In vitro evaluation of azole fungicide sensitivity in Fusarium langsethiae , F. tricinctum , F. poae and F. sporotrichioides populations from Irish oats

Abstract

BACKGROUND

Fusarium head blight (FHB) in oats represents a significant challenge to crop production and food safety, primarily because of mycotoxin contamination. In this study, 286 Fusarium isolates, representing in majority F. langsethiae, and small numbers of F. tricinctum, F. poae and F. sporotrichioides, obtained from commercial Irish oat crops in 2022 were evaluated in vitro for their sensitivity to three azole fungicides: prothioconazole‐desthio (PDZ), tebuconazole (TBZ) and mefentrifluconazole (MFZ). A microtitre plate assay was used to generate dose–response curves and to determine their half‐maximal effective concentrations (EC₅₀) to each of the three fungicides.

RESULTS

Differences in fungicide sensitivity and evidence of moderate cross‐resistance among the azole fungicides tested, particularly in F. langsethiae, were detected. Specifically, isolates were overall more sensitive to PDZ, while sensitivity differences between fungicides and partial cross‐resistance between MFZ and TBZ were statistically significant. These results were supported by a principal component analysis and a cluster analysis that confirmed F. langsethiae isolates were least sensitive to MFZ, this fungicide being responsible for the highest amount of variability in the population.

CONCLUSION

Given that oats are used in human consumption and animal feed, these findings underscore the critical need for ongoing resistance monitoring to maintain food safety standards aligned with European Union regulatory requirements. © 2026 The Author(s). Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Microtitre plate assays of 286 Irish oat isolates show species‐specific azole sensitivity. In Fusarium langsethiae, mefentrifluconazole and tebuconazole sensitivities correlate, defining four phenotypes and supporting resistance monitoring and integrated control.

1. INTRODUCTION

Oats (Avena sativa) are an economically important cereal ¹ used for human consumption, animal feed and selected industrial applications. ² , ³ , ⁴ In food, demand is driven by their high beta‐glucan content, ² whereas in feed oats remain important particularly in equine nutrition. ³ Industrial uses include skincare, where colloidal oatmeal is incorporated into formulations. ⁴

However, the productivity and safety of oats are increasingly compromised by Fusarium head blight (FHB). In oats, FHB is caused by pathogenic fungi belonging primarily to the Fusarium genus, which colonise cereal heads, resulting in shrivelled grains, lower weight and reduced yield causing significant economic loss. ⁵ , ⁶ Moreover, some of the Fusarium species lead to harmful mycotoxins contamination of grains. ⁷ The most known Fusarium mycotoxins are deoxynivalenol (DON) produced by F. graminearum and F. culmorum; nivalenol (NIV) produced by F. graminearum, F. cerealis, F. culmorum and F. poae in northern regions; T‐2 toxin and HT‐2 toxin produced mainly by F. sporotrichoides, F. poae and F. langsethiae. ⁸ , ⁹ , ¹⁰ Surveys in northern Europe have reported oats as having some of the highest levels of DON, T‐2 and HT‐2 mycotoxins, which pose serious food safety concerns. ¹¹

The genus Fusarium encompasses numerous species, the prevalence and dominance of which vary significantly with climatic and regional conditions, as well as agronomic practices. ⁸ Among the approximatively 17 Fusarium species most commonly associated with FHB in warmer and humid regions (southern Europe, the USA, and Asia), F. graminearum is the predominant FHB pathogen. It typically outcompetes other species because of its high aggressiveness and adaptability. Although F. culmorum and F. poae are typically associated with temperate and cooler regions, several studies have documented their occurrence in warmer or transitional climates, especially under dry or variable weather conditions. ⁶ , ⁷ , ¹² , ¹³ In cooler, temperate climates characteristic of northern Europe, which includes Ireland, a distinct Fusarium community plays a more significant role in disease development. This community primarily consists of F. langsethiae, F. sporotrichioides and F. tricinctum, frequently causing FHB outbreaks in oats and other cereals. ⁶ , ¹⁴

In Ireland, F. langsethiae emerges as the primary Fusarium species associated with oats, contributing to the production of T‐2 and HT‐2 mycotoxins. This species, accompanied by other Fusarium spp. contributes to the overall Fusarium disease complex and mycotoxin profile found in harvested oats. ¹ , ¹⁵

Effective management of FHB in cereal crops is challenging because of the complexity of the disease, variability of the Fusarium species involved, and diverse environmental factors influencing pathogen prevalence and mycotoxin production. ¹⁶ , ¹⁷ , ¹⁸ Current FHB management strategies typically include tillage practices such as crop rotation, ploughing and residue management, biological control agents, use of resistant cultivars and the application of fungicides. ¹⁹ , ²⁰

Azole fungicides, including prothioconazole and tebuconazole (TBZ), act by blocking ergosterol biosynthesis, thereby disrupting fungal membrane integrity. ²¹ , ²² Extensive field work and meta‐analyses show that prothioconazole‐ and TBZ‐based programmes reduce Fusarium infections and mycotoxin contamination in cereals, such as wheat and barley, crops historically dominated by F. graminearum and F. culmorum. ²³ , ²⁴ However, information on azole efficacy against F. langsethiae remains scarce in Ireland and comparable north European regions where it is prevalent, ¹⁵ , ¹⁷ despite its increasing importance. Comparable knowledge gaps exist for other oat‐associated species, such as F. tricinctum, F. poae and F. sporotrichioides. Moreover, chemical control strategies are complicated by variability in sensitivity both within and between different Fusarium species, ²⁵ reinforcing the need for detailed evaluation of azole sensitivity in regionally dominated Fusarium populations.

