Functional Profiling of Kiwifruit Phyllosphere Bacteria: Copper Resistance and Biocontrol Potential as a Foundation for Microbiome-Informed Strategies

TL;DR

This study explores bacteria on kiwifruit leaves to find natural solutions for fighting a harmful disease while resisting copper, a common pesticide.

Contribution

The study identifies native copper-resistant bacteria with plant health benefits and biocontrol potential against Pseudomonas syringae pv. actinidiae.

Findings

53.3% of isolates showed copper resistance, including a highly resistant non-pathogenic strain related to Erwinia iniecta.

25 isolates exhibited all four plant growth-promoting traits and some showed antagonism against Psa.

Bacillus pumilus consistently suppressed Psa growth in vitro, suggesting potential for biocontrol.

Abstract

Bacterial canker, caused by Pseudomonas syringae pv. actinidiae (Psa) is a major threat to global kiwifruit production. Copper-based bactericides remain widely used, but increasing resistance highlights the urgency of developing sustainable alternatives. Understanding the functional capabilities of phyllosphere bacteria under copper pressure is critical for designing microbiome-informed management strategies. This study provides a culture-based functional inventory of bacteria associated with Actinidia chinensis var. deliciosa leaves from Portuguese orchards under long-term copper management, aiming to identify native taxa with traits relevant to plant health and resilience. A total of 1058 isolates were recovered and grouped into 261 Random Amplification of Polymorphic DNA (RAPD) clusters, representing 58 species across 29 genera. Representative strains were screened for Plant Growth-Promoting (PGP) traits (Indole-3-acetic acid (IAA), siderophore production, phosphate solubilization, ammonia production), copper tolerance, and in vitro antagonism against Psa. Copper resistance was widespread (53.3% of isolates with MIC ≥ 0.8 mM), including the first evidence of a highly copper-resistant PSA strain in Portuguese kiwifruit orchards and an exceptionally resistant non-pathogenic strain closely related to Erwinia iniecta (MIC 2.8 mM). A subset of 25 isolates combined all four PGP traits, and several also exhibited antagonism against Psa in vitro, among them Bacillus pumilus consistently supressed pathogen growth. Notably, antagonistic and multifunctional traits co-occurred in some isolates, highlighting promising candidates for integrated biocontrol strategies. Overall, the findings reveal a functionally diverse and copper-resilient collection of cultured bacteria, offering both challenges and opportunities for microbiome-based disease management. This work establishes a robust functional basis for subsequent in planta validation and the development of sustainable, microbiome-informed approaches for Psa control.

Figures

Full text

1. Introduction

The cultivation of kiwifruit (Actinidia spp.) in Portugal dates back to the 1970s, when the ‘Hayward’ variety (A. chinensis var. deliciosa) adapted well to local conditions [1]. Encouraged by a favourable climate and a growing market, production grew steadily in the 1980s and 1990s, particularly in the northern and central regions. By the early 2000s, national yields had surpassed 34,000 tonnes per year, placing Portugal among Europe’s key producers [2]. According to UNECE projections, output may increase by another 10%, reaching around 58,000 tonnes [3]. Today, the crop is mainly grown in Entre Douro e Minho, Beira Litoral, and parts of Beira Interior and the Tagus Valley, where edaphoclimatic conditions are favourable for kiwifruit growth [4].

Kiwifruit was considered a relatively problem-free crop, with fungal pathogens such as Botrytis cinerea and Diaporthe actinidiae being the primary concerns [5]. However, this changed dramatically following the spread of bacterial canker, caused by Pseudomonas syringae pv. actinidiae (Psa). Psa causes characteristic symptoms, including leaf spots, brown-black cankers with reddish exudates, flower necrosis, and dieback of shoots and leaders, often leading to severe yield losses and reduced vine longevity. Psa was first reported in Europe in 1992, but the severe outbreaks that turned it into a global threat began in Italy in 2008 [6]. It was detected in Portugal two years later, initially in orchards located in Santa Maria da Feira and Valença [7,8]. The pathogen quickly spread across northern and central regions, severely impacting orchards in the regions of Entre Douro e Minho and Beira Litoral.

Figueira et al. (2020) [9] analysed 604 Psa isolates from four orchards under different abiotic conditions and found substantial genetic diversification since the initial clonal outbreaks in 2010. This diversity varied according to the orchard, season and leaf niche, with a marked decline in diversity during autumn, when a few dominant genotypes prevailed—likely reflecting adaption to environmental and management conditions. This persistence highlights the adaptability of Psa and reinforces the need for site-specific, integrated management strategies. In this context, microbiome-based approaches are emerging as new avenues for sustainable and resilient disease control.

Since the earliest outbreaks of Psa, kiwifruit growers have relied on integrated strategies to manage the disease, combining orchard hygiene, pruning care, and irrigation control with chemical treatments and, where available, tolerant cultivars [6,10]. Copper-based bactericides are the basis in most regions, due to their broad activity and low cost. In some countries, antibiotics, such as streptomycin, are also used; however, this is not the case in the EU, where such products are banned due to concerns about antimicrobial resistance [11].

The problem is that copper treatments are not always effective, particularly under high disease pressure. Repeated application may contribute to residues on plant tissues and exert selective pressure on bacterial populations [12,13]. Resistance is no longer just a concern for the future: studies from Greece have already identified Psa strains capable of tolerating copper concentrations up to 400 µg/mL [13]. Moreover, resistance genes like copABCD have been detected in isolates from both Italy and New Zealand [12,14]. These copper-tolerant Psa isolates were identified in orchards subjected to repeated copper sprays over multiple seasons, highlighting the selective pressure imposed by long-term copper-based management. Within the EU, growers are mostly left with copper and plant defence inducers like acibenzolar-S-methyl [15,16,17,18,19]. These tools can reduce symptoms but rarely control the pathogen. The pressure for alternatives is mounting. Policies like the European Green Deal and the Farm-to-Fork Strategy call for a 50% pesticide reduction by 2030 and promote biological solutions as a way to protect food systems, ecosystems, and biodiversity [20,21]. In this context, understanding copper tolerance locally—including in Portuguese orchards—is not just useful, but essential for informed decision-making. Importantly, no copper-resistant Psa strains had been documented in Portugal prior to this study. Given the long-term use of copper-based sprays in Portuguese orchards, assessing whether copper tolerance may already have emerged locally—either in Psa or in co-occurring phyllosphere bacteria that can function as reservoirs of tolerance genes—was therefore particularly relevant.

