Effect of Epichloë gansuensis endophyte on rhizosphere bacterial communities and nutrient concentrations and ratios in the perennial grass species Achnatherum inebrians during three growth seasons

Keywords: different growth phase, endophyte, rhizosphere bacterial diversity, stoichiometry.

Maintaining high agricultural productivity and forage yield in the face of increased global food demand and natural resource consumption is a principal problem (Cassman 1999; Tilman et al. 2002). Over the last few decades, crop yields have not increased in proportion with increasing fertiliser inputs, resulting in increasing environmental risks and inefficient use of nutrients (Zhang et al. 2010a). The need to develop strategies to overcome this challenge has drawn attention to the importance of harnessing the biological potential inherent in rhizosphere processes for improving the use efficiency of crop nutrients. In the rhizosphere zone, important interactions occur among the plant, soil and microorganisms, and rhizosphere microbes can be regarded as an important factor for plant health. Rhizosphere microbial communities could govern plant nutrient chemistry and regulate plant growth and root health (Marschner et al. 2004; Dotaniya and Meena 2015). In addition, rhizosphere microorganisms can also impact nutrient uptake and plant growth by releasing growth-inhibiting or growth-stimulating substances that influence root structure and root physiology (Calvo et al. 2017), and stimulate hormonal regulation (Nihorimbere et al. 2011). Marschner et al. (2004) used PCR-DGGE (denaturing gradient gel electrophoresis) technology to analyse specific microbial groups in the microbial community, and found that many different factors contribute to shaping species composition and numerical predominance of specific bacteria; these factors include pH, soil type, soil fertility, plant nutrition status, plant species and plant age. Therefore, plant microbial ecologists are interested in the extent to which these communities are associated with changes in various environmental factors (Marschner et al. 2004; Ryu et al. 2005; Govindasamy et al. 2009). The biological activity of both the microorganisms and the root structure could play a central role in nutrient availability (Marschner 2012).

The rhizosphere microbial communities of different plants in the same soil are distinct (Ibekwe and Kennedy 1998; Schlatter et al. 2019). Much evidence has shown that the rhizosphere bacterial community will change during the growth and development of the plant (Houlden et al. 2008; Ziegler et al. 2013), which indicates that the plant growth phase has a key function in regulating the structure of the rhizosphere soil bacterial community (Kourtev et al. 2002; Fierer et al. 2013). Studies also showed that rhizosphere microbial structure and function are influenced by discrepancies of rhizodeposition and root exudation in the different zones in root, related to soil types, cultural practices, crop rotation and tillage (Miller et al. 1989; Westover et al. 1997; Grayston et al. 1998; Horwath et al. 1998; Lupwayi et al. 1998; Latour et al. 1999; Maarastawi et al. 2018; Man et al. 2020). Interactions between rhizosphere bacteria and plants also have an important effect on the mineral content of plants (Pii et al. 2015), and nutrient deficiencies might be overcome by rhizosphere bacteria (Zhang et al. 2010a). Microorganisms such as plant-growth-promoting rhizobacteria and mycorrhizal fungi decompose organic matter by releasing hydrolytic enzymes, thereby increasing soil nutrient availability (Miransari 2011; Ollivier et al. 2011; Meena et al. 2017). If the benefits of rhizosphere communities to plants are to be positively manipulated, a better understanding of the importance of plant factors and soil for the rhizosphere microbiome is required.

Plant nutrients play an important function in stoichiometry, ecosystem nutrient cycling and their relationship (Yang and Luo 2011; Reed et al. 2012). Carbon (C) accounts for 99% of the living biomass of plants. Nitrogen (N), potassium (K) and phosphorus (P) are indispensable macronutrients in the process of plant growth; however, in ecosystems they are usually a major factor limiting plant growth and development (Reef et al. 2010). Nitrogen, P, and K govern both larger scale biogeochemical patterns and physiological constraints of organisms (Sterner and Elser 2002). The demand for these elements is not fully understood in different plant growth stages and plant organs (Ågren 2008; Yang and Luo 2011). Some studies demonstrated that plant age and tissue are key factors regulating stoichiometry and nutrient levels (Niklas and Cobb 2006; Yang and Luo 2011). In addition, increasing plant age alters the distribution pattern of metabolites, which leads to changes in nutrient ratios and levels (Ågren 2008; Niklas and Cobb 2006; Yang and Luo 2011). The P : C, K : C and N : C ratios of leaves all decrease over a growth season (Ågren 2008).

Achnatherum inebrians (Hance) Keng (Stipa inebrians Hance) is a perennial grass species, and is widespread in natural grasslands of north-west China, including the subalpine and alpine grasslands of Gansu, Inner Mongolia, Xinjiang, Tibet and Qinghai (Li et al. 2004b). In western China, A. inebrians has become a dominant invasive species and plays an important role in grassland ecological restoration. A major characteristic of A. inebrians is that almost all plants are hosts of asexual endophytic fungi transmitted by seeds. This finding led to ongoing research into this grass–endophyte association, and included the initial assigning of the name Neotyphodium gansuense to the endophyte (Li et al. 2004b; Nan and Li 2004). In 2015, a re-examination of the endophytes in A. inebrians found that there were in fact two species, and these were assigned the names Epichloë gansuensis and E. inebrians (Chen et al. 2015). Li et al. (2004a) reported the characteristics of the Epichloë gansuensis endophyte: conidia solitariae, subglobosae, ellipsoidea, semiglobularea, reniformia, hyalina, saepe transversaliter affixa, hyalinae, laterales vel intercalares.

