Ash leaf microbiomes share a core bacterial community shaped by climate and site

Across 12 forest sites in Germany, the bacterial communities living on the leaves of European ash trees share a small core of common genera, yet the way those core bacteria connect to the rest of the community differs markedly from place to place, according to a new study published in Environmental Microbiome. The research, led by Valentin Burghardt, Sonja Wende and colleagues including Andreas Ulrich, sampled 30 ash trees at each of 12 intensive monitoring plots established under the FraxForFuture framework, sampling in June 2021. The work set out to identify a core microbiome for Fraxinus excelsior and to weigh how strongly geography, climate, soil properties, tree characteristics and leaf chemistry shape the phyllosphere, the microbial habitat formed by aerial plant surfaces.

The team collected around 20 leaves and petioles from each tree, taking samples from the mid-canopy using methods that varied by site, including shooting down leaf tufts with a specially prepared rifle at ten sites, climbing at one site, and using a slingshot and saw at another. Samples were frozen on dry ice, stored at minus 80 degrees Celsius, lyophilised and ground before analysis. The researchers extracted total DNA from two grams of compound leaves and amplified the V5–V6 region of the 16S rRNA gene using primers that exclude chloroplast and mitochondrial DNA of the host plant. Library preparation and paired-end sequencing were performed on an Illumina MiSeq platform, and the raw data were processed with the DADA2 pipeline for quality filtering, denoising and chimera removal.

After removing residual plant DNA sequences, the study obtained more than 36.7 million high-quality bacterial reads. Samples were rarefied to the minimum sequence count of 19,468, yielding 321 samples with between 126 and 770 amplicon sequence variants per sample, with a mean of 314. In total, 755 genera were detected across the dataset, distributed among 26 phyla. Three phyla dominated everywhere: Proteobacteria, with a mean relative abundance of 58.4 percent across all sites; Bacteroidetes, at 21.8 percent; and Actinobacteria, at 16.0 percent. The remaining 23 phyla together accounted for an average of just 3.8 percent of the communities.

At the genus level, three taxa stood out as the most abundant across all sites: Hymenobacter, with a mean relative abundance of 20.2 percent; Massilia, at 18.4 percent; and Sphingomonas. Their abundances varied considerably among sites and individual samples. Hymenobacter ranged from 1.8 percent in one sample to 50.1 percent in another, while Massilia ranged from 0.6 percent to 66.8 percent, reaching its highest site-level mean of 36.3 percent at SN1 and its lowest, 4 percent, at BY3. Differential abundance analysis using the LEfSe method identified 87 genera whose relative abundances differed significantly among sites, with 21 showing strong effect sizes, including Massilia, Methylobacterium, Sphingomonas, Hymenobacter and Kineococcus.

Community composition differed significantly among sites. A principal coordinate analysis based on Bray–Curtis dissimilarity showed that the first two axes explained 13.3 percent and 6.3 percent of the total variance, and analysis of similarity confirmed that sites generally differed significantly from each other, with an overall R value of 0.42 and a p value of 0.001. Alpha diversity also varied: Chao1 diversity was highest at BY3, at 540 plus or minus 166, and significantly greater than at most other sites, while SN1, ST1 and TH1 showed the lowest values. A Random Forest classification model based on genus-level abundances achieved a cross-validated accuracy of 62.8 percent in assigning samples to their sites, with samples from SN1 classified most accurately at 89.7 percent and TH1 least accurately at 31 percent.

To identify the drivers of these differences, the researchers performed a distance-based redundancy analysis testing 59 explanatory variables covering geographical position, climate, soil chemistry, tree characteristics and leaf chemistry, including phenols, procyanidins and amino acids. Geographical position, precipitation, tree age and leaf and soil chemistry emerged as significant factors. Variation partitioning showed that these variables together explained 22.1 percent of the total variation in bacterial community composition, with climate accounting for 6.7 percent and local soil and host-related variables for 6.6 percent as independent fractions, while geographical position alone explained only 0.8 percent. Geography shared a substantial 7.9 percent of explained variation with climate and local environmental variables, indicating that spatial differences in microbiome composition were largely associated with underlying climatic and environmental gradients rather than geographical distance itself. The remaining 77.9 percent of variation remained unexplained, suggesting that factors such as host genotype, microenvironmental conditions, dispersal, temporal dynamics and stochastic processes also contribute to community assembly.

