# Source paper archive - Source title: Seed mycobiome assembly varies across mother trees with contrasting levels of resistance to Dutch elm disease - Author(s): Martinez-Arias, C., Fernandez-Quiros, S., Rodriguez-Calcerrada, J., Martin, J. A. - Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.11.765837v1 - PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.11.765837v1.full.pdf - PDF SHA-256: 5f1000de01fb735d02425709c143ed96be3786b5c39947b9c3c5a907b552eaa3 - Converter: pymupdf4llm 0.0.27 - Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version. --- [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 1 **Seed mycobiome assembly varies across mother trees with contrasting levels of** 2 **resistance to Dutch elm disease** 3 Martínez-Arias, Clara [1,2*;] Fernández-Quirós, Susana [3] ; Rodríguez-Calcerrada, Jesús [3,4] ; 4 Martín, Juan Antonio [3,4] . 5 1 Departamento de Biotecnología-Biología Vegetal, Escuela Técnica Superior de Ingeniería 6 Agronómica, Alimentaria y de Biosistemas, Universidad Politécnica de Madrid (UPM), 7 Madrid, 28040, Spain. 8 2 Centro de Biotecnología y Genómica de Plantas (CBGP), Universidad Politécnica de 9 Madrid (UPM/CSIC), Campus de Montegancedo UPM, Pozuelo de Alarcón, Madrid 28223, 10 Spain. 11 3 Departamento de Sistemas y Recursos Naturales, ETSI Montes, Forestal y del Medio 12 Natural, Universidad Politécnica de Madrid, c. José Antonio Novais 10, 28040, Madrid, 13 Spain 14 4 Centro para la Conservación de la Biodiversidad y el Desarrollo Sostenible, Universidad 15 Politécnica de Madrid, C/José Antonio Novais 10, 28040, Madrid, Spain 16 ORCID iDs: 17 Clara Martínez-Arias 0000-0002-0964-1125 18 Jesús Rodríguez-Calcerrada 0000-0001-7408-7653 19 Juan Antonio Martín 0000-0002-3207-1319 20 21 *Corresponding author: clara.martinez.arias@upm.es 22 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 23 **Abstract** 24 Seed-associated fungal communities may influence plant performance from the earliest 25 stages of development, yet their potential association with disease resistance in long-lived 26 forest trees remains poorly understood. Here, we investigated whether _Ulmus minor_ trees 27 with different levels of resistance to Dutch elm disease (DED) exhibit consistent differences 28 in their seed-associated mycobiome. We characterized the fungal communities inhabiting 29 samaras from nine _U. minor_ genotypes with different levels of DED resistance (i.e. resistant, 30 intermediate and susceptible genotypes) using high-throughput sequencing of the internal 31 transcribed spacer 1 (ITS1) region. We identified 522 fungal amplicon sequence variants 32 (ASVs), including 74 core members of the samara mycobiome. The community was 33 dominated by Dothideomycetes and Tremellomycetes, but both diversity and community 34 composition differed among resistance groups. Susceptible genotypes harbored richer 35 fungal communities, whereas resistant genotypes supported less diverse but more distinct 36 assemblages. Differential abundance analysis identified 21 genera associated with host 37 resistance level, including an enrichment of _Protofenestella_ in resistant genotypes and of 38 _Taphrina_ in intermediate and susceptible genotypes. Co-occurrence network analysis 39 further revealed contrasting patterns of community organization, with resistant genotypes 40 showing more modular and less connected networks than susceptible genotypes. Together, 41 these findings indicate that DED resistance is associated with distinct patterns of seed 42 mycobiome assembly, diversity and taxonomic composition in _U. minor_ . The samara 43 mycobiome may therefore represent a previously overlooked component associated with 44 elm resistance, and provides candidate fungal taxa for future functional validation, with 45 potential applications in elm conservation and mycobiome-informed breeding. 46 **Keywords** 47 Seed, fungi, metabarcoding, forest tree, aggressive disease. [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 48 **Introduction** 49 Seeds are vital reservoirs of bacterial and fungal assemblages and a fundamental 50 component of plant–microbe interactions [1]. Seed colonization occurs through three main 51 pathways: (i) an internal maternal route, via the xylem or the non-vascular tissues of the 52 mother plant (maternal vertical transmission); (ii) a floral route, through the stigma (paternal 53 vertical transmission) [2]; and (iii) an external route, from the surrounding environment 54 (horizontal transmission) [3]. Vertically transmitted microbes seem to exhibit evolutionary 55 stability, shaped by natural selection favoring mutualistic interactions [4–6], and may 56 enhance plant survival and fitness under biotic and abiotic stresses [6]. 57 Nevertheless, seed microbiome research has focused mainly on bacteria, while seed 58 associated fungi remain underexplored, particularly in perennial woody species from 59 natural ecosystems [4]. Plant-associated microbes confer numerous benefits to their hosts 60 [7,8], and fungal endophytes can suppress pathogens directly, through antibiosis, 61 mycoparasitism or niche competition, and indirectly, through nutrient facilitation, hormone 62 modulation or priming of immune responses such as induced systemic resistance [9–12]. 63 Seeds are thus also vectors of microbial inheritance, transmitting fungal endophytes that 64 may influence plant establishment, stress tolerance and disease resistance [13]. The 65 structure and diversity of seed-borne communities depend strongly on host genotype, 66 suggesting that the seed mycobiome may contribute to the differences in disease 67 resistance observed among genotypes [14], and specific taxa have been proposed as 68 indicators of disease-suppressive seed microbiomes [15]. Characterizing the seed 69 mycobiome could therefore provide insights into early-life plant resilience and into the role 70 of the inherited microbiome in long-term resistance against pathogens. 71 This is particularly relevant for forest species threatened by diseases for which chemical 72 control is unfeasible, such as Dutch elm disease (DED). This vascular wilt disease, caused 73 by _Ophiostoma ulmi_ and _O. novo-ulmi_, has devastated _Ulmus_ populations in Europe and 74 North America [16,17], where elms were once keystone components of riparian 75 ecosystems [18]. The pathogen, disseminated by elm bark beetles, rapidly colonizes the 76 vascular tissues, causing xylem blockage, foliage wilting and eventually tree death [17]. 77 Identifying and propagating resistant _Ulmus minor_ genotypes has therefore become a 78 cornerstone of elm conservation and reforestation. Variation in DED symptoms within 79 natural populations dominated by a single elm genotype, together with differences in 80 resistance among genotypes in controlled trials, suggests that factors beyond host [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 81 genetics, such as the plant microbiome, may be involved in DED resistance [19,20]. 82 Moreover, DED resistance may be transmitted to a greater extent through the maternal 83 lineage [21–23]; although maternal and paternal effects are difficult to disentangle, 84 transmission of the maternal microbiome to the offspring could be one of the underlying 85 mechanisms. Indeed, certain fungal endophytes isolated from stems of resistant 86 genotypes can mitigate DED by activating the plant immune response, reducing symptoms 87 and buffering the pathogen-induced loss of mycobiome diversity [12,20,24]. 88 Here, we characterize for the first time the samara mycobiome of _U. minor_ genotypes with 89 documented differences in susceptibility to _O. novo-ulmi_ . We hypothesize that samaras 90 from DED-resistant genotypes harbor fungal assemblages distinct from those of 91 susceptible genotypes, enriched in protective endophytic taxa. Using high-throughput 92 sequencing, we aimed to: (i) compare fungal diversity and community composition in 93 samaras from _U. minor_ genotypes with different resistance levels to DED; (ii) identify fungal 94 indicators associated with resistant and susceptible genotypes; and (iii) characterize co 95 occurrence patterns and potential interactions among fungal taxa. These insights could 96 support microbiome-assisted breeding and sustainable disease mitigation strategies for 97 elm conservation. 