The potential for the emergence of fungicide resistance in Fusarium further threatens the sustainable management of FHB. ²¹ , ²⁶ In well‐studied species such as F. graminearum, azole resistance has been associated primarily with the repeated and intensive use of azole fungicides, and is caused by three main mechanisms: target‐site alterations such as point mutations that lower azole binding affinity ²⁷ ; regulatory mutations such as promoter insertions that overexpress the sterol‐14a‐demethylase (CYP51) gene ²⁸ ; and upregulation of multidrug efflux transporters that actively export azole molecules. ²⁹ Spolti et al. ²⁵ reported substantial geographic variability in TBZ sensitivity among F. graminearum isolates collected from North America and compared it with early surveys from Europe and South America, highlighting the adaptive potential of Fusarium populations to fungicide pressure. ²⁵ Moreover, the temporal and spatial dynamics of Fusarium species distribution and toxin production further complicate chemical management, necessitating region‐specific management recommendations. ³⁰

Comparable baseline data for oat‐associated F. langsethiae are almost non‐existent, apart from a small in vitro study showing growth inhibition by TBZ, prochloraz and fenpropimorph, ³¹ with no regional half‐maximal effective concentration (EC₅₀) distributions having been published. Information is equally sparse for F. tricinctum, F. poae and F. sporotrichioides.

Azole cross‐resistance is another concern. In F. graminearum, a survey based on New York isolates showed a strong positive correlation between TBZ and metconazole, ²⁵ suggesting that reduced sensitivity to one azole could cause a similar effect in other azole fungicides. Such data are lacking entirely for F. langsethiae, F. tricinctum, F. poae and F. sporotrichioides. Nevertheless, cross‐resistance patterns in other cereal pathogens underline this risk. In Zymoseptoria tritici, isolates with complex CYP51 haplotypes show concomitantly reduced sensitivity to TBZ and mefentrifluconazole (MFZ), but not necessarily between prothioconazole‐desthio (PDZ; the active metabolite of prothioconazole) and either TBZ or MFZ. ³² , ³³ Whether similar relationships occur in the Fusarium species that dominate oats remains unknown.

Given the rising incidence of F. langsethiae in oats across Europe ³⁴ and its role as the principal T‐2/HT‐2 producer in northern regions, there is a clear need to establish baseline azole sensitivity for the Fusarium complex that dominates oat production and to detect any early signs of cross‐resistance. Accordingly, this study evaluates the in vitro azole sensitivity profiles of 286 Irish Fusarium isolates recovered from Irish oats in 2022, focusing on the species most frequently detected and of greatest toxigenic concern: predominantly F. langsethiae, but also smaller numbers of F. sporotrichioides, F. poae and F. tricinctum isolates.

2. MATERIAL AND METHODS

2.1. Collection of Fusarium isolates

The single spore fungal isolates used in this study were identified as Fusarium spp. and obtained as described by Bucur et al. ¹⁵ Briefly, 286 fungal isolates obtained from oat crops having different cropping systems in 2022 and identified as Fusarium spp. based on colony and spore morphology and TEF‐α sequencing were included in this study. Among the Fusarium species included are F. langsethiae (n = 227) F. tricinctum (n = 22), F. poae (n = 22), F. sporotrichioides (n = 7) and F. avenaceum (n = 2) (Supporting Information, Table S1). These isolates were stored as culture plugs in 30% glycerol at −70 °C.

2.2. Fungicide sensitivity assay in vitro

2.2.1. Microtitre plate assay

The sensitivity of the Fusarium isolates to the azole fungicides PDZ (the fungicidal active metabolite of prothioconazole), MFZ and TBZ was determined using a microtitre plate assay, as described by Bucur et al. ³⁵ with some modifications. Fusarium isolates were grown from the 30% glycerol stocks stored at −70 °C on 0.25× potato dextrose agar (Merck, Rahway, NJ, USA), amended with 50 mg L⁻¹ of ampicillin sodium (Apollo Scientific Ltd, Stockport, UK) and 100 mg L⁻¹ of neomycin sulphate (Cayman Chemical, Ann Arbor, MI, USA) and were incubated for 7 days at 25 °C in dark conditions. On the morning of the experiment, 1 mL of sterile distilled water was added onto each plate, and the colonies were gently scraped using a sterile T‐shaped spreader. The spore and mycelial suspensions were pipetted into a 2‐mL Eppendorf tube and kept on ice. After settling for 10 min, the upper part of the suspension was transferred to new tubes and spores were quantified using KOVA® Glasstic® Slides (Kova International, Garden Grove, CA, USA), and subsequently adjusted to 100 000 spores mL⁻¹. Flat‐bottomed microtitre plates (Sarsted AG & Co., Nümbrecht, Germany) were filled with 200 μL of fungicide and spore suspension mixture, of which 100 μL of 2× PDB was amended with a 1.5‐fold dilution series of PDZ or MFZ and 2‐fold dilution series of TBZ (Sigma‐Aldrich, St. Louis, MO, USA) from stock solutions in dimethylsulfoxide (DMSO) and 100 μL of suspension (100 000 spores mL⁻¹) in sterile distilled water. This way, the 96‐well plates were filled with eight final concentrations of fungicides ranging from 0.15 to 0.0131 μg mL⁻¹ for PDZ, from 4 to 0.351 μg mL⁻¹ for MFZ and from 3 to 0.0469 μg mL⁻¹ for TBZ. Isolates were tested in batches of 55 isolates, with 11 isolates and a negative control tested in each 96‐well plate. Two technical replicates were prepared for all the isolates tested. Plates were sealed with parafilm and wrapped in cling film to prevent evaporation, and incubated in the dark at 25 °C.

2.2.2. Data analysis

Fungal growth was assessed after 3 days of incubation (~72 h) as a measure of light absorbance at 450 nm on a Synergy‐HT plate reader (BioTek Instruments, USA) and Gen5™ microplate software. For each plate, the absorbance obtained for the negative control was subtracted from all other isolate–fungicide combinations to account for the background noise. The adjusted values obtained for absorbance were then used to estimate the fungicide sensitivity via a dose–response curve, fitted with a four‐parameter log‐logistic model (LL.4) from the drc package. ³⁶ For each isolate, the half‐maximal effective concentration (EC₅₀) was extracted, and because technical replicates were used, their EC₅₀ values were averaged to produce a single value per isolate.

