Synthetic microbial communities derived from native niches enhance the high-temperature adaptability of Pinus yunnanensis seedlings

Abstract

The microbiome rewilding hypothesis suggests that understanding and reconstructing the microbial communities lost through domestication is vital for enhancing seedling quality and adaptability. Therefore, we investigated the structure and assembly of symbiotic microbial communities associated with Pinus yunnanensis, the most significant conifer species in southwestern China, and its dwarf variant, P. yunnanensis var. pygmaea. Subsequently, the functions of these microbes were characterized by inoculating dominant microbes and constructing synthetic communities, and examined their colonization status post-inoculation using amplicon sequencing. The results indicate that: (a) microbial communities are primarily differentiated by niche (soil, roots, needles), followed by geographical location, while trunk form variation has a minimal impact; (b) fungi are influenced by both chemistry and geographical factors, showing dispersion limitation, while bacteria are mainly affected by chemistry, exhibiting homogeneous diffusion; (c) single endophyte inoculation has a neutral to slightly negative impact on seedling growth but enhances resistance to high temperatures; (d) synthetic microbial communities (SynComs), constructed based on the strains' origin and initial functional screening, enhanced seedling growth and provided better protection against high-temperature stress than single strains. (e) one SynCom (SC5), composed of the dominant root isolates Phialocephala sp. (Fun6) and Paraburkholderia sp. (Bac7), significantly increased total seedling biomass by 62% and improved thermotolerance. These findings enhance our understanding of the symbiotic microbial communities of P. yunnanensis and demonstrate the potential of using specific SynComs, such as SC5, as bio-inoculants to improve seedling quality and stress tolerance in nursery production.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40793-026-00924-5.

Introduction

Plants and their associated microorganisms collectively constitute a "holobiont," engaging in mutual interactions and coevolution [1]. In recent years, the intensification of artificial cultivation and domestication processes has led to significant alterations in plant traits, including the enhancement or loss of specific traits, which may be correlated with changes in microbial composition. The microbiome rewilding hypothesis proposes that restoring key members of the diverse ancestral microbial communities lost during domestication and industrialization can improve the health of both plants and animals [2]. Theoretically, achieving rewilding can be facilitated by adhering to a straightforward roadmap: identifying beneficial microbial communities from wild ancestors in their native habitats, quantifying their beneficial impacts on multiple generations of modern crops through experimental validation, and elucidating the plant genetic factors underlying microbial colonization [2]. However, how the breeding process influences the assembly of the microbiome, particularly across different plant niches (e.g., roots vs. needles) and in the context of plant domestication, is an area that requires further investigation. A deeper understanding of these assembly rules could significantly enhance the success of rewilding approaches. The domestication process, however, has been shown to reduce the diversity of plant-associated microbial communities and disrupt beneficial plant-microbiome alliances, a phenomenon often referred to as part of the "domestication syndrome" [3].

Accumulating evidence has demonstrated that plant phenotypic traits and genetic variation can significantly shape the associated microbial communities. Host genotype has been shown to influence the composition of both root and phyllosphere microbiomes in various plant species, including Arabidopsis, barley, and poplar [4, 5]. Specific plant traits, such as immune system status, root exudation profiles, and developmental stages, act as selective filters that determine which microbial taxa can colonize different plant compartments [6, 7]. For instance, salicylic acid signaling pathways have been found to modulate the assembly of root microbiota by affecting the abundance of specific bacterial taxa [6]. Furthermore, plant circadian rhythms can influence the rhythmicity of rhizosphere microbiota, indicating a dynamic and reciprocal relationship between host physiology and microbial communities [8]. Understanding these host-microbe interactions is crucial for predicting how plant domestication and breeding may inadvertently alter the symbiotic microbiome.

Conifers are prominent components in north temperate biomes and constitute over 39% of the world’s forests [9]. The Northern Hemisphere conifers do not follow a latitudinal diversity gradient and show a unique distribution pattern, that is, most species are distributed in the mountainous areas at middle latitudes [10]. Despite the mystery of the distribution of coniferous forests remaining uncovered, they play a crucial role in ecological services and the economy by providing timber and wood products [11–13], enhancing carbon sequestration [14, 15], and promoting and facilitating ecological restoration [16, 17] in mid-latitude mountainous regions. Pinus yunnanensis Franch. is a typical endemic coniferous species that constitutes the bulk of subtropical coniferous forests in the mountainous regions of Southwest China. It ranks among the top ten dominant tree species in China in terms of both area and volume [18], serving as a primary tree species for afforestation and timber production in the region.

As a key species for afforestation, P. yunnanensis has been subject to artificial cultivation and selective breeding to improve seedling quality and timber yield. Superior families and individuals are selected for propagation, followed by nursery cultivation before outplanting for afforestation purposes. While this domestication process has promoted the application of improved varieties, large-scale artificial selection and intensive cultivation have also introduced several challenges. Recent studies have documented that trunk distortion is a prominent phenomenon in P. yunnanensis, with the proportion of trees exhibiting twisted trunks reaching approximately 48% in progeny testing forests [19]. This trunk distortion significantly affects growth traits, particularly plant height and diameter at breast height, resulting in volume loss rates ranging from 18.06% to 56.75% [19]. These findings indicate that domestication may also be associated with challenges during the seedling stage, such as slow early growth and increased disease susceptibility, which can further restrict seedling production and afforestation success.

The Yunnan Plateau, situated in the western part of southwest China, serves as the central growth region for P. yunnanensis and is identified as a core drought-prone area, characterized by seasonal drought conditions with recurrent trends [18]. Furthermore, as a temperature-sensitive species, the effect of temperature on P. yunnanensis growth shifts from positive to negative along a cold-to-warm climate gradient within its distribution range [20]. Higher temperatures, decreased precipitation, alterations in cloud cover, and frequent wildfires pose threats to tropical montane pine forests by overriding the environmental heterogeneity that typically acts as a buffer against catastrophic fires in these ecosystems [21].

