During a survey from July to November 2017, twigs and branches of oak trees showing external disease symptoms, dieback, wilting, canker, gummosis and twig/branch discoloration, and bark cracking were collected from Zagros forests of five provinces: Ilam, Kermanshah, Kurdistan, Lorestan, and West Azarbaijan. Cross-sections of the samples were examined to classify internal disease symptoms. To isolate fungi, small pieces of disinfested woody tissues (with 70% ethanol for 3 min) were transferred on potato dextrose agar (PDA) supplemented with 100 mg/L streptomycin sulfate and ampicillin and incubated at 20–25°C. Pure cultures were obtained using single spore or hyphal tip methods on tap water agar (2% WA). The isolates were stored on PDA at 4°C. Representative isolates were deposited in the culture collection of the Iranian Research Institute of Plant Protection (IRAN, Tehran, Iran) and the CBS collection of the Westerdijk Institute, Utrecht, The Netherlands. Isolates that were sequenced and studied morphologically are available in Table 1 .

Total DNA was extracted from isolates grown on potato dextrose broth (PDB) after 7–10 d at 25°C using the modified Raeder and Broda (1985) method as described by Abdollahzadeh et al. (2009). To reduce the number of isolates for sequencing and microscopy, fungal isolates were grouped at the species level using the inter-simple sequence repeat (ISSR) technique based on DNA profiles generated with M13 primer (data not shown), and representative isolates of each group were sequenced. According to fungal taxonomic groups or genera deduced from morphology, different gene regions were amplified and sequenced using the following primer pairs: LROR/LR5 for a part of 28S nrDNA (LSU) (Vilgalys and Hester, 1990; Cubeta et al., 1991), ITS5/ITS4 for the ITS1-5.8S nrDNA-ITS2 region (ITS) (White et al., 1990), ACT512F/ACT783R for a part of actin (act1) (Carbone and Kohn, 1999), 5F (or 5F2)/7cR for a part of RNA polymerase II second largest subunit (rpb2) (Woudenberg et al., 2013), EF1-728F/EF1-986R (Vilgalys and Hester, 1990) for a part of translation elongation factor 1-alpha (tef-1α), and Bt2a (or T1)/Bt2b for a part of β-tubulin (tub2) (O’Donnell and Cigelnik, 1997; Glass and Donaldson, 1999). The PCR mixtures (25 µL) consisted of 1×PCR buffer, 3 mM MgCl₂, 200 µM of dNTPs, 5 pmol of each primer, 1 U of Taq DNA polymerase, and 1 µL of template DNA (50–100 ng/µL). For amplification of ITS and tub2 regions, we followed the PCR conditions as described by Bashiri et al. (2022). The PCR conditions for tef-1α and rpb2 were as follows: an initial denaturation step of 5 min at 95°C followed by 35 cycles of 30 s at 95°C, 30 s at 52°C (tef-1α)/45 s at 54°C (rpb2)/1 min at 55°C (LSU)/30 s at 58°C (act1), and 90 s at 72°C, with a final extension of 7 min at 72°C. PCR products were purified and sequenced by Elim Biopharm (USA) via FAZABiotec Co. (Tehran, Iran) and BGI (China) via BMG (Bio Magic Gene) Co. (Karaj, Iran). Consensus sequences were prepared with BioEdit v. 7.0.0 (Hall, 2004). All new sequences generated in this study were submitted to GenBank ( Table 1 ).

The generated sequences together with the sequences retrieved from GenBank were aligned using Clustal X v. 1.83 (Thompson et al., 1997) or online MAFFT v. 7 and edited manually in BioEdit v. 7.0.0, where necessary. Each locus was aligned separately and the alignments were concatenated with Mesquite 2.75 (Maddison and Maddison, 2023). Both single and combined loci were analyzed by Maximum Parsimony (MP), Maximum Likelihood (ML), and Bayesian Inference (BI). MP was performed using PAUP v. 4.0b10 (Swofford, 2003). ML and BI were carried out through the online CIPRES Science Gateway (Miller et al., 2012) using RAxML-HPC BlackBox v. 8.2.10 (Stamatakis, 2014) and MrBayes v. 3.2.6 (Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003), respectively. MP analysis was executed according to Abdollahzadeh et al. (2013). Optimal nucleotide substitution models were detected for each locus using MrModelTest v. 2.3 (Nylander, 2004). The ML and Bayesian analyses were executed as described by Abdollahzadeh et al. (2020). Phylograms were plotted using FigTree v. 1.4.3 and edited in Adobe Illustrator CS2 v. 12.0.0. Alignments and trees were deposited in TreeBASE (www.treebase.org; S30794, S30795, and S30796) and taxonomic novelties registered in MycoBank (Crous et al., 2004).