All statistical analyses were done in R v.4.3.2 (2023‐10‐31 ucrt) ³⁷ and RStudio (v.1.2.5001 and 2023.12.0 + 369 ‘Ocean Storm’). ³⁸ First, the isolates tested were grouped by species and the normality and skewness for each group (species × fungicide) were tested using the Shapiro–Wilk test (shapiro.test). Because many groups deviated significantly from normality, particularly those with large numbers of isolates, a non‐parametric approach was chosen for species‐wise comparisons. The species where n was very small (F. avenaceum) were eliminated from the analysis. Outlier detection was carried out through a two‐step Linear Mixed Model (LMM) approach using the R packages lme4 ³⁹ and lmerTest ⁴⁰ to prevent skewed sensitivity estimates and misinterpretation of resistance patterns. First, replicate‐level EC₅₀ data (one curve‐fit per technical replicate) were analysed with an LMM; Pearson residuals were extracted and replicates whose residuals exceeded ±2.5 SD were flagged. Each flagged replicate was inspected in the raw growth curves and, where available, the plate adsorption profiles to verify bacterial contamination, pipetting error or reader saturation. Replicates confirmed as artefacts were discarded, and the isolate was re‐assayed whenever possible. The remaining replicates were then averaged to yield a single log‐transformed EC₅₀ for every isolate–fungicide combination. Second, the isolate means were re‐examined with a simplified linear model; means whose standardised residuals exceeded ±2.5 SD were retained but are highlighted in red in the sensitivity plot to indicate biologically extreme responses. All clustering, silhouette and principal component analyses (PCA) were performed on this same set of replicate‐screened means, independent of the visual flag.

The non‐parametric Kruskal–Wallis one‐way analysis of variance (ANOVA) test (kruska.test from ‘rstatix’ ⁴¹ ) was used to test whether overall significant differences between species existed. Where significant differences were detected (P < 0.05), subsequent post‐hoc Dunn's multiple comparison tests (kwAllPairsDunnTest from ‘PMCMRplus’) ⁴² with Holm's correction to adjust for multiple testing were done to identify where the differences are present.

To assess potential effects of geographic and agronomic factors, F. langsethiae mean sensitivity was assessed against Location as county, Crop and Rotation. For this, within F. langsethiae, records annotated ‘Organic’ in Crop were excluded because of their low number (n = 2). Location entries were harmonised to Kilkenny, Kildare, Laois, Tipperary and Other (minor sites collapsed). For each fungicide (TBZ, PDZ, MFZ), isolate‐level means (Mean_logEC₅₀) were tested against Location, Crop, and Rotation using a Kruskal–Wallis test (α = 0.05). Robustness and effect magnitude were then evaluated with confirmatory linear models, Mean_logEC₅₀ ~ Location + Crop + Rotation, reporting Type II ANOVA P‐values and partial η ² from sums of squares. For two‐level Crop, the median fold‐change was reported on the original EC₅₀ scale.

To examine potential cross‐resistance among the three fungicides (PDZ, TBZ, MFZ), Spearman's rank correlation ⁴³ was calculated using cor.test function adjusted to ‘spearman’ method both overall and within each species. A correlation matrix plot was used to illustrate relationships between fungicides and was generated using ggpairs from the ‘GGally’ package, ⁴⁴ which extends ‘ggplot2’. ⁴⁵ The correlation plot displays Spearman's rho values and P‐values in the upper panels, scatter plots (with simple linear fits) in the lower panels and kernel density distributions of logEC₅₀ on the diagonal panels.

Next, a PCA was performed on F. langsethiae logEC₅₀ data for all three fungicides (PDZ, TBZ, MFZ) to summarise overall fungicide sensitivity patterns and to corroborate cross‐resistance outcomes. PCA was conducted using the prcomp function with scaled variables, and results were visualised through PCA biplots created using ‘factoextra’. Finally, cluster analyses were done to further investigate the correlation patterns specifically between MFZ and TBZ sensitivity in F. langsethiae. Prior to clustering, isolate‐level logEC₅₀ data were scaled. For the distance‐based methods (hierarchical clustering and k‐means), the optimal cluster number was determined using the Gap Statistic method clusGap from the ‘cluster’ package ⁴⁶ and the NbClust function from the package with the same name. ⁴⁷ Subsequently, hierarchical clustering (HC) (hclust) using Ward's method (ward.D2) and k‐means clustering were fitted at the selected number of clusters, whereas Gaussian mixture model clustering using the mclust function ⁴⁸ was fitted with the number of components selected by Bayesian information criterion during model fitting. Clusters were evaluated and visualised through scatterplots, boxplots and silhouette analysis using the ‘factoextra’ ⁴⁹ and ‘cluster’ packages.

All graphical outputs, including boxplots overlaid with jittered points were created using ‘ggplot2’. ⁴⁵

3. RESULTS

3.1. In vitro fungicide sensitivity comparison of different Fusarium spp.

The distribution of logEC₅₀ values for three azole fungicides MFZ, PDZ and TBZ across the four Fusarium species is shown in Fig. 1. In each panel, the boxplots summarise the distribution of logEC₅₀ values for the four species, showing the median as a central line, the interquartile range (IQR) as the box, and the overall spread through whiskers that extend to the most extreme non‐outlier values, highlighted in red.

Figure 1.

Azole fungicides sensitivity profiles in Fusarium species. The mean logEC₅₀ values for four Fusarium species (F. langsethiae, F. sporotrichioides, F. poae and F. tricinctum) across three fungicides: prothioconazole‐desthio (PDZ), tebuconazole (TBZ) and mefentrifluconazole (MFZ). Each fungicide is presented in its own facet, with boxplots summarising the central tendency and dispersion of the isolates' responses. Isolate data points outside the typical range are highlighted in red as outliers using a two‐pass Linear Mixed Model (LMM) approach, indicating unusually high or low sensitivity. EC₅₀, half‐maximal effective concentration.