98 **Materials and methods** 99 **Seed sampling and storage** 100 Samaras were collected in April 2020 from nine healthy _Ulmus minor_ genotypes in an _ex-_ 101 _situ_ conservation plot at the “Puerta de Hierro” Forest Breeding Center (3°45’6” W, 102 40°27’23” N; 595 m a.s.l., Madrid, Spain). Genotypes were selected to represent a range of 103 DED resistance levels, previously determined in inoculation trials of the Spanish elm 104 breeding program on trees at least 4 years old, using a minimum of 6 ramets per genotype 105 and following the methodology described by [25]. After two consecutive years of artificial 106 inoculation with _O. novo-ulmi_, genotypes VAD2, JCA2 and MDV5 showed crown wilting 107 below 35 % (hereafter resistant genotypes), GRDF3, ABAL1 and MAPD2 between 35 % and 108 65 % (intermediate genotypes), and MDV1, TOPB1 and MAPD1 above 65 % (susceptible 109 genotypes) **(Table 1)** . All sampled trees grew under very similar micro-site conditions (soil 110 properties, surrounding vegetation and 4 × 4 m spacing) and received drip irrigation in spring 111 and summer to prevent drought stress. 112 Samaras were collected from one adult tree per genotype (16–29 years old, **Table 1** ; 6–12 m 113 in height; 17–32 cm in trunk diameter). Mature samaras at a similar phenological stage (light [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 114 beige to yellowish-brown pericarp) were sampled with a pole lopper from four fruit-bearing 115 branches per tree, cut from two canopy orientations (northwest and southeast) and from 116 the lower and upper crown. Samples from each tree were pooled, transported to the 117 laboratory at 4 °C, checked manually for fully developed and viable seeds, surface 118 sterilized, frozen in liquid N 2 and stored at –80 °C. For surface sterilization, 30–35 randomly 119 selected samaras per genotype were shaken for 1 min in 20 mL of filter-sterilized (0.22 µm 120 PES) 1 % Tween 20 solution (Merck KGaA, Darmstadt, Germany), rinsed with autoclaved 121 distilled water, shaken for 1 min in 70 % ethanol, rinsed twice with autoclaved distilled water 122 and air-dried on sterile filter paper in a laminar flow hood. 123 **Table 1:** DED resistance level of the nine _Ulmus minor_ genotypes included in the study. 124 Mean foliage wilting percentages were obtained from previous inoculation trials during 125 which an aggressive isolate of _Ophiostoma novo-ulmi_ was inoculated in two consecutive 126 years, and refer to 60 days after inoculation. These trials were performed during the regular 127 activities of the Spanish elm breeding program in “Puerta de Hierro” Forest Breeding Center 128 (Madrid, Spain) from 2001 to 2021. **Susceptibility test** **Resistance** **Genotype Provenance in Spain** **class** **Age of** **sampled tree** **Foliage wilting after** **Foliage wilting after** **N** **1st inoculation (%)** **2nd inoculation (%)** VAD2 Valencia, Ademúz Resistant 16 3.5 ± 2.5 8.3 ± 2.7 12 JCA2 Jaén, Cazorla Resistant 28 24.0 ± 5.4 30.8 ± 6.1 6 MDV5 Madrid, Dehesa de Amaniel Resistant 22 10.7 ± 4.7 25.7 ± 4.5 12 GRDF3 Granada, Deifontes Intermediate 28 23.1 ± 4.6 50.9 ± 5.1 9 ABAL1 Albacete, Alatoz Intermediate 22 43.7 ± 6.4 44.2 ± 7.8 6 MAPD2 Málaga, Pedrizas Intermediate 22 43.5 ± 6.8 49.5 ± 6.7 10 MDV1 Madrid, Dehesa de la Villa Susceptible 29 72.9 ± 4.5 94.1 ± 5.5 10 TOPB1 Toledo, Puebla de Montalbán Susceptible 29 64.5 ± 8.4 100 ± 0 6 MAPD1 Málaga, Pedrizas Susceptible 28 67.5 ± 6.7 90.0 ± 7.7 8 129 **DNA extraction and fungal metabarcoding** 130 From each genotype, three subsamples of 10–12 samaras were randomly selected (27 131 subsamples in total) and ground to a fine powder in a ball mill (Mixer mill MM 400, Retsch 132 GmbH, Haan, Germany). Genomic DNA was extracted from 50 mg of powder with the 133 Norgen Plant/Fungi DNA Isolation Kit (Norgen, Thorold, Canada). After quality control, the 134 first internal transcribed spacer (ITS1) region was amplified with primers ITS1F and ITS2, and 135 libraries were prepared with the Nextera XT Index Kit v2 and sequenced on an Illumina MiSeq 136 platform (2 × 300 bp paired-end reads) by Macrogen Inc. (Seoul, Republic of Korea). [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 137 **Analysis of sequencing data and metabarcoding** 138 Raw reads were processed with the DADA2 pipeline v1.22.0 [26] in R (R Development Core 139 Team, Posit Software, PBC, Boston, MA). Primers were removed with CUTADAPT 4.1 [27], 140 chimeras were removed with _removeBimeraDenovo_, and the resulting amplicon sequence 141 variants (ASVs) were taxonomically assigned with the naïve Bayesian classifier [28] against 142 the latest UNITE general fasta release for fungi [29]. ASV counts, taxonomy and sample 143 metadata were combined into a phyloseq object [30], and only fungal sequences were 144 retained. ASVs present in < 5 % of the samples with fewer than five reads were removed as 145 potential artifacts or low-abundance contaminants. For alpha and beta diversity analyses, 146 counts were normalized by scaling with ranked subsampling (SRS) using _srs_p_ from the 147 “QsRutils” package [31]. 148 **Data analysis** 149 _Alpha diversity_ 150 Observed richness and the Shannon and Simpson indices were calculated with the 151 “microbiome” package [32]; whereas richness considers only the number of ASVs, the 152 Shannon and Simpson indices also account for their relative abundances [33]. Indices were 153 compared with a nested ANOVA, with genotypes nested within resistance levels, using _aov_ 154 with genotype as the error stratum, after checking the normality of residuals (Shapiro–Wilk 155 and Kolmogorov–Smirnov tests, skewness, kurtosis and normal probability plots). We used 156 the functions _shapiro.test_ from the R package “stats” to perform the Shapiro-Wilks test and 157 the function _ks.test_ from the R package “dgof” for the Kolmogorov-Smirnov test. Skewness 158 and Kurtosis values were obtained with the function _skewness_ and _kurtosis_ from the R 159 package “moments,” and normal probability plots were drawn using the R packages 160 “ggplot” and “qqplotr.” When the effect of resistance level was significant (P < 0.05), groups 161 were compared with Fisher’s protected least significant difference (LSD) test on genotype 162 means. 163 _Beta diversity_ 164 Bray–Curtis dissimilarities, calculated on relative abundances, were visualized by principal 165 coordinate analysis (PCoA) with _plot_ordination_ (“phyloseq”) and tested by permutational 166 multivariate ANOVA (PERMANOVA) with _adonis2_ (“vegan”) _[34]_ . Because the three 167 subsamples of each genotype came from a single tree, permutations (9,999) were restricted 168 so that whole genotypes, rather than individual samples, were exchanged among resistance 169 levels ( _how_ function, “permute” package). Homogeneity of multivariate dispersion was [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 170 tested with _betadisper_ and _permutest_ under the same permutation scheme. Pairwise 171 comparisons between resistance levels used the same restricted permutations; as only ten 172 distinct permutations are possible when comparing two groups of three genotypes, 173 pairwise results are reported as effect sizes (R [2] ) and interpreted descriptively. 174 _Relative abundance of taxa_ 175 Relative abundances at the class level were represented for ASVs with a relative abundance 176 higher than 1 % and after filtering free sequence artifacts and low abundant contaminants, 177 using the _tax_glom_ and _plot_bar_ functions (“phyloseq”). 