In recent years, there has been a significant shift in our understanding of the plant microbiome, now recognised as a dynamic, responsive ecosystem that contributes to nutrient cycling, stress buffering, disease suppression, and immune regulation [22,23,24]. Consequently, the focus has moved away from simply replacing agrochemical inputs with biological ones. Instead, the aim is to manage this complex microbial community as a functional component of the plant holobiont.

Innovative technologies have helped open that door. Synthetic microbial communities (SynComs), microbiome engineering, and microbiome-assisted plant breeding are all being explored as ways to design more stable and site-specific solutions. In practical terms, microbial consortia adapted to local environmental conditions may offer better results than single-strain applications, particularly in open-field systems [23,24].

Kiwifruit is a good example of where this shift matters. Recent works have shown that the phyllosphere microbial communities can play a protective role, helping to buffer disease pressure from pathogens like Psa [25,26]. Studies by Correia et al. [27] and Patterson et al. [28] looked at how these communities respond to different production systems, though their findings are mostly based on culture-independent methods. Consequently, there is limited understanding of the functional capabilities of strains that can be isolated, characterized, and potentially deployed. This gap becomes even more relevant if we consider the plant as a holobiont—a partnership between the host and its microbiota [29,30]. In this frame, pathogens can destabilise the microbial balance, triggering shifts in community composition that may either support recovery or hasten plant decline [31,32]. Understanding these shifts is key to boosting microbial allies that can persist under pressure and promote resilience from within.

To address this gap, the present study aimed to explore the culturable bacterial microbiota associated with the leaves of A. chinensis var. deliciosa in organic orchards affected by Psa infection in Northwest Portugal. Specifically, we aimed to (i) isolate and identify epiphytic and endophytic bacterial strains from healthy and diseased male and female plants; (ii) provide a preliminary description of the structural diversity of the culturable bacteriome concerning plant health status and sex; (iii) conduct an initial screening of representative isolates for plant growth-promoting (PGP) traits; (iv) assess their antagonistic activity against Psa strains; and (v) evaluate copper tolerance among bacterial isolates. By integrating taxonomic, ecological, and functional insights, we further explore how environmental filtering and microbial plasticity shape the resilience of plant-associated communities under biotic stress conditions. Altogether, this preliminary but comprehensive in vitro functional triage was designed to identify robust native candidate taxa for subsequent in planta and field validation. As such, the present work should be viewed as a initial screening to an initial screening to pinpoint robust native taxa that warrant further testing under greenhouse and orchard conditions.

Through the isolation, functional characterisation, and preservation of native culturable bacteria, this work contributes to expanding the microbial resource base available for future research on microbiome-based approaches for plant health. By targeting bacteria naturally adapted to the kiwifruit phyllosphere, this study provides a starting point for the development of site-specific microbial resources with potential relevance for sustainable disease management strategies.

2. Materials and Methods

2.1. Sampling and Processing

Sampling was carried out to obtain comparable phyllosphere bacterial communities from four field-defined groups, representing two orchards (healthy vs. diseased) and two plant-associated categories (male- and female-associated vines, corresponding to different cultivars in this production system). Composite sampling was applied to integrate within-plant and between-plant variability and to obtain a representative cultured community profile for each group.

This study was conducted in two neighbouring organic kiwifruit orchards located in Northwest Portugal (near Vila Verde—41°41′19.4″ N; 8°24′29.6″ W), one classified as healthy and the other as diseased [33] based on prior screening for Pseudomonas syringae pv. actinidiae (Psa) following European and Mediterranean Plant Protection Organisation (EPPO) standards [34]. Diseased plants showed characteristic bacterial canker symptoms, and Psa infection was confirmed by molecular detection according to the EPPO protocol.

Leaves from five Actinidia chinensis cv. deliciosa ‘Hayward’ and five cv. ‘Tomuri’ plants were collected from the healthy orchard (H, Psa not detected) and the diseased orchard (D, Psa detected). In these orchards, ‘Hayward’ corresponded to female vines and ‘Tomuri’ to male vines; therefore, sex and cultivar effects are confounded in the present design, and any observed differences should be interpreted with caution. For each group, ten leaves were sampled from various canopy positions and pooled to form four composite samples: Healthy Males (HM), Healthy Females (HF), Diseased Males (DM), and Diseased Females (DF). Each group was represented by a single composite sample (n = 1 pooled biological unit per condition); therefore, results represent descriptive patterns within the cultured fraction rather than statistically replicated comparisons between groups.

All samples were collected on the same day (post-bloom, June) to minimise temporal variability in the phyllosphere microbiota. Samples were placed in sterile plastic bags, stored at 4 °C, and processed on the same day. Approximately 150 g of leaves from each composite sample were cut into small pieces and blended with 600 mL of sterile ultrapure water. The resulting extract was filtered through sterile gauze to remove debris and used for isolating epiphytic and endophytic microorganisms through culture-dependent methods, and for independent culture methods as previously described [33].

