From the leaf to the gut and back again: the fate and influence of phyllosphere bacteria in a gnotobiotic Arabidopsis – Pieris brassicae system

From the leaf to the gut and back again: the fate and influence of phyllosphere bacteria in a gnotobiotic Arabidopsis – Pieris brassicae system | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article From the leaf to the gut and back again: the fate and influence of phyllosphere bacteria in a gnotobiotic Arabidopsis – Pieris brassicae system Moritz Müller, Maryse A. P. Huve, Michael Kunzler, Mitja N. P. Remus-Emsermann, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8337060/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 13 You are reading this latest preprint version Abstract Background: The leaf surface, or phyllosphere, hosts abundant and diverse bacterial communities that interact with both the host plant and herbivorous insects, yet their collective influence on plant–insect interactions remains poorly investigated. We established a gnotobiotic insect-plant system combining Arabidopsis thaliana and Pieris brassicae larvae. Using defined synthetic phyllosphere communities (SynComs) of increasing richness (5, 10, or 20 members), we investigated how phyllosphere bacteria influence herbivore performance, plant defence responses, and bacterial colonisation of both leaves and the insect gut. Results: While larval weight tended to decrease with increasing community richness, only the most diverse SynCom (20 members) caused significant weight reductions without affecting survival. Likewise, plants harbouring the most diverse community showed enhanced jasmonic acid (JA) levels during feeding, whereas salicylic acid (SA) remained unchanged, suggesting the specific induction of JA-associated defences. Compared to plants experiencing no herbivory, feeding strongly reshaped bacterial colonisation on leaves, increasing total bacterial loads about fourfold and driving dominance of Pantoea eucalypti 299R as shown by 16S rRNA gene amplicon sequencing. Larvae acquired a distinct subset of bacteria, primarily recruited from the genera Methylobacterium , Microbacterium , Williamsia , and Curtobacterium . Conclusions: Together, these findings suggest that resident phyllosphere bacteria modulate plant defences and thereby affect herbivore performance, while herbivory restructures the leaf microbiota and bacterial filtering occurs during passage through the insect gut. Gut microbiota Jasmonic acid Lepidoptera Phyllosphere microbiota Synthetic community Insect performance Community diversity Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 INTRODUCTION The phyllosphere—the aerial surface of plants—represents one of the largest terrestrial habitats for microbial life on Earth, with leaves alone covering an estimated 6.4 × 10⁸ to 1 × 10⁹ km² globally and hosting up to 10 7 bacterial cells per gram of leaf tissue [1]. Leaves are consumed by a wide range of herbivorous insects, many of which are major pests in agricultural and natural ecosystems [2, 3]. The phyllosphere therefore represents an ecologically important interface where microbes and insects interact, with direct consequences for plant health and productivity. Microbes and herbivorous insects can influence each other indirectly through the plant’s immune system, which integrates multiple inducible defence pathways that are largely regulated by phytohormones [4, 5]. Jasmonic acid (JA) typically mediates responses to chewing herbivores and necrotrophic pathogens, while salicylic acid (SA) mediates responses to biotrophic pathogens and piercing–sucking insects [4, 6, 7]. Because these pathways often act antagonistically, activation of one often suppresses the other, creating scope for plant-mediated microbe–insect interactions. Accordingly, microbes and herbivores influence each other by modulating JA-SA cross-talk [8]. For instance, when leaves are damaged by chewing herbivores, JA-mediated signalling suppresses SA-dependent defences, allowing biotrophic pathogens to proliferate [7, 9]. Some pathogens enhance herbivore performance by activating SA-mediated defences or, conversely, suppress feeding by producing JA mimics [9–11]. Insect mutualists, which can be either pathogenic or nonpathogenic to plants, may suppress plant anti-herbivore defences by eliciting SA responses or by preventing recognition of the insect, thereby reshaping insect-plant interactions [12, 13]. Beyond pathogens and mutualists, nonpathogenic phyllosphere bacteria also elicit plant immune responses when they reach high densities [14–16]. Despite these insights, the broader ecological role of nonpathogenic phyllosphere bacteria in shaping plant–insect interactions is still poorly understood. Besides hormonal cross-talk, herbivorous insects and phyllosphere bacteria also interact through a range of direct physical and ecological processes. Herbivory damage releases nutrients into the otherwise nutrient-poor phyllosphere, creating transient hotspots that promote bacterial growth and entry into the apoplast [17, 18]. Insects further impact microbial dispersal by vectoring bacteria onto leaves via frass, saliva, or body surfaces [12]. These effects operate alongside other ecological factors shaping the leaf microbiota, including host identity, plant immune activity, cuticle composition, and plant metabolism [19]. Together, these processes shape the composition and dynamics of phyllosphere bacterial communities and may influence how microbes and herbivores affect each other through the plant. Most studies on plant–microbe–insect interactions in the phyllosphere have focused on individual mutualistic or pathogenic species [12, 20, 21]. By contrast, the ecological roles of resident nonpathogenic microbial communities and their effect on plants and herbivorous insects remain poorly understood. This raises key questions: Does the composition and diversity of microbial communities influence herbivory, and how does herbivore feeding, in turn, restructure these communities? Do phyllosphere microbes affect insect performance indirectly by modulating host defences, or directly through ingestion and gut colonisation? To address these questions, we established a gnotobiotic insect-plant system combining Arabidopsis thaliana (thale cress), Pieris brassicae (Large White; Lepidoptera) larvae, and synthetic phyllosphere bacterial communities (SynComs) of different richness. Pieris brassicae is a specialist herbivore of Brassicaceae, widely distributed in Europe, Asia, and North Africa, and a major pest of crops such as cabbage, broccoli, and mustard [22]. Its larvae are foliar feeders that cause extensive tissue damage. Like most studied Lepidoptera, the larval gut of P. brassicae lacks a stable resident microbiota, with bacteria acquired transiently mainly through feeding [23, 24]. Under laboratory rearing conditions, the gut of P. brassicae is virtually sterile, with only minimal bacterial presence [24], making this species an ideal system for manipulating bacterial exposure. This gnotobiotic insect-plant system enables investigation of how phyllosphere bacterial communities influence caterpillar performance, plant responses, and microbial colonisation patterns within a defined tripartite plant–microbe–insect interaction. To explore how resident phyllosphere bacteria shape the interplay between plants and herbivores, we combined insect performance assays, phytohormone analysis, and 16S rRNA gene amplicon sequencing. We investigated (1) how phyllosphere bacteria affect herbivorous insect performance, (2) how phyllosphere bacteria modulate levels of anti-herbivore defence-associated phytohormones, (3) how herbivory shapes the leaf-associated bacterial community, and (4) which of these bacteria are recruited by the larval gut environment. This work uncovers emergent properties of plant–microbe–insect interactions in the phyllosphere and establishes a tractable gnotobiotic model for dissecting their underlying mechanisms. MATERIALS AND METHODS Plant growth conditions Arabidopsis thaliana Col-0 seeds were surface-sterilised with 70% v/v ethanol for 2 min, followed by a 50% v/v household bleach solution (DanKlorix, Germany) containing 0.2% v/v Tween-20 for 7 min. Seeds were thoroughly washed with sterile water three times and stratified for two days at 4 °C. Single seeds were sown onto approximately 0.2 cm thick, 0.5 cm x 0.5 cm wide slices of half-strength Murashige and Skoog (½ MS) medium (Duchefa) with 1% w/v plant agar (Duchefa) or onto cut off pipette tips filled with the same medium [16], transferred into Petri dishes wrapped with Parafilm (Bemis Parafilm PM-996) and put into a plant growth chamber (poly klima GmbH, M5Z- TDL+) at 21 °C, 80% relative humidity, ~120 µmol m -2 s -1 light intensity, and a 11 h light/13 h dark photoperiod. Then, nine-day-old seedlings were transferred into sterile plant tissue culture boxes (size: 7 x 7 x 8 cm) (Bioworld Magenta GA-7 Plant Culture Box) filled with 90 g zeolite (0.2-0.5 mm grain ZeoLith Klinoptolith, labradorit.de) supplemented with 40 ml of sterile ¾ MS medium (Duchefa). Box lids had four holes with a diameter of roughly 0.5 cm sealed with a double layered micropore strip (3M Deutschland GmbH) to prevent contamination while allowing for air exchange. Four seedlings were transferred under aseptic conditions into each box and put back into the growth chamber. Each box represented one biological replicate. Synthetic bacterial communities SynCom composition was selected from a pool of 20 bacterial strains from prevalent genera of the A. thaliana phyllosphere microbiota [25]. All strains used were originally isolated from the phyllosphere and are nonpathogenic to A. thaliana (Figure 1 and Table S1). From the species pool, SynComs with three levels of species richness were designed, comprising 5, 10, and 20 strains (SynCom5, SynCom10, and SynCom20). For SynCom5 and SynCom10, members were randomly drawn out of the total pool and therefore unique in their composition in each of four independent experimental replicates (Figure S1). Strains were individually grown in Reasoner's 2A (R2A) medium or on R2A agar (HiMedia) for 1-4 days at 30 °C. Cultures were harvested by centrifugation for 7 min at 8,500 × g , resuspended in 10 mM MgCl 2 , diluted to a predetermined optical density at 600 nm (OD 600 ), and assembled in equal CFU ratios (Table S1). SynComs were stored as glycerol stocks (20% v/v glycerol) at -80 °C in either 0.5- or 1-ml aliquots. Inoculation of A. thaliana with bacterial communities SynCom glycerol stocks were thawed and centrifuged at 8,500 × g for 7 min. The supernatant was discarded and the pellet was resuspended in 1 ml phosphate-buffered saline (PBS; 8 g/L NaCl, 1.44 g/L Na 2 HPO 4 , 0.24 g/L KH 2 PO 4 , 0.2 g/L KCl; pH 7.4). Bacterial suspensions were then diluted to an OD 600 of 0.005. Plants were sprayed four weeks after sowing with an airbrush (Harder & Steenbeck Ultra) at 10 psi using a sterile plant culture box lid containing a single micropore-sealed hole (Figure 2). Before and between treatments, the airbrush was sterilised with 70% v/v ethanol and rinsed with sterile PBS. All boxes received 200 µl of spray volume; untreated control boxes were sprayed with sterile PBS instead of SynCom. Insect rearing and neonate collection for experiments Pieris brassicae larvae were reared on greenhouse-grown Brussels sprout plants ( Brassica oleracea var. gemmifera ) in a temperature-controlled room at 22 °C, 75% relative humidity, ~120 µmol m⁻² s⁻¹ light intensity and an 18 h light/6 h dark photoperiod, in 50 × 50 × 60 cm rearing cages containing approximately 100 individuals, until pupation and emergence. Adults were transferred to a greenhouse (20–24 °C day / 18–22 °C night, 75% relative humidity, ~120 µmol m⁻² s⁻¹ light intensity, 16 h light/8 h dark photoperiod) and fed with a 15% v/v aqueous honey solution. For egg collection, seven-week-old Brussels sprout plants were prepared by spraying the underside of the leaves with 70% v/v ethanol and allowing them to dry. One plant was then placed in a cage containing one-week-old butterflies and left overnight for oviposition. The egg-laden plant was then transferred to the larval temperature-controlled room. To prevent the hatching larvae from feeding on the egglaying-plants, eggs were gently removed from the leaves using sterilised tweezers and placed in sterile Petri dishes one day before hatching (four days after egg laying). On the day of hatching, timed to coincide with eight days after SynCom inoculation, all neonates were pooled and randomly assigned to the plant treatments. Neonates were carefully transferred using sterilised paintbrushes, with three neonates placed per plant tissue culture box (Figure 2). Each box constituted one biological replicate. Assessment of the gnotobiotic insect-plant system To evaluate the suitability of the gnotobiotic insect-plant system, sterile A. thaliana Col-0 seeds (prepared as described above) were sown on agar. After nine days, part of the seedlings were transplanted into pots (size: 9 x 9 x 8 cm) containing a 3:1 mixture of steamed soil (Einheitserde Typ P soil, Kausek) and vermiculite (Kausek). The remaining seedlings were transferred into sterile plant tissue culture boxes as described above. As in the plant tissue culture boxes, four plants were transferred in each pot. Plants were used 36 days after sowing and four neonate larvae were placed per pot. Larval biomass and survival were evaluated after seven days of feeding on plants grown in parallel either in pots or in tissue culture boxes. To that end, one neonate was placed on each plant within the boxes or pots, for a total of four neonates per box or pot. To prevent larvae from escaping, pots were enclosed in microperforated bags (Packseller) and sealed with tape. Plants and boxes were then placed in a plant growth chamber (poly klima GmbH, M5Z- TDL+) set to 21 °C, 80% relative humidity, ~120 µmol m -2 s -1 light intensity, and a 11 h light/13 h dark photoperiod. Each box or each pot constituted one biological replicate. All larvae from each replicate were weighed together using a precision balance (Precisa H_225SM-DR; resolution 0.01/0.1 mg) and divided by the number of weighed larvae. The number of larvae per replicate was also recorded. Larval sampling For all SynCom experiments , P. brassicae larvae were sampled one day after being placed on the plants for phytohormone measurements, or after seven days for all other experiments (Figure 2). Larvae from each box were pooled and treated as one biological replicate. The number of surviving larvae and their total weight were recorded for each box using a precision balance (Precisa H_225SM-DR; resolution 0.01/0.1 mg). Larvae were placed at –20 °C for 2 min to induce lethargy, then homogenised in 250 µl PBS in 1.5 ml Eppendorf tubes using sterilised metal pestles. An aliquot of the homogenate was stored at –20 °C for DNA extraction. From 40 µl of homogenate, a tenfold serial dilution was prepared and plated on R2A agar using a 96-pin replicator to quantify total CFU per larval weight. Leaf sampling Leaves were collected seven days after larval placement, in parallel with larval sampling. Two larvae-fed leaves per plant were taken from each box, excised at the petiole, and transferred into sterile 1.5 ml Eppendorf tubes. Fresh weight was recorded using a precision balance (Precisa H_225SM-DR; resolution 0.01/0.1 mg). To detach phyllosphere bacteria, each tube was filled with 1 ml PBS, sonicated for 5 min (75% ultrasound efficiency; EMAG Emmi-12HC), and vortexed for 10 s. A tenfold serial dilution of each leaf wash was then prepared and inoculated onto R2A agar using a 96-pin replicator. Each sample was plated in duplicate and incubated at 30 °C. Total CFU counts were normalised to leaf fresh weight, and the remaining leaf wash was stored at –80 °C for DNA extraction. DNA extraction and 16S rRNA gene amplicon sequencing DNA was extracted from larval, leaf, and SynCom20 inoculum samples using the MasterPure™ Complete DNA & RNA Purification Kit, with adjusted volumes of Tissue and Cell Lysis Solution. For larvae, 300 µl lysis solution was added to 150 µl homogenate and transferred to a 2-ml tube containing seven sterile zirconia beads. For leaf samples, 450 µl lysis solution was added per sample. For the SynCom20 inoculum, glycerol stocks were centrifuged at 8,500 × g for 7 min in a 2-ml tube; the supernatant was discarded, and seven zirconia beads plus 450 µl lysis solution were added. Larval and inoculum samples were homogenised in a Retsch MM400 bead mill for four cycles of 2 min at 30 Hz. Extraction negative controls consisting of reagent-only tubes were processed in parallel. Subsequent steps followed the kit protocol. DNA quality and quantity were measured using a DeNovix DS-11 FX+ spectrophotometer/fluorometer with the dsDNA Broad Range kit. Samples were stored at –20 °C until processing. Amplicon sequencing of the V3-V4 region of the 16S rRNA gene was done using the primers 341F (5’-CCTACGGGNGGCWGCAG-3’) and 806R (5’-GGACTACNVGGGTWTCTAAT-3’) with barcodes for pooling. PCR reaction mix was prepared per sample with 10 µl 5x Phusion GC Buffer, 1 µl 10 mM dNTPs, 2.5 µl forward and reverse primers at 10 µM, 50 ng template DNA, 1.5 µl DMSO, 0.5 µl Thermo Scientific Phusion High-Fidelity DNA Polymerase and filled up to 50 µl with water. The PCR cycling protocol included an initial denaturation for 30 s at 98 °C, then 30 cycles of denaturation at 98 °C for 10 s, annealing at 58 °C for 30 s and extension at 72 °C for 30 s. A final extension was done at 72 °C for 5 min. Amplicons were checked by agarose gel electrophoresis and purified using the Monarch PCR & DNA Cleanup Kit (New England Biolabs). DNA quality and concentration of purified amplicons were assessed on a DeNovix DS-11 FX+ spectrophotometer/fluorometer, and samples were pooled and concentrated to meet sequencing requirements. The pooled amplicon library was then sent to Novogene (Germany) for library preparation (PCR-free protocol) and sequencing on an Illumina NovaSeq X Plus platform (250-bp paired-end reads). Analysis of 16S rRNA gene amplicon data Paired-end reads were merged using PEAR with a minimum overlap of 10 bp and a p -value threshold of 0.01 [26]. Merged reads were demultiplexed using Cutadapt, and sequences for each sample were quality-filtered using USEARCH v11.0.667 (Edgar, 2010), retaining reads >350 bp with Phred scores >19. Filtered reads were dereplicated and clustered into OTUs at 97% sequence similarity. Representative sequences and OTU abundance tables were generated for taxonomic classification and diversity analysis, respectively. This analysis was performed with the High-Performance Computer “CURTA” of FUB-IT, Freie Universität Berlin [27]. Further analyses were conducted in RStudio [28], using the DECIPHER, Phyloseq, and Biostrings packages [29–35]. OTUs were classified using the SILVA rRNA database, and chloroplast, mitochondria, and unclassified OTUs were removed (Figure S2). Members of SynCom20 were identified by global alignment of OTU sequences with a custom V3-V4 reference database using USEARCH v11.0.667 [36], with at least 98% sequence identity and full-length coverage. For each query, only the best hit was reported. Phylogenetic relationships among Syncom20 members were generated based on orthogroup inference of annotated whole-genome sequences using OrthoFinder [37, 38]. Phylogenetic structure was evaluated by calculating the Faith’s phylogenetic diversity (PD) and mean pairwise phylogenetic distances (MPD), and visualised using the packages ggtree, picante, ape and ggstance [39–46]. Analysis of the bacterial communities was done with the R package vegan [47]. To minimise sampling bias, samples with fewer than 