In order to adapt to heterogeneous environment, P. yunnanensis exhibits high genetic diversity, with its morphological characteristics displaying distinct variation patterns across different ecogeographical contexts [22]. Taking trunk form as an example, P. yunnanensis with a straight trunk is considered normal, while twisted and spiral trunk forms also occur within the stand. Notably, P. yunnanensis suffers from degradation, manifested by a high proportion of twisted trunks, leading to the emergence of mutated variants such as P. yunnanensis var. pygmaea. P. yunnanensis var. pygmaea maintains its dwarfism even when grown in suitable habitats, exhibiting no prominent main trunk and forming a shrub-like appearance [23] (Fig. 1, Supplementary Fig. 1). The reasons for the trunk form differentiation in pines remain unknown at present. They may be caused by environmental factors such as wind, Earth's rotation, injuries, directional aspects, exposure to the sun and moon, and soil conditions, or they could be the combined effects of genetics and the environment [24–26]. Alternatively, it might be that the environment enhances the manifestation of genetic traits in phenotypes [26]. As the proportion of curved and twisted individuals gradually increases, it severely restricts the production of P. yunnanensis seedlings and diminishes their value for afforestation as well as timber use post-planting.

Fig. 1.

An overview of the experimental workflow for this study

Improving the quality of P. yunnanensis seedlings can be approached from both breeding and cultivation perspectives. Due to the large genome size, high complexity, and the lack of an established genetic system in pine species, genetic improvement of P. yunnanensis is challenging [27]. Given these challenges, improving cultivation practices from a microbial perspective offers a promising alternative. Recent advances in plant-microbiome research have demonstrated that beneficial microorganisms can enhance nutrient uptake, stress tolerance, and overall plant performance. A recent global meta-analysis demonstrated that endophytic inoculation significantly promotes seed germination, seedling growth, and nutrient uptake in trees, supporting the potential of leveraging symbiotic microbial communities to enhance reforestation and ecological restoration efforts [28]. Studies have demonstrated that plant growth-promoting bacteria can enhance drought tolerance in tree seedlings through biochemical, physiological, and biometrical parameters [29]. Furthermore, synthetic microbial communities (SynComs) have been developed to support drought tolerance in tree species, with tailored bacterial consortia outperforming single-strain inoculants [30]. Therefore, conducting research on the symbiotic system of P. yunnanensis and developing microbial-based strategies represents an important approach to effectively enhance seedling cultivation quality, increase afforestation survival rates, and ensure its wood properties.

In this study, we identified areas in Yunnan where P. yunnanensis coexisted with P. yunnanensis var. pygmaea. We evaluated microbial community differences and drivers by analyzing soil, roots, and needles of these pines. We isolated and re-inoculated dominant endophytes based on high-throughput abundance, and verified their functions in synthetic communities. We aimed to: (a) identify the primary drivers of microbial community assembly and quantitatively compare the magnitude of each driver; (b) assess the similarities and differences in the microbial community structures of soil, roots, and needles; (c) reveal the biological functions of dominant endophytes; and (d) construct synthetic communities, quantitatively evaluate their colonization patterns in different plant niches using amplicon sequencing, and comprehensively assess their biological functions in promoting seedling growth and high-temperature resistance.

Materials and methods

Site description and sample collection

This study comprises two parts: firstly, collecting pine trees with different trunk forms in the field to investigate microbial communities, and analyzing the driving forces behind microbial assembly qualitatively and quantitatively; secondly, isolating dominant microorganisms and inoculating them into P. yunnanensis seedlings to explore their functions (Fig. 1).

Samples were collected in Jianchuan (JC), Xiangyun (XY), and Xishan (XS) in Yunnan Province (April 13–17, 2022), at altitudes of 2000–2300 m. These areas have a subtropical plateau monsoon climate with small annual temperature variations, large daily differences, wet (May-Oct) and dry (Nov-Apr) seasons, and significant vertical temperature changes [31].

Three 15 m × 15 m plots were laid at each site. Each sample plot needed to include three types of pine trees with a diameter at breast height (DBH) of 6–10 cm, indicative of their 10 to 15 years of growth: namely, the straight-trunked P. yunnanensis (SP), the twisted P. yunnanensis (TP), and P. yunnanensis var. pygmaea (PP), all of which are included to facilitate the sampling process. Trees within 5 m of each other were selected for soil, needle, and fine root sampling (Supplementary Fig. 1). When collecting fine roots, the taproots of pines were first found from the bottom of pines, and the fine roots with diameter less than 2 mm were collected along the taproots of pines. The mature and healthy needles were collected. Rhizosphere soil was collected from root-dense areas within 10–30 cm depth. Fine roots were cut with 75% alcohol-sterilized scissors and placed in frozen tubes. Three bulk soil samples (10–30 cm depth) were collected as controls. All samples were stored in dry ice.

Chemical properties of roots, needles, and soil

Collected samples were divided into three parts: air-dried for chemical analysis, stored at -80 °C for microbial analysis, and used for microorganism isolation. Roots, needles, and soil were dried to constant weight at 80 °C, powdered using an MM400 ball mill (Retsch, Germany), and sieved to 0.2 mm. The chemical properties, encompassing organic carbon, total nitrogen, total phosphorus, and total potassium, were determined by following the methods described in the literature [32–35].

DNA extraction and sequencing from field experiment

After being rinsed with sterile water to remove adherent soil particles and sonicated for 10 min [36], the fine roots and needles were surface-disinfected with 0.5% sodium hypochlorite (NaClO, bleach) for 5 min and subsequently with 75% ethanol (CH₃CH₂OH) for another 5 min, and then washed with sterile water. The final wash water was inoculated onto tryptic soy agar (TSA) and potato dextrose agar (PDA) media to verify the success of surface sterilization.

