Native plants change the endophyte assembly and growth of an invasive plant in response to climatic factors

ABSTRACT

Climate change, microbial endophytes, and local plants can affect the establishment and expansion of invasive species, yet no study has been performed to assess these interactions. Using a growth chamber, we integrated the belowground (rhizosphere soils) and aboveground (mixture of mature leaf and leaf litter) microbiota into an experimental framework to evaluate the impacts of four native plants acting as microbial inoculation sources on endophyte assembly and growth of the invasive plant Ageratina adenophora in response to drought stress and temperature change. We found that fungal and bacterial enrichment in the leaves and roots of A. adenophora exhibited distinct patterns in response to climatic factors. Many fungi were enriched in roots in response to high temperature and drought stress; in contrast, many bacteria were enriched in leaves in response to low temperature and drought stress. Inoculation of microbiota from phylogenetically close native plant species (i.e., Asteraceae Artemisia atrovirens) causes the recipient plant A. adenophora (Asteraceae) to enrich dominant microbial species from inoculation sources, which commonly results in a lower dissimilar endophytic microbiota and thus produces more negative growth effects when compared to non-Asteraceae inoculations. Drought, microbial inoculation source, and temperature directly impacted the growth of A. adenophora. Both drought and inoculation also indirectly impacted the growth of A. adenophora by changing the root endophytic fungal assembly. Our data indicate that native plant identity can greatly impact the endophyte assembly and host growth of invasive plants, which is regulated by drought and temperature.

IMPORTANCE

There has been increasing interest in the interactions between global changes and plant invasions; however, it remains to quantify the role of microbial endophytes in plant invasion with a consideration of their variation in the root vs leaf of hosts, as well as the linkages between microbial inoculations, such as native plant species, and climatic factors, such as temperature and drought. Our study found that local plants acting as microbial inoculants can impact fungal and bacterial enrichment in the leaves and roots of the invasive plant Ageratina adenophora and thus produce distinct growth effects in response to climatic factors; endophyte-mediated invasion of A. adenophora is expected to operate more effectively under favorable moisture. Our study is important for understanding the interactions between climate change, microbial endophytes, and local plant identity in the establishment and expansion of invasive species.

INTRODUCTION

Plants are exposed to highly abundant and diverse soil microbes (1), which can help plants absorb nutrients, promote plant growth, and improve plant stress resistance (2–4). In introduced ecosystems, the local microbial community interacts with invasive plants (5) and has been considered to be an important driver of plant invasion (6). In recent years, increasing evidence has shown that microbial endophytes can be a key factor in the competitive success of invasive species (7, 8). Therefore, successful plant invasion greatly depends on endophytes obtained from neighboring plants (9–11). We assumed that the varied invasion resistance or facilitation of the local plant community reported previously (12, 13) may be mediated by native plants functioning as microbial inoculation sources of invasive plants in the introduced range.

Endophytes associated with invasive plants originate either from the plant species’ native (by cointroducing) or nonnative range (by horizontal transmission from local surroundings), or both (14, 15). Some invasive species, e.g., ectomycorrhizal woodies (16) and legumes (17), can coinvade with their endophytes. However, most introductions, especially dicot hosts, are from seeds (7). Seeds commonly contain a very limited number of microorganisms (18, 19); therefore, most alien plant species introduced by seeds invade exotic ecosystems with a very limited abundance of microbes [but also see references (16) and (17)].

The availability and quality of endophytes in the introduced range thus may greatly contribute to their range expansion. For example, by transplanting seeds, Stanton-Geddes and Anderson found low densities of suitable rhizosphere bacteria beyond the edge limit expansion range of the legume Chamaecrista fasciculata (20). Nonetheless, successful plant invaders may not be limited by a lack of mutualistic fungi; for example, Moora et al. indicated that alien palm seems to have replaced “lost” fungal mutualists from its native range with new mutualists in the introduced ranges (21). It is evident that some plant invasions depend on endophytes obtained from neighboring plants. For example, neighborhood tree species can act as a reservoir of fungal inoculum to alleviate mycorrhizal limitation at the invasion front of Pseudotsuga menziesii (9). Fungal endophytes associated with the roots of nurse cushion species have positive effects on plant invasion in an alpine ecosystem (10). The plant Centaurea stoebe, only in combination with diverse root endophytic fungi obtained from soils in the expansion range, may have an increased defense against specialized soil-borne organisms (22). Therefore, it is very interesting to characterize how local plant species impact the enrichment of endophytes of invasive plants in the introduced range.

Classical experiments have been designed to explore the interaction between local soil microbes and invasive plants (23, 24). Although phyllosphere microbiota, the communities of bacteria and fungi living on and in plant leaves, profoundly affect the health of their hosts and the entire ecosystem (25–27), their role in plant invasion has been less explored. Indeed, aboveground microbes interact with belowground microbes to impact plant growth in a community (28); moreover, invasive Impatiens glandulifera displays that a positive soil microbial effect can affect foliar endophytes to enhance resistance to herbivory (29). It is necessary to consider both soil and phyllosphere microbiota together in exploring the microbial role in plant invasion.