Depending on the fungal morphological group, culture characteristics and microscopic fungal structures were examined and recorded from cultures grown on PDA, oat meal agar (OA), malt extract agar (MEA), and yeast malt agar (YMA) at room temperature. Sporulation and fruiting body production were induced under a combination of near-UV and cool-white fluorescent lights under 12-h light/12-h dark conditions. The structure and dimensions of microscopic features were determined and measured in 100% lactic acid or distilled water using an Olympus DP72 camera on an Olympus BX51 microscope and a Cell Sense Entry measurement module. To compute the dimensions of each fungal structure, mean, standard deviation, and 95% confidence intervals were estimated based on at least 30 microscopic measurements. Dimensions are presented as the range of measurements with extremes in brackets followed by mean ± standard deviation.

Pathogenicity tests of recognized species were performed on leaves and stems of 2-year-old oak (Q. brantii) seedlings in Petri plates and under greenhouse conditions, respectively. The surface of leaves and stems was disinfested with 70% ethanol. Small pieces of bark (≈ 3 × 3 mm) were cut from the stems of potted seedlings and small scratches (≈3 mm) were made on the leaves. Mycelial plugs of 7-d-old colonies on PDA were inoculated on wounded leaves and stems. Controls were inoculated with sterile PDA plugs. To evaluate the phytotoxic activity of fungal isolates, inoculated leaves were incubated at 25°C for a period of 7 days. The inoculated seedling wounds were wrapped with Parafilm and placed under greenhouse conditions (22–28°C) and watered as needed. As the seedlings declined, the external symptoms were recorded and the extent of vascular discoloration (lesion length) was measured. To confirm Koch’s postulates, fungal isolates were re-isolated from inoculated leaves and stems on PDA at 25°C and morphologically re-examined. The greenhouse trials were performed as a completely randomized nested design (CRND) with three replications per treatment. To determine differences in lesion lengths caused by inoculated fungi, one-way ANOVA was executed. Homogeneity of variance and normality assumption were examined using Bartlett and Shapiro–Wilk’s tests, respectively. Least significant difference (LSD) values were calculated (p = 0.05) using SAS v. 9.1.3.

A total of 462 fungal isolates were collected from oak trees (Q. brantii, Q. infectoria, and Q. libani), of which 184 isolates (40%) were accommodated in coelomycetous Dothideomycetes. Various external and internal symptoms ( Figures 1 , 2 ) were observed on trees’ and twigs’ cross-sections listed in Table 1 . Primarily, borer feeding sites (e.g., Buprestidae) were observed in association with irregular and central wood necrosis and no correlation was found between fungal species and type of disease symptoms.

According to DNA profiles generated with M13 primer (data not shown), 36 isolates as representatives of recognized DNA banding patterns were selected for morphological and phylogenetic analyses.

Based on morphology and DNA sequence data (LSU, ITS, rpb1, rpb2, tef-1α, tub2, acl1, act1, and caM), 24 fungal species corresponding to 19 genera were recognized ( Figures 3 , 4 ). Of these, eight species with 184 isolates (40%) were placed in coelomycetous Dothideomycetes, namely, four known species—Botryosphaeria dothidea (19 isolates/4.1%), Didymella glomerata (16 isolates/3.5%), Kalmusia variispora (37 isolates/8.1%), and Neoscytalidium dimidiatum (20 isolates/4.3%)—and four new species that are described and named Alloeutypa iranensis sp. nov. (18 isolates/3.9%), Cytospora hedjaroudei sp. nov. (22 isolates/4.8%), Cytospora zagrosensis sp. nov. (24 isolates/5.2%), and Gnomoniopsis quercicola sp. nov. (28 isolates/6.1%) ( Table 1 , Figure 4 ). Biscogniauxia, Neocosmospora, and Cytospora were the most prevalent fungi with frequencies of 12.3%, 11.5%, and 10%, respectively ( Figure 3 ). Among the coelomycetous fungi, Cytospora and Kalmusia were the most common fungi associated with the decline of oak trees followed by Gnomoniopsis, Neoscytalidium, Botryosphaeria, Alloeutypa, and Didymella ( Figure 3 ). Moreover, Biscogniauxia rosacearum, K. variispora, and Neocosmospora sp. were the most prevalent fungi at the species level followed by G. quercicola, Fimetariella rabenhorstii, C. zagrosensis, and C. hedjaroudei ( Figure 4 ).

