The study was carried out at the Fukang Desert Ecosystem Observation and Experiment Station, Chinese Academy of Sciences, located in the southern edge of the Gurbantonggut desert in China (44°17'N–44°22'N, 87°55'E–87°56'E, 448–461 m above sea level). The site has a continental arid temperate climate, with an annual mean temperature of 6.6 °C (a maximum of 44.2 °C in hot, dry summer and a minimum of -42.2 °C in freezing winter) (Dai et al. 2015). The annual mean precipitation is about 160 mm with annual pan evaporation of 2000 mm, resulting in soil with high salinity (0.45–2.25%) (Xu et al. 2007).

On 30^th July 2015, we selected ten halophyte species Bassia dasyphylla (Fisch. et C. A. Mey.) Kuntze, Ceratocarpus arenarius L., Kalidium foliatum (Pall.) Moq., Salsola nitraria Pall., Suaeda acuminata (C. A. Mey.) Moq., Su. salsa (L.) Pall. (Chenopodiaceae), Eragrostis minor Host (Poaceae), Reaumuria songarica (Pall.) Maxim. (Tamaricaceae), Seriphidium santolinum (Schrenk) Poljak (Asteraceae), and Peganum harmala (L.) (Zygophyllaceae) at the site. Ten healthy individuals of each plant species were uprooted to collect twig and root samples at the location. All sampled individuals of the same species were more than 50 m away from each other, in order to reduce the spatial autocorrelation and recover representative local endophyte community (Li et al. 2016, Yao et al. 2019). The collected samples were immediately placed in autoclaved paper bags, labeled, and transported to the laboratory in an ice-box. Samples were stored at 4 °C and processed within 4 days.

Since most of the plant species involved in the current study (except for E. minor) possess reduced leaves, which are hard to discern from the stems, we selected only stems to isolate endophytes colonized aerial parts of the plants. Roots and stems of individual plants were cut into 5 mm long segments (ca. 2 mm in diameter). Eight root segments and 8 stem segments were randomly selected from each sample. In total, 1600 segments (10 plant species × 10 individuals × 2 tissue types × 8 segments) were used for endophyte isolation in this study.

Surface sterilization was conducted according to Guo et al. (2000). Segments were surface sterilized by consecutive immersion for 1 min in 75% ethanol, 3 min in 3.25% sodium hypochlorite, and 30 sec in 75% ethanol. Sets of four segments were then evenly placed in a 90 mm Petri dish containing potato dextrose agar (PDA, 2%). Benzylpenicillin sodium (50 mg/L, North China Pharmaceutical Group Corporation, China) was added to suppress bacterial growth. Petri dishes were sealed, incubated for 2 months at 25 °C, and examined periodically. When fungal colonies developed, they were transferred to a new PDA containing Petri dishes for purification. The purified strains were transferred to PDA slants for further study.

Subcultures on PDA were examined periodically, and the sporulated isolates were identified based on their morphological characteristics. The non-sporulated cultures were designated as mycelia sterilia, which were divided into different “morphotypes” according to colony color, texture, and growth rate on PDA (Guo et al. 2000). One representative strain of each morphotype or sporulated strain was selected for further molecular identification. The living cultures are deposited in China General Microbiological Culture Collection Center (CGMCC) in Beijing, China.

Genomic DNA was extracted from fresh cultures following the protocol of Guo et al. (2000). Fresh fungal mycelia (ca. 50 mg) were scraped from the surface of the PDA plate and transferred into a 1.5 mL microcentrifuge tube with 700 µL of preheated (65 °C) 2 × CTAB extraction buffer (2% CTAB, 100 mM Tris-HCl, 1.4 M NaCl, 20 mM EDTA, pH 8.0), and 0.2 g of sterilized quartz sand. The mycelium was ground using a glass pestle and then incubated in a 65 °C water bath for 30 min with occasional gentle swirling. Five hundred microliters of phenol:chloroform (1:1) were added into each tube and mixed thoroughly to form an emulsion. The mixture was spun at 12,000 g for 15 min at room temperature in a microcentrifuge, and the aqueous phase was transferred into a fresh 1.5 mL tube. The aqueous phase containing DNA was re-extracted with chloroform:isoamyl (24:1) until no interface was visible. Thirty microliters of 5 M KOAc was added into the aqueous phase followed by 200 µL of isopropanol and inverted gently to mix. The genomic DNA was precipitated at 9200 g for 2 min in a microcentrifuge at 4 °C. The DNA pellet was washed twice with 70% ethanol and dried using SpeedVac (AES 1010, Savant, Holbrook, NY, USA) for 10 min or until dry. The DNA pellet was then re-suspended in 65 µL ultrapure sterilized water.