Interestingly, endophyte community composition and its role in host plant growth largely depend on climatic factors, such as temperature (30) and drought (31–33). For example, endophytic bacterial communities from the roots and stems of Vitis vinifera differ in terms of composition and dynamic response to temperature (30). Root-associated fungi are less sensitive to drought than soil fungi (34); in contrast, a greenhouse drought experiment demonstrated greater sensitivity of bacterial root endophytes than bacteria in soil across 30 angiosperm species (31). Against the background of climate change, there has been increasing interest in the interactions between global changes and plant invasions (35, 36); however, the role of microbial endophytes in plant invasion with a consideration of their variation in the root vs leaf of hosts, as well as the linkages between microbial inoculations, such as native plant species, and climatic factors, such as temperature and drought remains to be determined.

Ageratina adenophora (Sprengel) R.M. King and H. Robinson is a perennial herb in the Asteraceae family, which is native to Central America but is a noxious weed in Asia, Africa, Oceania, and Hawaii (37). Previous reports have indicated that the belowground beneficial microbial association formed in local ranges greatly contributes to its successful invasion (38–40); in contrast, A. adenophora has also been reported to obtain foliar microbes primarily from surrounding plants (41), and its growth is negatively regulated by some foliar fungi (42). In particular, A. adenophora expands its range mainly into disturbance habitats by seeds (37, 43), where the bare habitats may give a chance for dispersal of A. adenophora propagules. This means that the availability and quality of endophytes in the expanded sites should greatly influence the establishment of A. adenophora at the early invasion stage.

It is well-established that endophytic microbiota differ across plant host species (22), and climate change impacts microbial diversity in the natural environment (30–33). We hypothesized that the assembly of the endophytic microbiota in the model invasive plant species (A. adenophora) and host growth are driven by critical interactions between the locally sourced plant-associated microbial community and climatic factors (i.e., temperature and drought). In this study, we thus integrated rhizosphere soil (RS) and leaf tissue as inoculation sources (simulating local microbial exposures) to evaluate how plant identity (conspecific vs heterospecific) impacts endophyte (bacteria vs fungi) assembly in the plant compartment (leaf vs root) and the growth of A. adenophora in response to climatic factors (temperature vs drought). In addition, there is a particularly strong association between the evolutionary relatedness of host plants and endophytic bacterial diversity and composition, which causes highly dissimilar microbiota to produce more positive effects on focal plant growth than highly similar microbiota in plant–soil feedbacks (31). We further predicted that native phylogenetically close plant species (i.e., Asteraceae) as inoculation sources can cause the recipient plant A. adenophora (Asteraceae), due to more similar evolutionary history, to enrich dominant microbial species from inoculation sources, which results in a lower dissimilar endophytic microbiota and thus produces more negative growth effects when compared to non-Asteraceae inoculation.

MATERIALS AND METHODS

Experimental design

The experimental design is shown in Fig. S1: (i) Considering that aboveground microbes interact with microbes below to impact plant growth in a community (28), we used both soil and phyllosphere microbiota as inoculation sources together to explore the endophyte assembly and growth role in plant invasion. (ii) We focused on the microbial effects on focal plants, i.e., the invasive plant A. adenophora. Thus, we exposed A. adenophora to only conspecific and heterospecific soils and leaf tissues [including mature leaf (ML) and leaf litter (LL)] without performing a full-factorial experimental design. (iii) Considering that A. adenophora expands its range mainly into disturbance habitats by seeds (37, 43), where direct competition with native plants is assumed to rarely occur at seedling establishment, we did not include the direct competition of native plants with A. adenophora (4). Four common local plants included three non-Asteraceae plants, i.e., Reinwardtia indica (Linaceae), Arthraxon prionode (Poaceae), Achyranthes bidentata (Amaranthaceae), and one Asteraceae plant Artemisia atrovirens. These native plant species are distributed frequently with A. adenophora at the same site; therefore, we assumed that the potential microbe-mediated interactions between these species can actually occur in the wild (5). Climatic factors involved two temperatures, i.e., 28/24°C (light/dark) (high temperature, H) and 22/18°C (light/dark) (low temperature, L), and two drought powers, i.e., 100 mL/cup sterile water (normal moisture, 0) and 100 mL/cup 20% PEG-6000 solution (drought stress, 20), with an osmotic potential of −4.906 bars to produce drought stress (44). A total of 20 treatments, including five microbial inoculation sources (the combination inoculation of RS, ML, and LL from five plant species) × two temperatures × two moisture contents, were designed in this experiment. Each treatment had 12 replicates.