Megablast search of the GenBank nucleotide database with ITS sequences revealed that our isolates are close to the members of Diatrypaceae (Eutypa/Eutypella), Cytosporaceae (Cytospora), and Gnomoniaceae (Discula/Gnomoniopsis). Thus, based on blast searches, preliminary phylogenetic analyses, literature, and fungal databases (indexfun-gorum/speciesfungorum/mycobank), three datasets were provided and analyzed.

The first DNA sequence dataset (ITS: 71, tub2: 47 sequences) consisted of our selected isolate IRAN 4323C, 68 isolates belonging to 66 species of the Diatrypaceae family, and two outgroups Xylaria hypoxylon CBS 122620/Kretzschmaria deusta CBS 826.72. The aligned datasets of ITS (691) and tub2 (454) were combined and subjected to MP, ML, and BI. After alignment, the combined dataset consisted of 1,145 characters including alignment gaps. Of these, 490 were constant, 192 were variable and parsimony-uninformative, and 463 were parsimony-informative. MP analysis of the remaining 463 parsimony-informative characters resulted in 729 most parsimonious trees (TL = 2,925, CI = 0.41, RI = 0.6, HI = 0.59). MrModelTest revealed that the general time-reversible model of evolution (Rodríguez et al., 1990), including estimation of invariable sites and assuming a discrete gamma distribution (GTR+I+G) with six rate categories (lsetnst = 6, rates = invgamma) and dirichlet (1,1,1,1) base frequencies, is the best nucleotide substitution model for both loci (ITS, tub2). The Bayesian analyses of the concatenated alignments of two loci generated 3,682 trees from which 920 trees were discarded as burn-in. The consensus tree and posterior probability values (PP) were calculated from the remaining 2,762 trees. The average standard deviation of split frequencies was 0.009962 at the end of the run. The RAxML search of the dataset with 760 distinct alignment patterns produced a best-scoring ML tree (lnL = −14,040.126555). The ML and Bayesian phylogenetic trees were mapped on the MP tree shown in Figure 5 with MP/ML/BI bootstrap support and posterior probability values at the nodes. In these analyses, our isolate IRAN 4323C was placed in a distinct clade in Alloeutypa, a new genus recently introduced in Diatrypaceae, which is recognized here as a new species named Alloeutypa iranensis sp. nov. close to A. flavovirens CBS 272.87 and isolate MFLU 19-0911 ( Figure 5 ). A. iranensis showed remarkable differences in nucleotide sequences with A. flavovirens CBS 272.87 (ITS: 7 substitutions; tub2: 7 substitutions, 21 deletions/insertions) and MFLU 19-0911 (ITS: 5 substitutions; tub2: 6 substitutions, 25 deletions/insertions). It is also significantly distinct from the type species A. milinensis FSATAS 4309 based on ITS (12 substitutions, 1 deletion/insertion) and tub2 (16 substitutions, 25 deletions/insertions) sequence data. Isolate MFLU 19-0911 is placed in a distinct clade and is apparently a representative of a new Alloeutypa species.