250 reads, including the negative controls, were excluded from further analysis, resulting in ≥99.5% Good’s coverage across all retained samples. OTU tables were rarefied 100 times to 250 reads per sample, mean abundances were determined and OTUs with mean of zero were removed. Alpha diversity was estimated by richness, evenness and Shannon diversity index. Beta diversity was estimated on Bray-Curtis distances and used to visualise differences in community composition among treatments using non-metric multidimensional scaling (NMDS). To identify taxa whose abundance was associated with a treatment, OTU vectors were fitted onto the NMDS ordination space ( envfit , vegan package), retaining only significant correlations at p < 0.05. In addition, differential abundance analysis on relative abundances of SynCom20 was performed using DESeq2 to identify taxa that are significantly enriched or depleted across treatments [48]. P -values of fold changes were corrected using Benjamini-Hochberg (BH) procedure to control for false discovery rates. Phytohormone extraction and analysis Leaf material for phytohormone quantification was collected eight days after bacterial inoculation and before placing the larvae (T0), and 24 hours later (T1). One leaf was sampled per plant from each replicate box, prioritising larvae-fed leaves. Plant material was sampled in 2 ml tubes containing four zirconia beads, and immediately snap-frozen in liquid nitrogen before storing at -80 °C until processing. Phytohormones were extracted using a protocol outlined by Wang et al. 2007 [49]. Briefly, frozen samples were weighed and homogenised in 1 ml of ethyl acetate and 2 µl of internal standards containing deuterated phytohormones (10 ng/ul D4-SA, 10 ng/ul D6-ABA [abscisic acid], 30.2 ng/ul D6-JA, 10 ng/ul D6-JA-Ile; HPC Standards GmbH). Homogenisation was done in cooled adapters on a Retsch MM400 bead mill for nine cycles of 20 s each at 30 Hz, followed by centrifugation at 13,000 × g for 10 min at 4 °C. Supernatants were transferred to new tubes, while 1 ml ethyl acetate was added to the pellets and the extraction step was performed a second time without internal standard. The supernatants from the second extraction round were pooled with the first ones and concentrated under vacuum (Savant DNA Speed Vac DNA 110). The concentrates were re-eluted in 300 µl of re-elution buffer (70% methanol, 0.1% formic acid), vortexed for 10 min at 22 °C and 1,500 rpm on the BIOER Mixing Block MB-102, and centrifuged for 20 min at 13,000 × g at 22 °C, before 200 µl supernatant was transferred into a HPLC vial. Extraction blanks were processed alongside samples. Quantification of JA, JA-Ile, ABA, and SA was performed using LC-ESI-IMS/MS (Waters UPLC-Synapt G2-S HDMS). Peak identification and integration were done in MassLynx (Waters). Phytohormone concentrations were normalised to their corresponding deuterated internal standards and the leaf fresh weight. JA-Ile was normalised by the ratio of the deuterated JA-Ile and the deuterated JA-Ile impurity. Finally, phytohormone levels at T1 were normalised by the average level of that phytohormone at T0. Data from two independent experiments were analysed jointly. Statistical analysis All statistical analyses, regression models, and data visualisation were performed in RStudio [28]. Plots were generated using the packages ggplot2, ggsignif, smplot2, and patchwork [50–53]. Data wrangling was carried out using the tidyverse [54]. Unless stated otherwise, (generalised) linear mixed-effects models ((G)LMM) were fitted to account for variation among independent experiment replicates, using the nlme or glmmTMB package [55–57]. Depending on the experiment, response variables included larval weight, log 10 -transformed CFU counts, larval survival, or normalised phytohormone levels. SynCom treatments and/or larval feeding were treated as fixed effects, while experimental replicate was included as a random intercept. Model assumptions were evaluated by visual inspection of residual plots. The effect of different treatments were evaluated using analysis of variances (ANOVA) for linear mixed models or a Type II Wald chi-square test for generalised models. Then, estimated marginal means were calculated using the emmeans package [58], and pairwise comparisons were adjusted with the BH procedure for multiple comparisons. Compact letter displays were generated to indicate statistically different treatment groups. For comparisons between two groups, parametric or non-parametric tests were applied after inspecting assumptions of normality (Shapiro-Wilk normality test) and homoskedasticity (Levene’s test) using the rstatix package [59]. Differences in larval biomass between open-pot and gnotobiotic systems were evaluated using Wilcoxon signed-rank test, whereas larval survival across these growth systems was compared using Fischer’s exact test. Bacterial abundance on the leaves after a seven day P. brassicae feeding period on either SynCom20 or axenic control plants was compared using Student’s t -test. To determine whether bacterial community composition differed among sample types (inoculum, fed leaf, non-fed leaf, larvae) after the feeding period, PERMANOVA was conducted on a Bray-Curtis distance matrix using the vegan package [60]. Differences in alpha diversity metrics (richness, evenness, Shannon diversity) and mean pairwise phylogenetic distances (MPD) were analysed using ANOVA followed by Tukey’s post hoc test. RESULTS Bacterial abundance on leaves and caterpillars across communities of different richness and its effect on caterpillar performance We established a gnotobiotic insect–plant system to study the interactions between A. thaliana , phyllosphere bacteria, and P. brassicae . To confirm that the system was suitable for larval rearing, we compared P. brassicae performance on gnotobiotic plants with that on plants cultivated in pots. While larvae were on average larger in pots compared to gnotobiotic plants, larval survival did not differ between conditions (Figure S3). To assess the effect of community richness in the interaction between A. thaliana and P. brassicae , plants were inoculated with SynComs of different richness. A pool of 20 phyllosphere strains was selected to capture the taxonomic diversity of natural leaf communities (Figure 1). Five or ten strains were randomly selected from the bacterial pool to assemble SynCom5 and SynCom10 for each of four independent experiments, respectively (Figure S1). A SynCom composed of the total strain pool was included to capture the highest richness level (SynCom20). Eight days after spray inoculation of plant leaves, P. brassicae neonates were introduced onto gnotobiotic or axenic control plants, and after seven days of feeding, bacterial abundance on fed leaves and caterpillars, as well as larval performance, were assessed. After seven days of feeding in the gnotobiotic system, larval survival rate was similar across treatments. On axenic control plants, an average of 96% of larvae survived, while mean survival on plants inoculated with different SynComs was 99%, 96%, and 98%, respectively, with no significant differences among treatments (Figure 3A, ꭓ 2 = 1.54, df = 3, p = 0.67). Phyllosphere bacterial abundance on A. thaliana leaves was similar across SynCom treatments, with only a modest increase in SynCom20 compared to SynCom10 (Figure 3B t (83) = 2.2, p adj = 0.045). Median bacterial loads were 1.6 × 10 8 , 1.2 × 10 8 , and 1.9 × 10 8 CFU per g of leaf fresh weight for SynCom5, SynCom10 and SynCom20, respectively. The effect of SynCom richness on bacterial loads was significant (ANOVA; F 3,83 = 209.5, p < 0.001). All plants that were inoculated with SynComs carried significantly higher bacterial loads than untreated controls ( p adj < 0.001). Bacterial abundance in P. brassicae larvae differed significantly among SynCom treatments (ANOVA; F 3,100 = 409.1, p < 0.001), with median loads of 1.6 × 10⁶, 1.1 × 10⁷, and 1.6 × 10⁷ CFU per g of larval fresh weight after feeding on plants inoculated with SynCom5, SynCom10, and SynCom20, respectively (Figure 3C). All SynCom treatments carried significantly higher bacterial loads than the untreated control ( p adj < 0.001). SynCom10 and SynCom20 had significantly higher bacterial loads than SynCom5 ( p adj < 0.01), whereas no significant difference was detected between SynCom10 and SynCom20. Larval weight was significantly influenced by SynCom treatment (Figure 3D, ANOVA on LMM; F 3,100 = 3.62, p = 0.0157), with larvae fed on plants inoculated with SynCom20 having a significant reduction in weight compared to larvae fed on control plants (Figure 3D, p adj = 0.0153). On average, larval weight decreased by 13.1%, 14.4%, and 17.6% when plants were inoculated with SynCom5, SynCom10, and SynCom20, respectively, compared to the mean larval weight on axenic control plants. However, no significant differences were detected among the three SynCom treatments (ANOVA on LMM, all comparisons: p adj > 0.05). Larval weight correlated negatively with SynCom richness (r = -0.27, p < 0.005; Figure S4A) and phylogenetic diversity (Faith’s PD, r = -0.30, p < 0.002; Figure S4B). Relative bacterial abundance on leaves (Figure S4C) and in larvae did not correlate with larval weight (Figure S4D). SynCom20 alters phytohormone levels of A. thaliana to caterpillar feeding We investigated how phyllosphere bacteria and herbivory influence anti-herbivore defence signalling in A. thaliana , by quantifying changes in the stress-related phytohormones JA, JA-Ile, SA, ABA. Plants were inoculated either with PBS (untreated control) or with SynCom20 and subsequently exposed to P. brassicae larvae. Phytohormone levels were measured at the onset of larval feeding (T0) and 24 hours later (T1) using gas chromatography–mass spectrometry (GC-MS) in two independent experiments. Feeding by P. brassicae induced strong changes in the phytohormone profiles of A. thaliana leaves (Figure 4). After 24 h, JA levels markedly increased in response to herbivory (GLMM, p < 0.05), with SynCom20-inoculated plants showing significantly higher accumulation than axenic control plants ( p adj < 0.0001). JA-Ile and ABA levels also increased significantly in response to herbivory, but were unaffected by the SynCom20 treatment. By contrast, SA concentrations remained unchanged across treatments. Together, these results show that P. brassicae herbivory primarily triggers JA-, JA-Ile-, and ABA-mediated defences, with SynCom20 further enhancing JA accumulation in A. thaliana . Larval feeding increases bacterial abundance on leaves and alters phyllosphere community composition To assess how larval feeding influences phyllosphere bacterial abundance, we compared SynCom20-inoculated plants that were fed upon for seven days to plants that did not experience feeding. As observed previously, SynCom20 inoculation reduced larval weight after feeding (Figure 5A; t -test, t (8) = 2.51, p < 0.05). Leaf bacterial loads were approximately fourfold higher on fed compared to non-fed plants, with mean abundances of 4.2 × 10 8 and 1 × 10 8 CFU per gram of leaf, respectively (Figure 4B; t -test, t (8) = 5.44, p < 0.05). These results indicate that larval feeding promotes bacterial proliferation in the phyllosphere. To examine how P. brassicae feeding affects phyllosphere bacterial composition and the acquisition of bacteria in the larval gut, we performed 16S rRNA gene amplicon sequencing (V3-V4 region). After seven days of feeding, we collected three types of samples: (i) SynCom20-inoculated leaves that had been fed upon, (ii) the larvae that fed on those leaves, and (iii) SynCom20-inoculated leaves that were not exposed to larvae. In addition, the original SynCom20 inoculum was sequenced to assess changes in community composition after the treatments. Sample type explained 73.9% of the variance between communities (PERMANOVA, p < 0.001). Relative to the original SynCom20 inoculum, both leaf and larval samples showed marked compositional shifts, with several strains becoming undetectable or persisting only at very low abundance (Figure 6A). Despite a strong overlap in taxa between leaves and larvae, differences in strain abundances resulted in distinct community compositions (Fig 6A, B). Larval feeding caused leaf communities to become dominated by Pantoea eucalypti 299R, except for two samples that clustered more closely with non-fed leaves or larval communities (Fig 6A, B). Given the strong increase in total bacterial load on fed leaves, P. eucalypti 299R was likely responsible for this proliferation (Fig 5B). To a lesser extent, Microbacterium sp. AC026 and Bacillus sp. AC266 also increased in relative abundance in fed compared to non-fed leaves, whereas communities of non-fed leaves were characterised by uniquely high abundances of Arthrobacter Leaf145 (Figure 6C). By contrast, larval samples were consistently enriched in Methylobacterium strains ( Methylobacterium radiotolerans 0-1 and Methylobacterium AC433), Microbacterium sp. AC026, Curtobacterium sp. AC273, and Williamsia sp. Leaf354, regardless of fed or non-fed leaves (Figure 6B, C). Sphingomonas strains, which were present at low abundance in both fed and non-fed leaves, were strongly underrepresented in larval samples (Figure 6C). Alpha diversity differed markedly among the bacterial communities from the inoculum, leaves, and larvae (Figure 7). Richness was highest in the SynCom20 inoculum, and decreased in both leaf and larval samples. Evenness was reduced in fed leaf communities compared to non-fed leaves ( p adj = 0.0256), reflecting the dominance of P. eucalypti 299R during herbivory (Figure 5). Shannon diversity followed a similar pattern, that is, an overall reduction in alpha diversity in fed leaf communities. The mean pairwise distance (MPD), describing the phylogenetic relatedness of strains within a community, was also highest in the inoculum and declined in plant- and larval-associated bacterial communities. Together, these results show that herbivory reshapes SynCom20 communities by increasing richness but reducing evenness and phylogenetic breadth, leading to dominance of specific strains within the leaf microbiota. DISCUSSION Using a gnotobiotic insect-plant system, we showed that phyllosphere bacteria and herbivory can jointly influence plant defences, the composition of plant and insect microbiota, and insect performance. Increasing bacterial richness in the phyllosphere reduced P. brassicae larval growth, with a significant effect observed only for the richest synthetic community tested: SynCom20. Herbivory reshaped the phyllosphere microbiota, increasing bacterial loads and enriching P. eucalypti 299R, while larvae acquired a filtered subset of leaf-associated bacteria dominated by Methylobacterium , Microbacterium , Williamsia , and Curtobacterium . Larval feeding on SynCom20-inoculated plants triggered a stronger induction of jasmonate-associated herbivore-responsive signalling than feeding on axenic control plants. These findings reveal reciprocal feedback between phyllosphere bacteria and herbivory, highlighting that leaf-associated bacteria shape plant defence responses and thereby modify herbivore performance. Although plant-associated microbes are known to modulate plant responses to herbivory and influence herbivore performance, most evidence comes from rhizosphere systems [61, 62]. In our system, P. brassicae larval growth tended to decrease with increasing bacterial richness in the phyllosphere, with a significant reduction observed only for the richest community (SynCom20) compared to axenic control plants. This pattern suggests that higher bacterial richness may enhance the community’s overall impact on plant–insect interactions, potentially through complementary or synergistic effects among strains influencing plant defences [63, 64]. In microbial systems, richer communities often exhibit metabolic complementarity, functional redundancy, and emergent properties that improve community performance and stability [63, 64]. Such emergent properties could enhance the ability of phyllosphere communities to modulate host physiology, thereby altering plant–herbivore interactions. Comparable patterns have been observed in rhizosphere microbiomes, where complex soil communities more reliably induce systemic resistance against herbivores than single-strain inocula [65–67]. However, it remains unclear whether the observed patterns in our system were driven by overall richness, emergent interactions among community members, or the activity of key taxa within SynCom20. Candidate strains contributing to these richness-dependent effects include those that became most abundant during larval feeding, such as P. eucalypti 299R, which is known to be a strong and metabolically versatile competitor in the phyllosphere [68]. These traits may promote its competitive success and capacity to shape community structure during herbivory, potentially affecting plant responses either directly through microbe–host signalling or indirectly by altering the abundance of taxa that influence defence signalling. Future work employing broader richness gradients, alternative community compositions, and targeted removal or enrichment of specific members could help disentangle these contributions. Overall, these findings show that even nonpathogenic phyllosphere bacterial communities affect herbivore growth, revealing an underappreciated ecological role of resident leaf microbiota in plant anti-herbivore responses. The ability of phyllosphere bacteria to alter herbivore performance in our experimental system likely arises from their influence on plant defence signalling rather than from direct effects on the larvae, as bacterial abundance on the leaf surface was not correlated with larval weight and no disease symptoms were observed. Moreover, inoculation of A. thaliana with SynCom20, the community associated with a significant decrease in larval growth, led to stronger JA induction during P. brassicae feeding, accompanied by similar, though weaker, trends for JA-Ile and ABA, while SA levels remained unchanged across treatments. Together, these results suggest that SynCom20 amplified JA-associated anti-herbivore responses during feeding, likely contributing to the observed reduction in larval growth and hinting at a priming effect on plant defences. Microbial priming of plant defences occurs when prior exposure to microorganisms enhances the plant’s capacity to respond more rapidly or intensely to subsequent challenges—such as pathogen infection or insect attack—compared with unprimed plants [69–71]. Rhizosphere-associated bacteria are well-known to induce such JA-dependent priming; for example, colonisation of A. thaliana roots by Pseudomonas simiae WCS417r primes JA/ET-regulated defences and enhances resistance against the chewing caterpillar Mamestra brassicae [72], and plant-growth promoting rhizobacteria exposure increases JA accumulation and JA-responsive gene expression in cotton, reducing damage and growth of Spodoptera exigua larvae [73, 74]. A similar mechanism may occur in the phyllosphere, where prior colonisation by nonpathogenic bacteria could sensitise JA signalling pathways without directly activating defences, allowing plants to mount a stronger response upon herbivory. Indeed, transcriptomic analysis of Arabidopsis leaves has shown that colonisation by the commensal Sphingomonas melonis Fr1, a member of our SynCom20, triggers the expression of hundreds of defence-associated genes in A. thaliana , including the JA/ethylene-responsive marker PDF1.2 [75]. This suggests that certain members of the phyllosphere actively maintain the plant immune system in an “alert” state. In addition to jasmonates, ABA also increased in response to