Sample collection and preparation for inoculations

In July 2019, we collected ML and standing LL, as well as RS, from five randomly selected individuals from each species, including the invasive plant A. adenophora and four native plants R. indica, A. prionodes, A. bidentata, and A. atrovirens from three plots, ~100 m from each other, in Xishan Forest (24°55′34″ N, 102°38′30″ E, 1890 m) of Kunming City, Yunnan Province, China (Fig. S1). The collected samples were transported to the laboratory. For each plant species, the collected mature leaves or leaf litters from five individuals in each plot were mixed as one replicate. Correspondingly, their rhizosphere soils were screened by a 1 mm sieve to remove roots, stones, and other debris and mixed as one replicate. A total of 45 samples [five plant species × three plant sample types (ML, LL, RS) × three replicates] were prepared as inoculation sources and stored at 4°C until utilization. For the inoculation usage, except for LL and RS, which were collected one time and stored at 4°C until utilization, ML was collected and used immediately to inoculate as needed (also see below description).

Inoculation experiment

First, A. adenophora seeds were collected in Xishan Forest (24°30′00″ N, 102°22′12″ E) in April 2018. The surface sterilized seeds (seeds were submerged in 75% ethanol for 30 s and 2% sodium hypochlorite for 3 min and rinsed with sterile water five times) were placed on the water agar plates and germinated for 21 days in an RXZ-380D growth chamber with a size of 59.4 × 59.4 × 197 cm (Ningbo Southeast Instrument Co., Ltd., Ningbo, China) at a temperature of 25/20°C (light/dark), a light intensity of 12,000 lux for a 12 h photoperiod, and a humidity of 60%. Subsequently, seedlings of similar size (shoot length approximately 1 cm) were randomly selected and transplanted into a 700 mL PP cup (one seedling per cup), which contained 50 g of autoclaved basal soil [Pindstrup sphagnum (Denmark) + vermiculite + perlite (vol/vol/vol/=8:1:1)] for further treatments.

Soil inoculation was carried out by placing 5 g of rhizosphere soil as a microbial inoculum 0.5 cm below the surface of 50 g autoclaved basal soil in each plastic cup (at a mass ratio of 1:10), which has been verified to exclude the potential soil nutritional effect (45). To simulate the air dispersal of microbes associated with leaves, we combined the intact ML and standing LL collected previously as one inoculation source. Leaf inoculation was performed as previously described (46). Briefly, on the 25th day after seedling transplanting, one mature leaf and one leaf litter were suspended 3 cm above A. adenophora seedlings with a plastic clip to effectively disperse microbial propagules, e.g., fungal spores or bacterial cells, from leaf tissue to the focal plant A. adenophora (see Fig. S1). Here, we did not inoculate mature and leaf litter separately because we assume that focal plants commonly receive microbes from both of them in the wild. Meanwhile, considering that microbial dispersal from leaf tissue may be related to leaf area, we standardized the suspended leaf tissues to a similar leaf area (~10 cm²) across plant species for inoculation. After inoculation, the seedlings were further cultured for 3 months in growth chambers. The leaf tissue was changed once a week for 6 weeks. Here, LLs were collected one time and stored at 4°C until utilization, while MLs were collected and used immediately to inoculate when we changed the leaf tissues. At harvest, three seedlings were randomly selected from each treatment to determine the enriched endophytic fungi and bacteria in the leaves and roots of A. adenophora by maximum-depth sequencing, and the remaining seedlings were used to determine the biomass.

Molecular sequencing of the microbial community

We focused on the microbial effects on one focal plant, i.e., the invasive plant A. adenophora. Thus, we exposed A. adenophora to only conspecific and heterospecific soils and leaf tissues (including mature leaf and leaf litter) and did not include a treatment without a microbial inoculation source. Therefore, we only characterized the fungi and bacteria associated with host plants as inoculum, as well as the enriched endophytes in A. adenophora under all treatments. For 45 samples [five plant species × three plant sample types (ML, LL, RS) × three replicates] collected as inoculation sources for the first time (July 2019), all these samples were immediately extracted for DNA and sent to the company for DNA sequencing after collection to represent the original microbial communities in inoculation materials. The inoculated leaf tissue was changed once a week for 6 weeks (see above description), and these subsequent tissues were not subject to sequencing; therefore, we ignored the possible changes in the microbial community associated with leaf litter over the storage period in the refrigerator, as well as those associated with mature leaves in the wild over the season.

The term “endophyte” has been controversial since it appeared (47); here, we referred to endophytes as fungi and bacteria that live inside a plant (48). Therefore, all leaves and roots of A. adenophora from each treatment were surface sterilized before determining the endophytic fungi and bacteria by DNA sequencing. Surface sterilization: plant tissues were submerged in 75% ethanol for 2 min and 2% sodium hypochlorite for 30 s and rinsed with sterile water five times.