The second dataset consisting of DNA sequences of six loci (LSU, ITS, rpb2, act1, tef-1α, and tub2) was analyzed to examine the taxonomic position of our Cytospora isolates. New generated sequences were aligned with available authentic sequences of Cytospora species and Diaporthe vaccinii CBS 160.32 as an outgroup (LSU: 43, ITS: 47, rpb2: 41, act1: 44, tef-1α: 39, tub2: 29 sequences). The concatenated alignment of LSU (803), ITS (593), rpb2 (664), act1 (311), tef-1α (420), and tub2 (503) was subjected to MP, ML, and BI. After alignment, the dataset consisted of 3,295 characters including alignment gaps. Of these, 2,283 were constant, 222 were variable and parsimony-uninformative, and 790 were parsimony-informative. MP analysis of the remaining 790 parsimony-informative characters resulted in 162 most parsimonious trees (TL = 3,345, CI = 0.48, RI = 0.67, HI = 0.52). MrModelTest revealed that the general time-reversible model of evolution (Rodríguez et al., 1990), including estimation of invariable sites and assuming a discrete gamma distribution (GTR+I+G) with six rate categories (lsetnst = 6, rates = invgamma) and dirichlet (1,1,1,1) base frequencies, is the best nucleotide substitution model for all loci (LSU, ITS, rpb2, act1, tef-1α, and tub2). The Bayesian analyses of the concatenated alignments of six loci generated 1,172 trees from which 292 trees were discarded as burn-in. The consensus tree and posterior probability values (PP) were calculated from the remaining 880 trees. The average standard deviation of split frequencies was 0.009890 at the end of the run. The RAxML search of the dataset with 1,272 distinct alignment patterns produced a best-scoring ML tree (lnL = −20,279.905622). The MP and ML trees were mapped on the Bayesian phylogenetic tree as shown in Figure 6 with MP/ML/BI bootstrap support and posterior probability values at the nodes.

Our isolates were placed in two distinct clades differentiated from all other Cytospora species that are recognized here as two new species. Two isolates, IRAN 4324C and IRAN 4326C, formed a well-supported clade as a sister group with C. macropycnidia CBS 149338 and named Cytospora zagrosensis sp. nov. This species is differentiated from C. macropycnidia in ITS (2 substitutions), rpb2 (2 substitutions), tef-1α (6 substitutions, 2 deletions/insertions), and tub2 (11 substitutions, 3 deletions/insertions) nucleotide sequences. The third isolate IRAN 4325C represents the second new species close to C. pistaciae CBS 144238 and C. parapistaciae CBS 144506, which is named Cytospora hedjaroudei sp. nov. ( Figure 6 ). Based on nucleotide sequences, C. hedjaroudei differs from two closely related species C. pistaciae (ITS: 5 substitutions, 1 deletion/insertion; tef-1α 1 substitution) and C. parapistaciae (ITS: 8 substitutions, 28 deletions/insertions; tef-1α: 4 substitutions, 1 deletion/insertion). It is noticeable that ITS sequence data are enough to separate C. hedjaroudei from both C. pistaciae and C. parapistaciae.

The third dataset contained three loci DNA sequences (ITS: 31, tef-1α: 23, and tub2: 26 sequences) of our selected isolate IRAN 4313C, 29 Gnomoniopsis species, and Melanconis stilbostoma CBS 109778 as an outgroup. The aligned single gene sequences were concatenated and subjected to MP, ML, and BI. The combined dataset consisted of 1,918 characters (ITS: 606, tef-1α: 370, and tub2: 942) including alignment gaps. Of these, 1,099 were constant, 278 were variable and parsimony-uninformative, and 541 were parsimony-informative. MP analysis of the remaining 541 parsimony-informative characters resulted in three most parsimonious trees (TL = 2,367, CI = 0.55, RI = 0.6, HI = 0.45). MrModelTest revealed that the general time-reversible model of evolution (Rodríguez et al., 1990), including estimation of invariable sites and assuming a discrete gamma distribution (GTR+I+G) with six rate categories (lsetnst = 6, rates = invgamma) and dirichlet (1,1,1,1) base frequencies, is the best nucleotide substitution model for all loci (ITS, tef-1α, and tub2). The Bayesian analyses of the concatenated alignments of three loci generated 4,892 trees from which 1,222 trees were discarded as burn-in. The consensus tree and posterior probability values (PP) were calculated from the remaining 3,670 trees. The average standard deviation of split frequencies was 0.009837 at the end of the run. The RAxML search of the dataset with 976 distinct alignment patterns produced a best-scoring ML tree (lnL = −13,465.491131). The ML and Bayesian phylogenetic trees were mapped on the MP tree shown in Figure 7 . Our isolate IRAN 4313C was clustered in a strongly supported clade close to G. paraclavulata CBS 123202 with significant nucleotide differences in all three loci ITS (5 substitutions, 2 deletions/insertions), tef-1α (35 substitutions, 8 deletions/insertions), and tub2 (58 substitutions, 9 deletions/insertions). Thus, it is recognized as a new species and named Gnomoniopsis quercicola sp. nov. ( Figure 7 ).