Extensive research has examined the consequences of the presence of the endophyte on A. inebrians and the ecosystem. The root system of A. inebrians is very developed, including the length of primary root and the numbers of lateral and fibrous roots. Xia (2018) showed that the root surface area of A. inebrians with endophyte (E+) was significantly greater than of A. inebrians without endophyte (E−). The root surface area of A. inebrians, both E+ and E−, increased first and then decreased with soil depth, and the maximum superficial area of root was at 10–30 cm soil depth under drought stress. The endophyte E. gansuensis dramatically promoted growth over the total root length in A. inebrians plants, with the greatest root length of both E+ and E− plants at 10–30 cm soil depth (Xia 2018). Epichloë gansuensis improved A. inebrians growth under low-N stress by modulating stoichiometry and increasing N-utilisation efficiency of the plant (Wang et al. 2018b). Epichloë gansuensis significantly affects bacterial diversity in A. inebrians rhizosphere soil; Li (2017) found the dominant bacteria of A. inebrians–E+ rhizosphere soil to be Bacillus, Hydrogenophaga and Pseudomonas spp. Several studies have demonstrated that the presence of E. gansuensis increases the tolerance of A. inebrians to biotic and abiotic stresses, including resistance to herbivores, diseases and pests (Li et al. 2007; Zhang et al. 2012), cadmium stress (Zhang et al. 2010b), NaCl stress (Wang et al. 2019, 2020a, 2020c), cold stress (Chen et al. 2016), drought stress (Xia et al. 2018), and low N stress (Wang et al. 2018a, 2018b). In addition, Epichloë symbionts in other grasses increase tolerance to abiotic stresses, including salt tolerance in tall fescue (Festuca arundinacea) (Reza and Mirlohi 2010) and Hordeum brevisubulatum (Song et al. 2015a), drought tolerance in tall fescue (White et al. 1992), and waterlogging tolerance in H. brevisubulatum (Song et al. 2015b). One study demonstrated that the interaction of grazing and Epichloë endophytes affected the contents of P, silicon and phenolic compounds in meadow fescue (Schedonorus pratensis), with silicon content ~16% higher in E+ than E− plants (Huitu et al. 2014).

Epichloë endophytes have powerful biological functions that affect the economic and ecological value of artificial and natural grasslands; therefore, further study to explore possible mechanisms and applications is warranted. Recent studies demonstrated that E. gansuensis influences rhizosphere bacterial community structure and root-associated fungal community structure, and clarified the effect of this endophyte on the relationship between soil nutrients and rhizosphere soil (Zhong et al. 2018; Ju et al. 2020). However, whether the presence of E. gansuensis can affect both the rhizosphere bacterial community structure and the leaf and root nutrient contents (C, N, P and K) in A. inebrians is poorly understood. The above information indicates that the rhizosphere microbial community plays an important role in nutrient absorption by plants. How E. gansuensis influences rhizosphere bacteria and regulates nutrient accumulation by the host grass is little understood. The effect of E. gansuensis on the rhizosphere bacterial community in host Achnatherum inebrians during different seasons of the yearly growing period remains unclear. Therefore, to study the effect of E. gansuensis on rhizosphere bacteria and nutrient accumulation by host grasses during different growth seasons, it is useful to gain understanding of the symbiotic relationship between A. inebrians and E. gansuensis.

The aims of this study are to explore: (i) the influence of E. gansuensis on rhizosphere bacterial community structure and diversity under A. inebrians in spring (May), autumn (August) and early winter (December); (ii) the influence of E. gansuensis on C, N, P and K contents in A. inebrians leaves and roots in May, August and December; and (iii) the influence of E. gansuensis on the relationship between rhizosphere bacteria and plant nutrients during the three growth seasons.

The experimental site was established at Yuzhong campus of Lanzhou University in 2011. Two regions of a site were established: one A. inebrians with E. gansuensis (E+), the other A. inebrians without E. gansuensis (E−). The seeds of E+ and E− A. inebrians plants derive from a single E+ plant. Half of the seeds were maintained as E+ for use in the study. The other half were treated with a 100-fold dilution of the fungicide thiophanate-methyl (active ingredient 70% methyltobizine; Jiangsu Longdeng Chemical Co., Ltd) for 2 h, and then the seeds were washed with sterile water. This is an effective method to kill the seedborne fungal endophyte of A. inebrians (Li et al. 2016; Hou et al. 2020). In order to ensure that the field plants in the E+ and E− plots were as required, aniline blue staining was used to check the infection of seedlings from the fungicide-treated and untreated seeds by microscopic examination of leaf-sheath pieces. The existence of endophyte on plants in the experimental areas was also evaluated by aniline blue staining to observe the presence of characteristic hyphae of E. gansuensis in leaf sheaths.

The study had three independent replicates. Each replicate comprised two blocks (one E+ and one E–) of dimensions 1.05 m by 2.45 m. Each block had 12 plants arranged in six rows × two lines, all 35 cm apart. Each plant represented one plot. The distance between E+ and E− blocks in a replicate was 2.5 m. There are 36 plants of each of E+ and E– per replicate, respectively. In the 6 years after establishment, the experimental area was affected by only natural environmental conditions, with no impact of livestock or wildlife. Three rhizosphere soil samples were collected from plants in E+ and E− plots in 2016; all samples from E+ plots showed E+ plants, and all samples from E− plots showed E− plants. The mean annual temperature was 11.6°C, and the annual rainfall was 400 mm year^–1 in 2016.

Rhizosphere soil, leaves and roots were collected on 15 May, 15 August and 15 December 2016. The leaves and roots were dried at 80°C to constant weight, and were used to measure contents of C, N, P and K. Samples were transferred to sterile and dry polyethylene bags with a sterilised spoon. Five individual E+ or E− A. inebrians plants were sampled for the rhizosphere soil (0–20 cm soil depth) and root samples in five random plots, and these five individual samples were blended into a composite E+ or E− sample. The leaves from the same five individual E+ or E− A. inebrians plants were blended into a composite E+ or E− leaf sample. Sampling was performed with three independent repetitions. All samples of rhizosphere soil were sieved through a 2-mm sieve, and stored 4°C for subsequent isolation of culturable rhizosphere soil bacteria.

Field soil nutrient compositions during the two respective growth seasons of May and August were found in our recently published paper (Hou et al. 2020), and the field soil nutrient compositions in December were as follows (mean ± s.e.): organic C 0.45 ± 0.05% in E+ soil, and 0.48 ± 0.03% in E– soil; total N (mg g^–1) 0.35 ± 0.01 in E+ soil, and 0.34 ± 0.01 in E– soil; total P (mg g^–1) 0.63 ± 0.08 in E+ soil, and 0.59 ± 0.02 in E– soil; nitrate-N (mg kg^–1) 1.57 ± 0.01 in E+ soil, and 1.48 ± 0.12 in E– soil; ammonium-N (mg kg^–1) 29.30 ± 1.87 in E+ soil, 23.34 ± 1.47 in E– soil.