For high-throughput sequencing, total DNA was extracted from 200 mg of fine roots and needles using the CTAB method [37], and from 0.5 g of soil using a commercial kit (FastDNA® SPIN Kit, MpBio). Bacterial diversity was analyzed using nested PCR to amplify the V5-V7 regions of the 16S rRNA gene with primer pairs 799F-1392R and 799F-1193R [4, 38]. Fungal diversity was analyzed by amplifying the ITS1 region using primers ITS1F and ITS2R [39]. The PCR system and amplification procedure followed the method described in reference [40]. Sequencing was performed on the Illumina MiSeq PE250 platform by Shanghai Majorbio, with raw sequences deposited in the Sequence Read Archive (PRJNA1184499 and PRJNA1185421). In this study, the sequencing results are presented in the form of OTU, which is conducive to more strains corresponding to the sequence alignment.

The raw FASTQ sequencing data underwent quality filtering [41], merging, denoising, and the generation of representative sequences, with all these processes being carried out by the Quantitative Insights Into Microbial Ecology (QIIME v1.9.1) tool [42]. Subsequently, the cleaned reads were processed to remove primer sequences and clustered into Operational Taxonomic Units (OTUs) with a 97% similarity threshold using UPARSE v11 [43]. These reads were annotated and compared against the Silva (v138) and Unite (v8.0) databases for bacterial and fungal identification, respectively, using the RDP classifier with a 70% confidence threshold.

Microbial isolation, correction, and comparison with high-throughput sequencing

For microbial isolation, fresh root and needle samples were collected from the same trees and immediately placed into sterile cryotubes. The cryotubes were then transferred into dry ice for rapid cooling and temporary preservation during transport to the laboratory. This approach was intended to quickly lower metabolic activity and minimize nucleic acid degradation, rather than to dehydrate or freeze-dry the tissues. Upon arrival, the samples were immediately processed for microbial isolation using fresh tissue sections following standard surface sterilization and culturing protocols.

To isolate endophytic bacteria and fungi from pine tree roots and needles, we first surface-sterilized and crushed the samples. Then, we conducted serial dilutions and inoculated them onto tryptone soy agar (TSA) for bacteria and potato dextrose agar (PDA) for fungi. After incubating at 28 °C for 14 days, colonies were monitored daily and purified by re-streaking 5–6 times. Bacterial 16S rRNA genes were amplified using primers 27F and 1492R [44], while fungal ITS regions were amplified using primers ITS5 and ITS4 [45]. PCR products were sequenced and processed using BioEdit 7.2.5 [46], then classified using Silva (for bacteria) and Unite (for fungi) databases. A total of 788 isolates were obtained from pine tissues, comprising 446 bacterial strains (93 from needles and 353 from roots) and 342 fungal strains (83 from needles and 259 from roots). To connect our cultivation efforts with the field community data, we first identified the top 50 most abundant bacterial and fungal OTUs from the high-throughput sequencing data of the field-collected root and needle samples. These top 50 OTUs, along with their relative abundances in roots, needles, and rhizosphere soil, are provided in Supplementary Table S1. For the matching process, we performed local BLAST searches using the V5-V7 regions of bacterial 16S rRNA gene sequences and the ITS regions of fungal sequences obtained from our isolates against the representative sequences of the top 50 OTUs. Isolates showing ≥ 96% sequence identity were considered matches. The matching results, including OTU IDs, isolate IDs, sequence identity percentages, and their abundances across different niches, are presented in Table 1. Detailed alignment results between isolated strains and representative OTU sequences are provided in Supplementary Fig. S7. Ultimately, 9 fungal and 7 bacterial isolates were chosen for functional validation and stored at -80 °C in glycerol stocks. Among the selected isolates, Bac3 and Fun1-Fun4 were originally isolated from needles, while the remaining strains were isolated from roots.

Table 1.

Bacterial and fungal isolates compared with OTUs and their abundance in different niches

OUT ID	Isolate ID	Phylum	Genus	Sequence Identity	Needles (%)	Roots (%)	Rhizosphere soil (%)
OTU13226	Bac1	Proteobacteria	Burkholderia	359/372 (96%)	0.00	0.37	0.70
OTU13219	Bac2	Proteobacteria	Achromobacter	369/372 (99%)	83.79	76.89	0.03
OTU14928	Bac3	Proteobacteria	Achromobacter	305/314 (97%)	0.00	0.11	0.00
OTU14942	Bac4	Proteobacteria	Stenotrophomonas	379/379 (100%)	0.11	0.08	0.00
OTU4981	Bac5	Firmicutes	Bacillus	377/377 (100%)	0.10	0.01	0.29
OTU16203	Bac6	Proteobacteria	Pantoea	377/377 (100%)	0.00	0.53	0.00
OTU4855	Bac7	Proteobacteria	Paraburkholderia	372/372 (100%)	0.01	0.57	0.90
OTU3746	Fun1	Ascomycota	Aspergillus	219/226 (96%)	2.48	0.00	0.00
OTU4	Fun2	Ascomycota	Penicillium	243/243 (100%)	1.05	0.14	0.07
OTU3203	Fun3	Ascomycota	Alternaria	235/235 (100%)	1.00	0.69	0.04
OTU3745	Fun4	Ascomycota	Alternaria	203/210 (96%)	0.93	0.57	0.01
OTU4281	Fun5	Ascomycota	Fusarium	217/218 (99%)	0.88	0.76	0.16
OTU2272	Fun6	Ascomycota	Phialocephala	244/244 (100%)	0.80	12.64	0.14
OTU3183	Fun7	Ascomycota	Cladosporium	224/224 (100%)	0.74	0.67	0.36
OTU11	Fun8	Ascomycota	Penicillium	244/245 (99%)	0.00	1.91	0.65
OTU2056	Fun9	Ascomycota	Phialocephala	246/246 (100%)	0.00	1.73	0.01