The internal transcribed spacer (ITS) region of rDNA was amplified using primer pairs ITS4 (White et al. 1990) and ITS1F (Gardes and Bruns 1993). Amplification was performed in a 50 µL reaction volume which contained PCR buffer (20 mM KCl, 10 mM (NH₄)₂SO₄, 2 mM MgCl₂, 20 mM Tris-HCl, pH 8.4), 200 µm of each deoxyribonucleotide triphosphate, 15 pmols of each primer, 100 ng template DNA, and 2.5 U Taq polymerase (Biocolor BioScience & Technology Company, Shanghai, China). The thermal cycling program was as follows: 3 min initial denaturation at 94 °C, followed by 35 cycles of 30-sec denaturation at 94 °C, 30-sec annealing at 52 °C, 1 min extension at 72 °C; and a final 10 min extension at 72 °C. A negative control using water instead of template DNA was included in the amplification process. Four microliters of PCR product from each PCR reaction were examined by electrophoresis at 80 V for 30 min in a 1% (w/v) agarose gel in 1 × TAE buffer (0.4 M Tris, 50 mM NaOAc, 10 mM EDTA, pH 7.8) and visualized under ultraviolet (UV) light after staining with ethidium bromide (0.5 µg/mL). PCR products were purified using Wizard SV Gel and PCR Clean-Up System (Promega, Madison, USA) and directly sequenced with primer pairs, as mentioned above in the ABI 3730-XL DNA sequencer (Applied Biosystems, Inc. USA).

A value of 97% of ITS region identity was used as a DNA barcoding threshold for OTU clustering (O’Brien et al. 2005). The taxonomical assignments for each OTU were determined according to the BLAST results against both UNITE+INSD (UNITE combined international nucleotide sequence databases) and GenBank public sequence databases. A representative sequence of each OTU was selected and searched against the UNITE+INSD fungal ITS databases (Kõljalg et al. 2013) using a basic local alignment search tool (BLAST) (Altschul et al. 1990). The DOIs of UNITE fungal Species Hypothese at 1.5% threshold (Nilsson et al. 2019) were also added to each of taxonomical assigments (Table 1). For reliable identification of the fungi, a representative sequence of each OTU was searched against the GenBank public sequence databases using BLASTN (Sun et al. 2011, Sun et al. 2012a). For further identification of these fungi, we select the most reliable sequence as a reference (the sequences originated from mycologists or taxonomists, yielded from taxonomical or phylogenetical studies, or were part of cultures or speciemens in famous collections, would be given higher credits). As for taxonomical levels higher than species, we typically relied on 90, 85, 80, and 75% sequence identity as a criterion for assigning OTUs with names of a genus, family, order, or class, respectively (Tedersoo et al. 2014). Nevertheless, the results of sequence-based identification were calibrated with morphological characteristics in our study given the strains within one OTU sporulated. The microscopic observation was applied with cultures mounted in sterile water using a compound microscope (Zeiss Axio Imager A2, Carl Zeiss Microscopy, Göttingen, Germany). The ITS sequences of endophytic fungi obtained in this study have been deposited in National Center for Biotechnology Information (NCBI) with GenBank accession no. KY114893 to KY114928 (Table 1).

All statistical analyses were carried out in R 3.3.1 (R Development Core Team 2016). The colonization rate of endophytic fungi was calculated as the total number of tissue segments infected by fungi divided by the total number of tissue segments incubated (Sun et al. 2011). The relative abundance was calculated as the number of isolates of a taxon divided by the total number of isolates of all taxa, and the fungal richness was defined as the number of fungal species in a sample.

One-way analysis of variance (ANOVA) was carried out to test the effect of plant species or tissue type (stem and root) on the colonization rate and species richness of endophytic fungi. Multiple comparisons were performed using post hoc Tukey’s HSD (Honest Significant Difference) tests to examine the significant differences among the plant species or tissue types at P < 0.05 level. All data were tested for normality and homogeneity of variance before ANOVA. In cases where satisfactory results of homogeneity of variance amongst plant species after square root and transformation were not observed (e.g., in stems), then nonparametric Kruskal-Wallis test followed by pairwise comparisons was applied to examine the significant difference among plant species at P < 0.05 level. T-test was applied to examine the significant difference of the colonization rate and species richness of endophytic fungi between stems and roots for each plant species at P < 0.05 level. Canonical correspondence analysis (CCA) was performed to observe the correlation between endophytic fungi and plant species or tissue types with the ‘cca’ function in the vegan package (Oksanen et al. 2019). The effects of plant species and tissue type on community composition of endophytic fungi were tested by permutational multivariate analysis of variance (PermANOVA) using the ‘adonis’ command in the vegan package (Oksanen et al. 2019).

