mon4x, Can4x, Cab6x, Can8x), considering both detection methodologies.

Exploratory mating-type composition screening. During field collections, no stromata were observed in F. rothmaleri individuals. Therefore, the sexual reproduction potential of the endophyte was evaluated through exploratory PCR screening for the MAT1-1 (785 bp) and MAT1-2 (215 bp) idiomorphs (Florea et al. 2015) in 4–6 F. rothmaleri individuals per sampling site (Mon, Cab, Can). Sample sizes were adjusted to ensure detection of both idiomorphs at each location, using total DNA extracted from leaf tissue. Primer sequences and PCR conditions are provided in Table S2.

Morphological analyses and growth rate. Asexual reproductive structures (conidia and conidiophores) have proven to be a valuable taxonomical trait for the identification of Epichloë endophytes (Gentile et al. 2005; Leuchtmann and Schardl 2005; McCargo et al. 2014; Campbell et al. 2017; Thünen et al. 2022). The dimensions of these structures have also been correlated with the putative ploidy level of the endophytes (Kuldau et al. 1999). To assess the natural variability in these traits and their potential link to ploidy, four Epichloë isolates from hosts with differing source populations and ploidy levels (i.e., Mon4x, Can4x, Cab6x, Can8x) were analyzed, with three biological replicates per isolate. Microscopic preparations consisted of 1 mm² sections of mycelium, grown for three weeks on water agar (WA; European Bacteriological Agar, Condalab), which were carefully melted and sealed under a coverslip. The length and width of 10 conidia, along with the total length and width at the base of 10 conidiophores, were recorded for each replicate, yielding a total of 30 entries per isolate. To morphometrically characterize these endophytes, a novel software –Epichloë conidia– was developed using Matlab version 9.3 (R2017b; see Appendix S1) to automatically detect and estimate conidial area. Only spores photographed in sagittal plane were analyzed to ensure comparability, using both ImageJ v154g (Schneider et al. 2012) and custom software, with a fixed scale of 15.8097 pixels/μm. Morphometric differences among groups were assessed using PERMANOVA based on Euclidean distances with 10,000 permutations (R package ‘vegan’ v2.6.10; Oksanen et al. 2025), which does not assume multivariate normality and is suitable for nested designs. Pairwise permutation tests (R package ‘coin’ v1.4.3, Hothorn et al. 2008) were also performed with 10,000 resamples, considering the clustering structure of the study: sampling sites (n = 3), host ploidy levels (n = 3), individuals per source population and host ploidy level (n = 4), and replicates within individuals (n = 10). P-values were adjusted using the Holm-Bonferroni method, with significance set at p* < 0.05.

Additionally, exploratory multivariate analyses, including Principal Component Analysis (PCA, R package ‘FactoMineR’ v2.11, Lê et al. 2008) and Linear Discriminant Analysis (LDA, R package ‘MASS’ v7.3.60.2), were carried out to identify potential patterns of endophyte diversification associated with population origin, host ploidy level, or their interaction. These analyses were performed under three data clustering schemes: (i) individual-level mean-pooled measurements (n = 16); (ii) replicate-level mean-pooled measurements (n = 48); and (iii) resampling-level individual measurements (n = 469). This hierarchical structuring of sample sizes was designed to help identify the main sources of variability within the system. To statistically evaluate the morphological variation revealed by principal component analysis (PCA), linear models (LM) and analyses of variance (ANOVA) were developed on the scores of the first two principal components (PC1 and PC2), including host ploidy level, population of origin, and their interaction as explanatory variables, using the R package “stats” v4.4.1. The performance of the LDA models was assessed using cross-validation procedures implemented with the R package “caret” v6.0.94 (Kuhn 2008).

Since the growth rate of fungal cultures on PDA medium has been used as a trait to characterize Epichloë species, 1.5 mm² sections of four two-week-old isolates per population were simultaneously grown on PDA plates at room temperature in the dark for 24 days to estimate endophyte growth rate. Three technical replicates per isolate were analyzed. Culture growth was monitored every eighth day by taking pictures and measuring the diameter of the culture. Growth front was considered the external limit for this measure and a digital vernier caliper was used to record culture diameter. Growth rate (GR; mm/day) was calculated for each individual sample following the equation GR = (Diameter_t´ - Diameter_t) / Δ_t, where Diameter_t is the initial diameter of the culture, Diameter_t’ is the final diameter and Δ~t ~is the total number of days. Mixed-effect models were fitted due to the dependency between the measurements registered (i.e., diameter measured through time) and the hierarchical structure of the experimental design using the R package ‘lme4’ v1.1.35.5 (Bates et al. 2015). The effect of the source population, host ploidy level, and their interaction (fixed effects) on fungal growth rate (mm/day) was analyzed, including individuals or replicates as random effects to account for intra-individual variability and make more precise estimates without inflating standard errors. ANOVAs and subsequent pos-hoc pairwise tests assessing differences by marginal means (EMMs) applying Tukey’s HSD correction were carried out to test the significance of these models and detect potential sources of variability using the R package ‘emmeans’ v1.10.5 (Lenth 2024). All these analyses were performed in R v4.4.1 (R Core Team 2024) using a custom script (script1.Rmd) created with R Markdown (rmarkdown’ v2.29; Allaire et al. 2024).