2.2. Bacterial Isolation and DNA Extraction

The isolation workflow was designed to maximise recovery of the culturable heterotrophic fraction of the phyllosphere across a broad range of physiological niches, thereby enabling downstream functional and taxonomic comparisons between plant conditions. To maximise recovery of facultative aerobic heterotrophic bacteria, serial dilutions (10^0^ to 10^−6^) of the plant extract were prepared. From each dilution, 100 µL was plated onto Alkaliphilic Buffered Medium 2 (ABM2) [35] and Reasoner’s 2A agar (R2A) [36] plates, each buffered to pH 5.5, 7.0 and 8.5 at a final concentration of 75 mM. Agar plates were incubated at 20 °C, 25 °C, and 37 °C. ABM2 and R2A were selected based on their contrasting nutrient content (high and low, respectively). Technical replication was performed for each dilution and condition to ensure robust recovery of representative culturable diversity. After five days of incubation, the Colony-Forming Units (CFUs) were counted. Colonies were characterised based on morphologic traits (pigmentation, shape, margin, size, and texture). All different morphological colony types were subsequently selected and sub-cultured in the same conditions to obtain an axenic culture. Isolates were cryopreserved in the isolation medium supplemented with 15% (w/v) glycerol and stored at −80 °C until further use. Each isolate was assigned an ID code (KWT001) for identification purposes during this study. Total DNA was extracted via thermal shock as previously described [9]. Briefly, a bacterial suspension was prepared in lysis buffer (2% Tween 20 in 0.1 M NaOH), followed by heat denaturation and sedimentation of cell debris. DNA concentrations were normalised using a Nanodrop spectrophotometer (Thermo Fisher Scientific Inc., Walthman, MA, USA).

2.3. Fingerprinting Analysis

Random Amplified Polymorphic DNA (RAPD) profiling was used to reduce redundancy among isolates and to identify unique genomic fingerprints, ensuring efficient selection of representative strains for sequencing and functional screening.

Bacterial isolates were fingerprinted by RAPD analysis using the primer OPA-03 (5′-AGTCAGCCAC-3′) as described by Costa et al. (2005) [37]. RAPD-PCR reactions were repeated for a subset of isolates whenever ambiguous profiles were obtained, and a negative control was included in each run. PCR products were separated by electrophoresis on 2% (w/v) agarose gels in 1× TAE buffer, stained with GreenSafe Premium (NZYTech, Lisbon, Portugal), and visualised with a Doc XR+ imaging system (Bio-Rad, Algés, Portugal). Gel images were imported into BioNumerics and normalized using the molecular weight marker (NZYDNA Ladder III, NZYTech, Lisbon, Portugal) as reference. Clustering was performed by visual inspection of band number and migration patterns, following the workflow described in Figueira et al. (2020) [9]. No fixed similarity threshold was applied, as RAPD types were defined as a dereplication criterion based on visually distinct banding patterns. All isolates were clustered according to their RAPD pattern profile and, at least, one strain of each RAPD group was selected as a representative strain for further analysis

2.4. Phylogenetic Analysis

Phylogenetic identification of representative isolates enabled taxonomic assignment of RAPD clusters and supported the description of major bacterial lineages across the sampled groups (HF/HM/DF/DM).

Representative strains from each RAPD cluster were selected for 16S rRNA gene sequencing; for clusters with three isolates, one strain was selected; for clusters with four or more isolates, two strains were selected. 16S rRNA gene fragments (~1500 bp) were amplified using primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1525R (5′-GGTTACCTTGTTACGACTT-3′) [38]. PCR products were purified and sequenced by StabVida (Portugal), using primer 519R (5′-GWATTACCGCGGCKGCTG-3′) [39]. Sequences were quality-checked and edited in UGENE [40]. Sequences were clustered at ≥98% similarity using CD-HIT Suite [41], and taxonomic identities were assigned through BLAST (v2.16) comparison against the NCBI GenBank database. In cases where the species-level assignment was ambiguous or required confirmation, phylogenetic trees were constructed in ARB software [42] using the neighbor-joining algorithm (NJ) with Jukes-Cantor correction and 1000-bootstrap resampling. Species-level identifications were retained only when supported by unambiguous BLAST results and/or phylogenetic placement; for taxonomically complex genera with limited 16S resolution (e.g., Pseudomonas, Pantoea and Enterobacterales), isolates were conservatively reported at genus level unless species assignment was clearly supported. To confirm Psa identification among Pseudomonas spp. isolates, specific testing was performed by the Laboratory for Phytopathology from Instituto Pedro Nunes (Portugal) according to EPPO standards [34].

2.5. Bacterial Plant Growth-Promoting Potential

These assays were selected to characterise core functional traits relevant to plant performance and to identify multifunctional bacterial isolates with potential agronomic relevance in kiwifruit systems.

The plant growth-promoting potential (PGP) of 135 isolates was assessed according to Martins et al. [38], including evaluation of phosphate solubilisation, siderophore production, ammonia production, and Indole-3-Acetic Acid (IAA) production. All tests were performed in triplicate. Qualitative phosphate solubilisation was evaluated following Almoneafy et al. [43]. Isolates were grown on GYA medium (GY broth plus 15 g/L agar), and phosphate solubilisation was indicated by the appearance of clear halos around colonies after seven days at 28 °C. Siderophore production was assessed using chrome azurol S (CAS) agar plates, with the detection of orange halos after three days at 28 °C, following Almoneafy et al. [43]. Ammonia production was evaluated as described by Singh and Kumar [44], based on colour change (yellow to brown) upon the addition of Nessler’s reagent to peptone water cultures after four days at 30 °C. Results were categorised into three levels: none (0), low (1), and high (2) production. Isolates were cultured in LB medium supplemented with L-tryptophan (40 µg/mL) at 30 °C, 160 rpm for 48 h. IAA production was quantified using Salkowski’s reagent and colorimetric measurement at 530 nm, following Almoneafy et al. [43]. IAA production was analysed by one-way ANOVA followed by Tukey’s multiple comparisons test (p < 0.05), using GraphPad Prism v8.4.3 (San Diego, CA, USA).

2.6. Antagonistic Activity Against Pseudomonas syringae pv. actinidiae

Antagonism assays were performed to determine whether native phyllosphere bacteria could inhibit Psa growth, providing insight into their potential role in natural disease suppression and biocontrol.