The host-fungus association preferences were evaluated based on a d’ interaction specialization index (Blüthgen et al. 2007) using the ‘dfun’ function in the bipartite package (Dormann et al. 2009) according to Toju et al. (2016). Briefly, a binarized sample × fungal taxon matrix (i.e., presence/absence) was converted into a ‘species-level’ matrix, in which rows depicted plant species, columns represented endophytic fungal taxa, and cell entries were the number of samples from which respective combinations of plants and fungi were observed. To perform a randomization analysis of the d’ index, plant species labels in the sample × fungal taxon matrix were shuffled, and then, the randomized species-level matrices were obtained (1000 permutations). The d’ value of each plant species or each fungal taxon was standardized as follows: standardized d’ = [d’_observed - Mean (d’_randomized)] /SD (d’_randomized), where the d’_observed was the d’ estimate of the original data, and Mean (d’_randomized) and SD (d’_randomized) were the mean and standard deviation of the d’ scores of randomized data matrices. Also, we evaluated the observed frequency (counts) of each plant-fungus association in the species-level matrix, and quantified with the two-dimensional preferences (2DP) in a pair of a plant species (i) and a fungal taxon (j) based on the species-level original and randomized matrices used in the d’ analysis: 2DP (i, j) = [N_observed (i, j) - Mean (N_randomized (i, j))] /SD (N_randomized (i, j)), where N_observed (i, j) denoted the number of samples from which a focal combination of a plant and a fungus was observed in the original data, and the Mean (N_ranodomized (i, j)) and SD (N_randomized (i, j)) were the mean and standard deviation of the number of samples for the focal plant-fungus pair across randomized matrices. The P values were adjusted based on the false discovery rate (FDR) (Benjamini and Hochberg 1995).

A total of 1046 fungal strains were recovered from 1600 tissue segments from ten halophyte species. The colonization rate of endophytic fungi ranged from 7.5 ± 3.33% to 83.75 ± 8.95% in stems, from 33.75 ± 11.19% to 97.5 ± 1.67% in roots, and from 38.75 ± 2.46% to 85.63 ± 2.28% overall for the entire plant among the ten halophyte species (Fig. 1). One-way ANOVA showed that the colonization rate of endophytic fungi was significantly affected by plant identity (F = 5.847, P < 0.001) and tissue type (F = 8.184, P < 0.001). In the entire plant, the colonization rate of endophytic fungi was significantly higher in Sa. nitraria than in other plants (except for Su. acuminata and Se. santolinum) and was significantly higher in Su. acuminata than in E. minor (Fig. 1). In the stem, the colonization rate of endophytic fungi was significantly higher in Sa. nitraria and Su. acuminata than in the other halophyte species (except for P. harmala). For P. harmala, the colonization rate of endophytic fungi was significantly higher than in B. dasyphylla and R. songarica (Fig. 1). In the root, the colonization rate of endophytic fungi was significantly higher in Se. santolinum, R. songarica, and Sa. nitraria than in E. minor and P. harmala, and was significantly higher in B. dasyphylla, C. arenarius, and Su. salsa than in P. harmala (Fig. 1). Furthermore, the colonization rate of endophytic fungi was significantly higher in roots than in stems in B. dasyphylla, C. arenarius, K. foliatum, Su. salsa, R. songarica, and Se. santolinum, but no significant difference was observed in the other four halophyte species (Fig. 1).

Figure 1. 

Colonization rate of endophytic fungi in stem, root, and total (stem + root) tissues of the ten halophyte species. Data are means ± SE (n = 10). Columns without shared lowercase, uppercase, and italic letters denote the significant difference in the stem, root, and total tissues among the halophyte species, respectively. Asterisks above bars indicate significant difference between stem and root tissues for each plant species (** P < 0.01, *** P < 0.001).

In total, 36 fungal taxa were isolated and identified based on morphological characters and ITS sequences (Table 1). The richness of endophytic fungi ranged from 0.5 ± 0.22 to 2.2 ± 0.2 in stems, from 1.2 ± 0.29 to 4 ± 0.3 in roots and from 1.2 ± 0.29 to 4 ± 0.3 (means ± SE) in overall among the ten halophyte species (Fig. 2). The richness of endophytic fungi was significantly affected by plant species (ANOVA, F = 4.635, P < 0.001) and tissue type (Kruskal-Wallis test, x² = 34.993, P < 0.001). In the stem, the fungal richness was significantly higher in Sa. nitraria than in B. dasyphylla and R. songarica, and significantly higher in E. minor than in R. songarica (Fig. 2). In the root, the fungal richness was significantly higher in Sa. nitraria than in B. dasyphylla, K. foliatum, Se. santolinum and P. harmala, and significantly higher in C. arenarius than in P. harmala (Fig. 2). Furthermore, the fungal richness was significantly higher in roots than in stems in ten plant species, except for E. minor, Se. santolinum and P. harmala (Fig. 2).