The culture media used in the study are shown in Table 1. Each rhizosphere soil sample (1 g) was added to 100 mL of the different sterilised liquid media, and cultured with constant shaking (150 rpm) at 28°C. After 12 h, the soil suspension was plated on the corresponding agar culture media using the dilution plate method, and incubated for another 24 h at 28°C until colonies appeared. Each colony was transferred to fresh liquid medium of the same type to establish a pure culture. After 12 h of constant shaking (150 rpm) at 28°C, bacterial suspensions were plated onto corresponding agar media, and then incubated for another 24 h at 28°C until colonies appeared. For preservation of bacteria, each colony from the different agar media was transferred to 20 mL KB liquid medium, which was subsequently cultured with constant shaking (150 rpm) at 28°C. After 24 h, 700 μL each bacterial suspension and 700 μL 50% glycerol were mixed for storage at –80°C. Identification of rhizosphere soil bacteria was performed by PCR with 16S primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-GGTTACCTTGTTACGACTT-3′) (Wang et al. 2015). The PCR products were sequenced, and the 16S rDNA sequences were compared with the GenBank databases by using BLAST.

Table 1.  Composition of the culture media

Total organic C in leaves and roots was measured according to the method of Sparks and Bartels (1996). Total P and N contents in the leaf and root were measured according to Song et al. (2015a), and K content in leaves and roots was measured using the method of Bao et al. (2016). Data on the contents of C, N, P and K from the leaves and roots were used to calculating C : N, C : P, N : P, N : K and P : K ratios of the leaves and roots.

Calculation of relative frequency of a taxon was by the formula of the number of colonies in a particular taxon divided by the total number of colonies in all taxa (Su et al. 2010). The diversity was assessed with the Shannon index (H′):

where k is the total number of each replicate in the genus level, and Pi is the relative frequency of each replicate in the genus level (Su et al. 2010).

SPSS 17.0 software (SPSS, Chicago, IL, USA) was used to analyse data. The significance difference between E+ and E− plant rhizosphere soils for all of the parameters was determined by independent t-tests. Two-way analysis of variance (ANOVA) was used to analyse the effects of endophyte and growth seasons on the Shannon index. Three-way ANOVA was used to analyse the effects of endophytic fungi, growth seasons and plant tissue on the contents of plant C, N), P and K. Three-way ANOVA tested the effects of endophyte, growth season and tissue on plant C : N, C : P, C : K, N : P, N : K and P : K ratio. Statistical significance was defined at the P = 0.05 confidence level, and means are evaluated with standard errors. RDA was performed using Canoco 4.5 software and PCA was performed using Statistic 6.0.

The composition of the culturable rhizosphere bacterial communities for the different growth seasons of E+ and E− plants is shown in Supplementary Material, Table S1 (available at the journal’s website). In total, 201 rhizosphere bacterial colonies were obtained from E+ and E− A. inebrians plant rhizosphere soil sampling in May, of which 116 were from E+ samples, and 85 from E− samples. In August, 275 rhizosphere bacterial colonies were obtained, of which 170 were from E+ samples and 105 from E− samples. In December, 225 rhizosphere bacterial colonies were obtained, of which 127 were from E+ samples, and 98 from E− samples.

The isolates of rhizosphere bacteria from E+ plants in May were classified into eight orders and 11 genera. The orders (with genera in parentheses) were Actinomycetales (Paenarthrobacter, Microbacterium), Streptomycetales (Streptomyces), Rhizobiales (Rhizobium), Enterobacterales (Enterobacter), Pseudomonadales (Acinetobacter, Pseudomonas), Bacillales (Bacillus), Sphingobacteriales (Sphingobacterium) and Flavobacteriales (Chryseobacterium, Flavobacterium) (Supplementary Table S1). Isolation of rhizosphere bacteria from E− plants in May found five orders and nine genera: Actinomycetales (Pseudarthrobacter, Microbacterium), Rhizobiales (Rhizobium, Ochrobactrum), Xanthomonadales (Stenotrophomonas), Pseudomonadales (Pseudomonas) and Bacillales (Bacillus, Brevibacillus, Lysinibacillus) (Supplementary Table S1). Bacillus, Rhizobium and Microbacterium were the common genera that occurred in rhizosphere soil of E+ and E− plants at the May sampling (Supplementary Table S1).

Rhizosphere bacteria from E+ plants in August belonged to nine orders and were classified into 19 genera (Supplementary Table S1). The orders and genera were Actinomycetales (Glutamicibacter, Arthrobacter, Pseudarthrobacter, Microbacterium), Corynebacteriales (Rhodococcus), Rhizobiales (Rhizobium), Burkholderiales (Variovorax), Enterobacterales (Enterobacter, Erwinia, Pantoea, Serratia), Pseudomonadales (Acinetobacter, Pseudomonas), Xanthomonadales (Stenotrophomonas), Bacillales (Bacillus, Brevibacillus, Lysinibacillus, Paenibacillus) and Sphingobacteriales (Olivibacter). By contrast, the rhizosphere bacteria from E− plants in August belonged to seven orders and were classified into 11 taxa: Actinomycetales (Arthrobacter, Microbacterium), Corynebacteriales (Rhodococcus), Rhizobiales (Ensifer), Enterobacterales (Enterobacter, Kluyvera, Pantoea), Pseudomonadales (Acinetobacter, Pseudomonas) and Xanthomonadales (Stenotrophomonas) (Supplementary Table S1). Kluyvera and Ensifer were unique to the E-plant rhizosphere soil in August. Acinetobacter, Arthrobacter, Bacillus, Enterobacter, Microbacterium, Pantoea, Pseudomonas and Stenotrophomonas were the common genera that occurred in rhizosphere soil of E+ and E− plants in the August sampling (Supplementary Table S1).