The antagonistic activity of 54 isolates against Psa strain CFBP7286 (Biovar 3) was evaluated using three methods: cross-streak, disk diffusion, and agar well diffusion. Overnight cultures of isolates and Psa were prepared in LB broth at 28 °C. All tests were performed in triplicate, and all three assays were applied to the same set of isolates. Antagonistic activity was analysed by one-way ANOVA followed by Tukey’s multiple comparisons test (p < 0.05), using GraphPad Prism v8.4.3 (San Diego, CA, USA).

In the Cross-Streak Method, following Lertcanawanichakul and Sawangnop [45], Psa was streaked at the centre of LB agar plates, and isolates were streaked perpendicularly. Then the inoculated plates were incubated for six days at 28 °C.

In the Single Disk Method, following Lertcanawanichakul and Sawangnop [45], LB agar plates were seeded with Psa suspension and filter discs impregnated with isolates were placed onto the surface. Inhibition zones were measured after six days at 28 °C.

In the Agar Well Diffusion Method, following Tontou et al. (2016) [46], bacterial inocula (10^8^ CFU/mL) were spotted onto LB agar plates and incubated for 48 h at 28 °C. Plates were then sprayed with Psa suspension (10^6^ CFU/mL) and re-incubated for 48 h at same conditions. The average inhibition area (AIA) was calculated following the equation: AIA = π (R^2^ − r^2^), where R is the radius of the inhibition zone and r is the radius of the colony.

2.7. Copper Tolerance

Copper tolerance assays were performed to evaluate the capacity of culturable phyllosphere taxa to withstand copper-based phytosanitary treatments, a key selective pressure in kiwifruit orchards and relevant for interpreting ecological resilience and biocontrol potential.

Copper tolerance was assessed on mannitol–glutamate yeast extract (MGY) agar following the plating procedure of Tontou et al. (2016) [46], with adaptations. Isolates were spotted on MGY supplemented with copper (II) sulphate pentahydrate (CuSO_4_·5H_2_O) at concentrations of 0.0, 0.8, 1.2, 2.0 and 2.8 mM and incubated at 25 °C for 96 h. The minimum inhibitory concentration (MIC) was defined as the lowest copper concentration at which no visible growth was observed. The 0.8 mM threshold was used as an operational cut-off for classifying isolates as copper-tolerant, consistent with thresholds commonly used in studies of phyllosphere- and Psa-associated bacteria [13,47]. Accordingly, isolates able to grow at ≥0.8 mM were classified as copper-resistant in vitro. Agar plates without added copper were used as growth controls, and all assays were performed in duplicate.

3. Results

3.1. Bacterial Community Structure Based on Culturable Isolates

To characterise the culturable phyllosphere bacteriome of Actinidia chinensis under Psa pressure, the taxonomic composition of bacterial isolates obtained from the four sampled groups (HF, HM, DF, DM) was examined. The culturable bacterial community isolated from kiwifruit leaves exhibited differences in structure across the four sampled groups, with disease status showing the strongest shift. In total, 1058 strains were isolated and grouped into 261 RAPD clusters and assigned to 58 species from 29 genera. RAPD clustering showed high correspondence with 16S-based identification, supporting the validity of this grouping strategy.

Figure 1 summarises the relative frequency of the recovered genera across the different sample types, highlighting both shared and condition-specific taxa. Because each condition was represented by a single composite sample, group comparisons are presented descriptively, and no formal statistical testing was performed. Five genera—Curtobacterium, Microbacterium, Pantoea, Pseudomonas, and Sphingomonas—were consistently detected across all sample types, representing the core culturable bacteriome of A. chinensis leaves. Among these, Pseudomonas and Pantoea were the most abundant genera across diseased plants (DF and DM), where Pseudomonas reached up to 56.5% and Pantoea up to 36.1%. In contrast, healthy plants (HF and HM) exhibited a more even distribution, with Actinomycetota genera such as Curtobacterium, Microbacterium and Frigoribacterium contributing substantially to community composition. Beyond the dominant taxa, additional shifts in the relative representation of several genera were observed across plant health status and sex. Sphingomonas and Erwinia tended to show higher relative frequencies in diseased plants, whereas Enterobacter and Pluralibacter were proportionally more represented in female plants under both health conditions. In addition, Frigoribacterium was particularly prominent in healthy female plants, contributing markedly to their overall community composition (Figure 1).

Several genera were exclusive to specific sample types: Chitinophaga, Enterococcus, and Terrabacter were found only in healthy female (HF) plants; Aurantimonadaceae (unclassified), Dermacoccus, and Geodermatophilus were exclusive to healthy male (HM) plants; Bacillus, Hafnia-Obesumbacterium, Pectobacterium, and Raoultella were unique to diseased female (DF) plants; and Brevundimonas, Cellulosimicrobium, Kocuria, and Streptococcus were exclusive to diseased male (DM) plants.

3.2. Phylogenetic Insights into the Culturable Bacteriome

To assess how major bacterial lineages were distributed across plant conditions, the phylogenetic composition of the representative isolates assigned to RAPD clusters was examined. A total of 218 representative isolates from RAPD clusters were sequenced for the 16S rRNA gene [∼1500 bp], resulting in the identification of 58 species distributed across 29 genera (Supplementary Table S1). Most species belonged to the phylum Pseudomonadota, particularly in diseased plants, where they accounted for 96.1% of isolates in DF and 87.0% in DM (Figure 1). The genus Pseudomonas was particularly abundant in both diseased groups, reflecting its dominance under canker pressure.

In contrast, healthy leaves exhibited a more phylogenetically diverse community, with Actinomycetota representing 28.2% and 45.1% of the isolates in HF and HM, respectively. Genera such as Curtobacterium, Microbacterium and Frigoribacterium were notably more frequent in healthy plants than in diseased ones. Minor proportions of Bacillota and Bacteroidota were also detected, the latter being exclusive to healthy female plants, suggesting that some low-abundance groups may be sensitive to disease-associated shifts in community structure.