herbivory, and this increase tended to be stronger in SynCom-inoculated plants, suggesting that bacterial colonisation may have enhanced ABA accumulation during feeding. ABA interacts extensively with JA signalling: ABA–JA crosstalk contributes to defence against insects, and ABA receptors such as PYL6 physically interact with the JA regulator MYC2 to modulate JA-responsive gene expression and resistance to herbivory [76]. Such hormonal crosstalk can alter both the magnitude and timing of JA-regulated defences under repeated or combined stresses [71], which could help explain the stronger inducible response observed with SynCom20. On the other hand, SA, typically associated with defence against biotrophic pathogens and systemic acquired resistance, remained unchanged across treatments [76]. Interestingly, SynCom20 alone did not induce SA accumulation in the absence of feeding, indicating that nonpathogenic phyllosphere bacteria do not activate strong pathogen-type defences. This pattern is consistent with the idea that commensal plant-associated microbes evolve strategies to limit costly immune activation—for example by dampening microbe-associated molecular pattern (MAMP)-triggered immunityor reducing recognition—while still modulating defence pathways [63, 67]. In our system, SynCom20 nevertheless enhanced JA accumulation during herbivory, consistent with a priming-like scenario in which nonpathogenic colonisers remain tolerated yet sensitise plants for stronger induced anti-herbivore defences [69–72]. However, the precise mechanisms by which SynCom20 enhances JA accumulation upon herbivory remain to be elucidated. While these hormonal patterns point to a defence amplification during feeding, the underlying triggers of the enhanced JA response could involve herbivory-induced shifts in phyllosphere community structure, as insect feeding is known to reshape leaf-associated bacterial assemblages [17], and as demonstrated by the strong compositional changes observed in our own system. Feeding by P. brassicae increased bacterial load on the leaf surface, suggesting that higher microbial densities achieved during herbivory could contribute to the enhanced JA response. This is consistent with recent findings that plant immune responses to nonpathogenic phyllosphere bacteria are density-dependent; while plants mount negligible responses to low populations, high bacterial titers elicit pathogen-like transcriptional reprogramming [77]. Chewing damage facilitates bacterial proliferation on leaves [17] and provides bacteria with entry points into internal tissues through wounds [78]. Once inside the apoplast, microbial molecules are perceived by pattern-recognition receptors located on the plasma membrane, making this compartment a major site for defence activation [79, 80]. Because bacterial densities within the apoplast are generally much lower than those found on the leaf surface, even limited internal colonisation may be sufficient to trigger strong immune responses. In this context, the intensified JA signalling observed in SynCom20-inoculated plants could result from the higher bacterial densities that accumulate during feeding and from immune activation following bacterial entry into wounded tissue. Effectively, the rapid proliferation of SynCom members may push bacterial titers beyond a detection threshold, triggering a defence “alert” that is driven by bacterial density, in line with previous studies [15, 77]. Such processes may underlie how phyllosphere bacteria, despite being nonpathogenic, modulate the magnitude of plant inducible defences during herbivory. The outcome of these interactions likely depends on which bacterial taxa gain access to internal tissues and how MAMPs are perceived once in the apoplast, where pattern-recognition receptor–mediated immune activation is strongest [79, 80]. While our results revealed a JA-dominated response, studies in other systems show that bacterial activation of plant defences can involve JA- or SA-associated pathways depending on the identity of the interacting microbe [11, 78, 81]. Beyond their effects on plant defence signalling, interactions between phyllosphere bacteria and herbivory influence microbial community assembly across the leaf and insect gut. The feeding process that activates defensive responses simultaneously reshapes the leaf microbiota, altering bacterial composition across the plant–insect gut continuum. Following SynCom20 inoculation and a subsequent feeding period, P. brassicae herbivory led to compositional shifts distinct from those observed in non-fed plants. This shift was characterised by reduced richness and evenness, suggesting that herbivory favoured a subset of strains better suited to the altered phyllosphere environment. Wounding during feeding likely modifies the physicochemical microenvironment of the leaf [17], potentially including the release of intracellular metabolites such as soluble sugars from damaged cells, while insect frass—known to contain labile carbon and nitrogen [82]—may further contribute nutrient-rich material to plant surfaces, together creating conditions that favour the proliferation of metabolically versatile taxa. In addition to these direct effects of wounding and nutrient availability, herbivory-induced JA induction itself can drive community restructuring [17]. Communities became phylogenetically more clustered compared to the inoculum, with most abundant taxa belonging to the phylum Pseudomonadota . Some members of this lineage—including the phyllosphere epiphyte P. eucalypti 299R—exhibit metabolic versatility and strong competitive ability in dynamic or resource-limited leaf environments [68]. However, whether such traits specifically explain the enrichment of Pseudomonadota under herbivory remains unclear. More broadly, these community-level shifts show that herbivory alters which bacterial taxa persist and proliferate on leaves, thereby influencing their relative abundance when ingested and entering the larval gut. Comparison of bacterial assemblages among the SynCom20 inoculum, leaf surfaces, and larval guts further revealed strong compositional shifts as bacteria transitioned from the inoculum to the leaf surface and subsequently to the larval gut. Strain abundances in the inoculum and in the phyllosphere or larval gut were only loosely correlated, and only a subset of strains persisted on leaves or in larvae, consistent with strong host- and environment-mediated filtering [23, 83]. Larval communities resembled those on both non-fed and fed leaves, but remained distinct from either, indicating that ingestion and passage through the gut further reshaped community composition. Distinct subsets of taxa with characteristic abundance patterns were consistently enriched in each environment, likely reflecting contrasting nutrient conditions, pH, and physiological environments of the leaf surface and the larval gut. The Lepidopteran larval gut is highly alkaline, has rapid food transit and a simple structure—conditions generally unfavourable for stable bacterial colonisation and explaining why persistent, species-specific gut communities are rarely established [23, 83, 84]. Nevertheless, the increased abundance of some bacterial species that were not promoted on leaves suggests that these taxa were able to proliferate within the gut, implying colonisation rather than passive passage. Although the consequences of such colonisation remain unclear, bacteria may interact with larval physiology or metabolic processes indirectly—for instance through competition, metabolite production, or modification of ingested plant material—while their survival and activity are also shaped by the plant’s defensive chemistry [84]. Previous studies have shown that Lepidopteran larvae can harbour transient, environmentally acquired bacteria without clear fitness benefits [23], although such transient gut associates may still suppress harmful taxa, including entomopathogens, through resource competition or antagonistic interactions [84]. Reintroduction of gut-associated bacteria onto the leaf surface via frass deposition or regurgitation, had only a minor effect on phyllosphere composition compared with the direct effects of herbivory. Bacterial communities on non-fed leaves were more similar to those found in larvae than to those on fed leaves, indicating that these gut-derived bacteria did not substantially alter leaf community structure. By contrast, herbivory itself caused major shifts, likely driven by tissue damage, nutrient leakage, and changes in leaf physiology and defence status [17]. These results suggest that herbivory reshapes the phyllosphere primarily by modifying habitat conditions and resource availability rather than by frass-associated gut bacteria. Overall, the feedback between herbivory and the phyllosphere microbiota appears asymmetric: while leaf-associated bacteria can influence plant defences and insect performance, herbivory exerts a stronger and more direct impact on microbial community structure. CONCLUSION This study reveals that nonpathogenic phyllosphere bacteria can modulate plant defences and herbivore performance, demonstrating an overlooked ecological role of resident leaf microbiota in plant–insect interactions. Increasing bacterial richness strengthened jasmonate-associated responses and reduced larval growth, suggesting that microbial diversity and composition influence the magnitude of inducible plant defences. Herbivory, in turn, reshaped the phyllosphere community and increased bacterial abundance, potentially creating a feedback loop in which microbial activity and feeding jointly shape plant physiology and microbiome structure. Unravelling the molecular and ecological mechanisms underlying these feedbacks, particularly how bacterial localisation, density, and signalling contribute to defence activation, will be key to understanding how microbiomes can enhance plant resilience to insect attacks. Finally, the gnotobiotic insect–plant system developed here provides a reproducible and scalable framework for disentangling the mechanisms linking microbial community structure, plant defences, and herbivore performance. Its controlled design enables systematic exploration of multitrophic interactions in the phyllosphere, paving the way for mechanistic and translational studies on the ecological functions of aboveground microbiomes. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Availability of data and material The datasets and materials supporting the conclusions of this article are available in the following repositories: The 16S amplicon sequencing data are in the European Nucleotide Archive (ENA) repository, accession number PRJEB104625 (https://www.ebi.ac.uk/ena/browser/view/PRJEB104625). The additional datasets are available in Zenodo (https://doi.org/10.5281/zenodo.17822324) [85]. The R scripts used for data analysis are accessible in the GitHub repository https://github.com/relab-fuberlin/mueller_arabidopsis_pieris_syncom. Competing interests The authors declare that they have no competing interests. Funding This work was financially supported by the Einstein Foundation Berlin (Einstein Independent Researcher Grant ESR-2023-784 to LPV) and the Investitionsbank Berlin (ProValid Grant VAL149/2023 to LPV). Authors' contributions Conceptualisation LPV, RS; Investigation: MM, MH, MK; Formal analysis: MM, LPV, RS, Funding acquisition LPV; Resources: MH, MRE; Writing – Original Draft: MM, MRE, RS, LPV; Writing – Review & Editing: MH, MK; Visualisation: MM, RS, LPV; Supervision: RS, LPV Acknowledgements We thank the Einstein Foundation Berlin (Einstein Independent Researcher Grant ratioESR-2023-784 to LPV) and the Investitionsbank Berlin (ProValid Grant VAL149/2023 to LPV) for financial support. We thank the HPC Service of FUB-IT, Freie Universität Berlin, for computing time. We also thank Sandra Hirsch and Mila Oeltjen for their valuable technical assistance and dedication to maintaining the day-to-day laboratory operations, Jasmin Jende, Yara Sophie Dietsch, Semira Romero Reyna, Franziska Schulze, Roxanne Elisabeth Esta Kahn-Cleland, Tia Hannah Teschner, Anastasia Yusupova, Johannes Lübke, and Nevin Zeyrek for their support in insect rearing, and Dr. Andreas Springer and Dipl.-Ing. Fabian Klautzsch (Schalley Group, Mass Spectrometry Core Facility) for phytohormone quantification. References Remus-Emsermann MNP, Schlechter RO. Phyllosphere microbiology: at the interface between microbial individuals and the plant host. New Phytol. 2018;218:1327–33. Angon PB, Mondal S, Jahan I, Datto M, Antu UB, Ayshi FJ, et al. Integrated pest management (IPM) in agriculture and its role in maintaining ecological balance and biodiversity. Adv Agric. 2023;2023:1–19. Agrawal AA, Maron JL. Long‐term impacts of insect herbivores on plant populations and communities. J Ecol. 2022;110:2800–11. Erb M, Meldau S, Howe GA. Role of phytohormones in insect-specific plant reactions. Trends Plant Sci. 2012;17:250–9. Vasantha-Srinivasan P, Noh MY, Park KB, Kim TY, Jung W-J, Senthil-Nathan S, et al. Plant immunity to insect herbivores: mechanisms, interactions, and innovations for sustainable pest management. Front Plant Sci. 2025;16:1599450. Li N, Han X, Feng D, Yuan D, Huang L-J. Signaling Crosstalk between Salicylic Acid and Ethylene/Jasmonate in Plant Defense: Do We Understand What They Are Whispering? Int J Mol Sci. 2019;20. Fang X, Xie Y, Yuan Y, Long Q, Zhang L, Abid G, et al. The role of salicylic acid in plant defense responses against biotic stresses. Plant Hormones. 2025;1:0–0. Thaler JS, Humphrey PT, Whiteman NK. Evolution of jasmonate and salicylate signal crosstalk. Trends Plant Sci. 2012;17:260–70. Lazebnik J, Frago E, Dicke M, van Loon JJA. Phytohormone mediation of interactions between herbivores and plant pathogens. J Chem Ecol. 2014;40:730–41. Dodueva IE, Lebedeva MA, Lutova LA. Phytopathogens and molecular mimicry. Russ J Genet. 2022;58:638–54. Chung SH, Rosa C, Scully ED, Peiffer M, Tooker JF, Hoover K, et al. Herbivore exploits orally secreted bacteria to suppress plant defenses. Proc Natl Acad Sci U S A. 2013;110:15728–33. Coolen S, Magda R der-Molen, Welte CU. The secret life of insect-associated microbes and how they shape insect-plant interactions. FEMS Microbiol Ecol. 2022;98. Frago E, Dicke M, Godfray HCJ. Insect symbionts as hidden players in insect-plant interactions. Trends Ecol Evol. 2012;27:705–11. Keppler A, Roulier M, Pfeilmeier S, Petti GC, Sintsova A, Maier BA, et al. Plant microbiota feedbacks through dose-responsive expression of general non-self response genes. Nat Plants. 2025;11:74–89. Maier BA, Kiefer P, Field CM, Hemmerle L, Bortfeld-Miller M, Emmenegger B, et al. A general non-self response as part of plant immunity. Nat Plants. 2021;7:696–705. Miebach M, Schlechter RO, Clemens J, Jameson PE, Remus-Emsermann MNP. Litterbox-A gnotobiotic Zeolite-Clay System to Investigate Arabidopsis -Microbe Interactions. Microorganisms. 2020;8. Humphrey PT, Whiteman NK. Insect herbivory reshapes a native leaf microbiome. Nat Ecol Evol. 2020;4:221–9. Humphrey PT, Nguyen TT, Villalobos MM, Whiteman NK. Diversity and abundance of phyllosphere bacteria are linked to insect herbivory. Mol Ecol. 2014;23:1497–515. Bodenhausen N, Bortfeld-Miller M, Ackermann M, Vorholt JA. A synthetic community approach reveals plant genotypes affecting the phyllosphere microbiota. PLoS Genet. 2014;10:e1004283. Brock AK, Berger B, Schreiner M, Ruppel S, Mewis I. Plant growth-promoting bacteria Kosakonia radicincitans mediate anti-herbivore defense in Arabidopsis thaliana . Planta. 2018;248:1383–92. Vacheron J, Péchy-Tarr M, Brochet S, Heiman CM, Stojiljkovic M, Maurhofer M, et al. T6SS contributes to gut microbiome invasion and killing of an herbivorous pest insect by plant-beneficial Pseudomonas protegens . ISME J. 2019;13:1318–29. Meenakshi, Thakur S, Choudhary K, Kumar R, Kumar S, Sharma D. Ecofriendly Management of Pieris brassicae in Brassica oleracea : A Review. Biol Bull Rev. 2023;13:691–702. Hammer TJ, Janzen DH, Hallwachs W, Jaffe SP, Fierer N. Caterpillars lack a resident gut microbiome. Proc Natl Acad Sci U S A. 2017;114:9641–6. Paniagua Voirol LR, Weinhold A, Johnston PR, Fatouros NE, Hilker M. Legacy of a butterfly’s parental microbiome in offspring performance. Appl Environ Microbiol. 2020;86. Bai Y, Müller DB, Srinivas G, Garrido-Oter R, Potthoff E, Rott M, et al. Functional overlap of the Arabidopsis leaf and root microbiota. Nature. 2015;528:364–9. Zhang J, Kobert K, Flouri T, Stamatakis A. PEAR: a fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics. 2014;30:614–20. Bennett L, Melchers B, Proppe B. Curta: A General-purpose High-Performance Computer at ZEDAT, Freie Universität Berlin. 2020. Posit team. RStudio: Integrated Development Environment for R. 2023. Glöckner FO, Yilmaz P, Quast C, Gerken J, Beccati A, Ciuprina A, et al. 25 years of serving the community with ribosomal RNA gene reference databases and tools. J Biotechnol. 2017;261:169–76. Huber W, Carey VJ, Gentleman R, Anders S, Carlson M, Carvalho BS, et al. Orchestrating high-throughput genomic analysis with Bioconductor. Nat Methods. 2015;12:115–21. McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One. 2013;8:e61217. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41 Database issue:D590–6. Wright E. Using DECIPHER v2.0 to analyze big biological sequence data in R. R J. 2016;8:352. Yilmaz P, Parfrey LW, Yarza P, Gerken J, Pruesse E, Quast C, et al. The SILVA and “All-species Living Tree Project (LTP)” taxonomic frameworks. Nucleic Acids Res. 2014;42 Database issue:D643–8. Pagès H, Aboyoun P, Gentleman R, DebRoy S. Biostrings: Efficient manipulation of biological strings. 2024. Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 2010;26:2460–1. Emms DM, Kelly S. OrthoFinder: phylogenetic orthology inference for comparative genomics. Genome Biol. 2019;20:238. Emms DM, Kelly S. OrthoFinder: solving fundamental biases in whole genome comparisons dramatically improves orthogroup inference accuracy. Genome Biol. 2015;16:157. Yu G, Smith DK, Zhu H, Guan Y, Lam TT-Y. Ggtree: An r package for visualization and annotation of phylogenetic trees with their covariates and other associated data. Methods Ecol Evol. 2017;8:28–36. Yu G, Lam TT-Y, Zhu H, Guan Y. Two Methods for Mapping and Visualizing Associated Data on Phylogeny Using Ggtree. Mol Biol Evol. 2018;35:3041–3. Yu G. Using ggtree to Visualize Data on Tree-Like Structures. Curr Protoc Bioinformatics. 2020;69:e96. Xu S, Li L, Luo X, Chen M, Tang W, Zhan L, et al. Ggtree: A serialized data object for visualization of a phylogenetic tree and annotation data. Imeta. 2022;1:e56. Yu G. Data integration, manipulation and visualization of phylogenetic trees. Boca Raton: Chapman and Hall/CRC; 2022. Kembel SW, Cowan PD, Helmus MR, Cornwell WK, Morlon H, Ackerly DD, et al. Picante: R tools for integrating phylogenies and ecology. Bioinformatics. 2010;26:1463–4. Paradis E, Schliep K. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics. 2019;35:526–8. Henry L, Wickham H, Chang W. ggstance: Horizontal “ggplot2” Components. 2024. Oksanen J, Simpson GL, Blanchet FG, Kindt R, Legendre P, Minchin PR, et al. Vegan: Community ecology package. CRAN: Contributed Packages. 2001. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. bioRxiv. 2014. Wang L, Halitschke R, Kang J-H, Berg A, Harnisch F, Baldwin IT. Independently silencing two JAR family members impairs levels of trypsin proteinase inhibitors but not nicotine. Planta. 2007;226:159–67. Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer Science & Business Media; 2009. Ahlmann-Eltze C, Patil I. ggsignif: R Package for Displaying Significance Brackets for “ggplot2.” PsyArXiv. 2021. Min SH, Zhou J. smplot: An R Package for Easy and Elegant Data Visualization. Front Genet. 