Pine trees seedling cultivation and single-strain inoculation

P. yunnanensis seeds, collected in Kunming City, Yunnan Province, were surface-sterilized with 0.5% NaClO for 20 min and rinsed 3–5 times with sterile water. The soil used for all inoculation experiments was a mixture of humus soil and vermiculite (2:1, v/v), which was sterilized by autoclaving at 121 °C for 60 min. This sterilization cycle was repeated three times, with a 24 h interval between each cycle, to ensure thorough sterility. The disinfected seeds were sown in sterile plastic cups (9 cm diameter) containing 150 mL sterile water and 50 g sterilized soil. The cups were then sealed with parafilm and placed in an artificial climate chamber set to a photoperiod of 12 h light (10,000 lx) at 26 °C and 12 h dark at 20 °C, with relative humidity maintained at 75% for 60 days (Supplementary Fig. 2). After 60 days, seedlings with similar height (around 2 cm variation) were selected for subsequent experiments to minimize initial variation and ensure experimental homogeneity (Supplementary Fig. 2). Seven bacterial strains and nine fungal strains were activated and fermented in TSB media at 37 °C and PDB media at 28 °C with shaking at 180 rpm. The bacterial solution was adjusted to OD600 = 0.5 and 0.1 mL of mycelium were resuspended in 10 mL of sterile PDB medium. Each plant received a 10 mL inoculum of bacteria or fungi near the rhizosphere. Controls received sterile media. After 30 days, seedlings were divided into normal (26 °C) and high-temperature stress (45 °C for 8 h, characterized by needle dehydration and wilting, with plants surviving after rewatering) (Supplementary Fig. 2). Growth parameters (plant height, ground diameter, dry weight, and root system traits) were measured only under normal conditions (26 °C), because under acute heat stress (45 °C for 8 h) rapid wilting and water loss would confound genuine growth effects. Therefore, only physiological indicators (peroxidase, superoxide dismutase, and malondialdehyde) were assessed under heat stress. For growth parameter measurement under normal conditions, five replicates (n = 5) were randomly selected to record plant height, ground diameter, and dry weight; the root systems were scanned using an LA‑S root scanner to measure root volume, root surface area, and other parameters (Wseen, China) [47]. For physiological assessment, ten additional replicates (five at 26 °C and five at 45 °C) were used to measure needle peroxidase (POD), superoxide dismutase (SOD), and malondialdehyde (MDA) levels using kits from Solarbio (China). At the same time point, to further confirm the establishment of inoculated strains, we performed re-isolation from plant tissues. Inoculated isolates were consistently re-isolated.

Design of synthetic microbial communities in vitro and measurement of their effects

Based on the composition of strains and the role of individual strains, six synthetic microbial communities (SynComs), SC1 to SC6, were assembled and tested for their impact on seedling growth under normal (26 °C) and high temperature (45 °C) conditions (Table 2). SC1 contained all 16 isolates, while SC2 and SC3 comprised only bacteria or fungi, respectively. SC4 consisted of isolates with predominantly negative effects on seedlings (Bac2, Bac3, Bac4, Fun3, Fun4). SC5 was a two-strain combination (Bac7 and Fun6) showing promising enzyme activities. SC6 comprised isolates with predominantly positive effects under high temperature (Bac1, Bac6, Bac7, Fun4-Fun7, Fun9). To establish the synthetic microbial community (SynCom), bacterial and fungal cells were quantified following the procedure described above. Equal volumes of each isolate were pooled to generate the SynCom working suspension. Two-month-old sterile P. yunnanensis seedlings were then inoculated with 10 mL of the suspension into the rhizosphere. The control group received 10 mL of a 1:1 mixture of sterilized TSB and PDB media. The number of replicates for growth morphological parameters was maximized based on available seedlings: CK (7), SC5 and SC6 (8 each), and other SynComs (5 each). Root scanning and enzyme activity assays were performed with five replicates for all treatments. The same response variables described above were measured. For amplicon sequencing, three independent samples from the SC4-6 treatment were randomly selected and subjected to high-throughput sequencing of pine needles, roots, and rhizosphere soil. The raw reads have been deposited in the NCBI Sequence Read Archive under BioProject PRJNA1188772.

Table 2.

Composition of synthetic communities

Syncoms	Composition	Number of strains
SC1	Bac1-7; Fun1-9	16
SC2	Bac1, 2, 3, 4, 5, 6, 7	7
SC3	Fun1, 2, 3, 4, 5, 6, 7, 8, 9	9
SC4	Bac2-4; Fun3, Fun4	5
SC5	Bac7; Fun6	2
SC6	Bac1, Bac6-7; Fun4-7, Fun9	8

Data and statistical analyses

To correct for uneven sequencing depth, reads from 90 samples were rarefied to the minimum number of reads per sample (24,312 for bacteria and 26,675 for fungi) for downstream analyses. Analysis was based on these subsampled reads. Principal coordinate analysis (PCoA) based on Bray–Curtis dissimilarities visualized microbial structures [48], followed by PERMANOVA to determine significance of composition variation [49]. Partial Least Squares Path Modeling (PLS-PM), using the R package "plspm", was used to calculate both direct and indirect effects of geography, climate, soil, and plant chemistry on microbial communities [50, 51]. Geography included longitude, latitude, and altitude. Climate data (annual mean temperature and precipitation) was extracted using R packages "geodata" and "terra" [52, 53]. Soil and plant chemistry encompassed K, C/N, C/P, and N/P. Fungi and bacteria were represented by PCoA1 and PCoA2. Beta nearest taxon index (βNTI) was calculated using R packages "ape", "ICAMP", "picante", and "NST" [54]. A neutral community model estimated stochasticity effects on community assembly using a nonlinear least-squares method [55]. Mann‒Whitney U tests were conducted in R to investigate significant differences among βNTI groups [56]. P. yunnanensis growth and resistance indicators outside the 1.5-fold interquartile range were considered as outliers [57] and excluded from statistical analysis. Independent samples t-tests were conducted to assess significant differences among multiple treatment groups and the single control group. Heatmaps of the top 20 OTUs were plotted using the R package pheatmap [58]. Other diagrams were plotted in GraphPad Prism 9 (GraphPad Software, Inc., CA, USA).