3.3. Distribution of Multifunctional Plant Growth Promoting Bacteria and Copper Tolerance

From the 135 bacterial isolates screened for plant growth-promoting (PGP) traits—indole-3-acetic acid (IAA) production, siderophore production, ammonia production, and phosphate solubilisation—and copper tolerance, 25 isolates exhibited all four PGP traits and were classified as multifunctional candidates. These isolates were recovered from all four plant conditions, with a higher proportion originating from DM and HF plants (Figure 2A, Table S2), indicating that multifunctionality is not restricted to healthy tissues and can persist under disease pressure.

Analysis of the taxonomic distribution of the multifunctional isolates revealed a strong concentration in a limited number of genera (Figure 2B). Pantoea and Pseudomonas together accounted for more than half of the multifunctional strains. Additional multifunctional isolates belonged to Enterobacter, Frigoribacterium and Rhodococcus, whereas the remaining genera (Buttiauxella, Brevundimonas, Microbacterium, Pluralibacter and Erwinia) were represented by a single isolate each, indicating a highly uneven distribution of multifunctionality across taxa. Notably, several genera that were abundant in the broader culturable community (e.g., Sphingomonas and Curtobacterium) were not represented among the multifunctional isolates, highlighting that numerical dominance does not necessarily translate into functional dominance.

Copper tolerance was widespread among the culturable phyllosphere community. Using the 0.8 mM cut-off commonly applied to phyllosphere bacteria associated with Psa [13,47], 46.7% (63/135) were classified as copper-sensitive (MIC < 0.8 mM), while 53.3% (72/135) were resistant (MIC ≥ 0.8 mM) (Table S2). Most resistant isolates displayed MIC values of 0.8–1.2 mM, although a small number reached 2.0 mM, indicating substantial tolerance within the phyllosphere community.

Several commensal taxa exhibited combinations of plant growth-promoting traits and copper tolerance, as revealed by the functional screening (Table S2). These included isolates taxonomic affiliated with Pantoea (closest match: P. agglomerans KWT1371, KWT1372, KWT1355), Enterobacter (closest match: E. asburiae KWT1367) and Erwinia (closest match: E. iniecta KWT9), which maintained relevant functional traits under copper stress conditions. Importantly, multifunctionality and copper tolerance were not strictly associated, as several multifunctional isolates were copper-sensitive, highlighting the independence of these functional dimensions within the phyllosphere community.

3.4. Antagonistic Activity Against Pseudomonas syringae pv. actinidiae and the Potential Application of Bacillus pumilus in Kiwifruit Disease Management

To determine whether native phyllosphere bacteria could suppress Psa activity, the antagonistic potential of 54 isolates was assessed using three complementary inhibition assays. Overall, approximately 10% of the isolates inhibited Psa infection in at least one method, indicating that antagonistic capacity is widespread within the culturable phyllosphere microbiota. In the cross-streak assay, two isolates, Pseudoclavibacter helvolus (KWT786) and Bacillus pumilus (KWT1381), produced visible growth inhibition against Psa, typically manifesting as narrow inhibition zones along the streak interface. The disk diffusion assay confirmed antimicrobial activity in a subset of isolates, which produced inhibition halos with an average diameter of 0.94 ± 0.12 mm. The agar well diffusion method provided quantitative estimates of antagonistic strength, with Average Inhibition Area (AIA) values ranging from 221.19 to 228.17 mm^2^, and several isolates displaying strong inhibition comparable to known biocontrol agents.

Among the tested strains, B. pumilus (KWT1381) consistently displayed strong antagonism across all three methods. In the agar well diffusion assay, this isolate produced a clear inhibition halo measuring 22 mm in diameter, corresponding to an AIA of 228.17 mm^2^. Notably, some potential antagonist isolates overlapped with the multifunctional subset described in Section 2.3, indicating that plant growth-promoting traits and pathogen suppression may co-occur in certain native strains, thereby highlighting particularly promising native candidates for integrated biocontrol strategies.

4. Discussion

Relative frequency analysis showed that Pseudomonas was the most frequently recovered genus within the culturable bacteriome in diseased plants, reaching more than half of the total isolates in diseased plants. In contrast, healthy samples displayed a more balanced distribution among genera, with Frigoribacterium and Curtobacterium contributing significantly to the community structure. Overall, disease presence was associated with a marked decrease in community evenness and diversity, with increased recovery of opportunistic genera such as Pseudomonas and Pantoea. These results are consistent with the Anna Karenina Principle in microbial ecology, which posits that stressed hosts tend to harbour more variable and less even microbial communities [32,48]. Our culture-based findings therefore support the broader pattern of Psa-associated dysbiosis previously reported in kiwifruit and other plant systems [25,26,27,31].

The timing of sampling (late spring) was particularly relevant, as in Portugal, Psa populations expand during spring under favourable temperatures (12–18 °C) and high humidity, while consecutive summer heat days lead to a sharp reduction in diversity and dominance of a few adapted clones [9]. Sampling during this peak of pathogen activity provided a representative snapshot of how strong infection pressure reshapes the culturable phyllosphere microbiota.

To contextualise our findings, the isolated bacteriome in this study was compared with the phyllosphere microbiome previously characterised by metabarcoding using the same plant material [33]. Despite methodological differences, both datasets identified Pseudomonadota as the dominant phylum, with Pseudomonas and Sphingomonas as key genera. Our culture-based approach recovered several taxa also reported by Ares et al. (2021) [33], including Curtobacterium, Frigoribacterium, Microbacterium, and Methylobacterium. In contrast, metabarcoding captured a broader diversity, including Hymenobacter and Massilia, reflecting the intrinsic bias of culture-based recovery towards fast-growing or nutritionally versatile bacteria.