2021;12:802894. Pedersen TL. patchwork: The Composer of Plots. CRAN: Contributed Packages. 2019. Wickham H, Averick M, Bryan J, Chang W, McGowan L, François R, et al. Welcome to the tidyverse. J Open Source Softw. 2019;4:1686. Pinheiro J, Bates DM. Mixed-effects models in S and S-PLUS. 1st edition. New York, NY: Springer; 2000. Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team. nlme: Linear and Nonlinear Mixed Effects Models. 2021. McGillycuddy M, Warton DI, Popovic G, Bolker BM. Parsimoniously fitting large multivariate random effects in glmmTMB. J Stat Softw. 2025;112. Lenth RV, Piaskowski J. emmeans: Estimated Marginal Means, aka Least-Squares Means. 2025. Kassambara A. rstatix: Pipe-Friendly Framework for Basic Statistical Tests. 2023. Oksanen J, Simpson GL, Blanchet FG, Kindt R, Legendre P, Minchin PR, et al. vegan: Community Ecology Package. 2023. Friman J, Pineda A, van Loon JJA, Dicke M. Bidirectional plant‐mediated interactions between rhizobacteria and shoot‐feeding herbivorous insects: a community ecology perspective. Ecol Entomol. 2021;46:1–10. Pirttilä AM, Brusila V, Koskimäki JJ, Wäli PR, Ruotsalainen AL, Mutanen M, et al. Exchange of microbiomes in plant-insect herbivore interactions. MBio. 2023;14:e0321022. Dubey A, Malla MA, Vimal SR, Kumar A, Prasad SM, Khan ML. Plant-microbiome engineering: synergistic microbial partners for crop health and sustainability. Plant Growth Regul. 2025. https://doi.org/10.1007/s10725-025-01385-5. Lei J, Feng J, Ding J, Yang Y. Revisiting the classical biodiversity-ecosystem functioning and stability relationships in microbial microcosms. PNAS Nexus. 2025;4:pgaf114. Pineda A, Kaplan I, Hannula SE, Ghanem W, Bezemer TM. Conditioning the soil microbiome through plant-soil feedbacks suppresses an aboveground insect pest. New Phytol. 2020;226:595–608. Pineda A, Kaplan I, Bezemer TM. Steering soil microbiomes to suppress aboveground insect pests. Trends Plant Sci. 2017;22:770–8. Rashid MH-O, Chung YR. Induction of systemic resistance against insect herbivores in plants by beneficial soil microbes. Front Plant Sci. 2017;8:1816. Schlechter RO, Kear EJ, Bernach M, Remus DM, Remus-Emsermann MNP. Metabolic resource overlap impacts competition among phyllosphere bacteria. ISME J. 2023;17:1445–54. Mauch-Mani B, Baccelli I, Luna E, Flors V. Defense priming: An adaptive part of induced resistance. Annu Rev Plant Biol. 2017;68:485–512. Martinez-Medina A, Flors V, Heil M, Mauch-Mani B, Pieterse CMJ, Pozo MJ, et al. Recognizing Plant Defense Priming. Trends Plant Sci. 2016;21:818–22. Hilker M, Schmülling T. Stress priming, memory, and signalling in plants. Plant Cell Environ. 2019;42:753–61. Pangesti N, Reichelt M, van de Mortel JE, Kapsomenou E, Gershenzon J, van Loon JJA, et al. Jasmonic acid and ethylene signaling pathways regulate glucosinolate levels in plants during rhizobacteria-induced systemic resistance against a leaf-chewing herbivore. J Chem Ecol. 2016;42:1212–25. Zebelo S, Song Y, Kloepper JW, Fadamiro H. Rhizobacteria activates (+)-δ-cadinene synthase genes and induces systemic resistance in cotton against beet armyworm ( Spodoptera exigua ). Plant Cell Environ. 2016;39:935–43. Aziz M, Nadipalli RK, Xie X, Sun Y, Surowiec K, Zhang J-L, et al. Augmenting sulfur metabolism and herbivore defense in Arabidopsis by bacterial volatile signaling. Front Plant Sci. 2016;7:458. Vogel C, Bodenhausen N, Gruissem W, Vorholt JA. The Arabidopsis leaf transcriptome reveals distinct but also overlapping responses to colonization by phyllosphere commensals and pathogen infection with impact on plant health. New Phytol. 2016;212:192–207. Wang Y, Mostafa S, Zeng W, Jin B. Function and mechanism of jasmonic acid in plant responses to abiotic and biotic stresses. Int J Mol Sci. 2021;22:8568. Miebach M, Faivre L, Schubert D, Jameson P, Remus-Emsermann M. Nonpathogenic leaf-colonizing bacteria elicit pathogen-like responses in a colonization density-dependent manner. Plant-Environment Interactions. 2024;5:e10137. Río-Álvarez I, Muñoz-Gómez C, Navas-Vásquez M, Martínez-García PM, Antúnez-Lamas M, Rodríguez-Palenzuela P, et al. Role of Dickeya dadantii 3937 chemoreceptors in the entry to Arabidopsis leaves through wounds. Mol Plant Pathol. 2015;16:685–98. Dora S, Terrett OM, Sánchez-Rodríguez C. Plant-microbe interactions in the apoplast: Communication at the plant cell wall. Plant Cell. 2022;34:1532–50. Farvardin A, González-Hernández AI, Llorens E, García-Agustín P, Scalschi L, Vicedo B. The Apoplast: A Key Player in Plant Survival. Antioxidants (Basel). 2020;9. Sugio A, Dubreuil G, Giron D, Simon J-C. Plant-insect interactions under bacterial influence: ecological implications and underlying mechanisms. J Exp Bot. 2015;66:467–78. Kagata H, Ohgushi T. Positive and negative impacts of insect frass quality on soil nitrogen availability and plant growth. Popul Ecol. 2012;54:75–82. Shao Y, Mason CJ, Felton GW. Toward an Integrated Understanding of the Lepidoptera Microbiome. Annu Rev Entomol. 2024;69:117–37. Paniagua Voirol LR, Frago E, Kaltenpoth M, Hilker M, Fatouros NE. Bacterial Symbionts in Lepidoptera: Their Diversity, Transmission, and Impact on the Host. Front Microbiol. 2018;9:556. Müller M, Huve MAP, Kunzler M, Remus-Emsermann M, Schlechter R, Paniagua Voirol LR. Gnotobiotic Arabidopsis– Pieris brassicae –Microbiota System Dataset. 2025. Remus-Emsermann MNP, Kim EB, Marco ML, Tecon R, Leveau JHJ. Draft Genome Sequence of the Phyllosphere Model Bacterium Pantoea agglomerans 299R. Genome Announc. 2013;1. Green PN, Bousfield IJ. Emendation of Methylobacterium Patt, Cole, and Hanson 1976; Methylobacterium rhodinum (Heumann 1962) comb. nov. corrig.; Methylobacterium radiotolerans (Ito and Iizuka 1971) comb. nov. corrig.; and Methylobacterium mesophilicum (Austin and Goodfellow 1979) comb. nov. Int J Syst Bacteriol. 1983;33:875–7. Loper JE, Lindow SE. Lack of evidence for in situ fluorescent pigment production by Pseudomonas syringae pv. syringae on bean leaf surfaces. Phytopathology. 1987;77:1449. Innerebner G, Knief C, Vorholt JA. Protection of Arabidopsis thaliana against leaf-pathogenic Pseudomonas syringae by Sphingomonas strains in a controlled model system. Appl Environ Microbiol. 2011;77:3202–10. Additional Declarations No competing interests reported. Supplementary Files Additionalfile1Mlleretal.docx Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 22 Apr, 2026 Reviews received at journal 21 Apr, 2026 Reviews received at journal 10 Apr, 2026 Reviews received at journal 18 Mar, 2026 Reviewers agreed at journal 12 Mar, 2026 Reviewers agreed at journal 10 Mar, 2026 Reviewers agreed at journal 08 Mar, 2026 Reviews received at journal 22 Feb, 2026 Reviewers agreed at journal 25 Jan, 2026 Reviewers invited by journal 24 Jan, 2026 Editor assigned by journal 04 Jan, 2026 Submission checks completed at journal 19 Dec, 2025 First submitted to journal 11 Dec, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8337060","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":569000166,"identity":"b18d7fab-dbb9-4872-bd4c-9401f19be3a3","order_by":0,"name":"Moritz Müller","email":"","orcid":"","institution":"Freie Universität Berlin","correspondingAuthor":false,"prefix":"","firstName":"Moritz","middleName":"","lastName":"Müller","suffix":""},{"id":569000167,"identity":"3da40910-de8d-4198-b7a2-c3acc0fa3f14","order_by":1,"name":"Maryse A. P. Huve","email":"","orcid":"","institution":"Freie Universität Berlin","correspondingAuthor":false,"prefix":"","firstName":"Maryse","middleName":"A. P.","lastName":"Huve","suffix":""},{"id":569000170,"identity":"91216e72-1890-443f-af9c-362a6192d37d","order_by":2,"name":"Michael Kunzler","email":"","orcid":"","institution":"Freie Universität Berlin","correspondingAuthor":false,"prefix":"","firstName":"Michael","middleName":"","lastName":"Kunzler","suffix":""},{"id":569000172,"identity":"c8905141-15e8-477b-a797-967489eb0ac1","order_by":3,"name":"Mitja N. P. Remus-Emsermann","email":"","orcid":"","institution":"Freie Universität Berlin","correspondingAuthor":false,"prefix":"","firstName":"Mitja","middleName":"N. P.","lastName":"Remus-Emsermann","suffix":""},{"id":569000173,"identity":"f9f172c8-df04-4d17-b408-4b6289cdde29","order_by":4,"name":"Rudolf O. Schlechter","email":"","orcid":"","institution":"Freie Universität Berlin","correspondingAuthor":false,"prefix":"","firstName":"Rudolf","middleName":"O.","lastName":"Schlechter","suffix":""},{"id":569000174,"identity":"41cd3f1f-bf49-48f3-97b8-bc3ede49d833","order_by":5,"name":"Luis R. Paniagua Voirol","email":"data:image/png;base64,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","orcid":"","institution":"Freie Universität Berlin","correspondingAuthor":true,"prefix":"","firstName":"Luis","middleName":"R. Paniagua","lastName":"Voirol","suffix":""}],"badges":[],"createdAt":"2025-12-11 13:08:56","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8337060/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8337060/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99850654,"identity":"86345ba8-edfa-4dbc-9b83-61cff304d0a4","added_by":"auto","created_at":"2026-01-09 03:41:43","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":271965,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePhylogenetic tree of bacterial strains used in the construction of SynComs. \u003c/strong\u003eMaximum-likelihood species tree of SynCom strains, which all belong to prevalent genera in natural \u003cem\u003eArabidopsis thaliana\u003c/em\u003e phyllosphere microbiomes, reconstructed from the concatenated alignment of single-copy orthogroups using OrthoFinder. Branch lengths represent the average number of substitutions per site across a large range of gene families.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/bc48a026a58fb66ef54e641e.png"},{"id":99850657,"identity":"e4b46070-817f-4278-8548-f2e9f9fc8809","added_by":"auto","created_at":"2026-01-09 03:41:43","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1786454,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGnotobiotic plant-insect system. \u003c/strong\u003e\u003cem\u003eArabidopsis thaliana\u003c/em\u003e plants were grown under axenic conditions on a zeolite–MS substrate for four weeks in sealed plant tissue culture boxes. Plants were inoculated with SynComs by foliar spraying. Control plants were mock-treated with PBS. In parallel, \u003cem\u003ePieris brassicae\u003c/em\u003eeggs were collected from surface-sterilised Brussels sprout plants, removed from the leaves one day before hatching, and placed in sterile Petri dishes to obtain neonates for the experiments. Eight days after plant inoculation, \u003cem\u003eP. brassicae\u003c/em\u003e neonates were introduced into the system. After the feeding period (one or seven days depending on the experiment), leaves and larvae were sampled for bacterial quantification, community profiling, and phytohormone quantification. One box containing four plants and three larvae constituted one biological replicate.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/2c86c35f5bf8da38fa51b8ab.png"},{"id":99850652,"identity":"90a06b19-fc7b-47c0-a821-1fdeec28eebb","added_by":"auto","created_at":"2026-01-09 03:41:42","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":217666,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEffect of SynComs on bacterial colonisation and larval performance.\u003c/strong\u003e Plants were inoculated with SynComs of increasing richness and incubated for eight days before neonates were placed to feed. After seven days of feeding, bacterial abundance (CFU/g) and larval weight were measured. (A) Survival of larvae on untreated control plants or plants inoculated with SynCom5, SynCom10, or SynCom20. Mean and 95% CI from non-parametric bootstrap resampling are shown in red. (B) Bacterial abundance on leaves and\u003cstrong\u003e \u003c/strong\u003e(C) in \u003cem\u003ePieris brassicae\u003c/em\u003e larvae fed on untreated control plants (n = 28) or plants inoculated with SynCom5 (n = 28), SynCom10 (n = 23), or SynCom20 (n = 28). (D) Normalised larval weight relative to the mean weight of the control samples within each experiment. Different letters indicate significant group differences among treatments (\u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e \u0026lt; 0.05) based on linear mixed-effects models followed by Benjamini–Hochberg-adjusted pairwise comparison of estimated marginal means.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/85934cfa52de11d6db1a1b9b.png"},{"id":100357350,"identity":"a8d298d5-29f5-427a-89ef-da215cd2a1b5","added_by":"auto","created_at":"2026-01-16 07:19:44","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":163528,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePhytohormone level changes in \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eA. thaliana\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e leaves after 24 h of \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eP. brassicae\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e larval feeding.\u003c/strong\u003e Plants were either mock-treated (control) or inoculated with SynCom20, and subsequently either exposed to \u003cem\u003ePieris brassicae\u003c/em\u003e larval feeding or not. Phytohormones jasmonic acid (JA), jasmonoyl-isoleucine (JA-Ile), abscisic acid (ABA) and salicylic acid (SA) were quantified at T0 and after 24 h of feeding (T1). Relative concentration changes after 24 hours of feeding are shown. Treatments included untreated control + no feeding (n = 18), untreated control + feeding (n = 20), SynCom20 + no feeding (n = 20), and SynCom20 + feeding (n = 20). Different letters indicate significant differences among treatments (\u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj \u003c/sub\u003e\u0026lt; 0.05) based on generalised linear mixed-effects models (GLMMs) followed by Benjamini–Hochberg–adjusted pairwise comparisons of estimated marginal means.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/685b5d848c5cc37063011b89.png"},{"id":100357375,"identity":"ab92b1f3-e4b9-419e-b7f4-5ee1778ac109","added_by":"auto","created_at":"2026-01-16 07:19:47","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":69718,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEffect of SynCom20 and herbivory in larval performance and bacterial colonisation.\u003c/strong\u003e (A) Larval weight of \u003cem\u003ePieris brassicae \u003c/em\u003eafter feeding for seven days on untreated control (n = 8) or on SynCom20-inoculated \u003cem\u003eArabidopsis thaliana\u003c/em\u003e plants (n = 8). (B) Total bacterial abundance on leaves (CFU/g) of SynCom20-inoculated plants exposed or not exposed to \u003cem\u003eP. brassicae\u003c/em\u003e larvae (n = 8, per treatment). Significant differences between groups are indicated by stars (*, 𝛼 = 0.05; Student’s \u003cem\u003et\u003c/em\u003e test).\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/8431817878082be6a8e84a28.png"},{"id":100357343,"identity":"30659fd0-5e3c-49fe-9e69-85ea5be893f7","added_by":"auto","created_at":"2026-01-16 07:19:42","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":355822,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRelative abundance and community composition of SynCom20 under herbivory.\u003c/strong\u003e \u003cem\u003eArabidopsis thaliana\u003c/em\u003e plants were inoculated with SynCom20, and after eight days \u003cem\u003ePieris brassicae\u003c/em\u003e neonates were introduced to feed for seven days before sampling. (A) Relative abundance of bacterial strains across treatments: Inoculum (SynCom20); inoculated non-fed leaves; inoculated fed leaves; larvae after feeding on inoculated plants. (B) NMDS plot on Bray-Curtis distances across treatments. Arrows indicate individual strains driving compositional differences (envfit, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05), with arrow direction positively correlated with abundance. (C) Differential abundance analysis of bacterial strains between treatments, shown as log\u003csub\u003e2\u003c/sub\u003e fold change (log\u003csub\u003e2\u003c/sub\u003eFC). Left: Herbivory effect on the phyllosphere (non-fed leaves vs. fed leaves); centre: Herbivory-altered phyllosphere vs. gut (fed leaves vs. larva); right: Overall gut filtering (non-fed leaves vs. larvae).\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/e80e43be3fb531e479e2381d.png"},{"id":100357120,"identity":"e541f4c6-a0eb-4a51-8793-432b22a4b06d","added_by":"auto","created_at":"2026-01-16 07:18:52","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":172961,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCommunity diversity metrics of SynCom20 in response to herbivory.\u003c/strong\u003e \u003cem\u003eArabidopsis thaliana\u003c/em\u003e plants were inoculated with SynCom20 and exposed or not exposed to \u003cem\u003ePieris brassicae\u003c/em\u003e feeding. Alpha diversity metrics—richness, Shannon’s evenness, Shannon’s diversity, and mean pairwise phylogenetic distance (MPD)—were compared across treatments: Inoculum (SynCom20); inoculated non-fed leaves; inoculated fed leaves; larvae after feeding on inoculated plants. MPD values were normalised to that of an even SynCom20 community where all strains had equal abundance. Different letters indicate significant differences among treatments based on ANOVA followed by Tukey’s \u003cem\u003epost hoc\u003c/em\u003e test for multiple comparisons.\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/bbdc4dc8f880b58f7b17b0f0.png"},{"id":100379926,"identity":"d48563e9-6d07-45c2-b395-4ac49e83f57b","added_by":"auto","created_at":"2026-01-16 09:54:58","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3669287,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/b569cf4f-b834-4743-adba-e5fe0d8443aa.pdf"},{"id":99850659,"identity":"69b0ebfc-c531-45fc-9156-04bc5b508da2","added_by":"auto","created_at":"2026-01-09 03:41:43","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":378713,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalfile1Mlleretal.docx","url":"https://assets-eu.researchsquare.com/files/rs-8337060/v1/36ed92e1ad98a25dd34e42a9.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"From the leaf to the gut and back again: the fate and influence of phyllosphere bacteria in a gnotobiotic Arabidopsis – Pieris brassicae system","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eThe phyllosphere\u0026mdash;the aerial surface of plants\u0026mdash;represents one of the largest terrestrial habitats for microbial life on Earth, with leaves alone covering an estimated 6.4 \u0026times; 10⁸ to 1 \u0026times; 10⁹ km\u0026sup2; globally and hosting up to 10\u003csup\u003e7\u003c/sup\u003e bacterial cells per gram of leaf tissue [1]. Leaves are consumed by a wide range of herbivorous insects, many of which are major pests in agricultural and natural ecosystems [2, 3]. The phyllosphere therefore represents an ecologically important interface where microbes and insects interact, with direct consequences for plant health and productivity.