We sequenced these samples by maximum-depth sequencing as previously described, in which the target gene is first labeled with a unique identifier (UID) and is amplified together with the DNA fragment to eliminate amplification bias and base errors in the PCR [see references (49) and (50)]. DNA extraction, PCR amplification, and Illumina sequencing of the microbial community were completed by Kangce Biotech Co., Ltd. (Wuhan, China). The total genomic DNA of the samples was extracted by the DP321 Kit (Tiangen, China). Primers ITS1 (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2 (5′-GCTGCGTTCTTCATCGATGC-3′) were used to amplify the ITS region of fungal DNA, and primers 515 (5′-GTGCCAGCAGCCGCGG-3′) and 806 (5′-TACCAGGGTATCTAATC-3′) were used to amplify the V4 region of bacterial DNA. The amount of DNA library in mature leaves was 500 ng, that in leaf litters was 100 ng, and that in rhizosphere soil was 150 ng. Each 50 µL PCR included 10 µL of 5× Hifi buffer, 1 µL of dNTPs (10 mM), 1 µL of each primer (10 µM), 1 µL of KAPA HiFi DNA polymerase (1 U/µL), 10 ng of template DNA, and 36 µL of ddH₂O. The amplification was run in a Bio-Rad T100 thermocycler (Bio-Rad, USA) (3 min at 95°C, followed by 28 cycles of 30 s at 95°C, 30 s at 60°C, 45 s at 72°C, and 10 min at 72°C). The resulting PCR products (~300 bp) were purified using AMPure XP Beads (Beckman, USA) and double-end sequenced at 2 × 150 bp on an Illumina HiSeq Sequencer (Illumina, USA).

Paired-end reads were preprocessed using Trimmomatic software v0.36 (51) to remove adapters and filter out low-quality sequences (average quality score <Q20). Flash software v1.2.11 (52) was used to assemble the paired-end reads. Multiple identical sequences marked by different UIDs represent the natural copies of microorganisms in each sample, while the sequences with the same UID were merged to remove duplications to quantify the microbial components in DNA samples. Therefore, this method partitioned operational taxonomic units (OTUs) at a unique level. According to the Unite-ITS database v 04.02.2020 (53) and Greengene database v13.8 (54), species annotation with an 80% confidence threshold for fungal OTUs and bacterial OTUs was performed using the Burrows–Wheeler Alignment tool. In total, 165 plant and soil samples were sequenced for their fungal and bacterial communities. However, seven samples failed to obtain fungal DNA sequences, and eight samples failed to obtain bacterial DNA sequences. We analyzed 158 samples of fungi and 157 samples of bacteria. The rarefaction curves indicated that the obtained sequences can represent the majority of the fungal and bacterial communities (Fig. S2). OTUs with relative abundances less than 1/10,000, as well as chloroplast and mitochondrial DNAs, were ignored in subsequent data analysis.

Data analysis

Principal coordinate analysis (PCoA) was used to visualize the microbial similarity among microbial inoculation sources and treatments. Permutational analysis of variance (PERMANOVA) was used to test the microbial community differences. All data used for the community analyses were microbial relative abundance, which represents the original copy number of fungi or bacteria in samples by maximum-depth sequencing [see references (49) and (50)], and distance matrices were constructed based on the Bray–Curtis dissimilarity index. A random forest algorithm was used to identify the enriched fungal and bacterial genera that significantly changed in relative abundance in leaves and roots of A. adenophora in response to climate change.

Due to the high A. adenophora seedling mortality under certain treatments, the dry biomass of seedlings used for microbial sequencing had to be calculated by the linear regression method based on the wet-biomass data to ensure sufficient samples for subsequent statistical analysis (Fig. S3a and b). The response index (RI) was used to evaluate the effects of native plants on the growth of A. adenophora and calculated as follows: (variable_treatments – variable_control)/variable_control (55), where variable_treatments and variable_control represent the seedling aboveground biomass and belowground biomass in each replicate, with native plant treatments (heterospecific exposure) and an A. adenophora control (conspecific exposure), respectively. Nonparametric analysis with the Mann–Whitney U-test was used to identify differences between the treatments and control, and the Kruskal–Wallis test was performed to compare the differences in RI between different microbial inoculation sources under the same hydrothermal conditions. PLS-PM was performed to illustrate the direct and indirect effects of the microbial inoculation source, temperature, and drought on the seedling biomass of A. adenophora. The model was analyzed using the goodness-of-fit statistic.

Visualization of PCoA was performed using CANOCO v5.0 software (56). Bray‒Curtis distance and PERMANOVA were conducted using the functions of “vegdist” and “adonis2” in the R v.3.6.3 package “vegan,” respectively (57). Nonparametric analysis, including the Mann–Whitney U-test and Kruskal–Wallis test, was performed using the R (v4.1.3) “coin” package (58). PLS-PM was executed using the function “inner plot” in the R (v4.1.3) “plspm” package (59). Random forest was constructed using the online analysis platform (https://www.cloudtutu.com). The remaining figures were drawn in GraphPad Prism v7.0 (GraphPad Software, Inc., San Diego, CA, USA).

