A total of six halophytes were collected at three sampling sites in China. In August 2011, the healthy roots of Phragmites australis (family Poaceae) were collected at the inland saline and arid soil of Changji, XinJiang Province (N 44°29′, E 87°93′), northwest of China. In July of 2014, roots and rhizosphere soils of Suaeda salsa were collected from the coastal region at the mouth of the Yellow River in DongYing, ShanDong Province (N 37°23′, E 118°55′). Furthermore, the rhizosphere soil, intact roots of S. salsa (family Amaranthaceae), P. australis, Calystegia soldanella (family Convolvulaceae), Carex scabrifolia (family Cyperaceae), Kochia scoparia (family Amaranthaceae) and Messerschmidia sibirica (family Boraginaceae) were collected from a saline coastal sandy soil in QingDao, ShanDong Province (N 35°51′, E 120°02′) in 2015. The plant species selected are considered to be good bio-indicators of saline environments. A map of collection sites and photos of plant samples was provided in Figure 1.

Fungi associated with rhizosphere soils (within a 1-mm vicinity of the roots) were isolated using the traditional serial dilution technique (Jogaiah et al., 2013). For collecting the soil samples, roots with adhering soil were mixed with 40 ml sterile phosphate buffer solution (100 mM PBS, pH = 7.2) in 50-ml tubes (BD Falcon). Tubes were vortexed at high speed for 5 min using a Vortex Genie 2 (Mo Bio Laboratories Inc., USA), allowing the release of most of the rhizosphere soils. The root samples were removed and soil suspensions were centrifuged for 15 min at 10°C at 6000 g to pellet the rhizosphere soils. Then, 1 g of each soil sample was re-suspended in 99 ml PBS and 10-fold serial dilution was made from the suspension (10^-3 to 10^-5). Dilution of 10^-4 and 10^-5 were used to isolate fungi. 100 μl of each diluted suspension was spread evenly onto Czapek’s agar plates supplemented with 0.05 g/L streptomycin sulfate and 0.02 g/L tetracycline hydrochloride to eliminate bacterial contamination. The composition of the medium was as follows: sucrose 30 g, NaNO₃ 2.0 g, K₂HPO₄ 1.0 g, MgSO₄^.7H₂O 0.5g, KCl 0.5 g, FeSO₄^.H₂O 0.01g, agar 20 g and 1 liter of distilled water, pH 5.6. All plates were incubated at 25°C in darkness for at least 1 week until colonies appeared. Colonies were picked up and transferred to fresh potato dextrose agar (PDA) plates for purification.

For isolating fungal endophytes, roots and seeds (excised from the utricles) were heavily surface sterilized by immersion in ethanol (75%, v/v) for 30 s, and then soaked in 2.0% NaClO (v/v) for 5–10 min depending on the type of tissues, followed by 90% ethanol (v/v) for 30 s to remove the residual NaClO. Finally, sterilized distilled water was used to rinse the samples at least five times. The roots were dried with sterile filter paper, cut into 3–5 mm segments (for seeds cut into two halves), and placed on 2% malt extract agar (MEA, Oxoid) plates supplemented with antibiotics as mentioned above. Totally, 70 seed segments from S. salsa and approximately 160 root segments from each plant species were used for isolation. The plates were incubated for 2 weeks at 20°C in the dark, after which emerging hyphae developed from the tissue fragments were transferred to PDA for purification.

The mycelia were scraped from the PDA plates and transferred to a sterile 1.5 ml microcentrifuge tube. Fungal DNA was extracted using DNeasy Plant Mini Kit (Qiagen) according to the manufacturer’s protocol. Four primer pairs, ITS1F and ITS4 (White et al., 1990), NS1 and NS4 (White et al., 1990), LR0R and LR5 (Vilgalys and Hester, 1990) as well as EF1-983F and EF1-2218R (Rehner and Buckley, 2005) were used for the amplification of internal transcribed spacer (ITS) of the rDNA cluster, partial 18S rRNA (small subunit, SSU) and 28S rRNA (large subunit, LSU), and partial transcription elongation factor 1-α (tef1) gene, respectively. The final volume of the PCR reaction was 50 μl which contained 2 μl template DNA (20–50 ng), 0.5 μl of each forward and reverse primer (50 μM), 25 μl 2x Taq MasterMix (Cwbio, Beijing), and 22 μl sterile deionized water.