\u003c/p\u003e\n\u003cp\u003eMicrobes and herbivorous insects can influence each other indirectly through the plant\u0026rsquo;s immune system, which integrates multiple inducible defence pathways that are largely regulated by phytohormones [4, 5]. Jasmonic acid (JA) typically mediates responses to chewing herbivores and necrotrophic pathogens, while salicylic acid (SA) mediates responses to biotrophic pathogens and piercing\u0026ndash;sucking insects [4, 6, 7]. Because these pathways often act antagonistically, activation of one often suppresses the other, creating scope for plant-mediated microbe\u0026ndash;insect interactions. Accordingly, microbes and herbivores influence each other by modulating JA-SA cross-talk [8]. For instance, when leaves are damaged by chewing herbivores, JA-mediated signalling suppresses SA-dependent defences, allowing biotrophic pathogens to proliferate [7, 9]. Some pathogens enhance herbivore performance by activating SA-mediated defences or, conversely, suppress feeding by producing JA mimics [9\u0026ndash;11]. Insect mutualists, which can be either pathogenic or nonpathogenic to plants, may suppress plant anti-herbivore defences by eliciting SA responses or by preventing recognition of the insect, thereby reshaping insect-plant interactions [12, 13]. Beyond pathogens and mutualists, nonpathogenic phyllosphere bacteria also elicit plant immune responses when they reach high densities [14\u0026ndash;16]. Despite these insights, the broader ecological role of nonpathogenic phyllosphere bacteria in shaping plant\u0026ndash;insect interactions is still poorly understood.\u003c/p\u003e\n\u003cp\u003eBesides hormonal cross-talk, herbivorous insects and phyllosphere bacteria also interact through a range of direct physical and ecological processes. Herbivory damage releases nutrients into the otherwise nutrient-poor phyllosphere, creating transient hotspots that promote bacterial growth and entry into the apoplast [17, 18]. Insects further impact microbial dispersal by vectoring bacteria onto leaves via frass, saliva, or body surfaces [12]. These effects operate alongside other ecological factors shaping the leaf microbiota, including host identity, plant immune activity, cuticle composition, and plant metabolism [19]. Together, these processes shape the composition and dynamics of phyllosphere bacterial communities and may influence how microbes and herbivores affect each other through the plant.\u003c/p\u003e\n\u003cp\u003eMost studies on plant\u0026ndash;microbe\u0026ndash;insect interactions in the phyllosphere have focused on individual mutualistic or pathogenic species [12, 20, 21]. By contrast, the ecological roles of resident nonpathogenic microbial communities and their effect on plants and herbivorous insects remain poorly understood. This raises key questions: Does the composition and diversity of microbial communities influence herbivory, and how does herbivore feeding, in turn, restructure these communities? Do phyllosphere microbes affect insect performance indirectly by modulating host defences, or directly through ingestion and gut colonisation?\u003c/p\u003e\n\u003cp\u003eTo address these questions, we established a gnotobiotic insect-plant system combining \u003cem\u003eArabidopsis thaliana\u003c/em\u003e (thale cress), \u003cem\u003ePieris brassicae\u003c/em\u003e (Large White; Lepidoptera) larvae, and synthetic phyllosphere bacterial communities (SynComs) of different richness. \u003cem\u003ePieris brassicae\u003c/em\u003e is a specialist herbivore of Brassicaceae, widely distributed in Europe, Asia, and North Africa, and a major pest of crops such as cabbage, broccoli, and mustard [22]. Its larvae are foliar feeders that cause extensive tissue damage. Like most studied Lepidoptera, the larval gut of \u003cem\u003eP. brassicae\u003c/em\u003e lacks a stable resident microbiota, with bacteria acquired transiently mainly through feeding [23, 24]. Under laboratory rearing conditions, the gut of \u003cem\u003eP. brassicae\u003c/em\u003e is virtually sterile, with only minimal bacterial presence [24], making this species an ideal system for manipulating bacterial exposure. This gnotobiotic insect-plant system enables investigation of how phyllosphere bacterial communities influence caterpillar performance, plant responses, and microbial colonisation patterns within a defined tripartite plant\u0026ndash;microbe\u0026ndash;insect interaction. \u003c/p\u003e\n\u003cp\u003eTo explore how resident phyllosphere bacteria shape the interplay between plants and herbivores, we combined insect performance assays, phytohormone analysis, and 16S rRNA gene amplicon sequencing. We investigated (1) how phyllosphere bacteria affect herbivorous insect performance, (2) how phyllosphere bacteria modulate levels of anti-herbivore defence-associated phytohormones, (3) how herbivory shapes the leaf-associated bacterial community, and (4) which of these bacteria are recruited by the larval gut environment. This work uncovers emergent properties of plant\u0026ndash;microbe\u0026ndash;insect interactions in the phyllosphere and establishes a tractable gnotobiotic model for dissecting their underlying mechanisms.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cp\u003ePlant growth conditions\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eArabidopsis thaliana \u003c/em\u003eCol-0 seeds were surface-sterilised with 70% v/v ethanol for 2 min, followed by a 50% v/v household bleach solution (DanKlorix, Germany) containing 0.2% v/v Tween-20 for 7 min. Seeds were thoroughly washed with sterile water three times and stratified for two days at 4 \u0026deg;C. Single seeds were sown onto approximately 0.2 cm thick, 0.5 cm x 0.5 cm wide slices of half-strength Murashige and Skoog (\u0026frac12; MS) medium (Duchefa) with 1% w/v plant agar (Duchefa) or onto cut off pipette tips filled with the same medium [16], transferred into Petri dishes wrapped with Parafilm (Bemis Parafilm PM-996) and put into a plant growth chamber (poly klima GmbH, M5Z- TDL+) at 21 \u0026deg;C, 80% relative humidity, ~120 \u0026micro;mol m\u003csup\u003e-2\u003c/sup\u003e s\u003csup\u003e-1\u003c/sup\u003e light intensity, and a 11 h light/13 h dark photoperiod. Then, nine-day-old seedlings were transferred into sterile plant tissue culture boxes (size: 7 x 7 x 8 cm) (Bioworld Magenta GA-7 Plant Culture Box) filled with 90 g zeolite (0.2-0.5 mm grain ZeoLith Klinoptolith, labradorit.de) supplemented with 40 ml of sterile \u0026frac34; MS medium (Duchefa). Box lids had four holes with a diameter of roughly 0.5 cm sealed with a double layered micropore strip (3M Deutschland GmbH) to prevent contamination while allowing for air exchange. Four seedlings were transferred under aseptic conditions into each box and put back into the growth chamber. Each box represented one biological replicate.\u003c/p\u003e\n\u003cp\u003eSynthetic bacterial communities\u003c/p\u003e\n\u003cp\u003eSynCom composition was selected from a pool of 20 bacterial strains from prevalent genera of the \u003cem\u003eA. thaliana\u003c/em\u003e phyllosphere microbiota [25]. All strains used were originally isolated from the phyllosphere and are nonpathogenic to \u003cem\u003eA. thaliana \u003c/em\u003e(Figure 1 and Table S1). From the species pool, SynComs with three levels of species richness were designed, comprising 5, 10, and 20 strains (SynCom5, SynCom10, and SynCom20). For SynCom5 and SynCom10, members were randomly drawn out of the total pool and therefore unique in their composition in each of four independent experimental replicates (Figure S1).\u003c/p\u003e\n\u003cp\u003eStrains were individually grown in Reasoner\u0026apos;s 2A (R2A) medium or on R2A agar (HiMedia) for 1-4 days at 30 \u0026deg;C. Cultures were harvested by centrifugation for 7 min at 8,500 \u0026times; \u003cem\u003eg\u003c/em\u003e, resuspended in 10 mM MgCl\u003csub\u003e2\u003c/sub\u003e, diluted to a predetermined optical density at 600 nm (OD\u003csub\u003e600\u003c/sub\u003e), and assembled in equal CFU ratios (Table S1). SynComs were stored as glycerol stocks (20% v/v glycerol) at -80 \u0026deg;C in either 0.5- or 1-ml aliquots.\u003c/p\u003e\n\u003cp\u003eInoculation of \u003cem\u003eA. thaliana\u003c/em\u003e with bacterial communities\u003c/p\u003e\n\u003cp\u003eSynCom glycerol stocks were thawed and centrifuged at 8,500 \u0026times; \u003cem\u003eg\u003c/em\u003e for 7 min. The supernatant was discarded and the pellet was resuspended in 1 ml phosphate-buffered saline (PBS; 8 g/L NaCl, 1.44 g/L Na\u003csub\u003e2\u003c/sub\u003eHPO\u003csub\u003e4\u003c/sub\u003e, 0.24 g/L KH\u003csub\u003e2\u003c/sub\u003ePO\u003csub\u003e4\u003c/sub\u003e, 0.2 g/L KCl; pH 7.4). Bacterial suspensions were then diluted to an OD\u003csub\u003e600\u003c/sub\u003e of 0.005. Plants were sprayed four weeks after sowing with an airbrush (Harder \u0026amp; Steenbeck Ultra) at 10 psi using a sterile plant culture box lid containing a single micropore-sealed hole (Figure 2). Before and between treatments, the airbrush was sterilised with 70% v/v ethanol and rinsed with sterile PBS. All boxes received 200 \u0026micro;l of spray volume; untreated control boxes were sprayed with sterile PBS instead of SynCom.\u003c/p\u003e\n\u003cp\u003eInsect rearing and neonate collection for experiments\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ePieris brassicae\u003c/em\u003e larvae were reared on greenhouse-grown Brussels sprout plants (\u003cem\u003eBrassica oleracea\u003c/em\u003e var. \u003cem\u003egemmifera\u003c/em\u003e) in a temperature-controlled room at 22 \u0026deg;C, 75% relative humidity, ~120 \u0026micro;mol m⁻\u0026sup2; s⁻\u0026sup1; light intensity and an 18 h light/6 h dark photoperiod, in 50 \u0026times; 50 \u0026times; 60 cm rearing cages containing approximately 100 individuals, until pupation and emergence. Adults were transferred to a greenhouse (20\u0026ndash;24 \u0026deg;C day / 18\u0026ndash;22 \u0026deg;C night, 75% relative humidity, ~120 \u0026micro;mol m⁻\u0026sup2; s⁻\u0026sup1; light intensity, 16 h light/8 h dark photoperiod) and fed with a 15% v/v aqueous honey solution. For egg collection, seven-week-old Brussels sprout plants were prepared by spraying the underside of the leaves with 70% v/v ethanol and allowing them to dry. One plant was then placed in a cage containing one-week-old butterflies and left overnight for oviposition. The egg-laden plant was then transferred to the larval temperature-controlled room. To prevent the hatching larvae from feeding on the egglaying-plants, eggs were gently removed from the leaves using sterilised tweezers and placed in sterile Petri dishes one day before hatching (four days after egg laying). On the day of hatching, timed to coincide with eight days after SynCom inoculation, all neonates were pooled and randomly assigned to the plant treatments. Neonates were carefully transferred using sterilised paintbrushes, with three neonates placed per plant tissue culture box (Figure 2). Each box constituted one biological replicate.\u003c/p\u003e\n\u003cp\u003eAssessment of the gnotobiotic insect-plant system\u003c/p\u003e\n\u003cp\u003eTo evaluate the suitability of the gnotobiotic insect-plant system, sterile \u003cem\u003eA. thaliana\u003c/em\u003e Col-0 seeds (prepared as described above) were sown on agar. After nine days, part of the seedlings were transplanted into pots (size: 9 x 9 x 8 cm) containing a 3:1 mixture of steamed soil (Einheitserde Typ P soil, Kausek) and vermiculite (Kausek). The remaining seedlings were transferred into sterile plant tissue culture boxes as described above. As in the plant tissue culture boxes, four plants were transferred in each pot. Plants were used 36 days after sowing and four neonate larvae were placed per pot. Larval biomass and survival were evaluated after seven days of feeding on plants grown in parallel either in pots or in tissue culture boxes. To that end, one neonate was placed on each plant within the boxes or pots, for a total of four neonates per box or pot. To prevent larvae from escaping, pots were enclosed in microperforated bags (Packseller) and sealed with tape. Plants and boxes were then placed in a plant growth chamber (poly klima GmbH, M5Z- TDL+) set to 21 \u0026deg;C, 80% relative humidity, ~120 \u0026micro;mol m\u003csup\u003e-2\u003c/sup\u003e s\u003csup\u003e-1\u003c/sup\u003e light intensity, and a 11 h light/13 h dark photoperiod. Each box or each pot constituted one biological replicate. All larvae from each replicate were weighed together using a precision balance (Precisa H_225SM-DR; resolution 0.01/0.1 mg) and divided by the number of weighed larvae. The number of larvae per replicate was also recorded.\u003c/p\u003e\n\u003cp\u003eLarval sampling\u003c/p\u003e\n\u003cp\u003eFor all SynCom experiments\u003cem\u003e, P. brassicae \u003c/em\u003elarvae were sampled one day after being placed on the plants for phytohormone measurements, or after seven days for all other experiments (Figure 2). Larvae from each box were pooled and treated as one biological replicate. The number of surviving larvae and their total weight were recorded for each box using a precision balance (Precisa H_225SM-DR; resolution 0.01/0.1 mg). Larvae were placed at \u0026ndash;20 \u0026deg;C for 2 min to induce lethargy, then homogenised in 250 \u0026micro;l PBS in 1.5 ml Eppendorf tubes using sterilised metal pestles. An aliquot of the homogenate was stored at \u0026ndash;20 \u0026deg;C for DNA extraction. From 40 \u0026micro;l of homogenate, a tenfold serial dilution was prepared and plated on R2A agar using a 96-pin replicator to quantify total CFU per larval weight.\u003c/p\u003e\n\u003cp\u003eLeaf sampling\u003c/p\u003e\n\u003cp\u003eLeaves were collected seven days after larval placement, in parallel with larval sampling. Two larvae-fed leaves per plant were taken from each box, excised at the petiole, and transferred into sterile 1.5 ml Eppendorf tubes. Fresh weight was recorded using a precision balance (Precisa H_225SM-DR; resolution 0.01/0.1 mg). To detach phyllosphere bacteria, each tube was filled with 1 ml PBS, sonicated for 5 min (75% ultrasound efficiency; EMAG Emmi-12HC), and vortexed for 10 s. A tenfold serial dilution of each leaf wash was then prepared and inoculated onto R2A agar using a 96-pin replicator. Each sample was plated in duplicate and incubated at 30 \u0026deg;C. Total CFU counts were normalised to leaf fresh weight, and the remaining leaf wash was stored at \u0026ndash;80 \u0026deg;C for DNA extraction.\u003c/p\u003e\n\u003cp\u003eDNA extraction and 16S rRNA gene amplicon sequencing\u003c/p\u003e\n\u003cp\u003eDNA was extracted from larval, leaf, and SynCom20 inoculum samples using the MasterPure\u0026trade; Complete DNA \u0026amp; RNA Purification Kit, with adjusted volumes of Tissue and Cell Lysis Solution. For larvae, 300 \u0026micro;l lysis solution was added to 150 \u0026micro;l homogenate and transferred to a 2-ml tube containing seven sterile zirconia beads. For leaf samples, 450 \u0026micro;l lysis solution was added per sample. For the SynCom20 inoculum, glycerol stocks were centrifuged at 8,500 \u0026times; \u003cem\u003eg\u003c/em\u003e for 7 min in a 2-ml tube; the supernatant was discarded, and seven zirconia beads plus 450 \u0026micro;l lysis solution were added. Larval and inoculum samples were homogenised in a Retsch MM400 bead mill for four cycles of 2 min at 30 Hz. Extraction negative controls consisting of reagent-only tubes were processed in parallel. Subsequent steps followed the kit protocol. DNA quality and quantity were measured using a DeNovix DS-11 FX+ spectrophotometer/fluorometer with the dsDNA Broad Range kit. Samples were stored at \u0026ndash;20 \u0026deg;C until processing.\u003c/p\u003e\n\u003cp\u003eAmplicon sequencing of the V3-V4 region of the 16S rRNA gene was done using the primers 341F (5\u0026rsquo;-CCTACGGGNGGCWGCAG-3\u0026rsquo;) and 806R (5\u0026rsquo;-GGACTACNVGGGTWTCTAAT-3\u0026rsquo;) with barcodes for pooling. PCR reaction mix was prepared per sample with 10 \u0026micro;l 5x Phusion GC Buffer, 1 \u0026micro;l 10 mM dNTPs, 2.5 \u0026micro;l forward and reverse primers at 10 \u0026micro;M, 50 ng template DNA, 1.5 \u0026micro;l DMSO, 0.5 \u0026micro;l Thermo Scientific Phusion High-Fidelity DNA Polymerase and filled up to 50 \u0026micro;l with water. The PCR cycling protocol included an initial denaturation for 30 s at 98 \u0026deg;C, then 30 cycles of denaturation at 98 \u0026deg;C for 10 s, annealing at 58 \u0026deg;C for 30 s and extension at 72 \u0026deg;C for 30 s. A final extension was done at 72 \u0026deg;C for 5 min.\u003c/p\u003e\n\u003cp\u003eAmplicons were checked by agarose gel electrophoresis and purified using the Monarch PCR \u0026amp; DNA Cleanup Kit (New England Biolabs). DNA quality and concentration of purified amplicons were assessed on a DeNovix DS-11 FX+ spectrophotometer/fluorometer, and samples were pooled and concentrated to meet sequencing requirements. The pooled amplicon library was then sent to Novogene (Germany) for library preparation (PCR-free protocol) and sequencing on an Illumina NovaSeq X Plus platform (250-bp paired-end reads).\u003c/p\u003e\n\u003cp\u003eAnalysis of 16S rRNA gene amplicon data\u003c/p\u003e\n\u003cp\u003ePaired-end reads were merged using PEAR with a minimum overlap of 10 bp and a \u003cem\u003ep\u003c/em\u003e-value threshold of 0.01 [26]. Merged reads were demultiplexed using Cutadapt, and sequences for each sample were quality-filtered using USEARCH v11.0.667 (Edgar, 2010), retaining reads \u0026gt;350 bp with Phred scores \u0026gt;19. Filtered reads were dereplicated and clustered into OTUs at 97% sequence similarity. Representative sequences and OTU abundance tables were generated for taxonomic classification and diversity analysis, respectively. This analysis was performed with the High-Performance Computer \u0026ldquo;CURTA\u0026rdquo; of FUB-IT, Freie Universit\u0026auml;t Berlin [27].\u003c/p\u003e\n\u003cp\u003eFurther analyses were conducted in RStudio [28], using the DECIPHER, Phyloseq, and Biostrings packages [29\u0026ndash;35]. OTUs were classified using the SILVA rRNA database, and chloroplast, mitochondria, and unclassified OTUs were removed (Figure S2). Members of SynCom20 were identified by global alignment of OTU sequences with a custom V3-V4 reference database using USEARCH v11.0.667 [36], with at least 98% sequence identity and full-length coverage. For each query, only the best hit was reported.\u003c/p\u003e\n\u003cp\u003ePhylogenetic relationships among Syncom20 members were generated based on orthogroup inference of annotated whole-genome sequences using OrthoFinder [37, 38]. Phylogenetic structure was evaluated by calculating the Faith\u0026rsquo;s phylogenetic diversity (PD) and mean pairwise phylogenetic distances (MPD), and visualised using the packages ggtree, picante, ape and ggstance [39\u0026ndash;46].\u003c/p\u003e\n\u003cp\u003eAnalysis of the bacterial communities was done with the R package vegan [47]. To minimise sampling bias, samples with fewer than 250 reads, including the negative controls, were excluded from further analysis, resulting in \u0026ge;99.5% Good\u0026rsquo;s coverage across all retained samples. OTU tables were rarefied 100 times to 250 reads per sample, mean abundances were determined and OTUs with mean of zero were removed. Alpha diversity was estimated by richness, evenness and Shannon diversity index.\u003c/p\u003e\n\u003cp\u003eBeta diversity was estimated on Bray-Curtis distances and used to visualise differences in community composition among treatments using non-metric multidimensional scaling (NMDS). To identify taxa whose abundance was associated with a treatment, OTU vectors were fitted onto the NMDS ordination space (\u003cem\u003eenvfit\u003c/em\u003e, vegan package), retaining only significant correlations at \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05. In addition, differential abundance analysis on relative abundances of SynCom20 was performed using DESeq2 to identify taxa that are significantly enriched or depleted across treatments [48]. \u003cem\u003eP\u003c/em\u003e-values of fold changes were corrected using Benjamini-Hochberg (BH) procedure to control for false discovery rates.