Genome size estimation. DAPI-stained slides were prepared to confirm that the endophyte conidia were uninucleate before performing genome size estimations. Two-week-old 1 mm2 slices of mycelium were grown on a nutrient-deficient medium, consisting of water agar, for 15 days at room temperature and in the dark to facilitate sporulation. Photographs were taken with a fluorescence microscope (Miotic BA410) equipped with the Miotic MoticamPro 285D camera.

Existing protocols for other fungal species were adapted to the endophytes under study by flow cytometry (Ploidy Analyzer, Sysmex) to determine their genome sizes and infer the ploidy levels of the endophytes isolated from the different F. rothmaleri cytotypes. Colletotrichum acutatum strain PT812 (68 Mb or ~ 0.069 pg/1C) was included as primary standard following Talhinhas et al. (2017). Each Epichloë isolate and the standard were ground with a sterile mortar and pestle in 700 µl of ddH₂O and grown for 7–10 days in PDA plates with a sterile cellophane disk to increase the availability of active young mycelia. A 0.5 cm² portion was extracted and grounded in a glass petri dish with 1 ml of LB01 buffer (Doležel et al. 2007) using a sharp blade. The nuclear suspension was filtered through a 20 μm nylon mesh filter before being stained with 50 μl of propidium iodide. After a brief incubation on ice (3–5 minutes), samples were analyzed using the Ploidy Analyzer flow cytometer (Sysmex). Sample processing was carried out on ice (~ 4ºC) to ensure nuclear stability. Three isolates were considered to establish the genome sizes of endophytes from each source population and the ploidy level of the host (Mon4x, Can4x, Cab6x, Can8x), measuring 4 technical replicates in each case (n=12). Here, a minimum threshold of 5000 nuclei was set for each measurement, and coefficients of variation (CV) were considered acceptable when they were below 10%, following previous flow cytometry analysis standards in other fungi (Bourne et al. 2014).

Molecular and phylogenetic analyses. Epichloë isolates were grown for 7–10 days on PDA plates covered with sterile cellophane disks to facilitate mycelial recovery. DNA was extracted using a modified version of the CTAB-based DNA extraction protocol (Doyle and Doyle 1987) (see Appendix S2). Quality and concentration were assessed using a Biodrop spectrophotometer (µLite) and a Qubit 3.0 fluorometer. PCR amplified products from four nuclear loci [γ-actin (actG), calmodulin (CalM), translation elongation factor 1-α (tefA), and β-tubulin (tubB)], as well as the ITS region of rDNA, were subjected to bidirectional Sanger sequencing (Moon et al. 2004; McCargo et al. 2014; Chen et al. 2019; Thünen et al. 2022; Wang et al. 2022), using optimized primers and cycling programs detailed in Table S2. Two isolates per combination of source population and host ploidy level (Mon4x, Can4x, Cab6x, Can8x) were selected for molecular characterization.

Sequences were trimmed and aligned in Geneious Prime version 2024.0.5 (Biomatters Ltd, New Zealand) using MAFFT algorithm v7.490 (Katoh and Standley 2013) and subsequently processed manually. Single nucleotide polymorphisms (SNPs) calling was performed with a threshold value of a minimum variant frequency of 25% due to the small sample size of our dataset. Reference sequences for the five studied loci were obtained from published Epichloë genomes and two Claviceps purpurea strains that were used as outgroups (Table S3). Only haploid genomes (i.e., single-copy loci) were included, verified using a custom bash script (script2.sh). Single-gene maximum likelihood phylogenetic trees were obtained using IQ-TREE 2 (Minh et al. 2020) with 1000 UltraFast Bootstrap (BS) replicates (Hoang et al. 2017). For visualization, the lengths of all branches were transformed using x^¼. Subsequently, both the concatenated maximum likelihood tree and a coalescence-based species tree were obtained using IQ-TREE 2 and ASTRAL-III (Zhang et al. 2018), respectively. The likelihood-based site concordance factor (sCFL) was calculated using 1000 replicates for the concatenated ML tree (Mo et al. 2023), and quartet scores for the species tree and its two alternative topologies (Sayyari and Mirarab 2016) were inferred using ASTRAL-III. All trees were generated, edited, and formatted with two custom scripts created with bash and R Markdown (script3.sh, script4.Rmd).