Results

Profile of high-throughput sequencing from field experiments

In total, 90 samples, including 27 roots (3 locations × 3 trunk forms × 3 replicates), 27 needle samples, 27 rhizosphere soil samples, and 9 bulk soil samples (3 locations × 3 replicates), underwent high-throughput sequencing to analyze their microbial communities. The rarefaction curves basically reached saturation at the minimized reads, and the number of operational taxonomic units (OTUs) in the soil was higher than that of the endophytic OTUs in the plants (Supplementary Fig. 3).

In terms of bacterial composition, Proteobacteria is the most abundant bacterial phylum among plant endophytes, accounting for approximately 95% in needles and 90% in roots. Among them, OTU13219 (belonging to the Alcaligenaceae family) is the absolutely dominant OTU, with a proportion of about 77% in roots and 84% in needles. In soil, the top three dominant phyla are Proteobacteria, Acidobacteriota, and Actinobacteriota (Supplementary Fig. 4). In terms of fungal composition, Ascomycota and Basidiomycota are the dominant phyla, collectively accounting for more than 80% of the fungi present in all samples. The microbial communities at the genus level exhibit significant differences among samples from different niches (Supplementary Fig. 4).

Driving forces of symbiotic microorganisms in pine trees

PCoA was used to compare microbial communities across different niches (rhizosphere soil, bulk soil, roots, needles), sites (TJ, JC, XY), and pine trunk forms (TP, PP, SP). The results showed that fungal and bacterial communities could be classified by niche (Fig. 2a, e). Different niches explained 63.82% of fungal and 17.69% of bacterial variation, both of which were highly significant (Fig. 2d, h; P ≤ 0.001). Pine trunk forms showed no distinction in their microbial communities (Fig. 2c, g), consistent with the results of Adonis tests (Fig. 2d, h; P > 0.05). Additionally, locations and the interactions between niches and locations also had a significant effect on microbial communities (P_bacteria ≤ 0.05, P_fungi ≤ 0.001). Rhizosphere and bulk soils showed no significant differences (Supplementary Fig. 5).

Fig. 2.

Effects of different niches, locations (Jianchuan, JC; Xiangyun, XY; Xishan, XS) and trunk forms (straight-trunk P. yunnanensis, SP; twisted P. yunnanensis, TP; P. yunnanensis var. pygmaea, PP) on the community composition of bacteria (a-d) and fungi (e–h). PCoA was used to visualize the similarity in microbial community composition between different niches (a, e), locations (b, f), and trunk forms (c, g).The Venn diagram displays the variation in the bacterial (d) and fungal (h) communities explained by each given variable and their shared variation (PERMANOVA; * P < 0.05; ***P < 0.001) based on the samples excluding bulk soil samples

PLS-PM was used to explore relationships among location, climate, soil/plant chemistry, and microbial communities (Fig. 3). Both bacterial and fungal communities were mainly determined by chemistry (with path coefficients of 0.79, P ≤ 0.001 for bacteria, and 0.84, P ≤ 0.001 for fungi) (Fig. 3). Additionally, fungal communities were influenced by geography (path coefficient = 0.24, P = 0.001) (Fig. 3b). Furthermore, plot climate affected bacterial communities, while fungal communities were more sensitive to plot location (Fig. 3c, d). The total effects analysis (Fig. 3c, d) showed that the influence of geography on bacterial communities (path coefficient = -0.1) and climate on fungal communities (path coefficient = -0.01) were negative but weak, and neither reached statistical significance (P > 0.05).

Fig. 3.

Partial least squares path modelling (PLS-PM) show the effects of geography on microbial communities via climate, and soil chemistry (a, b), and standardized total effects of these variables on microbial communities (c, d). The Goodness of Fit (GOF) was utilized to evaluate the model. R² indicates the proportion of variance in the dependent variables explained by the model. The thickness of the arrows represents the strength of the path coefficient's influence. Red and blue lines depict positive and negative correlations, respectively. Solid and dashed lines indicate significant and nonsignificant correlations, respectively, with asterisks denoting significance levels (**P < 0.01; ***P < 0.001). Soil/Plant chemistry refers to the elements C, N, P, K and their stoichiometry

The βNTI, which measures the average phylogenetic distance between the taxa within a community, was used to assess microbial community assembly. Significant βNTI differences were found between fungal and bacterial communities from different niches and across sample locations (Fig. 4a, b, e, f). However, no significant variations were observed among different pine tree trunk forms (Fig. 4c, g). Furthermore, the proportions of deterministic processes (including homogeneous selection with βNTI < -2 and heterogeneous selection with βNTI > 2) and stochastic processes (including dispersal limitation with |βNTI|< 2 and RC > 0.95; homogenizing dispersal with |βNTI|< 2 and RC < -0.95; and drift with |βNTI|< 2 and RC < 0.95) varied among different microbial communities (Fig. 4d, h). Bacterial communities were influenced by homogenizing dispersal and drift, while fungal communities were affected by dispersal limitation and drift (Fig. 4d, h).

Fig. 4.

The effects of different niches, locations, and trunk forms of pine trees on microbial communities’ assembly and diversity. a-c, e–g βNTI calculations of phylogenetic turnover in niches (needles, roots and rhizosphere soil), locations (JC, TJ and XY) and trunk forms (SP, TP and PP) samples. *P < 0.05; **P < 0.01; ***P < 0.001 in the Dunn's multiple comparisons test (n = 352). d, h The relative influence of five community assembly processes in microbial communities with different niches, locations, and trunk forms of pine trees

Functional validation of dominant strains isolated from pine trees

The results suggest that sample niches are the primary factor influencing microbial community structure. We compared 788 isolates (446 bacteria, 342 fungi) with the top 50 abundant OTUs from roots and needles (Table 1, Supplementary Fig. 6). Seven bacterial and nine fungal isolates matched OTUs from sequencing with > 96% similarity (Table 1, Supplementary Fig. 7). These isolates were selected for subsequent inoculation experiments.