\u003c/p\u003e\n\u003cp\u003ePhytohormone extraction and analysis\u003c/p\u003e\n\u003cp\u003eLeaf material for phytohormone quantification was collected eight days after bacterial inoculation and before placing the larvae (T0), and 24 hours later (T1). One leaf was sampled per plant from each replicate box, prioritising larvae-fed leaves. Plant material was sampled in 2 ml tubes containing four zirconia beads, and immediately snap-frozen in liquid nitrogen before storing at -80 \u0026deg;C until processing. Phytohormones were extracted using a protocol outlined by Wang \u003cem\u003eet al.\u003c/em\u003e 2007 [49]. Briefly, frozen samples were weighed and homogenised in 1 ml of ethyl acetate and 2 \u0026micro;l of internal standards containing deuterated phytohormones (10 ng/ul D4-SA, 10 ng/ul D6-ABA [abscisic acid], 30.2 ng/ul D6-JA, 10 ng/ul D6-JA-Ile; HPC Standards GmbH). Homogenisation was done in cooled adapters on a Retsch MM400 bead mill for nine cycles of 20 s each at 30 Hz, followed by centrifugation at 13,000 \u0026times; \u003cem\u003eg\u003c/em\u003e for 10 min at 4 \u0026deg;C. Supernatants were transferred to new tubes, while 1 ml ethyl acetate was added to the pellets and the extraction step was performed a second time without internal standard. The supernatants from the second extraction round were pooled with the first ones and concentrated under vacuum (Savant DNA Speed Vac DNA 110). The concentrates were re-eluted in 300 \u0026micro;l of re-elution buffer (70% methanol, 0.1% formic acid), vortexed for 10 min at 22 \u0026deg;C and 1,500 rpm on the BIOER Mixing Block MB-102, and centrifuged for 20 min at 13,000 \u0026times; \u003cem\u003eg\u003c/em\u003e at 22 \u0026deg;C, before 200 \u0026micro;l supernatant was transferred into a HPLC vial. Extraction blanks were processed alongside samples. Quantification of JA, JA-Ile, ABA, and SA was performed using LC-ESI-IMS/MS (Waters UPLC-Synapt G2-S HDMS).\u003c/p\u003e\n\u003cp\u003ePeak identification and integration were done in MassLynx (Waters). Phytohormone concentrations were normalised to their corresponding deuterated internal standards and the leaf fresh weight. JA-Ile was normalised by the ratio of the deuterated JA-Ile and the deuterated JA-Ile impurity. Finally, phytohormone levels at T1 were normalised by the average level of that phytohormone at T0. Data from two independent experiments were analysed jointly.\u003c/p\u003e\n\u003cp\u003eStatistical analysis\u003c/p\u003e\n\u003cp\u003eAll statistical analyses, regression models, and data visualisation were performed in RStudio [28]. Plots were generated using the packages ggplot2, ggsignif, smplot2, and patchwork [50\u0026ndash;53]. Data wrangling was carried out using the tidyverse [54].\u003c/p\u003e\n\u003cp\u003eUnless stated otherwise, (generalised) linear mixed-effects models ((G)LMM) were fitted to account for variation among independent experiment replicates, using the nlme or glmmTMB package [55\u0026ndash;57]. Depending on the experiment, response variables included larval weight, log\u003csub\u003e10\u003c/sub\u003e-transformed CFU counts, larval survival, or normalised phytohormone levels. SynCom treatments and/or larval feeding were treated as fixed effects, while experimental replicate was included as a random intercept. Model assumptions were evaluated by visual inspection of residual plots. The effect of different treatments were evaluated using analysis of variances (ANOVA) for linear mixed models or a Type II Wald chi-square test for generalised models. Then, estimated marginal means were calculated using the emmeans package [58], and pairwise comparisons were adjusted with the BH procedure for multiple comparisons. Compact letter displays were generated to indicate statistically different treatment groups.\u003c/p\u003e\n\u003cp\u003eFor comparisons between two groups, parametric or non-parametric tests were applied after inspecting assumptions of normality (Shapiro-Wilk normality test) and homoskedasticity (Levene\u0026rsquo;s test) using the rstatix package [59]. Differences in larval biomass between open-pot and gnotobiotic systems were evaluated using Wilcoxon signed-rank test, whereas larval survival across these growth systems was compared using Fischer\u0026rsquo;s exact test. Bacterial abundance on the leaves after a seven day \u003cem\u003eP. brassicae\u003c/em\u003e feeding period on either SynCom20 or axenic control plants was compared using Student\u0026rsquo;s \u003cem\u003et\u003c/em\u003e-test.\u003c/p\u003e\n\u003cp\u003eTo determine whether bacterial community composition differed among sample types (inoculum, fed leaf, non-fed leaf, larvae) after the feeding period, PERMANOVA was conducted on a Bray-Curtis distance matrix using the vegan package [60]. Differences in alpha diversity metrics (richness, evenness, Shannon diversity) and mean pairwise phylogenetic distances (MPD) were analysed using ANOVA followed by Tukey\u0026rsquo;s \u003cem\u003epost hoc\u003c/em\u003e test.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003eBacterial abundance on leaves and caterpillars across communities of different richness and its effect on caterpillar performance\u003c/p\u003e\n\u003cp\u003eWe established a gnotobiotic insect\u0026ndash;plant system to study the interactions between \u003cem\u003eA.\u0026nbsp;thaliana\u003c/em\u003e, phyllosphere bacteria, and \u003cem\u003eP. brassicae\u003c/em\u003e. To confirm that the system was suitable for larval rearing, we compared \u003cem\u003eP. brassicae\u003c/em\u003e performance on gnotobiotic plants with that on plants cultivated in pots. While larvae were on average larger in pots compared to gnotobiotic plants, larval survival did not differ between conditions (Figure S3).\u003c/p\u003e\n\u003cp\u003eTo assess the effect of community richness in the interaction between \u003cem\u003eA. thaliana\u003c/em\u003e and \u003cem\u003eP.\u0026nbsp;brassicae\u003c/em\u003e, plants were inoculated with SynComs of different richness. A pool of 20 phyllosphere strains was selected to capture the taxonomic diversity of natural leaf communities (Figure 1). Five or ten strains were randomly selected from the bacterial pool to assemble SynCom5 and SynCom10 for each of four independent experiments, respectively (Figure S1). A SynCom composed of the total strain pool was included to capture the highest richness level (SynCom20). Eight days after spray inoculation of plant leaves, \u003cem\u003eP.\u0026nbsp;brassicae\u003c/em\u003e neonates were introduced onto gnotobiotic or axenic control plants, and after seven days of feeding, bacterial abundance on fed leaves and caterpillars, as well as larval performance, were assessed.\u003c/p\u003e\n\u003cp\u003eAfter seven days of feeding in the gnotobiotic system, larval survival rate was similar across treatments. On axenic control plants, an average of 96% of larvae survived, while mean survival on plants inoculated with different SynComs was 99%, 96%, and 98%, respectively, with no significant differences among treatments (Figure 3A, ꭓ\u003csup\u003e2\u003c/sup\u003e = 1.54, df = 3, \u003cem\u003ep\u003c/em\u003e = 0.67).\u003c/p\u003e\n\u003cp\u003ePhyllosphere bacterial abundance on \u003cem\u003eA. thaliana\u003c/em\u003e leaves was similar across SynCom treatments, with only a modest increase in SynCom20 compared to SynCom10 (Figure 3B \u003cem\u003et\u003c/em\u003e(83) = 2.2, \u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e= 0.045). Median bacterial loads were 1.6 \u0026times; 10\u003csup\u003e8\u003c/sup\u003e, 1.2 \u0026times; 10\u003csup\u003e8\u003c/sup\u003e, and 1.9 \u0026times; 10\u003csup\u003e8\u003c/sup\u003e CFU per g of leaf fresh weight for SynCom5, SynCom10 and SynCom20, respectively. The effect of SynCom richness on bacterial loads was significant (ANOVA; \u003cem\u003eF\u003c/em\u003e\u003csub\u003e3,83\u003c/sub\u003e = 209.5, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001). All plants that were inoculated with SynComs carried significantly higher bacterial loads than untreated controls (\u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e \u0026lt; 0.001).\u003c/p\u003e\n\u003cp\u003eBacterial abundance in \u003cem\u003eP. brassicae\u003c/em\u003e larvae differed significantly among SynCom treatments (ANOVA; \u003cem\u003eF\u003c/em\u003e\u003csub\u003e3,100\u003c/sub\u003e = 409.1, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001), with median loads of 1.6 \u0026times; 10⁶, 1.1 \u0026times; 10⁷, and 1.6 \u0026times; 10⁷ CFU per g of larval fresh weight after feeding on plants inoculated with SynCom5, SynCom10, and SynCom20, respectively (Figure 3C). All SynCom treatments carried significantly higher bacterial loads than the untreated control (\u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e \u0026lt; 0.001). SynCom10 and SynCom20 had significantly higher bacterial loads than SynCom5 (\u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e \u0026lt; 0.01), whereas no significant difference was detected between SynCom10 and SynCom20.\u003c/p\u003e\n\u003cp\u003eLarval weight was significantly influenced by SynCom treatment (Figure 3D, ANOVA on LMM; \u003cem\u003eF\u003c/em\u003e\u003csub\u003e3,100\u003c/sub\u003e = 3.62, \u003cem\u003ep\u003c/em\u003e = 0.0157), with larvae fed on plants inoculated with SynCom20 having a significant reduction in weight compared to larvae fed on control plants (Figure 3D, \u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e = 0.0153). On average, larval weight decreased by 13.1%, 14.4%, and 17.6% when plants were inoculated with SynCom5, SynCom10, and SynCom20, respectively, compared to the mean larval weight on axenic control plants. However, no significant differences were detected among the three SynCom treatments (ANOVA on LMM, all comparisons: \u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e \u0026gt; 0.05). Larval weight correlated negatively with SynCom richness (r = -0.27, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.005; Figure S4A) and phylogenetic diversity (Faith\u0026rsquo;s PD, r = -0.30, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.002; Figure S4B). Relative bacterial abundance on leaves (Figure S4C) and in larvae did not correlate with larval weight (Figure S4D).\u003c/p\u003e\n\u003cp\u003eSynCom20 alters phytohormone levels of \u003cem\u003eA. thaliana\u003c/em\u003e to caterpillar feeding\u003c/p\u003e\n\u003cp\u003eWe investigated how phyllosphere bacteria and herbivory influence anti-herbivore defence signalling in \u003cem\u003eA. thaliana\u003c/em\u003e, by quantifying changes in the stress-related phytohormones JA, JA-Ile, SA, ABA. Plants were inoculated either with PBS (untreated control) or with SynCom20 and subsequently exposed to \u003cem\u003eP. brassicae\u003c/em\u003e larvae. Phytohormone levels were measured at the onset of larval feeding (T0) and 24 hours later (T1) using gas chromatography\u0026ndash;mass spectrometry (GC-MS) in two independent experiments.\u003c/p\u003e\n\u003cp\u003eFeeding by \u003cem\u003eP. brassicae\u003c/em\u003e induced strong changes in the phytohormone profiles of \u003cem\u003eA. thaliana\u003c/em\u003e leaves (Figure 4). After 24 h, JA levels markedly increased in response to herbivory (GLMM, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05), with SynCom20-inoculated plants showing significantly higher accumulation than axenic control plants (\u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e \u0026lt; 0.0001). JA-Ile and ABA levels also increased significantly in response to herbivory, but were unaffected by the SynCom20 treatment. By contrast, SA concentrations remained unchanged across treatments. Together, these results show that \u003cem\u003eP. brassicae\u003c/em\u003e herbivory primarily triggers JA-, JA-Ile-, and ABA-mediated defences, with SynCom20 further enhancing JA accumulation in \u003cem\u003eA. thaliana\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003eLarval feeding increases bacterial abundance on leaves and alters phyllosphere community composition\u003c/p\u003e\n\u003cp\u003eTo assess how larval feeding influences phyllosphere bacterial abundance, we compared SynCom20-inoculated plants that were fed upon for seven days to plants that did not experience feeding. As observed previously, SynCom20 inoculation reduced larval weight after feeding (Figure 5A; \u003cem\u003et\u003c/em\u003e-test, \u003cem\u003et\u003c/em\u003e(8) = 2.51, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05). Leaf bacterial loads were approximately fourfold higher on fed compared to non-fed plants, with mean abundances of 4.2 \u0026times; 10\u003csup\u003e8\u003c/sup\u003e and 1 \u0026times; 10\u003csup\u003e8\u003c/sup\u003e CFU per gram of leaf, respectively (Figure 4B; \u003cem\u003et\u003c/em\u003e-test, \u003cem\u003et\u003c/em\u003e(8) = 5.44, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05). These results indicate that larval feeding promotes bacterial proliferation in the phyllosphere.\u003c/p\u003e\n\u003cp\u003eTo examine how \u003cem\u003eP. brassicae\u003c/em\u003e feeding affects phyllosphere bacterial composition and the acquisition of bacteria in the larval gut, we performed 16S rRNA gene amplicon sequencing (V3-V4 region). After seven days of feeding, we collected three types of samples: (i) SynCom20-inoculated leaves that had been fed upon, (ii) the larvae that fed on those leaves, and (iii) SynCom20-inoculated leaves that were not exposed to larvae. In addition, the original SynCom20 inoculum was sequenced to assess changes in community composition after the treatments.\u003c/p\u003e\n\u003cp\u003eSample type explained 73.9% of the variance between communities (PERMANOVA, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001). Relative to the original SynCom20 inoculum, both leaf and larval samples showed marked compositional shifts, with several strains becoming undetectable or persisting only at very low abundance (Figure 6A). Despite a strong overlap in taxa between leaves and larvae, differences in strain abundances resulted in distinct community compositions (Fig 6A, B).\u003c/p\u003e\n\u003cp\u003eLarval feeding caused leaf communities to become dominated by \u003cem\u003ePantoea eucalypti\u003c/em\u003e 299R, except for two samples that clustered more closely with non-fed leaves or larval communities (Fig 6A, B). Given the strong increase in total bacterial load on fed leaves, \u003cem\u003eP.\u0026nbsp;eucalypti\u003c/em\u003e 299R was likely responsible for this proliferation (Fig 5B). To a lesser extent, \u003cem\u003eMicrobacterium\u003c/em\u003e sp. AC026 and \u003cem\u003eBacillus\u003c/em\u003e sp. AC266 also increased in relative abundance in fed compared to non-fed leaves, whereas communities of non-fed leaves were characterised by uniquely high abundances of \u003cem\u003eArthrobacter\u0026nbsp;\u003c/em\u003eLeaf145 (Figure 6C).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBy contrast, larval samples were consistently enriched in \u003cem\u003eMethylobacterium\u0026nbsp;\u003c/em\u003estrains (\u003cem\u003eMethylobacterium radiotolerans\u003c/em\u003e 0-1 and \u003cem\u003eMethylobacterium\u003c/em\u003e AC433), \u003cem\u003eMicrobacterium\u0026nbsp;\u003c/em\u003esp. AC026, \u003cem\u003eCurtobacterium\u003c/em\u003e sp. AC273, and \u003cem\u003eWilliamsia\u0026nbsp;\u003c/em\u003esp. Leaf354, regardless of fed or non-fed leaves (Figure 6B, C). \u003cem\u003eSphingomonas\u003c/em\u003e strains, which were present at low abundance in both fed and non-fed leaves, were strongly underrepresented in larval samples (Figure 6C).\u003c/p\u003e\n\u003cp\u003eAlpha diversity differed markedly among the bacterial communities from the inoculum, leaves, and larvae (Figure 7). Richness was highest in the SynCom20 inoculum, and decreased in both leaf and larval samples. Evenness was reduced in fed leaf communities compared to non-fed leaves (\u003cem\u003ep\u003c/em\u003e\u003csub\u003eadj\u003c/sub\u003e = 0.0256), reflecting the dominance of \u003cem\u003eP. eucalypti\u003c/em\u003e 299R during herbivory (Figure 5). Shannon diversity followed a similar pattern, that is, an overall reduction in alpha diversity in fed leaf communities. The mean pairwise distance (MPD), describing the phylogenetic relatedness of strains within a community, was also highest in the inoculum and declined in plant- and larval-associated bacterial communities. Together, these results show that herbivory reshapes SynCom20 communities by increasing richness but reducing evenness and phylogenetic breadth, leading to dominance of specific strains within the leaf microbiota.\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eUsing a gnotobiotic insect-plant system, we showed that phyllosphere bacteria and herbivory can jointly influence plant defences, the composition of plant and insect microbiota, and insect performance. Increasing bacterial richness in the phyllosphere reduced \u003cem\u003eP. brassicae\u003c/em\u003e larval growth, with a significant effect observed only for the richest synthetic community tested: SynCom20. Herbivory reshaped the phyllosphere microbiota, increasing bacterial loads and enriching \u003cem\u003eP. eucalypti\u003c/em\u003e 299R, while larvae acquired a filtered subset of leaf-associated bacteria dominated by \u003cem\u003eMethylobacterium\u003c/em\u003e, \u003cem\u003eMicrobacterium\u003c/em\u003e, \u003cem\u003eWilliamsia\u003c/em\u003e, and \u003cem\u003eCurtobacterium\u003c/em\u003e. Larval feeding on SynCom20-inoculated plants triggered a stronger induction of jasmonate-associated herbivore-responsive signalling than feeding on axenic control plants. These findings reveal reciprocal feedback between phyllosphere bacteria and herbivory, highlighting that leaf-associated bacteria shape plant defence responses and thereby modify herbivore performance.\u003c/p\u003e\n\u003cp\u003eAlthough plant-associated microbes are known to modulate plant responses to herbivory and influence herbivore performance, most evidence comes from rhizosphere systems [61, 62]. In our system, \u003cem\u003eP. brassicae\u003c/em\u003e larval growth tended to decrease with increasing bacterial richness in the phyllosphere, with a significant reduction observed only for the richest community (SynCom20) compared to axenic control plants. This pattern suggests that higher bacterial richness may enhance the community\u0026rsquo;s overall impact on plant\u0026ndash;insect interactions, potentially through complementary or synergistic effects among strains influencing plant defences [63, 64]. In microbial systems, richer communities often exhibit metabolic complementarity, functional redundancy, and emergent properties that improve community performance and stability [63, 64]. Such emergent properties could enhance the ability of phyllosphere communities to modulate host physiology, thereby altering plant\u0026ndash;herbivore interactions. Comparable patterns have been observed in rhizosphere microbiomes, where complex soil communities more reliably induce systemic resistance against herbivores than single-strain inocula [65\u0026ndash;67]. However, it remains unclear whether the observed patterns in our system were driven by overall richness, emergent interactions among community members, or the activity of key taxa within SynCom20. Candidate strains contributing to these richness-dependent effects include those that became most abundant during larval feeding, such as \u003cem\u003eP. eucalypti\u003c/em\u003e 299R, which is known to be a strong and metabolically versatile competitor in the phyllosphere [68]. These traits may promote its competitive success and capacity to shape community structure during herbivory, potentially affecting plant responses either directly through microbe\u0026ndash;host signalling or indirectly by altering the abundance of taxa that influence defence signalling. Future work employing broader richness gradients, alternative community compositions, and targeted removal or enrichment of specific members could help disentangle these contributions. Overall, these findings show that even nonpathogenic phyllosphere bacterial communities affect herbivore growth, revealing an underappreciated ecological role of resident leaf microbiota in plant anti-herbivore responses.