178 _Core and exclusive seed fungal members_ 179 Core and exclusive members were from relative abundances identified 180 ( _transform_sample_counts_, “phyloseq”) for each resistance level. An ASV was considered 181 part of the core of a group when it exceeded a relative abundance of 0.001 % and a 182 prevalence of 50 % across replicates ( _core_members_, “microbiome” [32]). The core shared 183 _intersect_ by all groups and the members exclusive to each group were identified with the 184 and _setdiff_ functions, respectively, and visualized with Venn diagrams (“ggVennDiagram” 185 [35]). 186 _Comparison of_ U. minor _seed mycobiome with an available_ U. minor _stem mycobiome_ 187 Seed ASVs were compared with the OTUs previously obtained from stems of adult _U. minor_ 188 trees in the same plot [19], a study that included seven of the genotypes analyzed here. 189 Representative ITS sequences of the stem mycobiome were used as BLASTN queries (NCBI 190 BLAST+ v2.17.0+, R-based workflow) against a local database built from the samara 191 sequences. Matches with an E-value ≤ 10⁻⁵, an alignment length ≥ 100 bp and 100 % identity 192 were considered shared taxa and linked to their taxonomic assignment. Because the stem 193 dataset was generated by 454 pyrosequencing, with a lower sequencing depth than 194 Illumina, we assumed that it captured the most abundant and representative members of 195 the stem mycobiome; we therefore restricted the comparison to the taxa shared by both 196 tissues, and their relative abundances are presented for descriptive purposes. 197 _Differential abundance_ 198 Differences in the relative abundance of fungal genera between resistance levels were 199 tested with the “DESeq2” package [36]. ASVs were agglomerated at the genus level 200 ( _tax_glom_, “phyloseq” [30]) and converted with _phyloseq_to_deseq2_, with resistance level 201 as the design factor and subsamples as replicates. As DESeq2 does not accommodate 202 random effects, the non-independence of the three subsamples from each tree could not [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 203 be explicitly modelled. Pairwise comparisons between resistance levels were performed 204 with Wald tests and P-values were adjusted with the Benjamini–Hochberg method; genera 205 with adjusted P < 0.05 were considered differentially abundant. 206 _Co-occurrence networks_ 207 Co-occurrence networks were built for each resistance level from pairwise Spearman 208 correlations between ASVs, calculated on the SRS-normalized counts of the nine samples 209 of each level. Only robust and significant correlations (|ρ| > 0.75, P < 0.05) were retained, 210 and positive and negative correlations were interpreted as potential co-occurrence and co 211 exclusion relationships, respectively. Networks were analyzed with the “igraph” package; 212 modules were identified with the fast-greedy algorithm (checked with the walktrap 213 algorithm), and the modularity of each network was compared with that of 1,000 Erdős– 214 Rényi random networks with the same number of nodes and edges. 215 **Results** 216 **Fungal community and taxonomic composition** 217 Sequencing of the 27 subsamples yielded 4,233,864 reads (mean 156,809 per subsample; 218 range 146,543–193,866), distributed in 1,316 ASVs, of which 522 were retained after 219 filtering. These were assigned to 253 species, 166 genera, 101 families, 51 orders and two 220 phyla, Ascomycota (70 % of reads) and Basidiomycota (30 %). Dothideomycetes, 221 Tremellomycetes, Taphrinomycetes and Microbotryomycetes dominated in all genotypes 222 ( **Fig. 1** ). Although the same major classes were detected at all resistance levels, their 223 relative abundances differed markedly among them. 224 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 225 **Fig. 1** . Class-level relative abundance of the samara fungal communities in the three 226 replicates of samaras from nine _Ulmus minor_ genotypes with contrasting levels of 227 resistance to Dutch elm disease (DED). VAD2, JCA2 and MDV5 are DED-resistant 228 genotypes; GRDF3, ABAL1 and MAPD2 are DED-intermediate genotypes; and MDV1, TOPB1 229 and MAPD1 are DED-susceptible genotypes. 230 **Biodiversity analysis and shared seed fungal members between elm resistance levels** 231 Alpha diversity differed among resistance levels, particularly for observed richness and 232 Shannon diversity ( **Fig. 2A** ). Richness differed significantly among all three levels, being 233 highest in susceptible genotypes, intermediate in resistant genotypes and lowest in 234 intermediate genotypes. Shannon diversity was significantly higher in susceptible 235 genotypes than in resistant and intermediate ones, which did not differ from each other. The 236 Simpson index also tended to be higher in susceptible genotypes, although differences 237 were marginally non-significant (P = 0.07; **Table 2** ). Overall, susceptible genotypes 238 supported richer and more diverse fungal communities than resistant and intermediate 239 genotypes. 240 241 **Fig. 2** . **(A)** Alpha diversity of the _Ulmus minor_ samara mycobiome based on observed 242 amplicon sequence variant (ASV) richness, Shannon diversity and Simpson diversity 243 indices according to DED-resistance level. Different letters indicate significant differences 244 between groups according to Fisher’s post hoc test ( _P_ < 0.05) after nested one-way ANOVA 245 analysis in which resistance level was tested against the variation among genotypes. **(B).** [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 246 Differences in fungal community composition of the _Ulmus minor_ samara mycobiome 247 according to DED-resistance level. Principal coordinate analysis (PCoA) of mycobiome 248 composition based on Bray-Curtis dissimilarities among the three DED-resistance levels. 249 _R_ ² and _P_ values indicate the results of the PERMANOVA analysis. **(C)** Venn diagram showing 250 shared and exclusive amplicon sequence variants (ASVs) among the three DED-resistance 251 levels. 252 PERMANOVA revealed significant differences in community composition among resistance 253 levels, which explained 30.2 % of the variation (R [2] = 0.30; P = 0.0318; **Fig. 2B, Table 2** ). 254 Multivariate dispersion did not differ among groups (P = 0.37), indicating that these 255 differences reflect shifts in composition rather than in within-group variability. Pairwise 256 effect sizes were largest between resistant and susceptible genotypes (R [2] = 0.31), moderate 257 between resistant and intermediate genotypes (R [2] = 0.25) and smallest between 258 intermediate and susceptible genotypes (R [2] = 0.18; **Table 2** ). Consistently, resistant 259 genotypes clustered together along the first PCoA axis (29.1 % of the variation), susceptible 260 genotypes were displaced towards the opposite end, and intermediate genotypes spanned 261 a broader range of the ordination space ( **Fig. 2B** ). 262 **Table 2:** Statistical analyses of alpha and beta diversity of the _Ulmus minor_ samara 263 mycobiome. (A) Alpha diversity, measured as observed amplicon sequence variant (ASV) 264 richness and Shannon and Simpson indices. The effect of Dutch elm disease (DED) 265 resistance level was assessed with nested ANOVAs in which resistance level was tested 266 against the variation among genotypes within resistance levels. (B) Beta diversity, analyzed 267 with permutational multivariate analysis of variance (PERMANOVA) on Bray–Curtis 268 dissimilarities, with 9,999 permutations of whole genotypes among resistance levels. 269 Pairwise comparisons are reported as effect sizes (R²). SS, sum of squares; df, degrees of 270 freedom; MS, mean square; F, F statistic; R², proportion of variance explained. Red letters 271 indicate significant effects (P < 0.05). **A. Alpha diversity results** **Effect** **SS** **df** **MS** **F** **P** **ASV observed richness** Tolerance 16005.63 2 8002.81 23.72 0.001 **Shannon diversity index** Tolerance 1.78 2 0.89 4.88 0.05 **Simpson-index** Tolerance 0.06 2 0.03 4.22 0.07 **B. Beta diversity results** **PERMANOVA** **df** **SS** **r2** **F** **Pr(>F)** **sig** **Model** 2 1.4406 0.302 5.1959 0.0318 **Residuals** 24 3.327 0.697 **Total** 26 4.7675 1 **PAIRWISE COMPARISONS** **df** **SS** **f.model** **r2** **INTERMEDIATE VS RESISTANT** 1 0.708 5.18 0.245 **INTERMEDIATE VS SUSCEPTIBLE** 1 0.555 3.57 0.183 **RESISTANT VS SUSCEPTIBLE** 1 0.891 7.25 0.312 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 272 Of the 205 ASVs retained in the core analysis, 74 (36 %) formed the core seed mycobiome 273 ( **Fig. 2C; Supp. Table 1** ), which was dominated by Tremellomycetes (40.5 %) and 274 Dothideomycetes (35.5 %). Susceptible genotypes had the highest number of exclusive 275 ASVs (51), followed by resistant genotypes (24). 