Following the inoculation of a single bacterial isolate, the bacteria exhibited either neutral or adverse effects on the growth of P. yunnanensis compared to the non-inoculated control (Fig. 5). Except for a slight increase in ground diameter observed with Fun9, the microorganisms had neutral or slightly negative effects on the growth indicators of P. yunnanensis. Notably, the bacteria significantly reduced root surface area and total dry weight. Additionally, bacteria still exhibit a neutral-to-negative impact on the resistance of P. yunnanensis seedlings at normal temperatures. However, many fungi demonstrate significant increases in POD and SOD levels, although MDA also increases significantly (indicating damage to the needles). Of course, under high-temperature conditions, some isolates of bacteria and fungi show increased POD and SOD levels along with decreased MDA, which enhances the resistance of P. yunnanensis. But overall, fungi have a better effect on improving the resistance of P. yunnanensis seedlings compared to bacteria.

Fig. 5.

The effects on the growth and resistance of P. yunnanensis after inoculation with a single isolate, compared to the non-inoculated control. POD (peroxidase), SOD (superoxide dismutase), and MDA (malondialdehyde) are indicated with or without a ‘T‑’ prefix. Indicators with a “T‑” prefix (T‑POD, T‑SOD, T‑MDA) were measured under high‑temperature stress (45°C for 8 h). All other indicators were measured under normal conditions (26°C). B1-B7: Bac1-Bac7; F1-F9: Fun1-Fun9. The red dashed line represents the average value of each indicator for the control group. Error bars represent mean ± SD. *P < 0.05; **P < 0.01; *** P < 0.001 in the independent-samples t-test compared to the control group

Functional validation of synthetic microbial communities

We constructed six synthetic communities, named as SC1 through SC6, by assembling dominant isolates (Table 2). SC1 contained all 16 isolates. SC2 comprised seven bacterial isolates. SC3 comprised nine fungal isolates. SC4 consisted of three bacterial (Bac2, Bac3, Bac4) and two fungal isolates (Fun3, Fun4). SC5 contained two isolates (Bac7 and Fun6). SC6 consisted of three bacterial (Bac1, Bac6, Bac7) and five fungal isolates (Fun4-Fun7, Fun9).

After inoculating the synthetic communities, SC5 and SC6 significantly increased the biomass and root surface area of P. yunnanensis seedlings while significantly reducing MDA, demonstrating protective effects on the seedlings (Fig. 6). Additionally, SC1-3 significantly increased POD levels at high temperatures, also exhibiting protective effects against heat stress. Although SC4 consists of isolates with predominantly negative effects, its overall impact was neutral. We found that SC5 and SC6 significantly increased the biomass and root surface area of P. yunnanensis seedlings while significantly reducing MDA, demonstrating protective effects on the seedlings (Fig. 6).

Fig. 6.

Effects of synthetic microbial communities on growth and physiological responses of P. yunnanensis seedlings. POD (peroxidase), SOD (superoxide dismutase), and MDA (malondialdehyde) are indicated with or without a ‘T‑’ prefix. Indicators with a “T‑” prefix (T‑POD, T‑SOD, T‑MDA) were measured under high‑temperature stress (45°C for 8 h). All other indicators were measured under normal conditions (26°C). The red dashed line indicates the average value of each indicator for the control group. Error bars represent mean ± SD.*P < 0.05; **P < 0.01; ***P < 0.001 in the independent-samples t-test compared to the control group

To understand the colonization patterns, we sequenced the microbial communities of P. yunnanensis seedlings inoculated with SC4-6 using amplicon sequencing (Fig. 7). Both bacterial and fungal communities exhibited significant niche differentiation (Fig. 7a, b, P = 0.001). All re-inoculated isolates, except for Bac4 and Fun4 which had an abundance of less than 0.5% after SC4 re-inoculation, were detected. Notably, Fun5 was abundant post-SC6 in needles, roots, and soil (Fig. 7c, d). Compared to pine needles, the remaining isolates exhibited higher abundance in both rhizosphere soil and roots. Fungi generally colonized more effectively than bacteria, exhibiting higher abundance and consistent identities (Fig. 7, Supplementary Fig. 8). Additionally, non-inoculated fungi were also present, potentially originating from seed endophytes or environmental sources.

Fig. 7.

The colonization of synthetic microbial communities SC4-6 in the rhizosphere soil, roots, and needles of P. yunnanensis after inoculation. PCoA analysis revealed the differentiation of bacterial communities (a) and fungal communities (b) across different niches within P. yunnanensis. The heatmap illustrates the colonization patterns of the top 20 bacterial (c) and fungal (d) OTUs across various niches. The red-highlighted OTU information includes the genus, the corresponding inoculated isolate, and the name of the synthetic community that contains the isolate. The bar plot on the right displays the average abundance of each OTU across all samples. N: Needles, R: Roots, S: Rhizosphere soil. MM: Methylobacterium-Methylorubrum, BCP: Burkholderia-Caballeronia-Paraburkholderia, ANPR: Allorhizobium-Neorhizobium-Pararhizobium-Rhizobium, ES: Escherichia-Shigella

Discussion

The concept of microbiome rewilding aims to restore lost microbial communities through domestication to enhance host growth, development, and adaptability [2]. This study explores the composition, functions, and assembly of symbiotic microorganisms in P. yunnanensis. It is now widely accepted that the establishment of microbial communities associated with plants is a non-random process governed by specific assembly regulations [59]. Factors that influence the microbial community structure in plants encompass soil types [60, 61], plant parts [62], plant host species or genotypes [4, 5], the plant immune system [6], plant development stages [7], seasons [63, 64], and even plant circadian rhythms [8].