Our results complement these studies by providing a culture-based inventory of experimentally validated strains with plant-beneficial traits, thus bridging the gap between sequence-based inference and applied microbial selection. Unlike previous work that relied mainly on predictive functional profiling, such as Correia et al. (2025) [27] and Patterson et al. (2025) [28], the present collection offers concrete candidates for further testing as potential biocontrol agents, particularly under biotic stress conditions. Importantly, this work goes beyond descriptive cataloguing by incorporating a systematic functional screening, allowing the identification of concrete native strains with validated agronomic potential. It is important to note that the community profiles reported here reflect the distribution of taxa within the cultured fraction recovered under the isolation conditions applied, rather than the true relative abundance of bacteria in the natural phyllosphere. Culture-dependent methods are inherently selective and may bias recovery towards fast-growing and nutritionally versatile taxa, while underrepresenting slow-growing, dormant, or media-sensitive bacteria. Differences in growth rates, competitive effects on plates, and the choice of culture media and incubation conditions can therefore influence the apparent frequency of taxa among isolates. Although this limits direct ecological inference, culture-based approaches provide a complementary and application-oriented perspective by enabling functional validation and strain-level characterization, which are not accessible through amplicon sequencing alone.

This convergence with other culture-independent surveys of Actinidia phyllosphere [25,26,27,49] reinforces the view that Psa infection strongly reduces diversity and favours opportunistic taxa such as Pseudomonas. Genera such as Sphingomonas and Methylobacterium, consistently detected and proportionally more represented in healthy plants across both approaches, stand out as putative contributors to resilience. Certain Sphingomonas strains, such as S. sediminicola Dae20, have been shown to enhance systemic resistance against foliar pathogens through the activation of jasmonic acid and ethylene signalling, highlighting their role in promoting plant immunity and potentially stabilising phyllosphere microbiota under stress [50,51]. Similarly, Methylobacterium spp. are known to synthesise phytohormones such as cytokinins, contribute to plant stress alleviation, and exhibit biocontrol activity against several pathogens [52].

Overall, these findings suggest that Psa infection may act as a strong environmental filter, restructuring the culturable fraction of the phyllosphere microbiome in line with the holobiont framework [30,53]. Integrating culture-dependent and molecular approaches, therefore, provides complementary perspectives and helps identify robust microbial candidates for sustainable disease control in kiwifruit. This culture-based validation also resonates with recent perspectives emphasising that preserving and harnessing microbial diversity is central to future plant health strategies, positioning native strain collections as key resources for biocontrol innovation [22].

Regarding the cultivable bacteriome described in item 3.2, the results are consistent with previous studies showing that Psa infection reshapes phyllosphere microbial structure by reducing diversity and favouring Pseudomonadota dominance [25,26,27,33].

At the genus level, Pseudomonas, Curtobacterium, Microbacterium, Pantoea, and Sphingomonas were recovered from all sample types, forming the core culturable microbiota of A. chinensis. These genera have also been consistently reported in culture-independent surveys [33], reinforcing their ecological relevance. The persistent presence of Sphingomonas and Methylobacterium in healthy plants suggests their potential roles in plant protection and microbiome stability, given their known capacity to modulate plant immunity, produce antimicrobial compounds, and enhance plant growth under stress [50,51,52,54].

Several genera were restricted to a single sample type. For example, Terrabacter and Enterococcus were exclusive to HF samples, Geodermatophilus and Dermacoccus to HM, and Raoultella, Hafnia, and Pectobacterium to DF. This pattern indicates that both disease status and plant-associated group differences (HF/HM/DF/DM) are reflected in the culturable microbiota, although pathogen pressure appears to be the main driver [33,53]. Such selective pressure may destabilise the microbiome assemblies, leading to dysbiosis, consistent with models proposing that loss of microbial homeostasis compromises plant resilience [31]. Within the holobiont framework, this disruption extends beyond the plant host, affecting the entire plant–microbiota system [30]. Moreover, although some sex-related differences (HF vs. HM) were detected, disease status was the dominant factor shaping bacterial communities, consistent with host-mediated filtering processes [53].

Although sampling depth was not sufficient to capture the full taxonomic diversity, particularly in diseased plants, the combined RAPD and 16S rRNA profiling was adequate to reveal the dominant trends and taxonomic shifts across conditions. We acknowledge that relying solely on 16S rRNA sequencing limited species-level resolution for a subset of isolates; future work will apply higher-resolution markers to refine taxonomic assignments and strengthen ecological interpretations. These results align well with patterns described in culture-independent surveys, providing complementary culture-based insights into the effects of Psa infection on phyllosphere communities.

The analysis of the distribution of multifunctional PGPBs revealed that these isolates were not restricted to healthy environments and could persist or even proliferate under disease pressure, particularly in male-associated groups (HM/DM; ‘Tomuri’). At the genus level, multifunctionality was strongly concentrated in Pantoea and Pseudomonas, which together accounted for more than half of the multifunctional isolates. Both genera were also widely distributed across all sample types in the broader culturable dataset, reinforcing their ecological relevance within the kiwifruit phyllosphere. Their repeated representation among multifunctional isolates suggests that ecological success in this system is associated with functional versatility, particularly under contrasting host health conditions. Additional multifunctional isolates belonged to Enterobacter, Frigoribacterium and Rhodococcus, indicating that multifunctionality, although unevenly distributed, spans phylogenetically diverse taxa.

Interestingly, Enterobacter (including multifunctional strains) and Klebsiella (within the broader PGPB set)—both affiliated with the Enterobacteriaceae family—were identified among PGPB isolates from both healthy and diseased plants. Their functional versatility, coupled with their presence in copper-tolerant groups (described in Section 2.3), underscores their adaptability to contrasting environmental conditions. By contrast, Sphingomonas and Methylobacterium were most frequently recovered among cultured isolates of healthy samples in terms of abundance but were not represented among the multifunctional isolates, whereas Frigoribacterium was represented by a limited number of multifunctional strains. This suggests that numerical prevalence does not necessarily correspond to functional dominance. Such ecological-functional mapping highlights the resilience and potential of specific genera as bioinoculants in Actinidia cultivation, while also suggesting that diseased groups—particularly DM, may represent potential reservoirs of robust taxa selected under biotic and abiotic stress.