276 **Comparison with the stem mycobiome of trees from the same plot** 277 Forty-five ASVs/OTUs were shared between the samara and stem mycobiomes ( **Fig. 3;** 278 **Supp. Table 2** ). In both tissues, these shared taxa were dominated by Dothideomycetes, 279 although Lecanoromycetes were substantially less abundant and Tremellomycetes more 280 abundant in samaras ( **Fig. 3A** ). At the genus level, _Aureobasidium_ was a major constituent 281 in both tissues, while _Protofenestella_, _Cladosporium_ and _Alternaria_ were well represented 282 in samaras ( **Fig. 3B** ). The yeasts _Buckleyzyma_ (Cystobasidiomycetes) and _Genolevuria_ 283 (Tremellomycetes) were also notably present in both tissues. 284 **Fig. 3** . Class- and genus-level relative abundance of the 45 amplicon sequence variants 285 (ASVs) shared between all samples of the _Ulmus minor_ samara mycobiome analyzed in this 286 study and the stem mycobiome reported by Macaya-Sanz et al. (2023) [19]. 287 **Differentially abundant fungi between elm resistance levels** 288 DESeq2 identified 21 genera as differentially abundant in at least one of the three pairwise 289 comparisons between resistance levels ( **Fig. 4; Supp. Table 3** ). The most abundant of 290 these, with similar abundances across the three genotypes within each resistance level, 291 were _Protofenestella_, _Udeniomyces_, _Taphrina_, _Aureobasidium_, _Filobasidium_, 292 _Cryptococcus_, _Genolevuria_, _Buckleyzyma_ and _Kwoniella_ ( **Fig. 4** ). The potential plant 293 pathogen _Taphrina_ and the yeasts _Filobasidium_ and _Cryptococcus_ were significantly 294 enriched in intermediate and susceptible genotypes relative to resistant ones, whereas 295 _Aureobasidium_, _Buckleyzyma_, _Kwoniella_ and _Genolevuria_ were enriched only in 296 susceptible genotypes. Conversely, _Protofenestella_ was significantly more abundant in [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 297 resistant than in intermediate and susceptible genotypes. Thus, DED resistance was 298 associated with a distinct seed mycobiome signature, including a reduction of potentially 299 pathogenic and opportunistic taxa such as _Taphrina_ . 318 319 **Fig. 4.** Differentially abundant fungal genera among the samaras of the three DED320 resistance levels (resistant, intermediate and susceptible) identified with DESeq2. Different 321 letters indicate significant differences between resistance levels based on pairwise Wald 322 tests (Benjamini–Hochberg adjusted P < 0.05). Bubbles represent the mean number of 323 DESeq2-normalised reads of each genus in each genotype, colored by resistance level. 324 Owing to the large differences in read counts among genera, genera are grouped into blocks, 325 each with its own bubble size scale (right). 326 **Co-occurrence network analysis of the elm seed mycobiome** 327 Exploratory co-occurrence networks differed in structure among resistance levels ( **Fig. 5;** 328 **Table 3** ). The susceptible network was the most densely connected (89 edges among 34 329 ASVs; density = 0.16), whereas the resistant network was the least connected (68 edges 330 among 35 ASVs) and was split into three components; the intermediate network showed [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 331 intermediate values (92 edges among 36 ASVs). Negative associations were present in all 332 networks and accounted for 29.4 %, 47.8 % and 34.8 % of the edges in resistant, 333 intermediate and susceptible genotypes, respectively. The networks of resistant and 334 intermediate genotypes were more modular than expected by chance (modularity = 0.44 335 and 0.36, compared with 0.38 and 0.32 in random networks; P = 0.019 and 0.029, 336 respectively), whereas that of susceptible genotypes was not (0.29 compared with 0.31; P 337 = 0.78). 338 **Fig. 5.** Spearman co-occurrence networks of the samara mycobiome of _Ulmus minor_ 339 genotypes with contrasting levels of DED resistance. Networks are shown separately for 340 resistant, intermediate and susceptible genotypes. Each node represents a fungal taxon, 341 with node size proportional to its mean relative abundance and node color indicating fungal 342 class. Edges represent significant associations between fungal taxa, with positive and 343 negative correlations shown in black and red, respectively. Only strong associations (| _ρ_ | > 344 0.75) are shown. 345 **Table 3:** Topological parameters of Spearman co-occurrence networks constructed from 346 the fungal members of the _Ulmus minor_ samara mycobiome according to host DED 347 resistance. Networks were constructed separately for resistant, intermediate and 348 susceptible genotypes. Number of edges represents the total number of significant 349 associations between fungal members. Positive and negative edges indicate positive and 350 negative Spearman correlations, respectively, based on an absolute correlation coefficient 351 threshold of | _ρ_ | > 0.75 and _P_ < 0.05. Modularity of the co-occurrence networks of each 352 resistance level compared with that of 1,000 Erdős–Rényi random networks with the same 353 number of nodes and edges. P indicates the proportion of random networks with a 354 modularity equal to or higher than the observed value. **Topological parameter** **Resistant** **Intermediate** **Susceptible** **Number of edges** 68 92 89 **Positive edges** 48 (70.58%) 48 (52.17 %) 58 (65.16 %) **Negative edges** 20 (29.41 %) 43 (47.82 %) 31 (34.83 %) **Positive edges (ρ>0.75)** 29 (42.64 %) 25 (27.17 %) 38 (42.69 %) **Negative edges (ρ<-0.75)** 5 (7.35 %) 16 (17.39 %) 11 (12.35 %) [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) **Modularity** **Resistant** **Intermediate** **Susceptible** Modularity (fast-greedy) 0.435 0.359 0.292 Modularity of random networks 0.381 ± 0.025 0.317 ± 0.021 0.308 ± 0.021 (mean ± SD) P (observed > random) 0.019 0.029 0.781 355 **Discussion** 356 _U. minor_ This study provides the first characterization of the samara mycobiome of 357 genotypes with contrasting resistance to DED. By analyzing intact samaras, including both 358 the seed and the pericarp, we also captured fungi from the pericarp, a maternal tissue that 359 can act as a microbial reservoir and transfer microbes to the emerging seedling during 360 germination. Mycobiome composition, diversity and co-occurrence patterns varied among 361 samaras from mother trees with different resistance levels. As all mother trees shared 362 similar environmental conditions, these differences reinforce the view that the seed 363 microbiome can be strongly shaped by host genotype. 364 **The samara mycobiome shares core fungal taxa with the stem endosphere** 365 The samara mycobiome was dominated by Dothideomycetes, Tremellomycetes, 366 Taphrinomycetes and Microbotryomycetes, with Arthoniomycetes, Cystobasidiomycetes, 367 Leotiomycetes and Sordariomycetes at lower abundances ( **Fig. 1** ). Its core included 368 _Alternaria_, _Aureobasidium_ and _Cladosporium_ (Dothideomycetes) and basidiomycetous 369 yeasts such as _Naganishia_, _Vishniacozyma_, _Dioszegia_ and _Filobasidium_ (Tremellomycetes) 370 ( **Fig. 2C, Supp. Table 1** ). _Alternaria_, _Cladosporium_ and yeasts such as _Vishniacozyma_ are 371 consistently identified as ubiquitous members of the global seed core mycobiome across 372 plant species [37,38]. 373 The comparison with the stem mycobiome of _U. minor_ trees from the same plot [19] 374 revealed shared taxa able to colonize both niches. The overlap was most pronounced at 375 higher taxonomic ranks, with Dothideomycetes, Tremellomycetes and 376 Cystobasidiomycetes dominating across tissues. The 45 shared ASVs/OTUs were 377 dominated by _Aureobasidium_, _Protofenestella_, _Cladosporium_, _Alternaria_, _Buckleyzyma_ 378 and _Genolevuria_, and included _Knufia_ (Eurotiomycetes) and _Rhodotorula_ (Sporidiobolales), 379 previously isolated, cultured and functionally characterized in _U. minor_ [12,24,39]. 