For MFZ (Fig. 1(A)), the boxplots reveal divergence in sensitivity across species. The distribution of F. langsethiae isolates showed greater variability than the other species, with a lower median logEC₅₀ value indicating higher overall sensitivity to MFZ compared with F. poae and F. tricinctum. Despite this overall pattern, the broader distribution of F. langsethiae suggests that a subset of isolates exhibits reduced sensitivity to MFZ. This was confirmed by the Kruskal–Wallis test by species, which detected a significant difference among species (χ ²  = 28.25, df = 3, P = 3.2 × 10⁻⁶). Dunn's post‐hoc test revealed that F. langsethiae was significantly more sensitive than both F. poae (adjusted P = 9.2 × 10⁻⁴) and F. tricinctum (adjusted P = 3.3 × 10⁻³); no significant differences were observed among the other species pairs.

By contrast, for PDZ (Fig. 1(B)) highly consistent sensitivity profiles across all four Fusarium species were observed. The medians were tightly aligned, and the IQRs were narrow, indicating little within‐group variability. No major differences were visually evident between species in either spread or central tendency, as confirmed by the Kruskal–Wallis test, which suggested that sensitivity to PDZ was comparable among species, with no significant inter‐species differences (χ ²  = 2.39, df = 3, P = 0.495).

For TBZ (Fig. 1(C)), although the overall variability was similar to that observed for MFZ, the median logEC₅₀ value for F. langsethiae was lower than those of F. sporotrichioides, F. tricinctum and F. poae, indicating that F. langsethiae was the most sensitive species to this fungicide. The other three species showed comparable higher medians, with relatively tight IQRs. These differences in central tendency were supported by a highly significant Kruskal–Wallis result (χ ²  = 62.36, df = 3, P ≈ 1.8 × 10⁻¹³). Dunn's test confirmed that F. langsethiae was significantly more sensitive than F. poae (adjusted P = 8.2 × 10⁻⁷), F. sporotrichioides (adjusted P = 6.0 × 10⁻⁵) and F. tricinctum (adjusted P = 5.2 × 10⁻⁶). No significant differences were found among the other three species. However, the comparatively small number of F. poae isolates (n = 7) reduces statistical power for that species; results for this taxon should therefore be interpreted with caution.

3.2. Differences among fungicides

When pooling data across all species, the distribution of logEC₅₀ values revealed significant differences in overall sensitivity between the three fungicides. MFZ showed the highest median logEC₅₀ (median = 0.023, IQR = −0.112 to 0.288), followed by TBZ (median = −0.518, IQR = −0.648 to −0.394), whereas PDZ exhibited the lowest sensitivity values (median = −1.04, IQR = −1.35 to −0.897). The Kruskal–Wallis test confirmed a highly significant overall difference (χ ²  = 558.36, df = 2, P < 2.2 × 10⁻¹⁶). Dunn's post‐hoc comparisons showed that all three fungicides differed significantly from one another after adjustment for multiple comparisons: the logEC₅₀ values for MFZ were significantly higher than both for PDZ (adjusted P ≈ 1.1 × 10⁻¹²³) and for TBZ (adjusted P ≈ 3.6 × 10⁻²⁷), whereas the logEC₅₀ values for TBZ were higher than those for PDZ (adjusted P ≈ 1.4 × 10⁻³⁷). These results suggest that although all three azole fungicides share a similar mode of action, their relative potencies against Fusarium spp. vary.

3.3. Correlation analysis and cross‐resistance

To investigate potential cross‐resistance among the three fungicides (MFZ, PDZ, TBZ), a correlation matrix plot (Fig. 2) was generated across the four Fusarium species. The matrix presents kernel density distributions (diagonal panels), Spearman correlation coefficients with P‐values (upper panels) and pairwise scatterplots (lower panels). Among the three fungicide pairs, MFZ exhibited a moderate positive correlation with TBZ (ρ ≈ 0.658, P < 2 × 10⁻¹⁶) and a weak but significant relationship with PDZ (ρ ≈ 0.170, P = 0.004). The strongest correlation observed was between MFZ and TBZ; however, visual inspection of the corresponding scatterplot revealed a shift in the relationship: although logEC₅₀ values were coupled at low MFZ ranges, TBZ plateaued as MFZ continued to rise, resulting in vertical dispersion. The PDZ–TBZ correlation was weaker (ρ ≈ 0.098, P = 0.105) and did not reach significance after correction, indicating limited shared variance. These overall trends point to a partial cross‐resistance pattern among the fungicides, with MFZ at the centre of the observed relationships.

Figure 2.

Correlation matrix for the three fungicides used [prothioconazole‐desthio (PDZ), tebuconazole (TBZ) and mefentrifluconazole (MFZ)] evaluated against four Fusarium species. Each diagonal panel displays a kernel density plot of logEC₅₀ values for a given fungicide. The x‐axis represents logEC₅₀ values (higher = lower sensitivity), and the y‐axis indicates density, or the proportion of isolates exhibiting that level of response. Off‐diagonal lower panels show pairwise scatterplots between fungicides, with each point coloured by species and overlaid with a linear fit line. Upper panels report Spearman correlation coefficients and associated P‐values, summarising the strength and significance of the pairwise relationships. EC₅₀, half‐maximal effective concentration.