RESULTS

Characterization of the microbial community in plant species before inoculation

To trace the microbiota enriched by A. adenophora from the neighborhood, we characterized the microbial community of inoculation sources. Plant sample type (ML vs LL vs RS) and plant species interaction, plant sample type, and plant species, to a lesser degree, significantly explained fungal community variation (Fig. 1a); however, plant sample type, plant sample type and plant species interaction, and plant species, to a lesser degree, significantly explained bacterial community variation (Fig. 1b). Relatively, microbiota from RSs were more similar than those from leaf tissues (LL, ML) (Fig. 1). Both fungal and bacterial communities were partially shared among the five plant species, regardless of leaf tissues or rhizosphere soils (Fig. S4 and S5). At the OTU level, the shared part accounted for 39.5% in ML, 41.4% in LL, and 17.9% in RS for fungi and 26.6% in ML, 34.3% in LL, and 66.7% in RS for bacteria (Fig. S5). Our data confirmed that plant-associated microbes were host- and tissue-specific, and in particular, plant sample type (here leaf vs rhizosphere soil) explained the greatest difference of microbial community, regardless of native or invasive species.

Fig 1.

PCoA of the similarity among fungal communities (a) and bacterial communities (b) in A. adenophora and four native plants as inoculation sources. Percentages of total explained variation by PCoA axes in each plot are given in parentheses. The Venn diagram shows the variation in the fungal (a) and bacterial (b) communities, which is explained by each given variable and their shared variation (***P < 0.001). PERMANOVA test for differences in microbial community across samples and/or plant species, with residuals of 40.83% for fungi and 24.26% for bacteria. AAd, A. adenophora; RI, R. indica; AP, A. prionodes; AB, A. bidentata; AAt, and A. atrovirens. ML, mature leaf; LL, leaf litter; and RS, rhizosphere soil.

Accumulation of endophytes by the invasive plant A. adenophora after inoculation

By the inoculation of native plant microbiota, we further characterized the endophytes enriched in roots and leaves by A. adenophora under different temperatures and drought treatments. As expected, plant sample type (leaf vs root) primarily explained the community variation, followed by plant species, temperature, and drought (Fig. 2; Table S1). Relatively, the microbiota from roots were still more similar than those from leaf tissues (Fig. 2a and e). A. adenophora likely enriched rare species (with a relative abundance less than 0.15% of total microbial community occurred in inoculation sources) from three inoculation sources of non-Asteraceae plants, regardless of fungi (Fig. 3b, c, and d) or bacteria (Fig. 3g, h, and i), but several abundant OTUs from Asteraceae A. atrovirens (e.g., fungal OTU2 in both LL and ML; bacterial OTU1 in ML, see Fig. 3e and j), as well as from conspecific inoculation source (A. adenophora) (e.g., fungal OTU1 in both LL and ML; bacterial OTU1 in both roots and leaves, see Fig. 3a and f), as its dominant endophytes in roots and/or in leaves, depending on temperature and drought. In total, the enriched dominant fungal OTUs fluctuated more (Fig. 3) and shared less (Fig. S6) than bacterial OTUs, depending on inoculation sources, climatic factors (temperature and drought), and plant compartments. Analysis of the pairwise Bray–Curtis dissimilarity index of enriched endophytic communities between heterospecific and conspecific inoculation indicated that Asteraceae plant A. atrovirens as inoculation sources significantly caused a lower or equal dissimilarity of endophytes enriched in leaves and roots of the recipient plant A. adenophora than three non-Asteraceae plants as inoculation sources, regardless of fungi (Fig. 4a and b) or bacteria (Fig. 4c and d), depending on climate factors. The results suggested that, besides the plant host filter, climatic factors such as temperature and drought also greatly affected the endophytic community enriched by A. adenophora from inoculation sources.

Fig 2.

PCoA of the community structure of the enriched endophytic fungi (a–d) and bacteria (e–h) in A. adenophora by conspecific and heterospecific inoculations. Panels depict the same ordination but represent a given plant tissue (a and e), microbial inoculation source (b and f), temperature (c and g), and drought power (d and h). Percentages of total explained variation by PCoA axes in each plot are given in parentheses, and permutation variance for each treatment factor is marked in the panel. PERMANOVA test, see Table S1.

Fig 3.

The top 10 fungal OTUs (a–e) and bacterial OTUs (f–j) enriched in leaves and roots of A. adenophora by conspecific (A. adenophora) and heterospecific (R. indica, A. prionodes, A. bidentata, and A. atrovirens) inoculations under different hydrothermal conditions. For each panel, the bar shows the relative abundance of the top 10 OTUs (color) and other rare OTUs (gray) enriched in leaves and roots by A. adenophora under each hydrothermal condition; if these dominant OTUs enriched by A. adenophora also appeared as the top 10 OTUs in corresponding microbial sources (MS), they are in color; otherwise, they are in gray. MS represents microbial sources from ML, LL, and RS associated with different plant species, including conspecific (A. adenophora) and heterospecific (R. indica, A. prionodes, A. bidentata, and A. atrovirens) inoculations. H0 represents high temperature plus sterile water; H20 represents high temperature plus 20% PEG-6000; L0 represents low temperature plus sterile water; and L20 represents low temperature plus 20% PEG-6000.

Fig 4.