Notably, one Psa isolate tolerated copper up to 2.0 mM, matching the upper resistance levels reported for Greek Psa field populations [13]. To our knowledge, this constitutes the first evidence of highly copper-resistant Psa in Portuguese orchards.

Remarkably, an Erwinia isolate (KWT9; closest match: E. iniecta) displayed the highest resistance detected, with growth up to 2.8 mM (Table S2). This surpasses thresholds previously reported for Psa in Greek orchards (2.0 mM) [13] and exceeds values described for other phyllosphere-associated bacteria [47]. Such exceptional resistance in a commensal taxon highlights the potential of the kiwifruit phyllosphere to serve as a reservoir of high copper tolerance traits within the cultured fraction.

The recurrence of Pantoea (closest match: P. agglomerans) and Curtobacterium flaccumfaciens across both healthy and diseased plants highlights their ecological versatility, while Rahnella and Enterobacter isolates (closest match: R. inusitata and E. asburiae) were consistently associated with healthy plants, supporting their potential role as native bioinoculants. Their presence in both healthy and diseased plants—particularly in male plants—suggests that copper-managed orchards may act as reservoirs of functionally robust taxa capable of persisting under chemical selection pressure.

In comparative terms, the proportion of copper-resistant isolates recovered in this study (53.3%) is strikingly higher than values reported in previous work on plant-associated bacteria. Beltrán et al. (2021) [47] found that only 10–20% of Pseudomonas isolates from stone fruit phyllosphere in Chile were resistant at the 0.8 mM threshold, with the majority classified as sensitive. Similarly, Thomidis et al. (2025) [13] reported copper resistance in Greek Psa populations as a subset phenomenon, with resistant isolates reaching up to 2.0 mM but never representing the majority. Other studies in phytopathogenic bacteria, such as Xanthomonas euvesicatoria [55], likewise described resistance as a minority trait at the field level. In contrast, our results indicate that copper resistance is already predominant in the Portuguese kiwifruit phyllosphere, encompassing both pathogenic and non-pathogenic taxa.

Recent evidence confirms the increasing prevalence of copper-resistant Psa strains in European orchards, particularly in Italy and Greece, with minimum inhibitory concentrations [MICs] reaching or exceeding 1.2 mM, and in some cases up to 2.0 mM [12,13]. Although no published studies have yet reported copper-resistant Psa strains from Portugal, the selective pressure imposed by copper-based treatments in local orchards may favour their emergence. In this context, the identification of native non-pathogenic and multifunctional strains, including E. iniecta, which grew at concentrations exceeding 2.0 mM, reaching up to 2.8 mM in the case of strain KWT9, while Curtobacterium flaccumfaciens tolerated copper at levels up to 2.0 mM (Table S2)—highlights their added value as candidates for integrated management in high-copper environments.

These findings underscore that the kiwifruit phyllosphere acts as a reservoir of beneficial and copper-resilient bacteria adapted to agrochemical stress. The multifunctionality and resilience of these strains may reflect evolutionary responses to plant-derived selection cues, stress factors, and intermicrobial interactions within the phyllosphere. Such traits reinforce the value of culture-dependent approaches for identifying native microbial resources with agronomic potential and support the integration of microbial functional traits into sustainable disease management frameworks.

Genera such as Pantoea, Pseudomonas, and Enterobacter emerge as particularly promising for biotechnological use, combining plant growth promotion, stress resilience, and potential disease suppression. Their presence across different health statuses and plant sexes suggests broad ecological plasticity, making them suitable candidates for microbiome-based interventions in diverse orchard conditions. This supports the idea that microbial adaptability to host environments is a key feature of successful plant-associated taxa [29].

Our findings complement recent studies by Fu et al. (2024) [26] and Correia et al. (2025) [27], which used metagenomic and transcriptomic data to infer microbial responses to Psa infection. While those studies highlighted functional potential, our culture-based methodology provides in vitro validation of traits, such as copper tolerance and PGP activities and identifies concrete candidate strains for future in planta testing. This functional confirmation expands the microbial toolbox for tailored microbiome-based interventions. The resulting collection of resilient and functionally diverse strains represents an asset for applied microbiome research.

Finally, the discovery that common phyllosphere bacteria, including well-documented plant associates like Pantoea agglomerans but also less-studied taxa such as Rahnella spp. and Microbacterium testaceum, may act as reservoirs of copper resistance warrants further investigation. While most studies on the microbial resistome in agricultural settings have focused on soil [56,57,58], our results suggest that the selection pressure from copper-based treatments creates a similar genetic pool of resistance in the phyllosphere. Copper resistance in some plant pathogens is well-documented [14], but our findings highlight that the issue extends beyond specific phytopathogenic taxa. The presence of high-level resistance in ubiquitous, non-pathogenic species raises concerns about the potential for horizontal gene transfer, which can facilitate the dissemination of resistance traits to other members of the microbial community, including emerging pathogens [59]. The occurrence of resistance in genera like Pantoea, which are well-known for their plant-associated lifestyle and potential for gene exchange, is particularly concerning [47,60]. Consequently, the kiwifruit phyllosphere may contribute to persistence and dissemination of copper resistance, with broader implications for sustainable disease management across a range of crops [13]. While these findings highlight the adaptive capacity of phyllosphere communities, the potential application of copper-tolerant strains requires careful consideration. Any future use will necessitate a dedicated biosafety evaluation, including assessments of genetic stability and the potential for horizontal transfer of resistance determinants under field conditions.

Although MIC-based assays provide a useful comparative framework to screen copper tolerance among isolates, in vitro thresholds measured in MGY medium do not directly translate into copper concentrations or bioavailability on leaf surfaces, which are influenced by formulation type, application regime, environmental conditions, and phyllosphere microhabitats. Therefore, the 0.8 mM cut-off used here should be interpreted as a pragmatic classification criterion for laboratory screening rather than a direct proxy of field exposure or environmental risk. Future work should integrate formulation-specific assays and in planta validation under realistic copper regimes to better assess ecological and agronomic relevance.