These correlations were predominantly driven by F. langsethiae, the only species to show statistically significant associations between MFZ and both TBZ and PDZ. Notably, the MFZ–TBZ correlation was particularly strong (ρ = 0.707, P < 2 × 10⁻¹⁶), reinforcing the idea of shared resistance mechanisms or selective pressure in this species. Nevertheless, the MFZ–TBZ scatterplot suggests a breakpoint: some isolates with high MFZ logEC₅₀ values no longer show proportional increases in TBZ EC₅₀ values. This implies that changes in sensitivity to MFZ may develop independently in a subgroup of F. langsethiae isolates, separate from TBZ. The MFZ–PDZ and PDZ–TBZ correlations were also significant in F. langsethiae, albeit weaker (ρ = 0.184 and 0.110, respectively), indicating less coordinated sensitivity shifts.

By contrast, none of the other species exhibited significant correlations between fungicide pairs. In F. tricinctum, all pairwise correlations were weak and non‐significant (PDZ–TBZ: ρ = 0.319, P = 0.158; PDZ–MFZ: ρ = −0.155, P = 0.502; MFZ–TBZ: ρ = 0.061, P = 0.793). Fusarium poae showed no significant association either (PDZ–TBZ: ρ = −0.048, P = 0.837; PDZ–MFZ: ρ = 0.344, P = 0.127; MFZ–TBZ: ρ = 0.138, P = 0.538), whereas F. sporotrichioides displayed weak and inconsistent trends (PDZ–TBZ: ρ = −0.357, P = 0.444; PDZ–MFZ: ρ = 0.214, P = 0.662; MFZ–TBZ: ρ = −0.5, P = 0.267). These results suggest that although F. langsethiae demonstrates consistent cross‐resistance, which is significant among TBZ–MFZ (ρ = 0.707) and weaker between TBZ–PDZ (ρ = 0.110) and MFZ–PDZ (ρ = 0.184), the correlations in the data set are not broadly species‐wide, but rather reflect specific resistance dynamics within F. langsethiae, because the other species do not show a clear or statistically supported relationship between fungicide sensitivities. This discrepancy may reflect smaller sample sizes or greater variability in those species, limiting the power to detect a meaningful relationship at the sample sizes tested, which was smaller compared with F. langsethiae.

The kernel density plots on the diagonal further illustrate how logEC₅₀ values are distributed across species and fungicides. Fusarium langsethiae exhibited moderate spread in all three fungicides, with broader tails in TBZ and MFZ indicating within‐species variability. By contrast, F. poae and F. sporotrichioides formed compact, left‐skewed distributions, consistent with high and uniform sensitivity. Fusarium tricinctum showed a broader distribution for MFZ (−0.5 to 2.2) and TBZ (−1.3 to 1.2), although this was not reflected in significant correlations. These differences in distribution align with the scatterplot patterns: F. langsethiae and F. tricinctum show greater dispersion, whereas F. poae and F. sporotrichioides cluster tightly and contribute little to overall correlation.

Field‐parameter analysis within F. langsethiae showed no significant effects of Location (as county), Crop or Rotation on the mean sensitivity [mean log₁₀(EC50)] to TBZ (Location P = 0.504, Crop P = 0.831, Rotation P = 0.786) or to MFZ (Location P = 0.096, Crop P = 0.063, Rotation P = 0.130). For PDZ, Kruskal–Wallis indicated a small Crop difference (P = 0.002), but this was not supported by the confirmatory model (Type II ANOVA: Location P = 0.955, Crop P = 0.186, Rotation P = 0.725), and effect sizes were very small (partial η ² 0.00001–0.039; PDZ–Crop ≈ 0.008). Median fold‐changes between Crop levels were likewise small: TBZ 1.07×, PDZ 1.27×, MFZ 1.13×, indicating no biologically meaningful separation (Table 1).

Table 1.

Field‐parameter analysis for Fusarium langsethiae

Fungicide	Factor	No groups	n	Kruskal–Wallis P‐value	Type II ANOVA P‐value	Partial η 2	Median fold‐change (Crop)
TBZ	Location	5	225	0.504053	0.42802	0.01738	NA
TBZ	Crop	2	225	0.831433	0.05366	0.01697	1.07×
TBZ	Rotation	2	225	0.785603	0.05114	0.01734	NA
PDZ	Location	5	223	0.07235	0.95474	0.00309	NA
PDZ	Crop	2	223	0.00223	0.18639	0.00807	1.27×
PDZ	Rotation	2	223	0.08897	0.72505	0.00057	NA
MFZ	Location	5	224	0.09645	0.06813	0.03927	NA
MFZ	Crop	2	224	0.06271	0.36176	0.00383	1.13×
MFZ	Rotation	2	224	0.12987	0.96302	0.00001	NA

Location levels: Kildare (n = 100), Kilkenny (n = 21), Laois (n = 21), Tipperary (n = 71), other (n = 12). Crop levels: spring oats (n = 139), winter oats (n = 86). Rotation levels: cereal (n = 131); non‐cereal (n = 94).

ANOVA, analysis of variance; TBZ, tebuconazole; PDZ, prothioconazole‐desthio; MFZ, mefentrifluconazole; NA, not applicable.

3.4. PCA of Fusarium langsethiae

To further explore multivariate patterns in azole fungicide sensitivity for F. langsethiae, a PCA was done for the three azole fungicides used: MFZ, TBZ and PDZ (Fig. 3). The first two principal components (PC1 and PC2) explained approximately 85% of the total variance, with PC1 alone accounting for 52.3%. The biplot revealed that MFZ and TBZ contribute similarly to PC1, with arrows pointing diagonally to the left, indicating the fungicides are positively correlated and suggesting that variation in response to these two fungicides was the primary driver of overall sensitivity differences. By contrast, PDZ pointed downward along PC2, capturing a smaller, orthogonal source of variability.

Figure 3.

Principal components analysis (PCA) of Linear Mixed Model‐filtered logEC₅₀ values for prothioconazole‐desthio (PDZ), tebuconazole (TBZ) and mefentrifluconazole (MFZ) in Fusarium langsethiae isolates. Arrows indicate variable loadings and their contributions to the principal components (PC). Points are coloured by the fungicide for which each isolate shows the highest logEC₅₀ (i.e. the dominant fungicide). The MFZ and TBZ vectors point in a similar direction, reflecting strong correlation and shared variance, whereas PDZ contributes more to the second PC. Most isolates are MFZ‐dominant and cluster along PC1, revealing substantial variation in MFZ sensitivity across the population. EC₅₀, half‐maximal effective concentration.