At the December sampling, bacteria in the rhizosphere soil of E+ plants belonged to four order and 13 genera. These were: Actinomycetales (Arthrobacter, Microbacterium, Paenarthrobacter), Burkholderiales (Massilia, Limnobacter), Enterobacterales (Enterobacter, Leclercia, Erwinia, Pantoea), Pseudomonadales (Pseudomonas, Acinetobacter) and Bacillales (Bacillus and Lysinibacillus). The rhizosphere bacteria from E− plants in December belonged to five orders and 10 genera: Actinomycetales (Paenarthrobacter, Glutamicibacter), Burkholderiales (Massilia), Enterobacterales (Enterobacter, Leclercia, Citrobacter, Kosakonia), Pseudomonadales (Acinetobacter, Pseudomonas) and Bacillales (Bacillus). Acinetobacter, Bacillus, Enterobacter, Leclercia, Massilia, Paenarthrobacter and Pseudomonas were the genera that were present in rhizosphere soil of both E+ and E− plants at the December sampling (Supplementary Table S1). Isolates of Arthrobacter, Erwinia, Pantoea, Microbacterium, Limnobacter and Lysinibacillus were confined to rhizosphere soil of E+ plants in December. Isolates of Glutamicibacter, Kosakonia and Citrobacter were isolated only in rhizosphere soil of E− plants during this sampling season.

In summary, the presence of E. gansuensis markedly influenced the A. inebrians rhizosphere soil bacterial community, and Bacillus was dominant genus in rhizosphere soil of E+ plants and E− plants during the three different growth seasons.

The Shannon index is used to reflect the alpha diversity and can be used to estimate the microbial diversity in the sample. The larger the Shannon index, the higher diversity of the community. Here, use of the Shannon index can reflect the richness and uniformity of each species in the community. Our data showed that two genera, Pseudomonas and Bacillus, both had high relative frequency in rhizosphere soil of E+ and E− plants in May (Fig. 1a), and that the relative frequency of Pseudomonas was highest (27.51%) and Acinetobacter lowest (2.59%) in rhizosphere soil of E+ plants in May (Fig. 1a). The relative frequency of Bacillus was highest (49.38%) and Pseudarthrobacter lowest (3.53%) in rhizosphere soil of E− plants in May (Fig. 1a). Two genera, Acinetobacter and Bacillus, both had high relative frequency in rhizosphere soil of E+ and E− plants in August (Fig. 1b), and the relative frequency of Bacillus was highest (14.71%) and Olivibacter lowest (1.82%) in rhizosphere soil of E+ plants in August (Fig. 1b). The relative frequency of Arthrobacter was highest (18.00%) and Ensifer lowest (2.82%) in rhizosphere soil of E− plants in August (Fig. 1b). Three genera, Pseudomonas, Bacillus and Enterobacter, had high relative frequency in rhizosphere soil of E+ and E− plants in December (Fig. 1c), and the relative frequency of Bacillus was highest (19.60%) and Massilia lowest (2.36%) in rhizosphere soil of E+ plants in December (Fig. 1c). The relative frequency of Pseudomonas was highest (28.74%) and Citrobacter lowest (3.30%) in rhizosphere soil of E− plants in December (Fig. 1c).

Fig. 1.  Relative frequencies of different bacterial community structures at the genus level in rhizosphere soil of A. inebrians with (E+) and without (E−) E. gansuensis endophyte in (a) May, (b) August and (c) December. (d) Shannon index of the rhizosphere bacteria at the different growth periods of E+ and E− A. inebrians. Data are the mean of three replicates, capped lines are standard errors. *P < 0.05: significant difference between E− and E+ by independent t-tests.

We also found that the Shannon index was different between the rhizosphere bacteria communities of E+ and E− A. inebrians in May, August and December. The Shannon index of the rhizosphere bacteria of E+ plants was higher than that of E− plants in all seasons, and was enhanced by 28.5% in May, 21.4% in August and 11.4% in December compared with E− plants (Fig. 1d). ANOVA showed that endophyte treatment had a significant (P < 0.001) impact on the Shannon index, as did season (Table 2). There was also a significant (P = 0.007) endophyte × season interaction effect on the Shannon index (Table 2). In summary, E. gansuensis significantly increased the Shannon index of the community of rhizosphere soil bacteria in A. inebrians during the three different growth seasons.

Table 2.  Two-way ANOVA for the effects of endophyte presence (with, without) and growth season (May, August, December) on Shannon index of rhizosphere bacterial community

The presence of E. gansuensis had significant effects on the C, N, P and K contents of leaves and C, P and K contents of roots during one or more of the growth seasons (Fig. 2). Leaf C content was significantly different between E+ and E− plants in May, being 18.9% higher in E+ plants (Fig. 2a). Similarly, E+ leaves had a higher content of N than E− leaves in May and December, with N content enhanced by 11.9% in May and 41.0% in December (Fig. 2b). Leaf P content was significantly different between E+ and E− plants in May and December, being higher in E+ than E− plants, with a relative increase of 20.8% in May (Fig. 2c). Leaf K content was significantly different between E+ and E− plants in August, being 12.5% higher in E+ plants (Fig. 2d). Root C content of E+ plants was higher than of E− plants in only August, with a relative enhancement of 9.8% (Fig. 2e). Root P and K contents of E+ plants were increased by 35.0% and 22.6%, respectively, compared with E− plants in May (Fig. 2g, h). However, there was no clear difference in root N content between E+ and E− plants in any season (Fig. 2f). Three-way ANOVA showed a significant (P = 0.002) main effect of endophyte treatment on content of N (but not C, P or K), whereas season and tissue effects were significant (P < 0.001) on contents of all four elements (Table 3). Significant two-way interaction effects included endophyte × season on N content (P = 0.001); endophyte × tissue on N (P < 0.001) and K (P = 0.030) contents; and season × tissue on N (P < 0.001), P (P < 0.001) and K (P < 0.001) contents (Table 3). The three-way interaction effect of endophyte × season × tissue was significant on K content (P < 0.001, Table 3). In summary, E. gansuensis had significantly different effects on the contents of C, N, P and K in A. inebrians leaves and roots during the three growth seasons.