The PCR reaction consisted of the following steps: a pre-denaturation at 94°C for 4 min, followed by 35 cycles of denaturation at 94°C for 40 s, annealing at 55°C for 50 s and elongation at 72°C for 1 min (2 min for LSU and SSU), and final extension at 72°C for 10 min. For tef1 amplification, we used a touchdown PCR cycle that started with an annealing temperature of 66°C in the first cycle, reducing by 1°C in each successive cycle over the next nine cycles until it reached 56°C, which was used in the remaining 30 cycles. All purified PCR products were sent to Shanghai Sangon Biological Engineering Technology and Services Co., Ltd (Shanghai, China) and sequenced with the above primer pairs. All sequences from this study have been deposited in the GenBank database under the accession numbers presented in Table 1. ITS sequences were subjected to BLASTn searches¹ for initial identification.

Preliminary ITS sequence-based identification supports the existence of a wide range of pleosporalean taxa. Thus, we aim to infer their phylogeny using three loci (SSU, LSU, and tef1), which have been used to give sufficient phylogenetic resolution within the Pleosporales (Tanaka et al., 2015), to provide insights into the relationships between our isolates and known pleosporalean fungi. To construct the phylogenetic tree, we retrieved the corresponding gene sequences from recently published papers focusing on the molecular phylogeny of the Pleosporales (Lumbsch and Lindemuth, 2001; de Gruyter et al., 2006, 2010; Ariyawansa et al., 2014; Phookamsak et al., 2014; Knapp et al., 2015; Phukhamsakda et al., 2015). A concatenated set of three genes from the strains listed in Table 1 and Supplementary Table S1, was used to create an alignment using muscle v3.8.425 (Edgar, 2004) tool integrated in AliView (Larsson, 2014). The conserved concatenated alignment based on nucleotides containing 2604 characters was subjected to Bayesian analysis using program MrBayes v3.2.5 (Huelsenbeck and Ronquist, 2001). The chain was run for 5 million generations by applying Generalized time reversible substitution model (Waddell and Steel, 1997). Two simultaneous, completely independent analyses starting from different random trees were run, using three heated chains and one “cold” chain. Once the analysis was finished, 37500 trees were summarized after discarding the first 25% of obtained 50,000 trees, and a consensus tree was obtained. Branch color corresponding to the posterior probability percentage of each clade was generated using FigTree v1.4.2². Further image improvement was performed in CorelDraw Graphics Suite X8.

To explore the adaptability of the pleosporalean fungi to adverse environmental conditions, their growth rates under various conditions (temperatures, pH, salt and osmolyte concentrations) were examined. Colony diameter was used as parameter. Mycelial plugs (5.0 mm in diameter) cut from 7-day-old PDA colonies were transferred to fresh PDA plates.

To evaluate fungal growth responses to different levels of temperatures and pH, the inoculated PDA plates were incubated at 10, 15, 20, 25, 28, and 30°C in the dark for 7 days. Similarly, fungi were cultured on PDA with a wide range of pH gradients (ranging from 6 to 11). The pH values were measured with a Mettler Toledo pH meter. All plates were incubated at 25°C. The salt stress was induced by adding different concentrations of ionic osmolytes (NaCl and KCl), and the non-ionic osmotic stress was imposed by using sorbitol. First, the PDA medium was supplemented with salt (KCl and NaCl) at concentrations of 2, 4, 6, 8, 10, and 12% (w/v), respectively. In parallel, PDA medium with non-ionic osmotic treatment was prepared using sorbitol (0.2–2.0 M) (Nikolaou et al., 2009). All plates were incubated at 25°C. Three to five replicates were performed for each treatment. This experiment was terminated after 10 days. Colony diameters of these fungi were recorded. Values were means of replications and shown in Radar charts.