In this study, we analyzed the drivers of microbial assembly in P. yunnanensis and its variety P. yunnanensis var. pygmaea, considering both qualitative (locations, niches, trunk forms) and quantitative (geography, climate, chemistry) factors. We found that niches are the main driver, followed by location, while trunk forms have little impact (Fig. 2, 4). Soil and plant chemistry, which represents the microenvironment of the plants, significantly influenced the microbial communities of P. yunnanensis (Fig. 3). The weak negative total effects of geography on bacteria and climate on fungi (Fig. 3c, d), although not statistically significant, hint at a potential trend where these factors influence community composition in a direction opposite to the main gradient. This observation is consistent with the predominant role of local chemistry in shaping microbial communities across our study sites, reinforcing that geographical and climatic factors play secondary roles at this spatial scale.

Recent studies on microbial community formation and differentiation in various niches, such as in grape, agave, Boechera stricta, and Arabidopsis thaliana, show overlap between rhizosphere and phyllosphere microbial communities [62, 65, 66], with soil serving as a reservoir for these microorganisms [65, 67]. Experiments with sterile A. thaliana and the vertical transmission pattern of oak seeds have both demonstrated the occurrence of reciprocal migration and ectopic colonization, as well as niche differentiation and divergent transmission routes, between rhizosphere and phyllosphere bacterial communities [62, 68]. We observed that the impact of different factors on microbial communities varies with the study scale. Globally, soil fungi are mainly influenced by climate, followed by soil and spatial patterns. [69]. Plant-associated fungal communities are also shaped by biogeography [1]. However, in our study with sites 100–400 km apart, niche-based differentiation of plant symbiotic microbial communities often exceeds that caused by location. This is due to similar climate types and minor variations in precipitation and temperature at this scale [70]. Certainly, the driving forces for fungal and bacterial communities are not entirely consistent. Fungi are more influenced by geography and dispersion limitation (Fig. 4), promoting endemic fungal populations in various locations [1]. In contrast, bacteria, dominated by Proteobacteria, are mainly influenced by homogeneous dispersion, resulting in similar bacterial community structures across different locations (Fig. 2–4).

Despite previous studies showing no differences in microbial communities between straight-trunk and twisted P. yunnanensis [26], the current study, including P. yunnanensis var. pygmaea with a notable morphological distinction, also found no significant differences. We hypothesize that the occurrence of P. yunnanensis var. pygmaea could be due to a variant of P. yunnanensis shaped by environmental influences like soil, wind, and climate, or alternatively, it might arise from the complex interaction between ecological and genetic factors leading to enhanced phenotypic differences [24–26]. Importantly, the absence of significant microbial differentiation is attributed to the high variability in microbial communities across different individuals, reflecting the impact of environmental heterogeneity (Fig. 2, Supplementary Fig. 4). This finding aligns with the conclusion that microbial community structures are influenced by soil and plant chemistry, as well as stochastic processes, indicating pine trees' adaptability to diverse and heterogeneous environments (Fig. S2 and S4).

The assembly of root microbial communities involves a two-step selection process mediated by root exudates, leading to enrichment transitions from rhizosphere soil to the root surface and interior [1]. Phyllosphere microbial communities can share a common origin with soil microbial communities, which can act as a primary reservoir [65, 67]; however, they are also assembled from other sources, including the atmosphere and other surrounding plants [1], contributing to their overall diversity. Based on observed significant differences in microbial community structure among different niches but not among pine trees with different trunk forms in this study, we chose to focus on dominant endophytes from roots and needles for functional validation.

The genera Bacillus (Bac5) and Burkholderia (Bac1) are well known to contain numerous plant growth-promoting rhizobacteria (PGPR) strains that maintain allelopathic interactions with host plants, aiding in resistance to abiotic and biotic stresses and fostering growth [71]. Recent studies have also found that members of the Achromobacter (Bac2, Bac3), Stenotrophomonas (Bac4), Paraburkholderia (Bac7) and Pantoea (Bac6) genera can promote plant growth or enhance plant resistance [72–75]. In this study, we found that inoculating single bacteria generally had a neutral to slightly negative impact on the growth of P. yunnanensis seedlings, but offered protection under high temperatures (Fig. 5). Notably, Bac2, the most abundant, had minimal effect on the growth and resistance of P. yunnanensis under high temperatures but reduced protection under ambient temperatures. Fungi exhibited similar trends, with a neutral to slightly negative impact on seedling growth post-inoculation, yet demonstrated a more protective effect under high temperatures compared to bacteria (Fig. 5). The interactions among plants, symbiotic microorganisms, and the environment form complex networks encompassing genetics, biochemistry, physics, and metabolism [76]. These interactions vary with the host, environment, and microbial species. However, overall, inoculating dominant microorganisms exhibited a protective effect against high temperature stress, indicating that the microbial community tends to counteract adverse environmental factors, consistent with the environmental attributes of P. yunnanensis growing in infertile plateau mountainous regions [77]. Post-inoculation, P. yunnanensis seedlings exhibited slightly negative traits, suggesting several hypotheses: Firstly, dominant endophytes may mitigate plant diseases by occupying niches and forming new phenotypes through trait reorganization, unrelated to host resistance and growth [76, 78]. Secondly, this phenomenon may be linked to a soil negative feedback mechanism, where negative feedback observed in controls disappears in sterile soil, enhancing their performance [79]. Additionally, significant individual differences in growth and resistance among seedlings may obscure the role of microorganisms. In conclusion, further exploration of microorganisms at genetic, metabolic, and transcriptional levels is crucial to elucidate their interactions with the host. We noted that some strains originally isolated from needles (e.g., Fun1-Fun4) showed lower relative abundances in roots based on field sequencing data (Table 1). However, after inoculation, these strains were detected in SynCom-treated plants (Fig. 7) and induced phenotypic changes (Fig. 5, 6), suggesting that they can establish and function in the rhizosphere despite their original niche preference. This aligns with the view that microbial colonization is influenced by both strain characteristics and host-mediated selection, and that niche differentiation primarily manifests as differences in relative abundance rather than absolute presence or absence. The functional consequences of such niche-specific recruitment, however, remain to be further elucidated.