Fig. 2.  Influence of E. gansuensis endophyte (E+, E–) on A. inebrians nutrient contents in the different growth seasons: (a) leaf organic C, (b) leaf N, (c) leaf P, (d) leaf K, (e) root organic C, (f) root N, (g) root P, (h) root K. Data are means ± standard error. *P < 0.05: significant difference between E− and E+ by independent t-tests in the same growth season.

Table 3.  Three-way ANOVA for the effects of endophyte presence (with, without), growth season (May, August, December) and plant tissue (root or leaf) on plant carbon, nitrogen, phosphorus and potassium contents
P-values in bold are significant

The ratios C : N, C : P and C : K increased over the growing season. The C : K ratio of E+ leaves was significantly (P < 0.05) higher than of E− leaves in May, but lower in August and December. The C : P ratios of E+ and E− leaves were not significantly different at any sampling. In December, the C : N ratio of E + leaves was remarkably (P < 0.05) lower than of E– leaves (Fig. 3a–c). The N : K ratio of E+ leaves was higher than of E− leaves in May and December, and the P : K ratio of E+ leaves was higher than of E− plants in May (Fig. 3e, f). The C : N and C : K ratios of E+ roots were significantly higher than of E− roots in August (Fig. 3g, i). Three-way ANOVA showed no significant main effect of endophyte treatment on the ratios C : N, C : P, C : K, N : P, N : K and P : K; however, season and tissue effects were significant on C : N, C : P, C : K, N : P and P : K (Table 4). Significant two-way interaction effects included endophyte × tissue on N : P ratio (P = 0.039), and season × tissue on C : N, C : K, N : P, N : K and P : K ratios (Table 4). The three-way interaction effect of endophyte × season × tissue was significant on C :  K ratio (P = 0.043, Table 4).

Fig. 3.  Effect of E. gansuensis endophyte (E+, E−) on A. inebrians nutrient ratios in the different growth seasons: (a) leaf C : N, (b) leaf C : P, (c) leaf C : K, (d) leaf N : P, (e) leaf N : K, (f) leaf P : K, (g) root C : N, (h) root C : P, (i) root C : K, (j) root N : P, (k) root N : K, and (l) root P : K. Data are means ± standard error. *P < 0.05: significant difference between E− and E+ by independent t-tests in the same growth season.

Table 4.  Three-way ANOVA for the effects of endophyte presence (with, without), growth season and plant tissue (root, leaf) on plant C : N, C : P, C : K, N : P, N : K and P : K ratios
P-values in bold are significant

The first axis of the redundancy analysis explained 28.6% of the variance, and the second axis explained 27.4% of the variance (Fig. 4). According to the analysis, E+ was an important factor affecting the N, P and K contents of leaves and the P content of roots, whereas E− was a key factor affecting the C, N and K contents of roots and the C content of leaves (Fig. 4). Further, our results showed that May was positively correlated with the N, P and K contents of leaves and with P and N contents of roots; however, August was positively correlated with the C and K contents of leaves and with C and K contents of roots. December had a positively correlation with the C, N and K contents of roots (Fig. 4). The presence of E. gansuensis clearly affected the different nutrients of leaves and roots, and the growth season affected the different nutrients of leaves and roots in A. inebrians.

Fig. 4.  Redundancy analysis of E. gansuensis endophyte (E+, E−), growth season, and leaf and root nutrient contents (C, N, P and K).

Effects of all of the genera of rhizosphere bacteria of E+ and E− A. inebrians on the plant contents of C, N, P and K during the different growth seasons were estimated.

We found significant effects of rhizosphere bacteria on the C, N, P and K contents of leaves and roots between E+ and E− A. inebrians plants in May (Fig. 5). For example, the C, N, P and K contents of leaves of E+ plants were positively correlated with Bacillus, Acinetobacter, Paenarthrobacter and Streptomyces; and the N and P contents of leaves of E− plants were positively correlated with Bacillus and Stenotrophomonas (Fig. 5). The C and N contents of roots of E+ plants were positively correlated with Microbacterium and Flavobacterium; and the N and K contents of roots of E− plants were positively correlated with Bacillus and Stenotrophomonas (Fig. 5). The P content of roots of E+ plants was positively correlated with Pseudomonas, Enterobacter, Sphingobacterium, Chryseobacterium and Microbacterium; and the P content of roots of E− plants was positively correlated with Pseudomonas (Fig. 5).

Fig. 5.  Redundancy analysis of rhizosphere bacteria and the leaf and root nutrient contents (C, N, P and K) of A. inebrians with (E+) and without (E−) E. gansuensis endophyte in May.

Similarly, our results also demonstrated significant effects of rhizosphere bacteria on the contents of C, N, P and K in leaves and roots between E+ and E− A. inebrians plants in August (Fig. 6). For example, the C and P contents of E+ leaves were positively correlated with Arthrobacter, Pantoea, Stenotrophomonas, Rhizobium, Paenibacillus and Bacillus; and the C, N and P contents of E− leaves were positively correlated with Microbacterium, Arthrobacter, Acinetobacter and Ensifer (Fig. 6). The N content of E+ leaves was positively correlated with Olivibacter and Lysinibacillus. The K content of E− leaves was positively correlated with Pseudomonas, Bacillus, Pantoea and Rhodococcus (Fig. 6). The C, N and K contents of E+ roots were positively correlated with Variovorax, Brevibacillus and Rhodococcus. The N and P contents of E− roots were positively correlated with Microbacterium, Arthrobacter, Acinetobacter and Ensifer (Fig. 6). The P content of E+ roots was positively correlated with Glutamicibacter and Acinetobacter. The C and K contents of E− roots were positively correlated with Kluyvera, Enterobacter and Stenotrophomonas (Fig. 6).

Fig. 6.  Redundancy analysis of rhizosphere bacteria and the leaf and root nutrient contents (C, N, P and K) of A. inebrians with (E+) and without (E−) E. gansuensis endophyte in August.