The detection of both highly resistant Psa and exceptionally resistant non-pathogenic isolates such as E. iniecta underscores the dual ecological role of copper tolerance in the phyllosphere. On the one hand, it facilitates pathogen persistence under chemical management; on the other, it highlights the adaptive capacity of commensal taxa that may be harnessed as biocontrol allies in copper-managed environments. However, copper resistance in Psa has been linked to mobile genetic elements, including integrative conjugative elements (ICEs) and plasmids carrying copABCD and cus/czc, whose transfer has been demonstrated both in vitro and in planta [14]. This highlights the need for vigilance when deploying copper-tolerant commensals as biocontrol agents [59] and underscores the broader ecological implications of copper resistance in phyllosphere microbial communities.

After evaluating the isolates for antagonistic activity and differences between plant conditions, we observed that Pseudomonas and Pantoea isolates showed a higher proportion of active strains in the DM group, while healthy female plants contributed fewer antagonistic isolates. These patterns suggest that pathogen pressure may enrich antagonistic taxa in certain host niches. Notably, potential antagonist isolates also exhibited copper tolerance or overlapped with the multifunctional subset, reinforcing their potential as resilient candidates for integrated disease management strategies. Overall, these antagonism assays represent a primary in vitro screening step to pre-select candidates for subsequent in planta validation. Any downstream application will require biosafety assessment, particularly for Enterobacterales/Enterobacteriaceae candidates.

B. pumilus is already employed in commercial biocontrol formulations targeting a range of phytopathogens in crops such as soybean, maize, strawberry, and sugarcane. However, to our knowledge, this is the first report documenting its antagonistic potential against Psa in kiwifruit. This finding highlights the opportunity to broaden the repertoire of biocontrol candidates available for Actinidia cultivation and reinforces the importance of culture-based approaches for identifying novel applications of well-characterised microbial strains.

The observed antagonism supports B. pumilus as a promising candidate for further evaluation within integrated disease management strategies for kiwifruit, particularly in scenarios where copper-resistant Psa strains emerge, and consumer demand for agrochemical-free solutions intensifies. Although B. pumilus KWT1381 did not exhibit copper tolerance, its strong antagonistic effect against Psa supports its potential as a complementary agent in microbiome-based strategies, particularly valuable under reduced-copper or agrochemical-free management systems. These antagonistic potentials complement previous reports of biocontrol against Psa using Aureobasidium pullulans or Actinobacteria [46,61], situating our findings within a growing portfolio of microbial solutions.

The presence of active strains in both healthy and diseased plants suggests that natural antagonists may contribute to microbiome-mediated suppression of Psa, even under pathogen pressure. Genera such as Pantoea and Pseudomonas have previously been reported as promising biocontrol candidates against diverse phytopathogens, and our results extend their relevance to the kiwifruit–Psa pathosystem. Notably, several antagonistic isolates originated from diseased plants, indicating that pathogen-challenged environments may act as reservoirs of functionally robust strains adapted to stress.

Importantly, vine sex was not independent from cultivar in the sampling design (female vines corresponded to ‘Hayward’ and male vines to ‘Tomuri’). Therefore, differences observed between HF/HM/DF/DM groups should be interpreted as group-level patterns rather than evidence of a cultivar-independent ‘sex effect’. In addition, because sampling relied on a single composite sample per group, inter-individual variability could not be quantified and statistically replicated comparisons between groups were not possible. Future work, including multiple cultivars per sex and replicated sampling, will be required to disentangle these factors and strengthen ecological inference.

5. Conclusions

This study provides an integrative assessment of the culturable phyllosphere bacteriome of Actinidia chinensis under pressure from Pseudomonas syringae pv. actinidiae (Psa) infection. By combining taxonomic, ecological and functional analyses, we demonstrate that disease status strongly influences microbial structure, reducing diversity and promoting the dominance of opportunistic Pseudomonadota such as Pseudomonas and Pantoea. These compositional shifts align with the Anna Karenina Principle, suggesting that Psa-induced dysbiosis disrupts the holobiont balance of the phyllosphere.

Despite this disruption, a subset of bacterial isolates exhibited multifunctional plant growth-promoting traits, while copper tolerance and in vitro antagonistic activity against Psa were also detected across the culturable collection. Notably, these functional traits can co-occur in native strains, supporting the identification of promising candidates for integrated biocontrol strategies. Among the isolates showing in vitro antagonism, Bacillus pumilus (KWT1381) consistently inhibited Psa, broadening the repertoire of genera with biocontrol potential and representing a novel application for this well-known biocontrol agent.

In addition, we report a strikingly high prevalence of copper resistance within the phyllosphere microbiota, with more than half of the isolates classified as resistant at the 0.8 mM cut-off, including both pathogenic and non-pathogenic taxa. Notably, this study documents for the first time in Portugal, a highly copper-resistant Psa isolate (up to 2.0 mM), and an exceptionally resistant non-pathogenic Erwinia isolate (closest match: E. iniecta 2.8 mM), underscoring both the potential for resistance maintenance and the opportunity to harness robust native taxa under copper pressure. These findings underscore the strong selective pressure imposed by copper-based treatments and highlight the dual ecological role of copper tolerance: enabling pathogen persistence while also selecting commensal taxa that may be harnessed for biocontrol under copper stress.

Overall, the results highlight the value of culture-dependent approaches in identifying native microbial resources with agronomic potential. As this work is based on in vitro functional assays, the findings should be interpreted as an initial screening step rather than direct evidence of field efficacy. Accordingly, future work will prioritise in planta validation, the development of synthetic consortia, and field trials under copper-managed conditions to bridge the gap between laboratory screening and effective, sustainable disease management in kiwifruit production. Together, these results provide a functionally validated culture collection that can serve as a foundation for targeted microbiome-based interventions in future kiwifruit disease management.