\u003c/p\u003e\n\u003cp\u003eThe ability of phyllosphere bacteria to alter herbivore performance in our experimental system likely arises from their influence on plant defence signalling rather than from direct effects on the larvae, as bacterial abundance on the leaf surface was not correlated with larval weight and no disease symptoms were observed. Moreover, inoculation of \u003cem\u003eA. thaliana\u003c/em\u003e with SynCom20, the community associated with a significant decrease in larval growth, led to stronger JA induction during \u003cem\u003eP. brassicae\u003c/em\u003e feeding, accompanied by similar, though weaker, trends for JA-Ile and ABA, while SA levels remained unchanged across treatments. Together, these results suggest that SynCom20 amplified JA-associated anti-herbivore responses during feeding, likely contributing to the observed reduction in larval growth and hinting at a priming effect on plant defences. Microbial priming of plant defences occurs when prior exposure to microorganisms enhances the plant\u0026rsquo;s capacity to respond more rapidly or intensely to subsequent challenges\u0026mdash;such as pathogen infection or insect attack\u0026mdash;compared with unprimed plants [69\u0026ndash;71]. Rhizosphere-associated bacteria are well-known to induce such JA-dependent priming; for example, colonisation of \u003cem\u003eA. thaliana\u003c/em\u003e roots by \u003cem\u003ePseudomonas simiae\u003c/em\u003e WCS417r primes JA/ET-regulated defences and enhances resistance against the chewing caterpillar \u003cem\u003eMamestra brassicae\u003c/em\u003e [72], and plant-growth promoting rhizobacteria exposure increases JA accumulation and JA-responsive gene expression in cotton, reducing damage and growth of \u003cem\u003eSpodoptera exigua\u003c/em\u003e larvae [73, 74]. A similar mechanism may occur in the phyllosphere, where prior colonisation by nonpathogenic bacteria could sensitise JA signalling pathways without directly activating defences, allowing plants to mount a stronger response upon herbivory. Indeed, transcriptomic analysis of Arabidopsis leaves has shown that colonisation by the commensal \u003cem\u003eSphingomonas melonis\u003c/em\u003e Fr1, a member of our SynCom20, triggers the expression of hundreds of defence-associated genes in \u003cem\u003eA. thaliana\u003c/em\u003e, including the JA/ethylene-responsive marker \u003cem\u003ePDF1.2\u003c/em\u003e [75]. This suggests that certain members of the phyllosphere actively maintain the plant immune system in an \u0026ldquo;alert\u0026rdquo; state.\u003c/p\u003e\n\u003cp\u003eIn addition to jasmonates, ABA also increased in response to herbivory, and this increase tended to be stronger in SynCom-inoculated plants, suggesting that bacterial colonisation may have enhanced ABA accumulation during feeding. ABA interacts extensively with JA signalling: ABA\u0026ndash;JA crosstalk contributes to defence against insects, and ABA receptors such as PYL6 physically interact with the JA regulator MYC2 to modulate JA-responsive gene expression and resistance to herbivory [76]. Such hormonal crosstalk can alter both the magnitude and timing of JA-regulated defences under repeated or combined stresses [71], which could help explain the stronger inducible response observed with SynCom20. On the other hand, SA, typically associated with defence against biotrophic pathogens and systemic acquired resistance, remained unchanged across treatments [76]. Interestingly, SynCom20 alone did not induce SA accumulation in the absence of feeding, indicating that nonpathogenic phyllosphere bacteria do not activate strong pathogen-type defences. This pattern is consistent with the idea that commensal plant-associated microbes evolve strategies to limit costly immune activation\u0026mdash;for example by dampening microbe-associated molecular pattern (MAMP)-triggered immunityor reducing recognition\u0026mdash;while still modulating defence pathways [63, 67]. In our system, SynCom20 nevertheless enhanced JA accumulation during herbivory, consistent with a priming-like scenario in which nonpathogenic colonisers remain tolerated yet sensitise plants for stronger induced anti-herbivore defences [69\u0026ndash;72]. However, the precise mechanisms by which SynCom20 enhances JA accumulation upon herbivory remain to be elucidated. While these hormonal patterns point to a defence amplification during feeding, the underlying triggers of the enhanced JA response could involve herbivory-induced shifts in phyllosphere community structure, as insect feeding is known to reshape leaf-associated bacterial assemblages [17], and as demonstrated by the strong compositional changes observed in our own system.\u003c/p\u003e\n\u003cp\u003eFeeding by \u003cem\u003eP. brassicae\u003c/em\u003e increased bacterial load on the leaf surface, suggesting that higher microbial densities achieved during herbivory could contribute to the enhanced JA response. This is consistent with recent findings that plant immune responses to nonpathogenic phyllosphere bacteria are density-dependent; while plants mount negligible responses to low populations, high bacterial titers elicit pathogen-like transcriptional reprogramming [77]. Chewing damage facilitates bacterial proliferation on leaves [17] and provides bacteria with entry points into internal tissues through wounds [78]. Once inside the apoplast, microbial molecules are perceived by pattern-recognition receptors located on the plasma membrane, making this compartment a major site for defence activation [79, 80]. Because bacterial densities within the apoplast are generally much lower than those found on the leaf surface, even limited internal colonisation may be sufficient to trigger strong immune responses. In this context, the intensified JA signalling observed in SynCom20-inoculated plants could result from the higher bacterial densities that accumulate during feeding and from immune activation following bacterial entry into wounded tissue. Effectively, the rapid proliferation of SynCom members may push bacterial titers beyond a detection threshold, triggering a defence \u0026ldquo;alert\u0026rdquo; that is driven by bacterial density, in line with previous studies [15, 77]. Such processes may underlie how phyllosphere bacteria, despite being nonpathogenic, modulate the magnitude of plant inducible defences during herbivory. The outcome of these interactions likely depends on which bacterial taxa gain access to internal tissues and how MAMPs are perceived once in the apoplast, where pattern-recognition receptor\u0026ndash;mediated immune activation is strongest [79, 80]. While our results revealed a JA-dominated response, studies in other systems show that bacterial activation of plant defences can involve JA- or SA-associated pathways depending on the identity of the interacting microbe [11, 78, 81].\u003c/p\u003e\n\u003cp\u003eBeyond their effects on plant defence signalling, interactions between phyllosphere bacteria and herbivory influence microbial community assembly across the leaf and insect gut. The feeding process that activates defensive responses simultaneously reshapes the leaf microbiota, altering bacterial composition across the plant\u0026ndash;insect gut continuum. Following SynCom20 inoculation and a subsequent feeding period, \u003cem\u003eP. brassicae\u003c/em\u003e herbivory led to compositional shifts distinct from those observed in non-fed plants. This shift was characterised by reduced richness and evenness, suggesting that herbivory favoured a subset of strains better suited to the altered phyllosphere environment. Wounding during feeding likely modifies the physicochemical microenvironment of the leaf [17], potentially including the release of intracellular metabolites such as soluble sugars from damaged cells, while insect frass\u0026mdash;known to contain labile carbon and nitrogen [82]\u0026mdash;may further contribute nutrient-rich material to plant surfaces, together creating conditions that favour the proliferation of metabolically versatile taxa. In addition to these direct effects of wounding and nutrient availability, herbivory-induced JA induction itself can drive community restructuring [17]. Communities became phylogenetically more clustered compared to the inoculum, with most abundant taxa belonging to the phylum \u003cem\u003ePseudomonadota\u003c/em\u003e. Some members of this lineage\u0026mdash;including the phyllosphere epiphyte \u003cem\u003eP. eucalypti\u003c/em\u003e 299R\u0026mdash;exhibit metabolic versatility and strong competitive ability in dynamic or resource-limited leaf environments [68]. However, whether such traits specifically explain the enrichment of \u003cem\u003ePseudomonadota \u003c/em\u003eunder herbivory remains unclear. More broadly, these community-level shifts show that herbivory alters which bacterial taxa persist and proliferate on leaves, thereby influencing their relative abundance when ingested and entering the larval gut.\u003c/p\u003e\n\u003cp\u003eComparison of bacterial assemblages among the SynCom20 inoculum, leaf surfaces, and larval guts further revealed strong compositional shifts as bacteria transitioned from the inoculum to the leaf surface and subsequently to the larval gut. Strain abundances in the inoculum and in the phyllosphere or larval gut were only loosely correlated, and only a subset of strains persisted on leaves or in larvae, consistent with strong host- and environment-mediated filtering [23, 83]. Larval communities resembled those on both non-fed and fed leaves, but remained distinct from either, indicating that ingestion and passage through the gut further reshaped community composition. Distinct subsets of taxa with characteristic abundance patterns were consistently enriched in each environment, likely reflecting contrasting nutrient conditions, pH, and physiological environments of the leaf surface and the larval gut. The Lepidopteran larval gut is highly alkaline, has rapid food transit and a simple structure\u0026mdash;conditions generally unfavourable for stable bacterial colonisation and explaining why persistent, species-specific gut communities are rarely established [23, 83, 84]. Nevertheless, the increased abundance of some bacterial species that were not promoted on leaves suggests that these taxa were able to proliferate within the gut, implying colonisation rather than passive passage. Although the consequences of such colonisation remain unclear, bacteria may interact with larval physiology or metabolic processes indirectly\u0026mdash;for instance through competition, metabolite production, or modification of ingested plant material\u0026mdash;while their survival and activity are also shaped by the plant\u0026rsquo;s defensive chemistry [84]. Previous studies have shown that Lepidopteran larvae can harbour transient, environmentally acquired bacteria without clear fitness benefits [23], although such transient gut associates may still suppress harmful taxa, including entomopathogens, through resource competition or antagonistic interactions [84]. Reintroduction of gut-associated bacteria onto the leaf surface via frass deposition or regurgitation, had only a minor effect on phyllosphere composition compared with the direct effects of herbivory. Bacterial communities on non-fed leaves were more similar to those found in larvae than to those on fed leaves, indicating that these gut-derived bacteria did not substantially alter leaf community structure. By contrast, herbivory itself caused major shifts, likely driven by tissue damage, nutrient leakage, and changes in leaf physiology and defence status [17]. These results suggest that herbivory reshapes the phyllosphere primarily by modifying habitat conditions and resource availability rather than by frass-associated gut bacteria. Overall, the feedback between herbivory and the phyllosphere microbiota appears asymmetric: while leaf-associated bacteria can influence plant defences and insect performance, herbivory exerts a stronger and more direct impact on microbial community structure. \u003c/p\u003e"},{"header":"CONCLUSION","content":"\u003cp\u003eThis study reveals that nonpathogenic phyllosphere bacteria can modulate plant defences and herbivore performance, demonstrating an overlooked ecological role of resident leaf microbiota in plant\u0026ndash;insect interactions. Increasing bacterial richness strengthened jasmonate-associated responses and reduced larval growth, suggesting that microbial diversity and composition influence the magnitude of inducible plant defences. Herbivory, in turn, reshaped the phyllosphere community and increased bacterial abundance, potentially creating a feedback loop in which microbial activity and feeding jointly shape plant physiology and microbiome structure. Unravelling the molecular and ecological mechanisms underlying these feedbacks, particularly how bacterial localisation, density, and signalling contribute to defence activation, will be key to understanding how microbiomes can enhance plant resilience to insect attacks.\u003c/p\u003e\n\u003cp\u003eFinally, the gnotobiotic insect\u0026ndash;plant system developed here provides a reproducible and scalable framework for disentangling the mechanisms linking microbial community structure, plant defences, and herbivore performance. Its controlled design enables systematic exploration of multitrophic interactions in the phyllosphere, paving the way for mechanistic and translational studies on the ecological functions of aboveground microbiomes.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eEthics approval and consent to participate\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003eConsent for publication\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003eAvailability of data and material\u003c/p\u003e\n\u003cp\u003eThe datasets and materials supporting the conclusions of this article are available in the following repositories: The 16S amplicon sequencing data are in the European Nucleotide Archive (ENA) repository, accession number PRJEB104625 (https://www.ebi.ac.uk/ena/browser/view/PRJEB104625). The additional datasets are available in Zenodo (https://doi.org/10.5281/zenodo.17822324) [85].\u003cem\u003e\u0026nbsp;\u003c/em\u003eThe R scripts used for data analysis are accessible in the GitHub repository\u003ca href=\"https://github.com/relab-fuberlin/mueller_arabidopsis_pieris_syncom\"\u003e\u0026nbsp;\u003c/a\u003ehttps://github.com/relab-fuberlin/mueller_arabidopsis_pieris_syncom.\u003c/p\u003e\n\u003cp\u003eCompeting interests\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003eFunding\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis work was financially supported by the Einstein Foundation Berlin (Einstein Independent Researcher Grant ESR-2023-784 to LPV) and the Investitionsbank Berlin (ProValid Grant VAL149/2023 to LPV).\u003c/p\u003e\n\u003cp\u003eAuthors\u0026apos; contributions\u003c/p\u003e\n\u003cp\u003eConceptualisation LPV, RS; Investigation: MM, MH, MK; Formal analysis: MM, LPV, RS, Funding acquisition LPV; Resources: MH, MRE; Writing \u0026ndash; Original Draft: MM, MRE, RS, LPV; Writing \u0026ndash; Review \u0026amp; Editing: MH, MK; Visualisation: MM, RS, LPV; Supervision: RS, LPV\u003c/p\u003e\n\u003cp\u003eAcknowledgements\u003c/p\u003e\n\u003cp\u003eWe thank the Einstein Foundation Berlin (Einstein Independent Researcher Grant ratioESR-2023-784 to LPV) and the Investitionsbank Berlin (ProValid Grant VAL149/2023 to LPV) for financial support. We thank the HPC Service of FUB-IT, Freie Universit\u0026auml;t Berlin, for computing time. We also thank Sandra Hirsch and Mila Oeltjen for their valuable technical assistance and dedication to maintaining the day-to-day laboratory operations, Jasmin Jende, Yara Sophie Dietsch, Semira Romero Reyna, Franziska Schulze, Roxanne Elisabeth Esta Kahn-Cleland, Tia Hannah Teschner, Anastasia Yusupova, Johannes L\u0026uuml;bke, and Nevin Zeyrek for their support in insect rearing, and Dr. Andreas Springer and Dipl.-Ing. Fabian Klautzsch (Schalley Group, Mass Spectrometry Core Facility) for phytohormone quantification.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eRemus-Emsermann MNP, Schlechter RO. Phyllosphere microbiology: at the interface between microbial individuals and the plant host. New Phytol. 2018;218:1327\u0026ndash;33.\u003c/li\u003e\n\u003cli\u003eAngon PB, Mondal S, Jahan I, Datto M, Antu UB, Ayshi FJ, et al. Integrated pest management (IPM) in agriculture and its role in maintaining ecological balance and biodiversity. Adv Agric. 2023;2023:1\u0026ndash;19.\u003c/li\u003e\n\u003cli\u003eAgrawal AA, Maron JL. Long‐term impacts of insect herbivores on plant populations and communities. J Ecol. 2022;110:2800\u0026ndash;11.\u003c/li\u003e\n\u003cli\u003eErb M, Meldau S, Howe GA. Role of phytohormones in insect-specific plant reactions. Trends Plant Sci. 2012;17:250\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eVasantha-Srinivasan P, Noh MY, Park KB, Kim TY, Jung W-J, Senthil-Nathan S, et al. Plant immunity to insect herbivores: mechanisms, interactions, and innovations for sustainable pest management. Front Plant Sci. 2025;16:1599450.\u003c/li\u003e\n\u003cli\u003eLi N, Han X, Feng D, Yuan D, Huang L-J. Signaling Crosstalk between Salicylic Acid and Ethylene/Jasmonate in Plant Defense: Do We Understand What They Are Whispering? Int J Mol Sci. 2019;20.\u003c/li\u003e\n\u003cli\u003eFang X, Xie Y, Yuan Y, Long Q, Zhang L, Abid G, et al. The role of salicylic acid in plant defense responses against biotic stresses. Plant Hormones. 2025;1:0\u0026ndash;0.\u003c/li\u003e\n\u003cli\u003eThaler JS, Humphrey PT, Whiteman NK. Evolution of jasmonate and salicylate signal crosstalk. Trends Plant Sci. 2012;17:260\u0026ndash;70.\u003c/li\u003e\n\u003cli\u003eLazebnik J, Frago E, Dicke M, van Loon JJA. Phytohormone mediation of interactions between herbivores and plant pathogens. J Chem Ecol. 2014;40:730\u0026ndash;41.\u003c/li\u003e\n\u003cli\u003e Dodueva IE, Lebedeva MA, Lutova LA. Phytopathogens and molecular mimicry. Russ J Genet. 2022;58:638\u0026ndash;54.\u003c/li\u003e\n\u003cli\u003e Chung SH, Rosa C, Scully ED, Peiffer M, Tooker JF, Hoover K, et al. Herbivore exploits orally secreted bacteria to suppress plant defenses. Proc Natl Acad Sci U S A. 2013;110:15728\u0026ndash;33.\u003c/li\u003e\n\u003cli\u003e Coolen S, Magda R der-Molen, Welte CU. The secret life of insect-associated microbes and how they shape insect-plant interactions. FEMS Microbiol Ecol. 2022;98.\u003c/li\u003e\n\u003cli\u003e Frago E, Dicke M, Godfray HCJ. Insect symbionts as hidden players in insect-plant interactions. Trends Ecol Evol. 2012;27:705\u0026ndash;11.\u003c/li\u003e\n\u003cli\u003e Keppler A, Roulier M, Pfeilmeier S, Petti GC, Sintsova A, Maier BA, et al. Plant microbiota feedbacks through dose-responsive expression of general non-self response genes. Nat Plants. 2025;11:74\u0026ndash;89.\u003c/li\u003e\n\u003cli\u003e Maier BA, Kiefer P, Field CM, Hemmerle L, Bortfeld-Miller M, Emmenegger B, et al. A general non-self response as part of plant immunity. Nat Plants. 2021;7:696\u0026ndash;705.\u003c/li\u003e\n\u003cli\u003e Miebach M, Schlechter RO, Clemens J, Jameson PE, Remus-Emsermann MNP. Litterbox-A gnotobiotic Zeolite-Clay System to Investigate \u003cem\u003eArabidopsis\u003c/em\u003e-Microbe Interactions. Microorganisms. 2020;8.