Significant effects of rhizosphere bacteria on the C, N, P and K contents of leaves and roots between E+ and E− A. inebrians plants were also evident in December (Fig. 7). For instance, the C, P and K contents of E+ leaves were positively correlated with Pseudomonas, Lysinibacillus and Leclercia in December; and the C, N and K contents of E− leaves were positively correlated with Citrobacter, Pseudomonas and Leclercia (Fig. 7). The N content of E+ leaves was positively correlated with Bacillus, Erwinia, Pantoea, Microbacterium, Limnobacter and Leclercia. The P content of E− leaves was positively correlated with Enterobacter and Paenarthrobacter (Fig. 7). Further, the C and P contents of E+ roots were positively correlated with Enterobacter, Acinetobacter, Massilia, Arthrobacter and Paenarthrobacter. The N and P contents of E− roots were positively correlated with Enterobacter and Paenarthrobacter. The N and K contents of E+ roots were positively correlated with Bacillus and Pseudomonas. The C and K contents of E− roots were positively correlated with Leclercia and Kosakonia (Fig. 7).

Fig. 7.  Redundancy analysis of rhizosphere bacteria and the leaf and root nutrient contents (C, N, P and K) of A. inebrians with (E+) and without (E−) E. gansuensis endophyte in December.

Figure 8 summarises results of principal component analysis for endophyte treatment in May, August and December; it shows that factors 1 and 2 explained >78% of total variance.

Fig. 8.  Principal component analysis of each site (A. inebrians with (E+) and without (E−) E. gansuensis endophyte) and season on the factor plane.

Endophytes of the genus Epichloë have been found in many cool-season grasses including Achnatherum inebrians, Elymus dahuricus, Festuca sinensis, F. elata, Hordeum brevisubulatum and Lolium perenne (Schardl et al. 2004; Song et al. 2015a; Wang et al. 2020a). A previous study showed that the relationships between Epichloë endophytes and grasses are considered mutually beneficial (Müller and Krauss 2005). Studies have reported that the relative abundances of the fungal component were dramatically distinct in different endophyte-treated bulk soils or rhizosphere soils, and that the presence of endophyte influences the structure of soil fungal community, but not the fungal total biomass in fescue soils (Rojas et al. 2016). Similarly, Zhong et al. (2018) found that A. inebrians infected with wild E. gansuensis endophyte had higher root-associated fungal diversity than A. inebrians infected with cultivated E. gansuensis; however, the presence of E. gansuensis obviously decreased root-associated fungal diversity compared with E− A. inebrians. Guo (2014) reported that, compared with other endophytes, Epichloë coenophiala AR584E+ and E. coenophiala CTE+ increased soil organic C content in rhizosphere soil. Franzluebbers et al. (1999) also showed that Neotyphodium coenophialum (Epichloë coenophiala) endophyte leads to small differences in soil microbial community structure between tall fescue plants with 0% and 100% endophyte infection.

In our study, communities of rhizosphere bacteria, and A. inebrians nutrients, were mediated by E. gansuensis endophyte infection and these are critical for ecosystem function and productivity. This research provides insights into how the A. inebrians–E. gansuensis symbiosis influences the rhizosphere microbial community, the C, N, P and K contents of the host grass, and the relationship between rhizosphere bacterial community and nutrient content leaves and roots of A. inebrians during three seasons of the growth period. Epichloë gansuensis increased the Shannon index of the rhizosphere bacterial community and changed the rhizosphere bacterial community (Fig. 1, Table 1). Zhong et al. (2018) reported that the interaction between soil disturbance and E. gansuensis leads to changes in fungi communities in the root association. Ju et al. (2020) demonstrated by using Illumina pyrosequencing that E. gansuensis increased the Shannon index of the A. inebrians rhizosphere soil bacterial community, which was consistent with our results. Further, the study of Hou et al. (2020) showed that E. gansuensis increased the activities of soil invertase, alkaline phosphatase and urease, and the contents of soil organic C, total N, available N and available P. Therefore, E. gansuensis perhaps regulates soil enzyme activity through reprograming soil microbial structure. Meanwhile soil enzymes participate in the soil nutrient cycle, which probably increases nutrient accumulation of host grasses to improve plant growth. Bansal et al. (2014) showed that with the increase in microbial diversity and richness, the activity of soil urease was also enhanced, and this increased N cycling, which is beneficial to the accumulation of host grass nutrients and plant growth. Similarly, our results indicated that E. gansuensis increased the uptake of nutrients (N, P and K) during the different growth seasons (Fig. 2). The study of Wang et al. (2020b) also indicated that beneficial bacteria (plant-growth promoting rhizobacteria, PGPR) activate nutrients (N, P and K) and promote wheat growth under reduced fertiliser application conditions. These PGPR function as biofertilisers; and likewise, the presence of E. gansuensis has a role as a biofertilisers because it enhances the presence of bacterial genera regarded as PGPR. Pseudomonas and Bacillus are dominant genera in May, Acinetobacter and Bacillus in August, and Pseudomonas, Bacillus and Enterobacter in December (Fig. 1).

The effect of an Epichloë endophyte on root-associated and rhizosphere microorganisms differs in different grass–Epichloë associations. The presence of an Epichloë endophyte in perennial ryegrass (Lolium perenne) and A. sibiricum suppressed arbuscular mycorrhizal fungi (AMF) colonisation, although this effect depended on endophyte strain, ryegrass cultivar and P supply (Liu et al. 2011; Zhou et al. 2016). A previous study demonstrated that Acremonium coenophialum (E. coenophiala) endophyte modulated N metabolism in tall fescue (Lyons et al. 1990). It is also reported that under salt stress and soil-water deficit conditions, the presence of an Epichloë endophyte affected the N and P contents in shoot and root of H. brevisubulatum and A. inebrians, and altered the stoichiometry of host grasses (Song et al. 2015a; Wang et al. 2019; Xia et al. 2018). Similarly, we found that E. gansuensis infection influenced C, N, P and K contents in leaves and roots of A. inebrians in May, August and December (Fig. 2), and changed the stoichiometry (elemental ratios) of host grasses (Fig. 3). Therefore, E. gansuensis could promote growth of host A. inebrians probably by regulating soil microbial community structure, soil enzyme activity and nutrient content, to increase nutrient accumulation. We found that E. gansuensis increased the N content of leaves compared with E− leaves in December (Fig. 2). However, in a previous study, endophyte infection had no clear influence on N-use efficiency of chewings fescue (Festuca rubra) (Richardson et al. 1999). Li et al. (2012) found that the benefits to A. sibiricum from endophyte infection largely depended on the supply of N and/or P, and in the case of P supply and N deficiency, there was no dramatic difference of N content in E+ and E− roots. According to our results, there were different E. gansuensis treatment effects on the contents of C, N, P and K in leaves and roots during the three different growth seasons (Fig. 2). Ju et al. (2020) reported that soil moisture had a clear influence on soil nutrients, especially nitrate-N, available N and P and total N; and E. gansuensis had obvious impacts on soil nutrients, especially available P and nitrate-N. Differences in soil moisture and temperature in the different growth seasons would likely have regulated soil nutrient content, and may be why E. gansuensis has seasonal effects on the nutrient content of the host grass.