380 Although our design cannot discriminate between vertically and horizontally transmitted 381 taxa, this overlap suggests that a substantial proportion of stem endophytes can also 382 colonize samaras, increasing their likelihood of becoming primary colonizers of emerging [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 383 seedlings [40]. The recurrence of these taxa across independent studies and tissues 384 suggests that certain members of the _U. minor_ mycobiome are systematically maintained, 385 and the prevalence of yeasts and yeast-like fungi agrees with their frequent presence in 386 seed tissues [41,42]. While direct experimental validation (e.g. tracking fluorescently 387 labeled strains) is needed to confirm seed-borne inheritance, our findings provide 388 candidate seed-transmitted fungi. 389 **DED resistant elms are associated with reduced seed mycobiome diversity** 390 Alpha and beta diversity differed between DED-resistant and susceptible genotypes ( **Fig.** 391 **2A, B** ), suggesting that the host genetic background exerts a selective pressure on seed 392 fungal community assembly. Genotype-dependent seed mycobiomes have also been 393 described in _Triticum_ spp., _Brassica napus_, _Cucurbita pepo_ and _Quercus petraea_ [43–46]. 394 In _U. minor_ , distinct stem fungal communities have likewise been associated with different 395 levels of DED resistance, indicating that genotype-dependent filtering may operate 396 throughout plant development [19,47]. 397 Intriguingly, susceptible genotypes showed higher fungal richness and diversity than 398 resistant ones ( **Fig. 2A** ). While this contrasts with the classic ecological paradigm 399 associating higher diversity with greater community stability [48], it matches the findings by 400 Martín et al. (2013) [47] with culturable _U. minor_ xylem endophytes and observations in 401 other pathosystems. For instance, seeds of _Brassica napus_ cultivars resistant to 402 _Plasmodiophora brassicae_ harbor less diverse communities than susceptible ones [45], 403 and cultivated _Triticum aestivum_, which tends to be more susceptible to diseases than its 404 wild relatives, harbors more diverse and variable fungal assemblages [46]. We hypothesize 405 that resistant _U. minor_ genotypes do not rely on a broad, unselective microbial consortium, 406 but selectively recruit a more specialized community enriched in key taxa that confer 407 resilience against biotic stress [49]. These early colonizers could occupy key ecological 408 niches and create a disease-suppressive seed microbiome acting as a primary line of 409 defense from the onset of germination [15]. Conversely, the higher diversity in seeds of 410 susceptible genotypes points to a more permissive seed environment and a lower capacity 411 of the host to discriminate among beneficial, neutral and potentially harmful fungi [50]. 412 Host defense activation can act as an active filter of microbiome assembly, with immune 413 signaling and specialized metabolites shaping which microbes establish and persist [51]. 414 In _U. minor_, susceptible genotypes show weak or delayed defense activation against _O._ 415 _novo-ulmi_ [52], a permissive response associated with more frequent shifts in the stem [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 416 mycobiome [20]. This effect may extend to seeds through the maternal mycobiome or 417 through defense mechanisms operating within seeds. Moreover, maintaining a more 418 diverse fungal community may entail additional costs for the host, potentially 419 compromising the resilience of the new plant [53]. 420 **DED-resistant genotypes selectively filter fungi colonizing samaras** 421 The fungi exclusive to each resistance level ( **Fig. 2C, Supp. Table 1** ) further support a 422 potential contribution of the seed mycobiome to disease resistance in seedlings. The 423 exclusive mycobiome of resistant genotypes included putative beneficial taxa, but also taxa 424 considered pathogens in other systems, such as _Fusarium commune_ and species of 425 _Stemphylium_ and Didymellaceae [54–56], none of which has been documented as a 426 pathogen of _Ulmus_ . 427 Among putative beneficial taxa exclusive to resistant genotypes were the basidiomycetous 428 yeasts _Vishniacozyma dimennae_ and _Sporobolomyces reniformis_, well-documented 429 epiphytic and endophytic colonizers of the phyllosphere [41,57] and core members of the 430 global seed mycobiome [37,38]. _Vishniacozyma_ in seeds has been correlated with high 431 levels of disease suppression against _Globisporangium ultimum_ in spinach and against 432 _Tilletia_ and _Fusarium_ in wheat [15,41], and it is more abundant in _Fraxinus excelsior_ 433 genotypes less susceptible to ash dieback, where it may participate in pathogen 434 antagonism [58]. The exclusive presence of _Cladosporium cladosporioides_, a vertically 435 transmitted endophyte that may enhance seed vigor [59], further supports the idea of a 436 protective consortium that could limit pathogen establishment through niche preemption, 437 nutrient competition and mycoparasitism [41,60]. 438 This putative consortium also included the dark septate endophytes _Pseudogymnoascus_ 439 _appendiculatus_ and _Knufia cryptophialidica_, which typically enhance tolerance to abiotic 440 and oxidative stress, critical during germination and early establishment [61–63]. The 441 detection of _Knufia_ and _Rhodotorula_, previously isolated from stems of resistant elms [19], 442 reinforces the hypothesis of transmission of protective endosymbionts. 443 In contrast, the exclusive presence of phytopathogens such as _Taphrina carpini_, _Botrytis_ sp. 444 and _Didymella glomerata_ in susceptible genotypes points to a more permissive seed 445 environment. Although only _Taphrina_ species are known pathogens of _Ulmus_, _Botrytis_ and 446 _Didymella_ species are key indicators of low-performing seed lots that fail to suppress 447 damping-off [15]. Together with the higher number of exclusive taxa, this suggests that [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 448 susceptible genotypes exert weaker filtering against opportunistic fungi, resulting in a 449 broader seed fungal community with a greater presence of potential opportunistic parasites 450 [64]. 451 **Differential abundance of taxa across the resistance gradient** 452 Differential abundance analysis further supports host-mediated filtering of the seed 453 mycobiome ( **Fig. 4** ). Resistant genotypes were enriched in _Protofenestella_, a genus within 454 the Dothideomycetes, a class frequently dominant in seed-associated communities and 455 repeatedly implicated in biocontrol and plant protection [6]. Dothideomycetous 456 endophytes have been associated with DED resistance in adult elm tissues [19], and their 457 beneficial roles have been validated _in vivo_ _[12,24,39]_ . Their detection in samaras may 458 indicate transmission to juvenile plants, whose early microbiome is largely seed-derived 459 [65]. In contrast, seeds of susceptible and intermediate genotypes were enriched in 460 _Taphrina_, a genus associated with leaf galls and deformities in woody hosts [66], again 461 pointing to a less restrictive seed environment open to colonization by opportunists and 462 latent pathogens [46]. 