Isolates were coloured by dominant fungicide sensitivity. The vast majority were classified as MFZ‐dominant, reflecting higher logEC₅₀ values for MFZ compared with TBZ and PDZ. These isolates showed a wide spread along PC1, particularly towards the left of the plot, consistent with a subset of isolates exhibiting reduced sensitivity to MFZ. Only four isolates were PDZ‐dominant, and these clustered within the right‐upper region of the MFZ‐dominant cloud, near the direction of the PDZ loading. Two TBZ‐dominant isolates appeared towards the mid‐left, consistent with their TBZ loading.

Overall, the PCA confirms that variation in MFZ sensitivity is the major source of variability in the F. langsethiae fungicide response, with TBZ contributing similarly but secondarily along the same axis. The separation of PDZ in the biplot suggests that PDZ sensitivity behaves differently from the other two fungicides and is less variable across isolates.

3.5. Cluster analysis of MFZ and TBZ sensitivity in Fusarium langsethiae

Finally, to confirm patterns in fungicide response and explore potential subpopulations of F. langsethiae, unsupervised clustering to detect natural groupings in sensitivity without prior assumptions on resistance phenotypes was conducted for MFZ and TBZ fungicides, because PCA placed MFZ and TBZ together on PC1 with the largest explained variance, whereas PDZ loaded primarily on an independent component with lower variance. Distance‐based clustering (HC and k‐means) supported a four‐cluster solution, whereas the model‐based approach (mclust) suggested five clusters.

Silhouette analysis revealed that HC achieved the highest average silhouette width (≈0.538), indicating well‐separated and internally cohesive clusters. By contrast, k‐means showed moderate cohesion (≈0.447), and mclust produced a poor silhouette score (≈ −0.014), suggesting instability in its cluster assignments. These results strongly support HC as the most robust and biologically relevant method for interpreting MFZ–TBZ sensitivity patterns.

The HC classification map (Fig. 4) identified four well‐defined clusters based on logEC₅₀ values for MFZ and TBZ. Cluster one (blue), the most abundant and centrally positioned, comprised of isolates with moderate logEC₅₀ values for both MFZ and TBZ, suggesting partial cross‐resistance. Next, cluster two (red) showed high logEC₅₀ values for both fungicides, marking another subgroup with possible cross‐resistance to both MFZ and TBZ. Cluster three (yellow) grouped isolates with low sensitivity to both fungicides, showing low logEC₅₀ values for MFZ and TBZ, indicative of a small subpopulation highly sensitive. Finally, cluster four (purple) included isolates with reduced sensitivity to MFZ but intermediate TBZ values, suggesting potential MFZ‐specific resistance. The classification map showed clean separation between the four clusters with minimal overlap, reflecting clear biological differentiation among isolate subgroups. The high silhouette width supports that these groupings are not arbitrary but reflect distinct MFZ–TBZ sensitivity phenotypes.

Figure 4.

Visualisation of hierarchical clustering (Ward's method) for Fusarium langsethiae isolates, using scaled mefentrifluconazole (MFZ) and tebuconazole (TBZ) logEC₅₀ values. Isolates are grouped into four distinct clusters represented by different colours and enclosed by ellipses. The classification map reveals minimal overlap between groups, indicating distinct phenotypic patterns. This spatial arrangement mirrors the cross‐resistance profiles and supports the biological relevance of the identified clusters. EC₅₀, half‐maximal effective concentration.

The boxplots of MFZ and TBZ sensitivity by cluster (Fig. 5) further confirmed these patterns. Cluster one (blue), which was the most populated, displayed intermediate medians for both fungicides, consistent with partial cross‐resistance. Cluster two (red) exhibited the highest median logEC₅₀ values for both MFZ and TBZ, supporting its identification as a clearly cross‐resistant subgroup. Cluster three (yellow), representing highly sensitive isolates, had the lowest median logEC₅₀ values for both MFZ and TBZ. By contrast, cluster four (purple) showed a marked elevation in MFZ median values but moderate TBZ values, reinforcing the presence of MFZ‐specific reduced sensitivity in this group.

Figure 5.

Distribution of logEC₅₀ values for mefentrifluconazole (MFZ) and tebuconazole (TBZ) in Fusarium langsethiae isolates across the four hierarchical clusters identified in Fig. 3. Boxplots display medians, interquartile ranges and outliers. Cluster 1 (blue) shows intermediate sensitivity to both fungicides. Cluster 2 (red) has the highest median logEC₅₀ values, indicative of reduced sensitivity or potential cross‐resistance. Cluster 3 (yellow) exhibits the lowest values, representing highly sensitive isolates. Cluster 4 (purple) shows a marked increase in MFZ resistance while maintaining moderate TBZ sensitivity. EC₅₀, half‐maximal effective concentration.

Taken together, the clustering results confirm the presence of phenotypic subgroups within F. langsethiae that differ in azole sensitivity. Although most isolates remain sensitive to both MFZ and TBZ, a subset shows selective resistance to MFZ, and another distinct group exhibits cross‐resistance to both fungicides. These findings are consistent with the observed positive correlation between MFZ and TBZ and suggest that emerging cross‐resistance is driven by specific isolate clusters.