Based on phylogenetic analyses and morphology, eight coelomycetous fungal species were characterized and associated with declined oak trees in the central and northern part of Zagros forests in Iran. Of these, four species were recognized as new fungal species for science, which are described here as follows:

Alloeutypa iranensis S. Bashiri & Abdollahz., sp. nov. ( Figure 8 )

MycoBank: MB850401.

Diagnosis: A. iranensis is the third species to be described in Alloeutypa. Since the type species A. milinensis was described based on the sexual morph and we have only observed the asexual morph, it is only possible to morphologically distinguish A. iranensis from A. flavovirens based on the larger dimensions of conidiomata (0.9–3.9 mm vs. 0.7–1.3 mm), conidiophores (12.6 × 3 μm vs. 4.9 × 2.4 μm), and conidiogenous cells (32.4 × 2.6 μm vs. 11.5× 2.4 μm) and shorter conidia (28.5 μm vs. 34 μm).

In this survey, pathogenicity of representative isolates was examined under in vitro (on leaves) and greenhouse (on seedling stems) conditions ( Figures 12 – 15 , Supplementary Figures 1–4 ). In both experiments, 19 fungal species were determined as pathogenic, and Alternaria spp. (A. alternata, A. atra, and A. contlous), Chaetomium globosum, and Parachaetomium perlucidum were nonpathogenic.

In the phytotoxicity assay under in vitro conditions on oak leaves, we qualitatively evaluated and grouped pathogenic species in four classes as highly virulent (Biscogniauxia persica, B. rosacearum, Botryosphaeria dothidea, Gnomoniopsis quercicola, and Neoscytalidium dimidiatum), moderately virulent (Cosmospora butyri, Didymella glomerata, Fusarium anulatum, Kalmusia variispora, and Neocosmospora sp.), weakly virulent (Alloeutypa iranensis, Cytospora hedjaroudei, C. zagrosensis, Fimetariella rabenhorstii, Neocosmospora metavorans, Phaeoacremonium tuscanicum, and Stilbocrea anihashemiana), and very weak virulent (Apiospora intestini and Nigrospora sp.).

In pathogenicity tests under greenhouse conditions, Biscogniauxia rosacearum, B. persica, Botryosphaeria dothidea, G. quercicola, and N. dimidiatum were the most virulent species, which caused a quick decline and death of inoculated seedlings after 28–30 days. Thus, to evaluate the pathogenicity of these species in a distinct analysis, we recorded the external and internal symptoms and analyzed them after 30 days ( Table 2 ). Pathogenicity of the other examined species were recorded after 60 days and analyzed separately ( Table 3 ). In the first statistical analyses, significant differences were determined in lesion lengths between species (F = 54.22, p < 0.001) and B. rosacearum was the most virulent species and G. quercicola was the most virulent coelomycetous species. Gnomoniopsis quercicola showed high phytotoxic activity on leaves examined under in vitro conditions ( Figure 15D ). Severe leaf yellowing and defoliation on inoculated seedlings and extended brown wood necrosis in cross-sections of inoculated stems were observed 28–30 days after inoculation under greenhouse conditions ( Figure 15 ). Botryosphaeria dothidea and N. dimidiatum showed high phytotoxic activity on leaves ( Supplementary Figures 1–4D ) and inoculated seedlings showed severe leaf blight and defoliation together with wedge-shaped and irregular wood necrosis in cross-sections of inoculated stems 28–30 days after inoculation ( Supplementary Figures 1–4 ). The second statistical analysis also indicated significant differences in lesion lengths between species (F = 318.33, p < 0.001) and all coelomycetous species; A. iranensis, C. hedjaroudei, C. zagrosensis, D. glomerata, and K. variispora were pathogenic and caused leaf necrosis, blight, and defoliation on oak seedlings under greenhouse conditions 60 days after inoculation. Moreover, cross-sections of inoculated stems showed an intermediate vascular discoloration and irregular wood necrosis ( Table 3 , Figures 12 – 14 , Supplementary Figures 2, 3 ). All inoculated pathogens were re-isolated from inoculated leaves and seedling stems, confirming Koch’s postulates ( Figures 12 – 15 , Supplementary Figures 1–4 ).