463 The enrichment of _Aureobasidium_, _Filobasidium_, _Cryptococcus_, _Genolevuria_, 464 _Buckleyzyma_ and _Kwoniella_ in susceptible genotypes also reflects the accumulation of a 465 wider range of taxa than in the more restricted assemblages of resistant trees. However, 466 _Filobasidium_ and _Aureobasidium_ are ubiquitous core members of seeds with documented 467 biocontrol properties based on nutrient competition and secondary metabolite production 468 [38,67], and species of _Aureobasidium_, _Filobasidium_ and _Genolevuria_ belong to the core 469 stem mycobiome of _U. minor_ [19]. Thus, although seeds of susceptible trees also harbor 470 potentially beneficial taxa, their overrepresentation suggests weaker directional filtering by 471 the host. 472 **Mycobiome network architecture differs across the DED resistance gradient** 473 Co-occurrence networks revealed fundamental differences in mycobiome architecture 474 across the resistance gradient ( **Fig. 5** ). The dense, highly interconnected and poorly 475 modular network of susceptible genotypes is consistent with a weak subdivision into 476 ecologically cohesive subcommunities [68,69], which may render the mycobiome 477 vulnerable to cascading disruptions following pathogen invasion, as perturbations 478 propagate rapidly through such communities [70,71]. Resistant genotypes displayed a 479 more modular architecture, with a cohesive core cluster and topologically isolated 480 subgroups of taxa, as also observed in coffee varieties resistant to coffee leaf rust [72]. [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 481 Modularity may reflect the organization of the community into stable consortia and niche 482 differentiation that reduces competition and promotes complementary functions [68], 483 potentially enhancing community resilience and the ability of the host to resist pathogen 484 establishment, as described for disease-suppressive soils [73]. 485 Together with the diversity results, these network topologies suggest divergent ecological 486 strategies: susceptible genotypes harbor diverse but weakly structured and permissive 487 fungal assemblages, potentially reflecting reduced host-mediated filtering, whereas 488 resistant genotypes host more constrained and modular assemblages, suggesting stronger 489 host selection of a curated subset of taxa. 490 **Conclusions** 491 The distinct samara mycobiomes of _U. minor_ genotypes suggest that the seed mycobiome 492 may contribute to the assembly of the plant mycobiome, with potential physiological 493 consequences throughout the tree’s lifespan. Our findings support the hypothesis that DED 494 resistance is linked to the transmission or acquisition of a curated, highly structured and 495 potentially protective seed mycobiome, which could shape seedling mycobiome assembly 496 _O. novo-ulmi_ colonizes the vascular through priority effects, pre-empting niches before 497 system. Functional validation through synthetic communities and controlled inoculation 498 trials is still needed to establish causality. Nevertheless, the genotype-specific fungal 499 indicators and co-occurrence patterns identified here open promising avenues for 500 mycobiome-assisted breeding and sustainable elm conservation. 501 **Acknowledgements** 502 We thank the Spanish elm breeding program and Puerta de Hierro Forest Breeding Centre 503 (Ministerio para la Transición Ecológica y el Reto Demográfico) for providing the plant 504 material and Jorge Domínguez for technical assistance. 505 This study was funded by the project AGL-2015-66952-R (MINECO/FEDER). C.M.-A. was 506 supported by a ‘FPI’ pre-doctoral fellowship from the Spanish Ministry of Economy and 507 Competitiveness during 2017-2021. 508 **Author Contributions** 509 C.M.-A. performed the experiments and laboratory analysis, analyzed and interpreted the 510 data, and wrote the main manuscript text. S.F.-Q. contributed to laboratory and field 511 experiments, data collection, and data interpretation. J. R.-C. and J.A.M. conceived and [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 512 designed the study, contributed to data interpretation, and manuscript writing and revision. 513 All authors reviewed, edited, and approved the final manuscript. 514 **Data Availability** 515 Raw sequences of the mycobiome are available online in NCBI under the BioProject code 516 PRJNA1548562. 517 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 518 **References** 519 [1] Berg G, Raaijmakers JM (2018) Saving seed microbiomes. ISME J 12:1167-1170. 520 [https://doi.org/10.1038/s41396-017-0028-2](https://doi.org/10.1038/s41396-017-0028-2) 521 [2] Cardinale M, Schnell S (2024) Is the plant microbiome transmitted from pollen to seeds? 522 [Front Microbiol 15:1343795. https://doi.org/10.3389/fmicb.2024.1343795](https://doi.org/10.3389/fmicb.2024.1343795) 523 [3] Maude RB (1996) Seedborne diseases and their control: principles and practice. CAB 524 International, Wallingford 525 [4] Shade A, Jacques MA, Barret M (2017) Ecological patterns of seed microbiome diversity, 526 transmission, and assembly. Curr Opin Microbiol 37:15-22. 527 [https://doi.org/10.1016/j.mib.2017.03.010](https://doi.org/10.1016/j.mib.2017.03.010) 528 [5] Adam E, Bernhart M, Müller H, Winkler J, Berg G (2018) The _Cucurbita pepo_ seed 529 microbiome: genotype-specific composition and implications for breeding. Plant Soil 530 [422:35-49. https://doi.org/10.1007/s11104-016-3113-9](https://doi.org/10.1007/s11104-016-3113-9) 531 [6] Nelson EB (2018) The seed microbiome: origins, interactions, and impacts. Plant Soil 532 [422:7-34. https://doi.org/10.1007/s11104-017-3289-7](https://doi.org/10.1007/s11104-017-3289-7) 533 [7] Sánchez-Cañizares C, Jorrín B, Poole PS, Tkacz A (2017) Understanding the holobiont: 534 the interdependence of plants and their microbiome. Curr Opin Microbiol 38:188-196. 535 [https://doi.org/10.1016/j.mib.2017.07.001](https://doi.org/10.1016/j.mib.2017.07.001) 536 [8] Trivedi P, Batista BD, Bazany KE, Singh BK (2022) Plant–microbiome interactions under a 537 changing world: responses, consequences and perspectives. New Phytol 234:1951-1959. 538 [https://doi.org/10.1111/nph.18016](https://doi.org/10.1111/nph.18016) 539 [9] Latz MAC, Jensen B, Collinge DB, Jørgensen HJL (2018) Endophytic fungi as biocontrol 540 agents: elucidating mechanisms in disease suppression. Plant Ecol Divers 11:555-567. 541 [https://doi.org/10.1080/17550874.2018.1534146](https://doi.org/10.1080/17550874.2018.1534146) 542 [10] Fontana DC, de Paula S, Torres AG, de Souza VH, Pascholati SF, Schmidt D, et al. (2021) 543 Endophytic fungi: biological control and induced resistance to phytopathogens and abiotic 544 [stresses. Pathogens 10:570. https://doi.org/10.3390/pathogens10050570](https://doi.org/10.3390/pathogens10050570) 545 [11] Shoresh M, Harman GE, Mastouri F (2010) Induced systemic resistance and plant 546 responses to fungal biocontrol agents. Annu Rev Phytopathol 48:21-43. 547 [https://doi.org/10.1146/annurev-phyto-073009-114450](https://doi.org/10.1146/annurev-phyto-073009-114450) 548 [12] Martínez-Arias C, Sobrino-Plata J, Gil L, Rodríguez-Calcerrada J, Martín JA (2021) 549 Priming of plant defenses against _Ophiostoma novo-ulmi_ by elm ( _Ulmus minor_ Mill.) fungal 550 [endophytes. J Fungi 7:687. https://doi.org/10.3390/jof7090687](https://doi.org/10.3390/jof7090687) 551 [13] Li X, Song M, Yao X, Chai Q, Simpson WR (2017) The effect of seed-borne fungi and 552 _Epichloë_ endophyte on seed germination and biomass of _Elymus sibiricus_ . Front Microbiol 553 [8:2488. https://doi.org/10.3389/fmicb.2017.02488](https://doi.org/10.3389/fmicb.2017.02488) 554 [14] Abdelfattah A, Tack AJM, Lobato C, Wassermann B, Berg G (2023) From seed to seed: 555 the role of microbial inheritance in the assembly of the plant microbiome. Trends Microbiol 556 [31:346-355. https://doi.org/10.1016/j.tim.2022.10.009](https://doi.org/10.1016/j.tim.2022.10.009) 557 [15] Diakaki M, Jimenez BA, De Lange E, Butterbach P, Van Der Heijden L, Köhl J, et al. (2025) 558 Spinach seed microbiome characteristics linked to suppressiveness against [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 559 _Globisporangium_ _ultimum_ damping-off. FEMS Microbiol Ecol 101:fiaf004. 560 [https://doi.org/10.1093/femsec/fiaf004](https://doi.org/10.1093/femsec/fiaf004) 561 [16] Brasier CM (1991) _Ophiostoma novo-ulmi_ sp. nov., causative agent of current Dutch 562 elm disease pandemics. Mycopathologia 115:151-161. 