We acknowledge that our use of 97% OTU clustering (UPARSE) rather than ASV-based methods such as DADA2 represents a methodological choice worth noting. While DADA2 offers single-nucleotide resolution [80], our experimental design was specifically built around directly comparing sequencing data with Sanger-sequenced cultured isolates. The V5–V7 region we targeted (~ 200–250 bp) provides reliable genus-level assignment—a resolution not meaningfully enhanced by ASV splitting given read length constraints. Moreover, ASV methods can split sequences from a single strain into multiple variants due to minor intra-strain variations, which would have complicated our isolate-matching workflow. Recent benchmarking studies confirm that UPARSE and DADA2 yield comparable diversity estimates and support similar ecological conclusions [81]. For fungal ITS analysis, 97% OTU clustering has even been recommended to ensure homogeneous results across replicates [82]. Looking forward, as longer-read sequencing becomes more accessible, future studies on key strains will benefit from ASV-based workflows combined with genomic data. For the present study, however, the OTU-based pipeline provided a direct and effective path from community profiling to functional validation.

Based on this, we constructed synthetic cross-kingdom microbial communities and assessed their functions. Compared to single-strain systems, synthetic functional microbial communities exhibit characteristics such as communication (unidirectional, bidirectional, or multidirectional via signal molecule exchange, detection, and mutual response), consensus gene expression, and labor division among members [83, 84]. In this study, inoculating synthetic microbial communities shifted the growth indicators of P. yunnanensis from negative (as seen with single isolates) to neutral or positive, enhanced seedling growth under normal conditions and improved physiological tolerance to acute heat stress (Fig. 6). This exhibited the superiority of synthetic communities over single isolates. Recent studies also show that community inoculation, due to microbial diversity and synergistic interactions, enhances biological fertility and bioremediation efficacy [85]. In Arabidopsis, host preference is only discernible within microbial communities [86]. Regarding colonization after synthetic community inoculation, fungal colonization was more effective than bacterial colonization in this study. All inoculated fungi were detected with relatively high abundance, indicating good symbiotic relationships with P. yunnanensis. In contrast, bacterial colonization varied widely, possibly related to strong priority effects, which requires further verification [87]. Additionally, we observed microbial community differentiation across different ecological niches, consistent with previous findings. Interestingly, apart from SC6, where Fun5 (Fusarium) was enriched in needles, most enriched fungi in other needles belonged to Aspergillus and Talaromyces genera, not our inoculated isolates. Whether this is related to vertical seed transmission needs further investigation. Furthermore, whether the inoculated communities from rhizosphere soil spread to the phyllosphere over time also requires experimental validation. Despite these interesting colonization patterns, we acknowledge a methodological limitation in our SynCom experiment: non-inoculated controls were not sequenced, which limits our ability to definitively exclude background signals. However, recent evidence demonstrates that even after rigorous sterilization treatments including heat, fungicides, and tissue culture, plants retain low levels but diversity-rich endophytic communities that can be transmitted through seeds, making it impossible to create truly microbe-free plants [88]. Given that sequencing non-inoculated controls would not completely eliminate background ambiguity, and because our primary goal was functional validation, sequencing was performed only on SynCom-treated plants, where the establishment and relative abundance of inoculated strains could be directly assessed. The strong correspondence between inoculated isolates and dominant OTUs recovered from treated plants (> 97% identity; Fig. 7), together with their high relative abundances, provides reasonable evidence of successful colonization. Future studies should incorporate non-inoculated controls and quantitative methods such as strain-specific qPCR.

Furthermore, it's noteworthy that a larger member count doesn't guarantee better performance; member complexity can impose a significant metabolic burden, limiting system productivity [89]. In this study, the optimal SC5 had just two members: a dominant fungus, Fun6 (Phialocephala), in the roots, and a subdominant bacterium, Bac7 (Paraburkholderia). Both showed good colonization after inoculation. Identifying key, synergistic microorganisms can mimic complex synthetic communities' functionalities [90], challenging artificial community design. While our results demonstrate that SynCom inoculation enhances seedling growth and physiological tolerance to acute heat stress under controlled conditions, whether these benefits translate into improved long-term adaptability in heterogeneous field environments requires further investigation.

Conclusions

Understanding and reconstructing microbial communities lost due to plant domestication can enhance host growth, development, and adaptability. Elucidating microbial assembly processes is crucial for microbiome rewilding. This study explored the composition, function, and assembly of symbiotic microorganisms in P. yunnanensis and its variety, P. yunnanensis var. pygmaea. Results showed that ecological niches primarily drive microbial community differentiation, followed by location, while trunk form has little impact. Chemistry significantly affects P. yunnanensis's microbial community. Functionally, dominant single microbes generally had neutral-to-slightly negative effects on seedling growth but protected against high temperatures. In contrast, a synthetic microbial community positively enhanced growth and resistance. These findings demonstrate the potential of specific SynComs to improve seedling growth and physiological stress tolerance, offering a promising strategy for improving seedling quality in nursery production.

Author contributions

CL, CN and XY conceived the idea and designed the experiments. HG and PL performed experiments and data analysis and wrote the paper. MD assisted in sampling and the chemical analysis. CL, HG, CN, XY, DJ and WF contributed to the data interpretation and revision. All authors discussed the results and approved the final version of the manuscript. CL funded the work.

Funding

This work was supported by Applied Basic Research Programs of Science and Technology Department of Yunnan Province (202401AT070297, 202401BD070001-107, 202201AT070023), National Nature Science Foundation of China (32360394), Yunnan Provincial First-Class Discipline Construction Fund for Forestry at Southwest Forestry University (LXXK-2025D02), Yunnan Revitalization Talent Support Program Young Talent Project (XDYC-QNRC-2022–0250), Southwest Forestry University Research Project (112113), and Research Innovation Fund for Graduate Students of Yunnan Provincial Department of Education(2023Y0734), Yunnan Graduate Tutor Team Building Project (2024–61).

Data availability

The raw sequence data obtained in this study have been deposited in the NCBI database under the BioProject ID PRJNA1184499, PRJNA1185421, PRJNA1188772.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.