Bray–Curtis dissimilarity of endophytic fungi (a and b) and bacteria (c and d) accumulated by A. adenophora when exposed to native plant species vs when exposed to A. adenophora as inoculation sources. H0 represents high temperature plus sterile water; H20 represents high temperature plus 20% PEG-6000; L0 represents low temperature plus sterile water; L20 represents low temperature plus 20% PEG-6000. The nonparametric Kruskal–Wallis test was used to identify the differences between treatments (different lowercase letters indicate P < 0.05).

We further identified some temperature- and drought-related fungi and bacteria enriched by A. adenophora (Fig. 5). Facing temperature change, many fungi enriched in response to high temperature were identified, e.g., Tuber (Asc.) and Fusarium (Asc.) in leaves (Fig. 5a) and Claroideoglomus (Glo.) and Rhizophagus (Glo.) in roots (Fig. 5c). In contrast, many bacteria enriched in response to low temperature were identified, particularly in leaves (Fig. 5b and d). Facing drought stress, many fungi, including Rhizophagus (Glo.), Fusarium (Asc.), Entophlyctis (Chy.), Epicoccum (Asc.), and Penicillium (Asc.), were enriched in roots (Fig. 5g), and only Nigrospora (Asc.), Fusarium (Asc.) and Phaeosphaeria (Asc.) were enriched in leaves (Fig. 5e). In contrast, many bacteria were enriched in leaves, e.g., Mycobacterium (Act.), Legionella (Pro.), Paenibacillus (Fir.), Enhydrobacter (Pro.), Bosea (Pro.), Acinetobacter (Pro.), and Rhodococcus (Act.) (Fig. 5f), and only Iamia (Act.), Legionella (Pro.), and Mucilaginibacter (Bac.) were enriched in roots (Fig. 5h). It is thus very interesting to verify if these endophytes can actually provide host A. adenophora to adapt climate stress.

Fig 5.

Temperature (a–d) and drought (e–h) regulated the relative abundance of endophytes in the roots and leaves of A. adenophora at the genus level. The random forest algorithm was used to select the genus with significant changes in relative abundance (*P < 0.05, **P < 0.01, and ***P < 0.001). The mean decrease in Gini represents the contribution to community variation. For fungi, Asc., Ascomycota; Glo., Glomeromycota; Bas., Basidiomycota; Chy., Chytridiomycota; and Mor., Mortierellomycota. For bacteria, Pro., Proteobacteria; Fir., Firmicutes; Bac., Bacteroidetes; Act., Actinobacteria; Aci., Acidobacteria; Cya., Cyanobacteria; and Gem., Gemmatimonadetes.

Growth effects on the invasive plant A. adenophora

We calculated the growth RI to evaluate the distinct effects on A. adenophora growth by inoculation of different native plants under climate stress. Compared to conspecific inoculation (A. adenophora), both heterospecific A. prionodes and A. atrovirens negatively impacted inoculation, but R. indica positively impacted A. adenophora growth only without drought stress (Fig. 6a, c, e, and g). Pathway analysis indicated that drought (path coefficient = −0.93, P = 0.0002), microbial inoculation source (path coefficient = −0.48, P = 0.015), and temperature (path coefficient = 0.41, P = 0.052) directly impacted the growth of A. adenophora (Fig. 7). Both drought (path coefficient = −0.36, P = 0.090) and inoculation (path coefficient = −0.46, P = 0.035) indirectly affected the growth of A. adenophora (path coefficient = −0.62, P = 0.012) by regulating the richness and composition of endophytic fungi in roots (Fig. 7). Therefore, native plant species can greatly impact the endophyte assembly and host growth of invasive plants in response to drought and temperature change.

Fig 6.

Effects of native plant inoculation on A. adenophora growth. ADB, aboveground dry biomass (a–d) and BDB, belowground dry biomass (e–h). H0 represents high temperature plus sterile water; H20 represents high temperature plus 20% PEG-6000; L0 represents low temperature plus sterile water; and L20 represents low temperature plus 20% PEG-6000. Nonparametric Mann–Whitney U-tests were used to identify the differences between the treatment and control (*P < 0.05, **P < 0.01), and the Kruskal–Wallis test was performed to compare the differences in RI between different microbial inoculation sources under the same hydrothermal conditions (different lowercase letters indicate P < 0.05). The error bar represents the standard error.

Fig 7.

PLS-PM of the relationship between temperature, drought, microbial inoculation source, endophytes, and seedling biomass of A. adenophora. The coefficient adjacent to each arrow is the strength of the effect of each standardized path and its significance (+P < 0.1, *P < 0.05, **P < 0.01, and ***P < 0.001). Solid and dashed arrows indicate significant or insignificant path coefficients, respectively. R² values indicate the variance of dependent variables explained by the model.