563 [https://doi.org/10.1007/BF00462219](https://doi.org/10.1007/BF00462219) 564 [17] Martín JA, Sobrino-Plata J, Rodríguez-Calcerrada J, Collada C, Gil L (2019) Breeding and 565 scientific advances in the fight against Dutch elm disease: will they allow the use of elms in 566 [forest restoration? New For 50:183-215. https://doi.org/10.1007/s11056-018-9640-x](https://doi.org/10.1007/s11056-018-9640-x) 567 [18] Martín JA, Domínguez J, Solla A, Brasier CM, Webber JF, Santini A, et al. (2023) 568 Complexities underlying the breeding and deployment of Dutch elm disease resistant elms. 569 [New For 54:661-696. https://doi.org/10.1007/s11056-021-09865-y](https://doi.org/10.1007/s11056-021-09865-y) 570 [19] Macaya-Sanz D, Witzell J, Collada C, Gil L, Martín JA (2023) Core endophytic 571 mycobiome in _Ulmus minor_ and its relation to Dutch elm disease resistance. Front Plant Sci 572 [14:1125942. https://doi.org/10.3389/fpls.2023.1125942](https://doi.org/10.3389/fpls.2023.1125942) 573 [20] Martínez-Arias C, Sobrino-Plata J, Rodríguez-Calcerrada J, Martín JA (2023) Mycobiome 574 shifts in Ulmus minor stems after infection with beneficial and pathogenic fungi. 575 [Phytobiomes J 7:464-477. https://doi.org/10.1094/PBIOMES-04-23-0025-R](https://doi.org/10.1094/PBIOMES-04-23-0025-R) 576 [21] Solla A, López-Almansa JC, Martín JA, Gil L (2014) Genetic variation and heritability 577 estimates of _Ulmus minor_ and _Ulmus pumila_ hybrids for budburst, growth and tolerance to 578 _Ophiostoma novo-ulmi_ [. iForest 8:422. https://doi.org/10.3832/ifor1227-007](https://doi.org/10.3832/ifor1227-007) 579 [22] Domínguez J, Macaya-Sanz D, Gil L, Martín JA (2022) Excelling the progenitors: breeding 580 for resistance to Dutch elm disease from moderately resistant and susceptible native stock. 581 [For Ecol Manage 511:120113. https://doi.org/10.1016/j.foreco.2022.120113](https://doi.org/10.1016/j.foreco.2022.120113) 582 [23] Martín JA, Solla A, Oszako T, Gil L (2021) Characterizing offspring of Dutch elm disease583 resistant trees ( _Ulmus_ _minor_ Mill.). Forestry 94:374-385. 584 [https://doi.org/10.1093/forestry/cpaa040](https://doi.org/10.1093/forestry/cpaa040) 585 [24] Martínez-Arias C, Sobrino-Plata J, Ormeño-Moncalvillo S, Gil L, Rodríguez-Calcerrada 586 J, Martín JA (2021) Endophyte inoculation enhances _Ulmus minor_ resistance to Dutch elm 587 [disease. Fungal Ecol 50:101024. https://doi.org/10.1016/j.funeco.2020.101024](https://doi.org/10.1016/j.funeco.2020.101024) 588 [25] Martín JA, Solla A, Venturas M, Collada C, Domínguez J, Miranda E, et al. (2015) Seven 589 _Ulmus minor_ clones tolerant to _Ophiostoma novo-ulmi_ registered as forest reproductive 590 [material in Spain. iForest 8:172-180. https://doi.org/10.3832/ifor1224-008](https://doi.org/10.3832/ifor1224-008) 591 [26] Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP (2016) DADA2: 592 high-resolution sample inference from Illumina amplicon data. Nat Methods 13:581-583. 593 [https://doi.org/10.1038/nmeth.3869](https://doi.org/10.1038/nmeth.3869) 594 [27] Martin M (2011) Cutadapt removes adapter sequences from high-throughput 595 [sequencing reads. EMBnet J 17:10-12. https://doi.org/10.14806/ej.17.1.200](https://doi.org/10.14806/ej.17.1.200) 596 [28] Wang Q, Garrity GM, Tiedje JM, Cole JR (2007) Naive Bayesian classifier for rapid 597 assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol 598 [73:5261-5267. https://doi.org/10.1128/AEM.00062-07](https://doi.org/10.1128/AEM.00062-07) 599 [29] Abarenkov K, Nilsson RH, Larsson KH, Taylor AFS, May TW, Frøslev TG, et al. (2024) The 600 UNITE database for molecular identification and taxonomic communication of fungi and [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 601 other eukaryotes: sequences, taxa and classifications reconsidered. Nucleic Acids Res 602 [52:D791-D797. https://doi.org/10.1093/nar/gkad1039](https://doi.org/10.1093/nar/gkad1039) 603 [30] McMurdie PJ, Holmes S (2013) phyloseq: an R package for reproducible interactive 604 analysis and graphics of microbiome census data. PLoS One 8:e61217. 605 [https://doi.org/10.1371/journal.pone.0061217](https://doi.org/10.1371/journal.pone.0061217) 606 [31] Beule L, Karlovsky P (2020) Improved normalization of species count data in ecology by 607 scaling with ranked subsampling (SRS): application to microbial communities. PeerJ 608 [8:e9593. https://doi.org/10.7717/peerj.9593](https://doi.org/10.7717/peerj.9593) 609 [32] Lahti L, Shetty S (2017) Tools for microbiome analysis in R. microbiome package version 610 1.19.1 611 [33] Kim BR, Shin J, Guevarra RB, Lee JH, Kim DW, Seol KH, et al. (2017) Deciphering diversity 612 indices for a better understanding of microbial communities. J Microbiol Biotechnol 613 [27:2089-2093. https://doi.org/10.4014/jmb.1709.09027](https://doi.org/10.4014/jmb.1709.09027) 614 [34] Oksanen J (2007) vegan: community ecology package. R package version 1.8-5. 615 [http://www.cran.r-project.org](http://www.cran.r-project.org/) 616 [35] Gao CH, Yu G, Cai P (2021) ggVennDiagram: an intuitive, easy-to-use, and highly 617 customizable R package to generate Venn diagram. Front Genet 12:706907. 618 [https://doi.org/10.3389/fgene.2021.706907](https://doi.org/10.3389/fgene.2021.706907) 619 [36] Love MI, Huber W, Anders S (2014) Moderated estimation of fold change and dispersion 620 [for RNA-seq data with DESeq2. Genome Biol 15:550. https://doi.org/10.1186/s13059-014-](https://doi.org/10.1186/s13059-014-0550-8) 621 0550-8 622 [37] Simonin M, Briand M, Chesneau G, Rochefort A, Marais C, Sarniguet A, et al. (2022) 623 Seed microbiota revealed by a large-scale meta-analysis including 50 plant species. New 624 [Phytol 234:1448-1463. https://doi.org/10.1111/nph.18037](https://doi.org/10.1111/nph.18037) 625 [38] Guschinskaya N, Cordovez V, Guillemette T (2026) Unexplored yeast diversity in seed 626 microbiota. bioRxiv 2024.11.27.625647. https://doi.org/10.1101/2024.11.27.625647 627 [39] Martínez-Arias C, Sobrino-Plata J, Medel D, Gil L, Martín JA, Rodríguez-Calcerrada J 628 (2021) Stem endophytes increase root development, photosynthesis, and survival of elm 629 plantlets ( _Ulmus_ _minor_ Mill.). J Plant Physiol 261:153420. 630 [https://doi.org/10.1016/j.jplph.2021.153420](https://doi.org/10.1016/j.jplph.2021.153420) 631 [40] Zeng Q, Zhao Y, Shen W, Han D, Yang M (2023) Seed-to-seed: plant core vertically 632 transmitted microbiota. J Agric Food Chem 71:19255-19264. 633 [https://doi.org/10.1021/acs.jafc.3c07092](https://doi.org/10.1021/acs.jafc.3c07092) 634 [41] Vujanovic V (2021) Tremellomycetes yeasts in kernel ecological niche: early indicators 635 of enhanced competitiveness of endophytic and mycoparasitic symbionts against wheat 636 [pathobiota. Plants 10:905. https://doi.org/10.3390/plants10050905](https://doi.org/10.3390/plants10050905) 637 [42] Jeong E, Abdellaoui N, Lim JY, Seo JA (2024) The presence of a significant endophytic 638 fungus in mycobiome of rice seed compartments. Sci Rep 14: 2 3367. 639 [https://doi.org/10.1038/s41598-024-73550-1](https://doi.org/10.1038/s41598-024-73550-1) 640 [43] Rybakova D, Mancinelli R, Wikström M, Postma J, Ehlers R, Goertz S, et al. (2017) The 641 structure of the Brassica napus seed microbiome is cultivar-dependent and affects the [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 642 interactions of symbionts and pathogens. Microbiome 5:104. 643 [https://doi.org/10.1186/s40168-017-0310-6](https://doi.org/10.1186/s40168-017-0310-6) 644 [44] Fort T, Pauvert C, Zanne AE, Ovaskainen O, Caignard T, Barret M, et al. (2021) Maternal 645 effects shape the seed mycobiome in _Quercus petraea_ . New Phytol 230:1594-1608. 646 [https://doi.org/10.1111/nph.17153](https://doi.org/10.1111/nph.17153) 647 [45] Wassermann B, Abdelfattah A, Wicaksono WA, Kusstatscher P, Müller H, Cernava T, et 648 al. (2022) The _Brassica napus_ seed microbiota is cultivar-specific and transmitted via 649 [paternal breeding lines. Microb Biotechnol 15:2379-2390. https://doi.org/10.1111/1751-](https://doi.org/10.1111/1751-7915.14077) 650 7915.14077 651 [46] Malacrinò A, Abdelfattah A, Belgacem I, Schena L (2023) Plant genotype influence the 652 structure of cereal seed fungal microbiome. Front Microbiol 13:1075399. 