Figure 2. 

Endophytic fungal richness in stem, root and total (stem + root) tissues of the ten halophyte species. Data are means ± SE (n = 10). Columns without shared lowercase, uppercase, and italic letters denote significant difference in the stem, root, and total tissues among the plant species, respectively. Asterisks above bars indicate the significant difference between stem and root tissues for each halophyte species (* P<0.05, ** P < 0.01, *** P < 0.001).

Of the 36 endophytic fungi, 32 were recovered from roots, 27 from stems, and 23 were common in both roots and stems (Fig. 3). Among seven abundant endophytic fungi (relative abundance > 15% in certain plant species), Alternaria eichhorniae was the most abundant stem endophyte and was recovered from C. arenarius, K. foliatum, P. harmala, Sa. nitraria, Su. acuminata, and Su. salsa (Fig. 4B, D, E, G, I, J). In addition, Monosporascus ibericus was exclusively recovered from roots of B. dasyphylla, K. foliatum, Su. acuminata, and Su. salsa (Fig. 4A, D, I, J). In C. arenarius, Pezizomycotina sp.1 was exclusively recovered from roots, and in Se. santolinum Pezizomycotina sp.1 was mostly recovered from roots (Fig. 4B, H). Sarocladium kiliense and Aspergillus fumigatiaffinis were only found on the roots of R. songarica (Fig. 4F). Bipolaris prieskaensis was mostly isolated from roots, and Preussia sp.2 was mostly distributed in stems in E. minor (Fig. 4C).

The CCA results indicated that the endophytic fungal community composition was significantly different between stems and roots of the ten halophyte species (Fig. 5A), and significantly different among some plants, such as R. songarica, B. dasyphylla, P. harmala and E. minor (Fig. 5B). The PermANOVA results also showed that the endophytic fungal community composition was significantly affected by tissue type (R² = 0.212, P = 0.001) and plant species (R² = 0.082, P = 0.001).

Figure 3. 

Relative abundance of endophytic fungi in the stem and root tissues of the ten halophyte species.

Figure 4. 

Relative abundance of endophytic fungi in the stem and root of different halophyte species.

Figure 5. 

Canonical correspondence analysis (CCA) ordination plot of endophytic fungal communities of stem and root tissues (A) and halophyte species (B). Dotted ellipses indicate 95% confidence intervals around centroids of tissue type (A) and plant species (B), B. dasyphylla = Bassia dasyphylla, C. arenarius = Ceratocarpus arenarius, K. foliatum = Kalidium foliatum, Sa. nitraria = Salsola nitraria, Su. acuminata = Suaeda acuminata, Su. salsa = Suaeda salsa, E. minor = Eragrostis minor, R. songarica = Reaumuria songarica, Se. santolinum = Seriphidium santolinum, and P. harmala = Peganum harmala.

Host-fungus association preference analysis showed that five out of ten halophyte species showed significant preferences to endophytic fungi, especially strong preferences in E. minor, R. songarica, and Se. santolinum (Fig. 6A). Among the 36 endophytic fungi, 13 showed significant preferences for host species, particularly strong preferences were observed in Al. eichhorniae, M. ibericus, Pezizomycotina sp.1, Sr. kiliense, Pezizomycotina sp.2, As. fumigatiaffinis, B. prieskaensis, Trichocomaceae sp., and Xylaria hypoxylon (Fig. 6A). Furthermore, 26 out of 208 pairs of plants and fungi showed significant preferences, such as pairs Pezizomycotina sp.1 and Se. santolinum, Sr. kiliense and R. songarica, B. prieskaensis and E. minor, As. fumigatiaffinis and R. songarica (Fig. 6).

Figure 6. 

Preferences observed in the plant-fungus associations. A Preference scores. The standardized d’ estimate of preferences for fungal taxon is shown for each halophyte (column), and the standardized d’ estimate of preferences for plant species is indicated for each of the fungal taxon (row). Each cell in the matrix indicates a two-dimensional preference (2DP) estimate, which measures to what extent the association of a focal plant-fungus pair was observed more/less frequently than expected by chance. P values were shown as false discovery rates (FDRs) in the plant/fungus analysis. B Relationship between 2DP and FDR-adjusted P values, 2DP values larger than 2.5 and those smaller than -2.5 represented strong preference and avoidance, respectively (P_FDR < 0.05). Significance: *, P < 0.05, **, P < 0.01, ***, P < 0.001.