DISCUSSION

Root is a more stable environment than leaf for endophyte enrichment by invasive plant A. adenophora from native plant species

Our previous field investigation found that A. adenophora exerts a strong selective enrichment for foliar fungi and root bacteria from the surrounding environments (40, 60). In this study, we confirmed that plant tissue specificity still explained the most variation in the enriched endophytic community, regardless of fungi or bacteria (Fig. 2; Table S1). These findings are similar to previous reports (31, 61). This fact primarily reflects different microenvironments associated with soils and leaves, such as nutrient level, UV irradiation, moisture, etc. Moreover, it is also related to the distinct physiology of roots and leaves, where some biosynthetic pathways can dramatically change the composition of endophyte communities (62). For example, mutations in cuticle synthesis genes affect the composition of foliar bacterial communities (63, 64), and salicylic acid signaling and glucosinolate biosynthesis genes can alter root microbiota composition (65, 66). Compared to roots, leaves have a larger apoplast, which facilitates the gas exchange essential for photosynthesis and provides a largely air-filled internal space for microbiota colonization (27). Additionally, plants can exert distinct selection on microbial colonists of their tissues through plant immune signaling (27, 66, 67).

We also found that microbial communities belowground varied less than those aboveground among plant species because rhizosphere soils and roots were clustered more concentrated than leaf tissues (Fig. 1; Fig. 2a and e). Our data indirectly supported that roots are a more stable environment than leaves for endophyte assembly in A. adenophora. Indeed, in contrast to root-associated habitats, leaves are characterized by much harsher environmental conditions, such as oligotrophy, exposure to UV radiation, desiccation, and antimicrobial compounds (26, 68). Under temperature change, roots are also considered a more stable environment than aboveground organs for bacterial endophytes in Vitis vinifera (30).

There is a potential evolutionary match between microbes associated with native plants and the invasive plant A. adenophora

Previously, by analyzing 30 phylogenetically diverse host plant species, Fitzpatrick et al. concluded a particularly strong association between host evolutionary relatedness and endophytic diversity and verified that highly dissimilar microbiota can produce more positive effects on focal plant growth than highly similar microbiota in plant–soil feedbacks (31). Our study tested only four native plants and failed to conclude such a strong correlation between host evolutionary relatedness and endophytic diversity (data not shown). Interestingly, we found that inoculation of phylogenetically close Asteraceae (A. atrovirens) can cause the recipient plant A. adenophora (Asteraceae) to enrich dominant microbial species from inoculation sources (Fig. 3e and j), which results in a lower or equal dissimilarity of the endophytic microbiota and thus produces more negative growth effects when compared to three non-Asteraceae inoculations (see Fig. 4 and 6).

Moreover, we found that the enrichment and growth effect of endophytic bacteria and fungi by A. adenophora are both tissue- and climate-dependent. For example, A. atrovirens as an inoculation source, when compared to non-Asteraceae plants, significantly caused a low dissimilarity of endophytes enriched in leaves and roots of A. adenophora [e.g., bacteria in leaves (Fig. 4c) and fungi in roots (Fig. 4b) under H0, as well as fungi in leaves (Fig. 4a) under L0], where the microbe-mediated negative roles in A. adenophora growth were also significant without drought stress (Fig. 6a, c, e, and g). In contrast, the positive plant R. indica as an inoculation source commonly caused a high dissimilarity of enriched endophytes of A. adenophora (Fig. 4). These facts reflect the potential evolutionary match between microbes associated with local Asteraceae plants and the invasive plant A. adenophora. Nonetheless, non-Asteraceae A. prionodes as inoculation sources (heterospecific exposure) also caused a negative effect on A. adenophora growth compared with conspecific inoculation (Fig. 6a, c, e, and g), suggesting that other nonmicrobial mechanisms of heterospecific inhibition, e.g., allelopathic chemicals produced by A. prionodes, may be involved.

Both the endophyte assembly and growth effect of the native plants as inoculum on invasive plant A. adenophora depend on temperature and drought

In addition to the plant host filter, climatic factors such as temperature and drought also greatly affected the endophytic community of A. adenophora after inoculation (Fig. 2c, d, g, and h; Table S1), as previously reported (30–33). Our data indicated that the climatic effects varied depending on plant sample type (root vs leaf) and microbial taxonomic kingdom (bacteria vs fungi) (Fig. 5). For example, under high temperature and drought stress, Rhizophagus (Glo.) were enriched in roots (Fig. 5c and g), but Fusarium (Asc.) were enriched in leaves (Fig. 5a and e). Rhizophagus species have shown a positive effect on plants facing abiotic stress, including drought stress (69, 70). Fusarium is a common plant pathogen, but endophytic strains have been proven to alleviate water stress on tomato seedlings (71). Previously, Fusarium were the dominant endophytes associated with A. adenophora leaves and roots (60), but only a few were pathogenic to A. adenophora (41, 72). It is interesting to verify in the future whether these endophytic Fusarium species can provide A. adenophora with the ability to adapt to drought stress. For bacteria, drought is commonly associated with the enrichment of many actinobacterial taxa (31, 73). In this case, actinobacterial taxa, including Mycobacterium, were enriched in leaves (Fig. 5f), but Iamia was enriched in roots (Fig. 5h). Some cold-adapted bacteria, e.g., Azospirillum (74), were also observed to be enriched in the roots of A. adenophora under low temperatures (Fig. 5d). Similarly, a recent meta-analysis indicated that beneficial bacteria and fungi are especially different in regard to plant health in stressful environments (75).