The core microbiome analysis defined core taxa as genera present in at least 94 percent of samples at a site with a relative abundance above 0.1 percent. In total, 38 genera appeared among the site-level core microbiomes, ranging from 10 genera at SN1 to 30 at BY3. Seven genera were detected in the core microbiomes of all plots, including Hymenobacter and Sphingomonas, which were relatively abundant, and Curtobacterium and Pseudomonas, which were less so. Several genera appeared in the cores of most plots, such as Microlunatus at 10 sites and Roseomonas at 7. Some genera were specific to single sites or regions: an unclassified Enterobacterales genus and an unclassified Erwiniaceae genus occurred only at SN1, while BY3 stood out with seven unique core genera, including Aureimonas, Nakamurella and Subtercola, and shared genera such as Curvibacter, Acidovorax and Klenkia with the nearby site BY2.

Co-occurrence network analysis added a structural dimension to these findings. A combined network across all sites contained 53 genera and 224 edges, with 116 positive and 108 negative correlations. Six hub genera were identified on the basis of degree, betweenness centrality and closeness centrality: Amnibacterium, Deinococcus, Microlunatus, an unclassified Acetobacteraceae genus, Methylobacterium and Massilia. Methylobacterium and Deinococcus showed the highest connectivity, each forming 22 edges with other nodes. Notably, Deinococcus achieved this hub status despite its low relative abundance, and the authors suggest that its well-known resistance to radiation, oxidative stress and desiccation may suit it to life on leaf surfaces, where it may also facilitate other bacteria through biofilm formation. When networks were constructed separately for each of the 12 sites at the ASV level, complexity differed substantially: southern German sites were more complex, with BY3 forming the most complex network at 294 nodes and 1,342 edges, while ST1 and TH1 were the least complex and three sites had no hubs at all.

The authors conclude that an ash core microbiome exists on leaves at all sites but is locally interconnected in different ways, and that less frequent genera can play important roles as hub taxa. Host-related factors also mattered: ash diameter at breast height, chlorophyll content and procyanidin content were all associated with community composition, although the authors caution that diameter is only an approximate proxy for tree age and that host genotype was not assessed. Soil properties, including calcium and sodium stocks and cation-exchange capacity, were also significantly associated with community composition, potentially acting indirectly through host nutrition, physiology and leaf chemistry. Because the study is a cross-sectional survey conducted at a single time point in one year, it captures a snapshot rather than seasonal dynamics, and the authors note that the role of tree genotype and canopy strata, as well as microbiome dynamics over the year, remain questions for future work.

The findings carry potential relevance for ash dieback, a major disease threat to Fraxinus excelsior caused by the fungus Hymenoscyphus fraxineus, which has driven much prior research on ash microbiomes. The authors suggest that an optimized core microbiome may be considered crucial for improving tolerance against plant diseases in future, and that if microbiome members are to be used as biocontrol agents, regional factors must be accounted for to ensure successful establishment of introduced microorganisms on the host. They propose that the genotype of ash trees and the different strata of the canopy should be investigated next, to establish which bacterial taxa are crucial for a stable, protective and beneficial ash leaf microbiome.

Subject of Research: Bacterial leaf microbiome diversity and drivers in European ash trees across Germany

Article Title: The core taxa connectivity in leaf microbiomes in ash trees stands is modulated by host-related and environmental factors

Article References: Burghard, V., Wende, S., Kahlenberg, G., Häuser, H., Kätzel, R., Kolb, S., & Ulrich, A. (2026). The core taxa connectivity in leaf microbiomes in ash trees stands is modulated by host-related and environmental factors. Environmental Microbiome, 21(1), Article 119. https://doi.org/10.1186/s40793-026-00967-8

Image Credits: AI Generated

DOI: 10.1186/s40793-026-00967-8

Keywords: ash trees, Fraxinus excelsior, phyllosphere, microbiome, core microbiome, co-occurrence networks, climate, soil chemistry, ash dieback, forest ecology, 16S rRNA sequencing, Germany