\u003c/li\u003e\n\u003cli\u003e Humphrey PT, Whiteman NK. Insect herbivory reshapes a native leaf microbiome. Nat Ecol Evol. 2020;4:221\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003e Humphrey PT, Nguyen TT, Villalobos MM, Whiteman NK. Diversity and abundance of phyllosphere bacteria are linked to insect herbivory. Mol Ecol. 2014;23:1497\u0026ndash;515.\u003c/li\u003e\n\u003cli\u003e Bodenhausen N, Bortfeld-Miller M, Ackermann M, Vorholt JA. A synthetic community approach reveals plant genotypes affecting the phyllosphere microbiota. PLoS Genet. 2014;10:e1004283.\u003c/li\u003e\n\u003cli\u003e Brock AK, Berger B, Schreiner M, Ruppel S, Mewis I. Plant growth-promoting bacteria \u003cem\u003eKosakonia radicincitans\u003c/em\u003e mediate anti-herbivore defense in \u003cem\u003eArabidopsis thaliana\u003c/em\u003e. Planta. 2018;248:1383\u0026ndash;92.\u003c/li\u003e\n\u003cli\u003e Vacheron J, P\u0026eacute;chy-Tarr M, Brochet S, Heiman CM, Stojiljkovic M, Maurhofer M, et al. T6SS contributes to gut microbiome invasion and killing of an herbivorous pest insect by plant-beneficial \u003cem\u003ePseudomonas protegens\u003c/em\u003e. ISME J. 2019;13:1318\u0026ndash;29.\u003c/li\u003e\n\u003cli\u003e Meenakshi, Thakur S, Choudhary K, Kumar R, Kumar S, Sharma D. Ecofriendly Management of \u003cem\u003ePieris brassicae\u003c/em\u003e in \u003cem\u003eBrassica oleracea\u003c/em\u003e: A Review. Biol Bull Rev. 2023;13:691\u0026ndash;702.\u003c/li\u003e\n\u003cli\u003e Hammer TJ, Janzen DH, Hallwachs W, Jaffe SP, Fierer N. Caterpillars lack a resident gut microbiome. Proc Natl Acad Sci U S A. 2017;114:9641\u0026ndash;6.\u003c/li\u003e\n\u003cli\u003e Paniagua Voirol LR, Weinhold A, Johnston PR, Fatouros NE, Hilker M. Legacy of a butterfly\u0026rsquo;s parental microbiome in offspring performance. Appl Environ Microbiol. 2020;86.\u003c/li\u003e\n\u003cli\u003e Bai Y, M\u0026uuml;ller DB, Srinivas G, Garrido-Oter R, Potthoff E, Rott M, et al. Functional overlap of the \u003cem\u003eArabidopsis\u003c/em\u003e leaf and root microbiota. Nature. 2015;528:364\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003e Zhang J, Kobert K, Flouri T, Stamatakis A. PEAR: a fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics. 2014;30:614\u0026ndash;20.\u003c/li\u003e\n\u003cli\u003e Bennett L, Melchers B, Proppe B. Curta: A General-purpose High-Performance Computer at ZEDAT, Freie Universit\u0026auml;t Berlin. 2020.\u003c/li\u003e\n\u003cli\u003e Posit team. RStudio: Integrated Development Environment for R. 2023.\u003c/li\u003e\n\u003cli\u003e Gl\u0026ouml;ckner FO, Yilmaz P, Quast C, Gerken J, Beccati A, Ciuprina A, et al. 25 years of serving the community with ribosomal RNA gene reference databases and tools. J Biotechnol. 2017;261:169\u0026ndash;76.\u003c/li\u003e\n\u003cli\u003e Huber W, Carey VJ, Gentleman R, Anders S, Carlson M, Carvalho BS, et al. Orchestrating high-throughput genomic analysis with Bioconductor. Nat Methods. 2015;12:115\u0026ndash;21.\u003c/li\u003e\n\u003cli\u003e McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One. 2013;8:e61217.\u003c/li\u003e\n\u003cli\u003e Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41 Database issue:D590\u0026ndash;6.\u003c/li\u003e\n\u003cli\u003e Wright E. Using DECIPHER v2.0 to analyze big biological sequence data in R. R J. 2016;8:352.\u003c/li\u003e\n\u003cli\u003e Yilmaz P, Parfrey LW, Yarza P, Gerken J, Pruesse E, Quast C, et al. The SILVA and \u0026ldquo;All-species Living Tree Project (LTP)\u0026rdquo; taxonomic frameworks. Nucleic Acids Res. 2014;42 Database issue:D643\u0026ndash;8.\u003c/li\u003e\n\u003cli\u003e Pag\u0026egrave;s H, Aboyoun P, Gentleman R, DebRoy S. Biostrings: Efficient manipulation of biological strings. 2024.\u003c/li\u003e\n\u003cli\u003e Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 2010;26:2460\u0026ndash;1.\u003c/li\u003e\n\u003cli\u003e Emms DM, Kelly S. OrthoFinder: phylogenetic orthology inference for comparative genomics. Genome Biol. 2019;20:238.\u003c/li\u003e\n\u003cli\u003e Emms DM, Kelly S. OrthoFinder: solving fundamental biases in whole genome comparisons dramatically improves orthogroup inference accuracy. Genome Biol. 2015;16:157.\u003c/li\u003e\n\u003cli\u003e Yu G, Smith DK, Zhu H, Guan Y, Lam TT-Y. Ggtree: An r package for visualization and annotation of phylogenetic trees with their covariates and other associated data. Methods Ecol Evol. 2017;8:28\u0026ndash;36.\u003c/li\u003e\n\u003cli\u003e Yu G, Lam TT-Y, Zhu H, Guan Y. Two Methods for Mapping and Visualizing Associated Data on Phylogeny Using Ggtree. Mol Biol Evol. 2018;35:3041\u0026ndash;3.\u003c/li\u003e\n\u003cli\u003e Yu G. Using ggtree to Visualize Data on Tree-Like Structures. Curr Protoc Bioinformatics. 2020;69:e96.\u003c/li\u003e\n\u003cli\u003e Xu S, Li L, Luo X, Chen M, Tang W, Zhan L, et al. Ggtree: A serialized data object for visualization of a phylogenetic tree and annotation data. Imeta. 2022;1:e56.\u003c/li\u003e\n\u003cli\u003e Yu G. Data integration, manipulation and visualization of phylogenetic trees. Boca Raton: Chapman and Hall/CRC; 2022.\u003c/li\u003e\n\u003cli\u003e Kembel SW, Cowan PD, Helmus MR, Cornwell WK, Morlon H, Ackerly DD, et al. Picante: R tools for integrating phylogenies and ecology. Bioinformatics. 2010;26:1463\u0026ndash;4.\u003c/li\u003e\n\u003cli\u003e Paradis E, Schliep K. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics. 2019;35:526\u0026ndash;8.\u003c/li\u003e\n\u003cli\u003e Henry L, Wickham H, Chang W. ggstance: Horizontal \u0026ldquo;ggplot2\u0026rdquo; Components. 2024.\u003c/li\u003e\n\u003cli\u003e Oksanen J, Simpson GL, Blanchet FG, Kindt R, Legendre P, Minchin PR, et al. Vegan: Community ecology package. CRAN: Contributed Packages. 2001.\u003c/li\u003e\n\u003cli\u003e Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. bioRxiv. 2014.\u003c/li\u003e\n\u003cli\u003e Wang L, Halitschke R, Kang J-H, Berg A, Harnisch F, Baldwin IT. Independently silencing two JAR family members impairs levels of trypsin proteinase inhibitors but not nicotine. Planta. 2007;226:159\u0026ndash;67.\u003c/li\u003e\n\u003cli\u003e Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer Science \u0026amp; Business Media; 2009.\u003c/li\u003e\n\u003cli\u003e Ahlmann-Eltze C, Patil I. ggsignif: R Package for Displaying Significance Brackets for \u0026ldquo;ggplot2.\u0026rdquo; PsyArXiv. 2021.\u003c/li\u003e\n\u003cli\u003e Min SH, Zhou J. smplot: An R Package for Easy and Elegant Data Visualization. Front Genet. 2021;12:802894.\u003c/li\u003e\n\u003cli\u003e Pedersen TL. patchwork: The Composer of Plots. CRAN: Contributed Packages. 2019.\u003c/li\u003e\n\u003cli\u003e Wickham H, Averick M, Bryan J, Chang W, McGowan L, Fran\u0026ccedil;ois R, et al. Welcome to the tidyverse. J Open Source Softw. 2019;4:1686.\u003c/li\u003e\n\u003cli\u003e Pinheiro J, Bates DM. Mixed-effects models in S and S-PLUS. 1st edition. New York, NY: Springer; 2000.\u003c/li\u003e\n\u003cli\u003e Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team. nlme: Linear and Nonlinear Mixed Effects Models. 2021.\u003c/li\u003e\n\u003cli\u003e McGillycuddy M, Warton DI, Popovic G, Bolker BM. Parsimoniously fitting large multivariate random effects in glmmTMB. J Stat Softw. 2025;112.\u003c/li\u003e\n\u003cli\u003e Lenth RV, Piaskowski J. emmeans: Estimated Marginal Means, aka Least-Squares Means. 2025.\u003c/li\u003e\n\u003cli\u003e Kassambara A. rstatix: Pipe-Friendly Framework for Basic Statistical Tests. 2023.\u003c/li\u003e\n\u003cli\u003e Oksanen J, Simpson GL, Blanchet FG, Kindt R, Legendre P, Minchin PR, et al. vegan: Community Ecology Package. 2023.\u003c/li\u003e\n\u003cli\u003e Friman J, Pineda A, van Loon JJA, Dicke M. Bidirectional plant‐mediated interactions between rhizobacteria and shoot‐feeding herbivorous insects: a community ecology perspective. Ecol Entomol. 2021;46:1\u0026ndash;10.\u003c/li\u003e\n\u003cli\u003e Pirttil\u0026auml; AM, Brusila V, Koskim\u0026auml;ki JJ, W\u0026auml;li PR, Ruotsalainen AL, Mutanen M, et al. Exchange of microbiomes in plant-insect herbivore interactions. MBio. 2023;14:e0321022.\u003c/li\u003e\n\u003cli\u003e Dubey A, Malla MA, Vimal SR, Kumar A, Prasad SM, Khan ML. Plant-microbiome engineering: synergistic microbial partners for crop health and sustainability. Plant Growth Regul. 2025. https://doi.org/10.1007/s10725-025-01385-5.\u003c/li\u003e\n\u003cli\u003e Lei J, Feng J, Ding J, Yang Y. Revisiting the classical biodiversity-ecosystem functioning and stability relationships in microbial microcosms. PNAS Nexus. 2025;4:pgaf114.\u003c/li\u003e\n\u003cli\u003e Pineda A, Kaplan I, Hannula SE, Ghanem W, Bezemer TM. Conditioning the soil microbiome through plant-soil feedbacks suppresses an aboveground insect pest. New Phytol. 2020;226:595\u0026ndash;608.\u003c/li\u003e\n\u003cli\u003e Pineda A, Kaplan I, Bezemer TM. Steering soil microbiomes to suppress aboveground insect pests. Trends Plant Sci. 2017;22:770\u0026ndash;8.\u003c/li\u003e\n\u003cli\u003e Rashid MH-O, Chung YR. Induction of systemic resistance against insect herbivores in plants by beneficial soil microbes. Front Plant Sci. 2017;8:1816.\u003c/li\u003e\n\u003cli\u003e Schlechter RO, Kear EJ, Bernach M, Remus DM, Remus-Emsermann MNP. Metabolic resource overlap impacts competition among phyllosphere bacteria. ISME J. 2023;17:1445\u0026ndash;54.\u003c/li\u003e\n\u003cli\u003e Mauch-Mani B, Baccelli I, Luna E, Flors V. Defense priming: An adaptive part of induced resistance. Annu Rev Plant Biol. 2017;68:485\u0026ndash;512.\u003c/li\u003e\n\u003cli\u003e Martinez-Medina A, Flors V, Heil M, Mauch-Mani B, Pieterse CMJ, Pozo MJ, et al. Recognizing Plant Defense Priming. Trends Plant Sci. 2016;21:818\u0026ndash;22.\u003c/li\u003e\n\u003cli\u003e Hilker M, Schm\u0026uuml;lling T. Stress priming, memory, and signalling in plants. Plant Cell Environ. 2019;42:753\u0026ndash;61.\u003c/li\u003e\n\u003cli\u003e Pangesti N, Reichelt M, van de Mortel JE, Kapsomenou E, Gershenzon J, van Loon JJA, et al. Jasmonic acid and ethylene signaling pathways regulate glucosinolate levels in plants during rhizobacteria-induced systemic resistance against a leaf-chewing herbivore. J Chem Ecol. 2016;42:1212\u0026ndash;25.\u003c/li\u003e\n\u003cli\u003e Zebelo S, Song Y, Kloepper JW, Fadamiro H. Rhizobacteria activates (+)-\u0026delta;-cadinene synthase genes and induces systemic resistance in cotton against beet armyworm (\u003cem\u003eSpodoptera exigua\u003c/em\u003e). Plant Cell Environ. 2016;39:935\u0026ndash;43.\u003c/li\u003e\n\u003cli\u003e Aziz M, Nadipalli RK, Xie X, Sun Y, Surowiec K, Zhang J-L, et al. Augmenting sulfur metabolism and herbivore defense in \u003cem\u003eArabidopsis\u003c/em\u003e by bacterial volatile signaling. Front Plant Sci. 2016;7:458.\u003c/li\u003e\n\u003cli\u003e Vogel C, Bodenhausen N, Gruissem W, Vorholt JA. The \u003cem\u003eArabidopsis\u003c/em\u003e leaf transcriptome reveals distinct but also overlapping responses to colonization by phyllosphere commensals and pathogen infection with impact on plant health. New Phytol. 2016;212:192\u0026ndash;207.\u003c/li\u003e\n\u003cli\u003e Wang Y, Mostafa S, Zeng W, Jin B. Function and mechanism of jasmonic acid in plant responses to abiotic and biotic stresses. Int J Mol Sci. 2021;22:8568.\u003c/li\u003e\n\u003cli\u003e Miebach M, Faivre L, Schubert D, Jameson P, Remus-Emsermann M. Nonpathogenic leaf-colonizing bacteria elicit pathogen-like responses in a colonization density-dependent manner. Plant-Environment Interactions. 2024;5:e10137.\u003c/li\u003e\n\u003cli\u003e R\u0026iacute;o-\u0026Aacute;lvarez I, Mu\u0026ntilde;oz-G\u0026oacute;mez C, Navas-V\u0026aacute;squez M, Mart\u0026iacute;nez-Garc\u0026iacute;a PM, Ant\u0026uacute;nez-Lamas M, Rodr\u0026iacute;guez-Palenzuela P, et al. Role of \u003cem\u003eDickeya dadantii\u003c/em\u003e 3937 chemoreceptors in the entry to \u003cem\u003eArabidopsis\u003c/em\u003e leaves through wounds. Mol Plant Pathol. 2015;16:685\u0026ndash;98.\u003c/li\u003e\n\u003cli\u003e Dora S, Terrett OM, S\u0026aacute;nchez-Rodr\u0026iacute;guez C. Plant-microbe interactions in the apoplast: Communication at the plant cell wall. Plant Cell. 2022;34:1532\u0026ndash;50.\u003c/li\u003e\n\u003cli\u003e Farvardin A, Gonz\u0026aacute;lez-Hern\u0026aacute;ndez AI, Llorens E, Garc\u0026iacute;a-Agust\u0026iacute;n P, Scalschi L, Vicedo B. The Apoplast: A Key Player in Plant Survival. Antioxidants (Basel). 2020;9.\u003c/li\u003e\n\u003cli\u003e Sugio A, Dubreuil G, Giron D, Simon J-C. Plant-insect interactions under bacterial influence: ecological implications and underlying mechanisms. J Exp Bot. 2015;66:467\u0026ndash;78.\u003c/li\u003e\n\u003cli\u003e Kagata H, Ohgushi T. Positive and negative impacts of insect frass quality on soil nitrogen availability and plant growth. Popul Ecol. 2012;54:75\u0026ndash;82.\u003c/li\u003e\n\u003cli\u003e Shao Y, Mason CJ, Felton GW. Toward an Integrated Understanding of the Lepidoptera Microbiome. Annu Rev Entomol. 2024;69:117\u0026ndash;37.\u003c/li\u003e\n\u003cli\u003e Paniagua Voirol LR, Frago E, Kaltenpoth M, Hilker M, Fatouros NE. Bacterial Symbionts in Lepidoptera: Their Diversity, Transmission, and Impact on the Host. Front Microbiol. 2018;9:556.\u003c/li\u003e\n\u003cli\u003e M\u0026uuml;ller M, Huve MAP, Kunzler M, Remus-Emsermann M, Schlechter R, Paniagua Voirol LR. Gnotobiotic Arabidopsis\u0026ndash;\u003cem\u003ePieris brassicae\u003c/em\u003e\u0026ndash;Microbiota System Dataset. 2025.\u003c/li\u003e\n\u003cli\u003e Remus-Emsermann MNP, Kim EB, Marco ML, Tecon R, Leveau JHJ. Draft Genome Sequence of the Phyllosphere Model Bacterium \u003cem\u003ePantoea agglomerans\u003c/em\u003e 299R. Genome Announc. 2013;1.\u003c/li\u003e\n\u003cli\u003e Green PN, Bousfield IJ. Emendation of \u003cem\u003eMethylobacterium\u003c/em\u003e Patt, Cole, and Hanson 1976; \u003cem\u003eMethylobacterium rhodinum\u003c/em\u003e (Heumann 1962) comb. nov. corrig.; \u003cem\u003eMethylobacterium radiotolerans\u003c/em\u003e (Ito and Iizuka 1971) comb. nov. corrig.; and \u003cem\u003eMethylobacterium mesophilicum\u003c/em\u003e (Austin and Goodfellow 1979) comb. nov. Int J Syst Bacteriol. 1983;33:875\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003e Loper JE, Lindow SE. Lack of evidence for in situ fluorescent pigment production by \u003cem\u003ePseudomonas syringae\u003c/em\u003e pv. \u003cem\u003esyringae\u003c/em\u003e on bean leaf surfaces. Phytopathology. 1987;77:1449.\u003c/li\u003e\n\u003cli\u003e Innerebner G, Knief C, Vorholt JA. Protection of \u003cem\u003eArabidopsis thaliana\u003c/em\u003e against leaf-pathogenic \u003cem\u003ePseudomonas syringae\u003c/em\u003e by \u003cem\u003eSphingomonas\u003c/em\u003e strains in a controlled model system. Appl Environ Microbiol. 2011;77:3202\u0026ndash;10.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"environmental-microbiome","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"sigs","sideBox":"Learn more about [Environmental Microbiome](https://environmentalmicrobiome.biomedcentral.com)","snPcode":"40793","submissionUrl":"https://submission.nature.com/new-submission/40793/3","title":"Environmental Microbiome","twitterHandle":"@bmc","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Gut microbiota, Jasmonic acid, Lepidoptera, Phyllosphere microbiota, Synthetic community, Insect performance, Community diversity","lastPublishedDoi":"10.21203/rs.3.rs-8337060/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8337060/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eThe leaf surface, or phyllosphere, hosts abundant and diverse bacterial communities that interact with both the host plant and herbivorous insects, yet their collective influence on plant–insect interactions remains poorly investigated.\u003cstrong\u003e \u003c/strong\u003eWe established a gnotobiotic insect-plant system combining \u003cem\u003eArabidopsis thaliana\u003c/em\u003e and \u003cem\u003ePieris brassicae\u003c/em\u003e larvae. Using defined synthetic phyllosphere communities (SynComs) of increasing richness (5, 10, or 20 members), we investigated how phyllosphere bacteria influence herbivore performance, plant defence responses, and bacterial colonisation of both leaves and the insect gut.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e While larval weight tended to decrease with increasing community richness, only the most diverse SynCom (20 members) caused significant weight reductions without affecting survival. Likewise, plants harbouring the most diverse community showed enhanced jasmonic acid (JA) levels during feeding, whereas salicylic acid (SA) remained unchanged, suggesting the specific induction of JA-associated defences. Compared to plants experiencing no herbivory, feeding strongly reshaped bacterial colonisation on leaves, increasing total bacterial loads about fourfold and driving dominance of \u003cem\u003ePantoea eucalypti\u003c/em\u003e 299R as shown by 16S rRNA gene amplicon sequencing. Larvae acquired a distinct subset of bacteria, primarily recruited from the genera \u003cem\u003eMethylobacterium\u003c/em\u003e, \u003cem\u003eMicrobacterium\u003c/em\u003e, \u003cem\u003eWilliamsia\u003c/em\u003e, and \u003cem\u003eCurtobacterium\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions: \u003c/strong\u003eTogether, these findings suggest that resident phyllosphere bacteria modulate plant defences and thereby affect herbivore performance, while herbivory restructures the leaf microbiota and bacterial filtering occurs during passage through the insect gut.\u003c/p\u003e","manuscriptTitle":"From the leaf to the gut and back again: the fate and influence of phyllosphere bacteria in a gnotobiotic Arabidopsis – Pieris brassicae system","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-09 03:41:34","doi":"10.21203/rs.3.rs-8337060/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-04-22T05:10:56+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-21T11:53:12+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-10T13:28:39+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-19T01:43:31+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"240670431925427060888751915372661216395","date":"2026-03-12T13:28:05+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"156895074368223472621998923196168581433","date":"2026-03-10T08:55:47+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"90257362758152082838531743002643179498","date":"2026-03-08T16:52:24+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-22T21:29:33+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"180659855424523077668247387956108621124","date":"2026-01-25T15:31:49+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-01-24T18:46:03+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-04T10:59:17+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-12-19T07:05:55+00:00","index":"","fulltext":""},{"type":"submitted","content":"Environmental Microbiome","date":"2025-12-11T12:59:19+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"environmental-microbiome","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"sigs","sideBox":"Learn more about [Environmental Microbiome](https://environmentalmicrobiome.biomedcentral.com)","snPcode":"40793","submissionUrl":"https://submission.nature.com/new-submission/40793/3","title":"Environmental Microbiome","twitterHandle":"@bmc","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"1edd2bbc-6688-4d5d-ad59-da35089c8a43","owner":[],"postedDate":"January 9th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-04-22T05:24:50+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-09 03:41:34","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8337060","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8337060","identity":"rs-8337060","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}