4. DISCUSSION

This study provides a comprehensive analysis of azole fungicide sensitivity in Fusarium populations from commercial Irish oat crops, with particular focus on inter‐ and intra‐Fusarium variability in sensitivity to the azole fungicides PDZ, TBZ and MFZ, and emerging patterns of reduced sensitivity. Our results highlight significant variability in sensitivity across and within species, and between fungicides, underscoring the need to tailor disease management strategies to both the local pathogen populations and the specific chemical compound used. This supports previous research demonstrating that the performance of triazole fungicides varies across Fusarium species and environments, ¹⁸ , ¹⁹ , ²⁴ and extends these findings to oat‐associated Fusarium populations. In particular, the results build on recent evidence from European cereal systems that emphasise species‐specific fungicide responses in FHB complexes, ¹⁶ and further highlight F. langsethiae as a key species of concern in Irish oat systems. ¹⁴ , ¹⁵

In Ireland, summer 2022 was atypically warm and dry during heading–flowering and early grain fill, conditions that generally reduce classical FHB pressure yet do not exclude infection by species adapted to cooler or drier windows. Consistent with this, our national oat survey detected widespread F. langsethiae in 2022 (94% of sampled oat fields), albeit at low incidence (<6% of isolates). ¹⁵ This pattern aligns with northern European reports where F. langsethiae tends to predominate when wetter–warmer flowering periods that favour F. graminearum and F. avenaceum are absent. ¹⁶ , ¹⁷ , ³⁰ Previous Irish oat surveys detected multiple mycotoxins, with frequent quantifiable T‐2/HT‐2 alongside others, underscoring the risk from low‐symptom Fusaria in oats. ¹ Collectively, the 2022 weather and the observed species pattern provide the ecological context for the phenotypic findings reported here. ¹ , ¹⁵ , ¹⁶ , ¹⁷ , ³⁰ Among the three fungicides tested (PDZ, TBZ and MFZ), TBZ and PDZ are registered in Ireland for reduction of FHB in wheat, either as single‐active products or in mixtures. In oats, spray programmes target foliar diseases rather than Fusarium; however, applications around panicle emergence/flowering may incidentally expose panicle‐associated Fusarium population and impose selection pressure. MFZ is a recent introduction to cereals and was included here because of its documented cross‐resistance in other cereal pathogens. ³³ PDZ – used here instead of its pro‐fungicide prothioconazole (PTZ) because PTZ is enzymatically converted to PDZ in planta but undergoes incomplete and variable conversion in vitro, leading to inconsistent EC₅₀ estimates ⁵⁰  – consistently exhibited the greatest potency across all Fusarium species, with uniformly low logEC₅₀ values and minimal variability. This aligns with previous work reporting high intrinsic activity of PDZ‐containing formulations against a range of Fusarium spp. in wheat and barley. ¹² , ²⁴ By contrast, MFZ and TBZ showed broader logEC₅₀ distributions, particularly within F. langsethiae, suggesting the emergence of subpopulations with reduced sensitivity. Similar variability has been documented in European countries, where F. langsethiae showed differential responses to TBZ and in some cases elevated EC₅₀ values under certain conditions. ¹⁶ , ³¹ These findings underline the need to account for species‐ and compound‐specific differences when developing chemical control strategies for oats, especially in temperate environments where F. langsethiae is commonly detected. ¹⁷

To explore the potential for cross‐resistance among the azole fungicides used, a correlation analysis was done. MFZ showed a moderate positive correlation with both TBZ and PDZ, suggesting partial cross‐resistance, with the strongest relationship observed between MFZ and TBZ. However, these patterns were largely driven by F. langsethiae, the only species to show statistically significant correlations among all fungicide pairs. The remaining species either displayed weak associations or had sample sizes too small to support reliable conclusions. This discrepancy reflects the larger number of isolates available for F. langsethiae, which enhanced the statistical power to detect within‐species trends. Consequently, and because F. langsethiae also demonstrated the widest range of logEC₅₀ values, subsequent analyses focused specifically on this species.

The positive correlation between MFZ and TBZ sensitivity in F. langsethiae is particularly noteworthy, because it mirrors similar trends observed in Z. tritici; in this species, strong cross‐resistance between MFZ and TBZ has been well documented across European populations and linked to shared resistance mechanisms, including mutations in the CYP51 gene and potential efflux activity. ³³ Although the current study did not assess underlying mechanisms directly, the observed correlation between MFZ and TBZ responses in F. langsethiae may reflect comparable processes. This is consistent with previous reports of azole resistance in Fusarium species, which have also highlighted the role of CYP51 paralogue modifications and transporter overexpression in reduced sensitivity. ²² , ²⁶ , ⁵¹ , ⁵²

Taken together with prior work, these correlations could suggest a polygenic basis for azole response in oat‐associated Fusaria. Across F. graminearum populations, EC₅₀ varies widely with high heritability and limited explanatory power from CYP51 coding polymorphisms, ⁵³ and genome‐wide association signals point to regulatory candidates rather than a single target‐site driver. ⁵⁴ In addition, unequal roles of CYP51 paralogues allow sensitivity shifts via expression/regulatory change without coding mutations, ²⁸ , ⁵² and ATP‐binding cassette (ABC) transporters mediating multidrug resistance are linked to azole tolerance and virulence, consistent with an efflux contribution. ²⁹ Moreover, azole‐reduced sensitivity can occur in isolates lacking CYP51 mutations or constitutive CYP51 overexpression, ⁵⁵ because fungicide exposure can induce stress–metabolism rewiring and upregulation of transporter and central metabolic pathways; ⁵⁶ consistently, Hellin et al. showed that a laboratory adapted strain and field isolates of F. culmorum with reduced azole sensitivity overexpressed the ABC transporter (FcABC1) with cross‐azole effects. ⁵⁷ Population studies in F. culmorum report broad triazole EC₅₀ dispersion and chemotype‐linked differences with limited temporal increase in mean sensitivity, indicating diffuse, context‐dependent selection rather than stepwise resistance. ⁵⁸ These patterns align with our observations; confirming mechanisms in F. langsethiae will require genetic analyses. PCA supported the cross‐resistance results and provided a multivariate perspective on fungicide sensitivity patterns. The first principal component (PC1), which explained more than 50% of the total variance, was primarily driven by responses to MFZ and TBZ, reflecting their dominant role in shaping overall variability. By contrast, PDZ contributed more substantially to the second principal component (PC2), indicating a distinct and less variable response pattern.