653 [https://doi.org/10.3389/fmicb.2022.1075399](https://doi.org/10.3389/fmicb.2022.1075399) 654 [47] Martín JA, Witzell J, Blumenstein K, Rozpedowska E, Helander M, Sieber TN, et al. (2013) 655 Resistance to Dutch elm disease reduces presence of xylem endophytic fungi in elms 656 ( _Ulmus_ [spp.). PLoS One 8:e56987. https://doi.org/10.1371/journal.pone.0056987](https://doi.org/10.1371/journal.pone.0056987) 657 [48] Xun W, Liu Y, Li W, Ren Y, Xiong W, Xu Z, et al. (2021) Specialized metabolic functions of 658 keystone taxa sustain soil microbiome stability. Microbiome 9:35. 659 [https://doi.org/10.1186/s40168-020-00985-9](https://doi.org/10.1186/s40168-020-00985-9) 660 [49] Debray R, Herbert RA, Jaffe AL, Crits-Christoph A, Power ME, Koskella B (2022) Priority 661 effects in microbiome assembly. Nat Rev Microbiol 20:109-121. 662 [https://doi.org/10.1038/s41579-021-00604-w](https://doi.org/10.1038/s41579-021-00604-w) 663 [50] Barret M, Guimbaud JF, Darrasse A, Jacques MA (2016) Plant microbiota affects seed 664 transmission of phytopathogenic microorganisms. Mol Plant Pathol 17:791-795. 665 [https://doi.org/10.1111/mpp.12382](https://doi.org/10.1111/mpp.12382) 666 [51] Kudjordjie EN, Sapkota R, Nicolaisen M (2021) Arabidopsis assemble distinct root667 associated microbiomes through the synthesis of an array of defense metabolites. PLoS 668 [One 16:e0259171. https://doi.org/10.1371/journal.pone.0259171](https://doi.org/10.1371/journal.pone.0259171) 669 [52] Martín JA, Solla A, Coimbra MA, Gil L (2005) Metabolic distinction of _Ulmus minor_ xylem 670 tissues after inoculation with _Ophiostoma novo-ulmi._ Phytochemistry 66:2458-2467. 671 [https://doi.org/10.1016/j.phytochem.2005.08.004](https://doi.org/10.1016/j.phytochem.2005.08.004) 672 [53] Newton AC, Fitt BDL, Atkins SD, Walters DR, Daniell TJ (2010) Pathogenesis, parasitism 673 and mutualism in the trophic space of microbe–plant interactions. Trends Microbiol 18:365674 [373. https://doi.org/10.1016/j.tim.2010.06.002](https://doi.org/10.1016/j.tim.2010.06.002) 675 [54] Husna A, Zakaria L, Mohamed Nor NMI (2021) _Fusarium commune_ associated with wilt 676 [and root rot disease in rice. Plant Pathol 70:123-132. https://doi.org/10.1111/ppa.13270](https://doi.org/10.1111/ppa.13270) 677 [55] Luo X, Hu Y, Xia J, Zhang K, Ma L, Xu Z, et al. (2024) Morphological and phylogenetic 678 analyses reveal three new species of Didymella (Didymellaceae, Pleosporales) from Jiangxi, 679 [China. J Fungi 10:75. https://doi.org/10.3390/jof10010075](https://doi.org/10.3390/jof10010075) 680 [56] Das A, Dutta S, Jash S, Barman AR, Das R, Kumar S, et al. (2019) Current knowledge on 681 pathogenicity and management of _Stemphylium botryosum_ in lentils ( _Lens culinaris_ ssp. 682 _culinaris_ [Medik). Pathogens 8:225. https://doi.org/10.3390/pathogens8040225](https://doi.org/10.3390/pathogens8040225) [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 683 [57] Domka A, Jędrzejczyk R, Ważny R, Gustab M, Kowalski M, Nosek M, et al. (2023) 684 Endophytic yeast protect plants against metal toxicity by inhibiting plant metal uptake 685 through an ethylene-dependent mechanism. Plant Cell Environ 46:268-287. 686 [https://doi.org/10.1111/pce.14473](https://doi.org/10.1111/pce.14473) 687 [58] Long F, Doonan JM, Nielsen LR, Kjær ED, Kosawang C (2024) Mother trees of common 688 ash ( _Fraxinus excelsior_ ) disperse different sets of mycobiome through their seed wings. 689 [BMC Res Notes 17:213. https://doi.org/10.1186/s13104-024-06863-z](https://doi.org/10.1186/s13104-024-06863-z) 690 [59] Qin Y, Pan X, Yuan Z (2016) Seed endophytic microbiota in a coastal plant and 691 phytobeneficial properties of the fungus _Cladosporium cladosporioides._ Fungal Ecol 24:53692 [60. https://doi.org/10.1016/j.funeco.2016.08.011](https://doi.org/10.1016/j.funeco.2016.08.011) 693 [60] Gorordo MF, Lucca ME, Sangorrín MP (2022) Biocontrol efficacy of the _Vishniacozyma_ 694 _victoriae_ in semi-commercial assays for the control of postharvest fungal diseases of 695 [organic pears. Curr Microbiol 79:259. https://doi.org/10.1007/s00284-022-02934-1](https://doi.org/10.1007/s00284-022-02934-1) 696 [61] Kim JS, Romero E, Cho Y, Ahumada-Rudolph R, Núñez C, Gómez-Espinoza J, et al. 697 (2025) Fungal diversity from Fildes Peninsula (Antarctica) and their antibiosis bioactivity 698 against two plant pathogens. J Microbiol 63:e2411029. 699 [https://doi.org/10.71150/jm.2411029](https://doi.org/10.71150/jm.2411029) 700 [62] Li X, He X, Hou L, Ren Y, Wang S, Su F (2018) Dark septate endophytes isolated from a 701 xerophyte plant promote the growth of _Ammopiptanthus mongolicus_ under drought 702 [condition. Sci Rep 8:7896. https://doi.org/10.1038/s41598-018-26183-0](https://doi.org/10.1038/s41598-018-26183-0) 703 [63] Ogaki MB, Teixeira DR, Vieira R, Lírio JM, Felizardo JPS, Abuchacra RC, et al. (2020) 704 Diversity and bioprospecting of cultivable fungal assemblages in sediments of lakes in the 705 [Antarctic Peninsula. Fungal Biol 124:601-611. https://doi.org/10.1016/j.funbio.2020.02.015](https://doi.org/10.1016/j.funbio.2020.02.015) 706 [64] Shahzad R, Khan AL, Bilal S, Asaf S, Lee IJ (2018) What is there in seeds? Vertically 707 transmitted endophytic resources for sustainable improvement in plant growth. Front Plant 708 [Sci 9:24. https://doi.org/10.3389/fpls.2018.00024](https://doi.org/10.3389/fpls.2018.00024) 709 [65] Johnston-Monje D, Gutiérrez JP, Becerra Lopez-Lavalle LA (2022) Stochastic inoculum, 710 biotic filtering and species-specific seed transmission shape the rare microbiome of plants. 711 [Life 12:1372. https://doi.org/10.3390/life12091372](https://doi.org/10.3390/life12091372) 712 [66] Oliver RP (2024) Diseases caused by fungi. In: Oliver RP (ed) Agrios’ plant pathology, 6th 713 [edn. Academic Press, London, pp 339-427. https://doi.org/10.1016/B978-0-12-822429-](https://doi.org/10.1016/B978-0-12-822429-8.00013-3) 714 8.00013-3 715 [67] Bergmann GE, Leveau JHJ (2022) A metacommunity ecology approach to 716 understanding microbial community assembly in developing plant seeds. Front Microbiol 717 [13:877519. https://doi.org/10.3389/fmicb.2022.877519](https://doi.org/10.3389/fmicb.2022.877519) 718 [68] Hernandez D, David A, Menges E, Searcy C, Afkhami M (2021) Environmental stress 719 [destabilizes microbial networks. ISME J 15:1722-1734. https://doi.org/10.1038/s41396-](https://doi.org/10.1038/s41396-020-00882-x) 720 020-00882-x 721 [69] Milke F, Garcia S, Simon M, Pacheco-Valenciana A, Lennartz S (2026) Microbial cohorts: 722 bringing ecological meaning to the modularity concept of co-occurrence networks. ISME 723 [Commun 6:ycag037. https://doi.org/10.1093/ismeco/ycag037](https://doi.org/10.1093/ismeco/ycag037) [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.11.765837; this version posted November 12, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.11.765837) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) 724 [70] Gilarranz LJ, Rayfield B, Liñán-Cembrano G, Bascompte J, Gonzalez A (2017) Effects of 725 network modularity on the spread of perturbation impact in experimental metapopulations. 726 [Science 357:199-201. https://doi.org/10.1126/science.aal4122](https://doi.org/10.1126/science.aal4122) 727 [71] Mintrone C, Rindi L, Bertocci I, Maggi E, et al. (2025) Modularity buffers the spread of 728 spatial perturbations in macroalgal networks. Curr Biol 35:154-162.e4. 729 [https://doi.org/10.1016/j.cub.2024.11.038](https://doi.org/10.1016/j.cub.2024.11.038) 730 [72] Wu Y, Zhao X, Wang Z, Li X, Zhang X, Xie C, et al. (2025) The role of coffee microbiomes 731 in pathogen resistance across varieties and ecological niches. Microorganisms 13:1909. 732 https://doi.org/10.3390/microorganisms13081909 733 [73] Spooren J, van Bentum S, Thomashow LS, Pieterse CMJ, Weller DM, Berendsen RL 734 (2024) Plant-driven assembly of disease-suppressive soil microbiomes. Annu Rev 735 [Phytopathol 62:1-30. https://doi.org/10.1146/annurev-phyto-021622-100127](https://doi.org/10.1146/annurev-phyto-021622-100127)