Plants take up CO₂, soil available N, available K, available P and other nutrient elements, and synthesise metabolites required for plant growth. The C, N, P and K contents of leaves and roots and their respective ratios changed with the plant growth season (Fig. 3). Owing to changes in C, N, P and K contents in different growth seasons and tissues, the C : N, C : P and C : K ratios in the leaves increased with growth stage. It is possible that the reduction in leaf C : N ratio of E+ plants was due to increased synthesis of N content compared with E− plants in December. Many researchers have proposed the growth-rate hypothesis, and this is one of the important paradigms for ecological stoichiometry (Elser et al. 1996, 2010). Low ratios of C : P and N : P exist in high-growth-rate species (Main et al. 1997). Our results show that the C : P ratios of leaves and roots decreased in May, supporting the growth-rate hypothesis. The N : P ratio of leaves is a key indicator to define the nutrient element limitations (Koerselman and Meuleman 1996). Our results reflected that the stoichiometry of E+ and E− plants should consider other elements such as K, and its elemental ratios N : K, C : K and K : P, which may respond to changing environment (Rivas-Ubach et al. 2012). Calvo et al. (2017) demonstrated that a mixture of four strains of plant-growth-promoting Bacillus spp. could enhance nutrient contents (N, P and K) by influencing root morphology at the early stages of maize (Zea mays) growth. A previous study by Xia (2018) indicated that E. gansuensis can regulate the root-system structure of A. inebrians. Therefore, our research implies that E. gansuensis has the role of promoting plant growth and stimulating the absorption of nutrients for the host grass, probably in part by regulating the root morphology of A. inebrians.

Epichloë endophytes have important biological and ecological function (Franzluebbers et al. 1999; Reza and Mirlohi 2010; Zhang et al. 2010b, 2012; Song et al. 2015a, 2015b; Rojas et al. 2016; Tian et al. 2017; Xia et al. 2018; Wang et al. 2020a). It was reported that root-associated fungi of uncultivated, wild, endophyte-infected A. inebrians show positive correlation with soil properties, but root-associated fungi of cultivated A. inebrians show negative correlation with soil properties (Zhong et al. 2018). However, there was no previous study into the relationship between rhizosphere bacteria and plant nutrients under E. gansuensis endophyte infection during the different growth seasons, and this led to the present study. Our results showed that E+ was positively correlated with the N, P and K contents of leaves and the P content of roots; however, E− was positively correlated with the C, N and K contents of roots and the C content of leaves (Fig. 4). Interestingly, Ju et al. (2020) reported that Proteobacteria was positively related with pH, available N and K, and nitrate-N in the rhizosphere, and negatively related with soil organic C and total N in the rhizosphere. Therefore, E. gansuensis probably influenced the rhizosphere soil bacterial community, and the microbial community is highly correlated with soil nutrients, and this would influence the absorption of nutrients for plants. A critical characteristic of the association between grass and Epichloë is that the endophyte is systemic within host grasses, but importantly, hyphae of these endophytic fungi are not present in roots (Christensen et al. 2008). The evidence from this present study is that E. gansuensis is an important factor positively affecting the nutrient content of leaves, with the exception of C.

There was a significant difference of relationship between rhizosphere bacteria and nutrient content of E+ and E− plants in the three different growth seasons. The grass A. inebrians grows in poor soil environments and E+ plants are becoming increasingly abundant on the barren grassland in north-west China, with nearly all A. inebrians plants being endophyte-infected. This suggests that the endophyte is beneficial, or at least not disadvantageous, to the host grass. Our previous study also showed that E. gansuensis improved tolerance to low N by regulating enzyme activity related to N metabolism and glucose-6-phosphate dehydrogenase activity (Wang et al. 2018a; Wang et al. 2018b). The present results confirm that E. gansuensis also provides an advantage by enhancing rhizosphere bacterial diversity, which probably further influenced enzyme activity and nutrient contents of soils, and positively affects absorption of nutrients by plants in natural conditions. As noted earlier, E. gansuensis enhances the activity of soil enzymes and contents of available N and P (Hou et al. 2020), and influences the diversity of the community of rhizosphere soil bacteria (Ju et al. 2020). Importantly, a study investigating response of rhizosphere soil properties, enzyme activities and microbial diversity to intercropping types on the Loess Plateau of China found that activities of soil invertase, soil phosphatase and soil urease had a positive correlation with the diversity of rhizosphere bacteria (Gong et al. 2019). The growing conditions of A. inebrians are often harsh with low nutrient availability. The present research shows that E. gansuensis increased rhizosphere bacterial diversity and enabled the A. inebrians host grass to have better growth status by accumulating greater quantities of nutrient elements when exposed to natural conditions. A study involving tall fescue indicated that Epichloë endophyte could increase adaptability of the grass to a P-deficient environment by regulating nutrient absorption (Malinowski et al. 1999, 2000). Feedback in the plant–soil system shows that plants affect characteristics of the rhizosphere, and then the rhizosphere affects the growth of the same or other plants (Ehrenfeld et al. 2005). Further, positive plant–soil feedback could come from increased availability of nutrients (Wardle et al. 1999; Chapman et al. 2006), or the accumulation of a mutualistic symbiont in the rhizosphere (Klironomos 2002). Until now, the key focus of studies into plant–soil feedback has been the effects of different plant species on soil properties (Burke et al. 1998), or on soil community composition (Vandenkoornhuyse et al. 2003; De Deyn et al. 2004; Innes et al. 2004; Kennedy et al. 2005). Therefore, E. gansuensis probably plays a significant role in positive plant–soil feedback by changing rhizosphere bacterial communities, which in turn enhances the nutritional status of A. inebrians.