Most isolates were MFZ‐dominant and showed broad dispersion along PC1, highlighting variability in MFZ responses. Although this variation aligns with the trends observed in cross‐resistance analysis, the PCA alone does not clearly separate isolates into different sensitivity categories. Instead, it complements the correlation results by visually reinforcing the central role of MFZ in structuring the sensitivity profiles of F. langsethiae. In addition, the separation of PDZ along PC2 further supports the conclusion that PDZ sensitivity behaves independently of the patterns observed for MFZ and TBZ, which further highlights the complexity of resistance development in F. langsethiae.

Next, unsupervised clustering of F. langsethiae isolates based on MFZ and TBZ sensitivity revealed the presence of distinct subpopulations with differing sensitivity levels. HC produced four clusters, most of them displaying a coordinated increase in logEC₅₀ values for both fungicides, which confirms the cross‐resistance pattern. However, one group of isolates was particularly noteworthy, because it exhibited elevated MFZ logEC₅₀ values while maintaining moderate TBZ sensitivity, which indicates MFZ‐specific reduced sensitivity. This divergence points to the possibility that resistance may initially emerge towards MFZ before expanding to other azole fungicides. Such evolutionary trajectories have been documented in F. graminearum, where TBZ‐resistant isolates exhibit competitive fitness and may serve as precursors to broader resistance. ²⁵ Furthermore, broader reviews of fungicide resistance evolution in plant pathogens reinforce that progressive shifts in sensitivity are often driven by cumulative selection pressure, particularly under repeated azole fungicides usage. ²¹ The robustness of the clustering structure, supported by silhouette analysis, and the alignment with species‐specific trends observed in the correlation matrix lend further biological credibility to the existence of these sensitivity subgroups.

Field parameters did not account for EC₅₀ distribution. After harmonising locations and excluding organic crops because of their small number, Location and Rotation did not explain TBZ or MFZ variability, and PDZ was effectively uniform; a small PDZ–Crop signal detected by a rank test was trivial and not supported by the confirmatory model. Thus, within the 2022 Irish F. langsethiae sample, the MFZ–TBZ sensitivity structure reflects intrinsic population variability rather than geography or coarse agronomy. Further work should relate EC₅₀ to genotype markers and documented spray histories to address mechanism and selection pressure.

From a disease management perspective, these results carry important implications. The variability in MFZ and TBZ sensitivity observed in F. langsethiae, along with evidence of partial cross‐resistance, highlights the risk of relying on a limited set of azole fungicides. Integrated pest management (IPM) approaches that include chemical rotation, mixtures with non‐azole modes of action and the use of partially resistant oat cultivars are essential to slow the spread of resistance. ¹⁹ , ²⁶ , ³⁰ In this context, the strong MFZ–TBZ correlation in F. langsethiae indicates that sequential use or co‐application of the two actives would concentrate selection on a shared sensitivity axis. Alternation across Fungicide Resistance Action Committee groups, with inclusion of a non‐azole partner where available, distributes selection pressure more broadly. Consequently, reserving a single, best‐timed azole spray for peak risk further limits cumulative selection, whereas choice of active can be guided by comparative baseline sensitivity within the data set. Field evidence for azole control of F. langsethiae in oats remains limited; where tested, azoles variably suppress growth and T‐2/HT‐2 production in vitro, indicating only partial control potential. ¹⁸ , ³¹ Together with our EC₅₀ distribution and partial cross‐resistance patterns, this supports a gradual erosion risk rather than imminent uniform failure and justifies continued sensitivity and toxin monitoring in oats. ¹⁸ , ²¹ , ²³ , ²⁴ , ³¹ The consistent efficacy of PDZ suggests that it remains a valuable component in resistance management, although its use should still be part of a diversified strategy to prevent future shifts in sensitivity.

Moreover, given the association between F. langsethiae and the production of T‐2 and HT‐2 toxins in oats, ¹⁴ and the restrictive European Union regulatory limits for these mycotoxins, ⁵⁹ continued monitoring of fungicide sensitivity in this species is critical for both yield protection and food safety. Our findings reinforce the value of routine EC₅₀ testing and statistical screening for emerging resistance trends. As climate change and agronomic intensification alter pathogen dynamics in northern European cereal systems, adaptive resistance monitoring will become increasingly important. ⁹ , ⁶⁰

Nonetheless, these EC₅₀ values quantify relative inhibition under controlled conditions and only provide baseline sensitivity distributions for surveillance. Therefore, the results should not be over‐interpreted as predictors of field performance, because translation to efficacy depends on multiple factors, such as dose and formulation, spray deposition and retention on panicles, pathogen load and infection timing, weather during and after application, and host physiology. Accordingly, product‐level recommendations require glasshouse or field validation that links EC₅₀ distributions to disease and mycotoxin outcomes at label rates and timings. In practice, these data are most useful for resistance management, guiding choice of mixtures and rotations with non‐azole modes of action and application timing within IPM, rather than to estimate control levels.

In conclusion, although azole fungicides remain broadly potent against Fusarium spp. in Irish oats, our data highlight emerging variability in MFZ and TBZ sensitivity among F. langsethiae isolates. The identification of distinct subgroups with reduced sensitivity, partial cross‐resistance and species‐specific trends underscores the need for ongoing surveillance and integrated resistance management. Targeted fungicide deployment based on accurate sensitivity profiles will be key to maintaining both crop protection and grain safety in evolving cereal systems.

Supporting information

Table S1. List of Fusarium isolates included in this study, showing species identification, sampling location, crop type, rotation category, and EC₅₀ values (μg mL⁻¹) for prothioconazole‐desthio (PDZ), tebuconazole (TBZ), and mefentrifluconazole (MFZ).

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.