Because the negative effect on the growth of A. adenophora by inoculation of Asteraceae A. atrovirens was significant only without drought stress (Fig. 6a, c, e, and g), we assume that a successful A. adenophora invasion in a drought year may primarily depend on host intrinsic traits because the drought-mediated damage in this case outweigh the endophyte-mediated effects. Accordingly, our pathway analysis indicated that drought stress produces a stronger direct negative effect than both the microbial inoculation-mediated indirect effect and direct effect on the growth of A. adenophora (Fig. 7). Similarly, Wilschut and van Kleunen concluded that plant–soil feedbacks during a drought event may be limited in comparison with the direct effects of drought (76). Indeed, a water-limited rather than temperature-limited climate may have contributed to the climate-driven plant diversity decline (77). Therefore, endophyte-mediated invasion (i.e., invasion resistance or facilitation of the local plant community) for A. adenophora is expected to operate more effectively under favorable moisture.

Implications for endophyte-mediated plant invasion under climate change

Previous pairwise experiments between invader and native plant species support the viewpoint that invaders are good competitors (78). Nonetheless, the competition exclusion of invasive plants is not always the primary limiting factor for the occurrence of some native species (79), and multiple mechanisms may be involved in invader fate at a certain stage of plant invasion (80). In particular, many examples have verified that the success of some invaders is not due to their superior competitive ability but is related to habitat disturbance [see references (81, 82)] or niche preemption (83). In these situations, direct competition between invasive plants and native plants rarely occurs, but microbes associated with native plants can indirectly affect the growth and establishment of invasive plants by air transmission or soil legacy effects. For example, the phyllosphere bacterial community composition of focal plants is dependent on the identity of the neighboring plant (84); at the colonization stage, when the plant arrives in a community, plant mycobiota might be influenced by the spatial distribution of plants already present in the community (85). The invasive plant A. adenophora is a typical weedy species, and its range expands mainly into habitats disturbed by seeds (37, 43). The disturbed bare habitats provide a chance for the dispersal of A. adenophora propagules to establish. Regarding this point, our data are important for understanding the role of microbes associated with native plants in the invasion of exotic plants without direct competition with native plants, particularly in the early stage or expansion front of exotic plants.

Nevertheless, our current data should be interpreted with caution. First, it remains to be isolated from temperature- and drought-related fungi and bacteria to verify whether they can actually function in A. adenophora adaptation to temperature and drought stress. Second, in most cases, it is unavoidable for invasive species to compete with their native counterparts; for example, A. adenophora commonly grows immediately with local plants after establishment. Importantly, plant competitive interactions and outcomes can be changed by key microbial groups (86). Therefore, an ideal experiment would include direct competition with native partners to evaluate the microbe-mediated neighboring effect on A. adenophora invasion in the future. Finally, to determine if the A. adenophora invasion into a community with high evolutionary relatedness (e.g., Asteraceae species) is more difficult than that with low relatedness in the case of endophyte-mediated invasion (i.e., the microbe-mediated invasion resistance of local plant communities), more local congener species of A. adenophora should be involved in the experimental system.

Conclusions

Our study integrated climate change, endophyte assembly, and native–invasive plant interactions into a research framework. We found that the plant sample type (leaf vs root) primarily explained fungal and bacterial enrichment in A. adenophora. We identified some temperature- and drought-related fungi and bacteria. Many fungi were enriched in roots in response to high temperature and drought stress; in contrast, many bacteria were enriched in leaves in response to low temperature and drought stress. Drought, microbial inoculation source, and temperature directly impacted the growth of A. adenophora. Both drought and microbial inoculation indirectly affected the growth of A. adenophora by regulating the richness and composition of endophytic fungi in roots. We found that native phylogenetically close plant species (i.e., Asteraceae) as inoculation sources commonly cause the invasive plant A. adenophora (Asteraceae) to enrich a low dissimilar endophytic microbiota and thus produce negative growth effects without drought stress. Finally, our data indicated that endophyte-mediated invasion (i.e., invasion resistance or facilitation of the local plant community) for A. adenophora is expected to operate more effectively under favorable moisture.

DATA AVAILABILITY

The next-generation data were submitted to GenBank under the Bioproject accession numbers PRJNA754850 for fungi and PRJNA755464 for bacteria. Data supporting the results are attached to the supplementary information Dataset.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.01093-23.

Dataset 1. aem.01093-23-s0001.xlsx.

Endophytic bacteria.

Dataset 2. aem.01093-23-s0002.xlsx.

Endophytic fungi.

Dataset 3. aem.01093-23-s0003.xlsx.

Microbial inoculation source and plant biomass.

Supplemental figures and tables. aem.01093-23-s0004.pdf.

Supplemental materials including 6 figures and 1 table, as well as corresponding legends.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.