Rhizosphere bacterial composition in A. inebrians was significantly related to the nutrient content of leaves and root (Figs 5 7). Many studies have confirmed the effect of rhizosphere microorganisms on plant nutrients; for example, Höflich et al. (1995) isolated two rhizosphere bacteria (Pseudomonas and Rhizobium) from soil that stimulated root length and branching, and promoted shoot and root dry weight of pea (Pisum sativum). Inoculation of PGPR enhanced wheat height and biomass, and enhanced the content of N, P and K in wheat shoots; further, after combined application of N fertiliser and PGPR, the speed of increase in nutrient accumulation and wheat yield was even higher (Abbasi et al. 2011). This was consistent with our results. Inoculations of Bacillus cereus, Brevibacillus reuszeri and Rhizobium rubi with manure enhanced plant weight, chlorophyll content, and N, K, and P content of broccoli compared with control (Yildirim et al. 2011). The present research also indicated that Bacillus had a positive correlation with N content of leaves of E+ and E− plants at May (Fig. 5). Bacillus promotes the recovery and absorption of K in Sudan grass (Sorghum vulgare Pers. var. sudanense) (Basak and Biswas 2009). The mechanism by which rhizosphere bacteria promote nutrient absorption may be related to their function in nutrient availability, because some rhizosphere bacteria have special biological functions. For example, Brevibacillus reuszeri and Ochrobactrum cytisi have P-solubilising function; Bacillus cereus, Bacillus, Pseudomonas and Rhizobium have N₂-fixing function; and Rhizobium rubi has the of P-solubilising and N₂-fixing functions (Glick 2012; Yildirim et al. 2011; Meena et al. 2017). All of these functions can be beneficial for promoting nutrient absorption. In addition, some rhizosphere bacteria can produce indole-3-acetic acid (IAA), for example, strains WPR-32, WPR-42 and WPR-51 from wheat rhizosphere (Abbasi et al. 2011), and IAA can contribute to the growth of plants. Our results also showed that Pseudomonas had a positive correlation with P content of roots of E+ plants in August (Fig. 6). Strains of Pseudomonas and Bacillus significantly increased the content of N, P and K in the roots and leaves of cotton (Gossypium sp.) and peas, and both can produce IAA (Egamberdiyeva and Höflich 2004). However, one study proved that some rhizosphere microorganisms are beneficial to plant yield and growth but they do not increase plant nutrient absorption; for example, some of the P-solubilising rhizobacteria isolated from field-grown crops can increase plant height and seed yield of canola but cannot promote P absorption of plants (De Freitas et al. 1997). Pseudomonas alcaligenes PsA15, Mycobacterium phlei MbP18 and Bacillus polymyxa BcP26 have a much better stimulatory role on N, P and K absorption of maize in nutrient-deficient soil than fertile soil (Egamberdiyeva 2007). Betaproteobacteria are very important in the process of plant nutrient cycling and growth; different growth periods strongly affect communities of Betaproteobacteria in potato rhizospheres, and different potato cultivars have differing rhizosphere effects on their communities (Inceoglu et al. 2010). In our study, E. gansuensis increased P content of leaves and roots in May, but had no effects on P content of leaves and roots in August and December, possibly because E. gansuensis activity and quantity were greater in May and the host A. inebrians plants were experiencing rapid growth. The development and structure of the rhizosphere bacterial community are affected by different factors such as plant species, plant age and soil type (Canbolat et al. 2006; Roesti et al. 2006; Qiao et al. 2017). This is consistent with our results, with the diversity and community structure of rhizosphere soil being different at the three sampling dates. The N level of plant tissue and soil N availability were significantly related to the change of rhizosphere bacterial community structure. One study found that at the peak of plant growth, the absorption rate of N by plants was inversely proportional to the level of microbial biomass N, which caused a change of bacterial community structure in the rhizosphere (Bell et al. 2015). This may explain the differences observed in our study of the rhizosphere bacterial community between E+ and E– plants in different growth seasons, which led to changes of plant nutrients.

Important findings of this study are that E. gansuensis enhanced A. inebrians rhizosphere bacterial community diversity at three different growth phases, with the Shannon index highest in August. Contents of C, N, P and K in E+ leaves and of C, P and K in E+ roots were also greater than of E− plants during one or more of the growth seasons. Moreover, the N, P and K contents of leaves and the P content of roots were positively correlated with E. gansuensis endophyte infection, whereas the C, N and K contents of roots and the C content of leaves were positively correlated with E−. There were significant impacts of rhizosphere bacteria on the C, N, P and K contents of E+ and E− A. inebrians leaves and roots in May, August and December. Bacillus had positive correlation with N content of leaves of E+ and E− host grasses in May; Pseudomonas had positive correlation with P content of roots of E+ plants in August; Pseudomonas had positive correlation with the C content of leaves and roots in E+ and E− plants in December. Overall, our results indicate that the mutualistic grass–Epichloë symbiosis has potential to affect the community structure of rhizosphere bacteria and promote host-grass nutrient uptake, improving the yield and quality of grass. The results suggest that host grasses–Epichloë endophyte symbiosis could enhance the competitive ability of host growth and the economic value of grasses in natural rangelands.

Jianfeng Wang and Zhibiao Nan designed the research. Jianfeng Wang, Wenpeng Hou and Zhibiao Nan performed the research. Jianfeng Wang, Zhibiao Nan, Chao Xia, Tao Chen, Xiuzhang Li and Wenpeng Hou analysed the data. Jianfeng Wang, Wenpeng Hou and Michael J. Christensen wrote the paper.

The authors declare that there are no conflicts of interest.