We used the modified Melin and Norkrans (MMN) free of nitrogen as the basal medium, to determine the ability of the fungi to utilize different inorganic and organic forms of nitrogen (Midgley et al., 2004; Bizabani and Dames, 2016). The nitrogen-free MMN medium contained (L^-1): 5.0 g glucose, 0.30 g KH₂PO₄, 0.14 g MgSO₄^⋅H₂O, 50 mg CaCl₂, 25 mg NaCl, 3 mg ZnSO₄, 12.5 mg ferric EDTA and 0.13 mg thiamine-HCl, pH 4.5 prior to autoclaving. For all treatments, a final N concentration was adjusted to 50 mg^⋅L^-1. The inorganic N source, ammonium [0.25 g^⋅L^-1 (NH₄)₂HPO₄] and nitrate (0.36 g^⋅L^-1 KNO₃), were separately added into the basal medium. A mixture of acidic, neutral, and aromatic amino acid amino acids [glutamine (Glu), glycine (Gly), valine (Val), leucine (Leu), and phenylalanine (Phe)], as well as Bovine Serum Albumin (BSA, N content 16%), were chosen as the organic N source. The two solutions were filter-sterilized with 0.45 and 0.22 μm Millipore filters and added into the autoclaved nitrogen-free basal medium separately. Three mycelial plugs of every fungus (5.0 mm in diameter) were cut with a sterile cork-borer from the edge of actively growing colonies on PDA plates, and then inoculated into 50 ml of the liquid medium in 250 ml flasks. Three replicates were prepared for each treatment. All the cultures were incubated in the dark, at 26°C and shaking at180 r/min. After 7 days, mycelia were collected by filtration through filter paper, and the growth of fungi was measured as dried mycelial biomass (mg).

We established the plant–fungal co-culture system to investigate the effects of fungal inoculation on plant growth. We also ask if the nitrogen status in the plant growth substrate will influence the outcome of plant–fungal interactions. In this case, both organic nitrogen in the form of amino acids and inorganic nitrogen in the form of ammonium nitrate (NH₄NO₃) and potassium nitrate (KNO₃).

The gnotobiotic rice seedlings were generated using the following steps. The rice seeds (cultivar: zhongjiazao 17) were surface sterilized as mentioned above. Seeds were sown in Petri dishes (150 mm diameter) containing 1/5 Murashige and Skoog (MS) medium and incubated at 28 ± 1°C for 96 h. Thereafter, the germinated seeds without microbial contamination were transplanted into the sterile plastic container with 1/2 MS and placed on growth chamber with a photoperiod of 12 h of light/12 h darkness and temperature of 28 ± 1°C for 5 days (Rodriguez et al., 2008). Uniform sized seedlings were selected for inoculation.

To support the growth of these diverse pleosporalean fungi as well as maintain a moderate fungal growth rate, 2 g oatmeal. L^-1 (Difco) as the carbon source was added into the 1/5 Murashige and Skoog (MS medium) (Mahmoud and Narisawa, 2013). Inorganic N sources (KNO₃ and NH₄NO₃) were added to the basal medium before autoclaving (Finlay et al., 1992), while a mixture of five amino acids (Glu, Gly, Val, Leu, and Phe) was filter-sterilized with 0.22 μm Millipore filters and added to the autoclaved basal medium with a final concentration of 30 mg N.L^-1 (Usuki and Narisawa, 2007). A total of 70 ml medium was poured into a sterile glass tube (25 mm diameter, 250 mm height). One fresh mycelial plug (5 mm in diameter) cut from colony margins was placed mycelial side down in the center of each tube, and pure PDA plugs were used as control. Prior to seedling inoculation, fungi were cultured in the medium for 3–7 days.

The 8-days-old rice seedlings were transplanted to the tubes that were pre-inoculated with fungi (five seedlings per tube). Three–five replications of each treatment were set up. Tubes were double sealed with parafilm and arranged randomly and finally kept in the growth chamber at 28 ± 1°C with a photoperiod of 12 h of light/12 h darkness. After incubation for 15 days, fresh biomass, plant height, root length were measured. Statistical analysis was performed using GraphPad Prism v6.0 (GraphPad, San Diego, CA, USA). Statistically significant differences between treatment and control groups were analyzed using multiple t-tests comparison.

Trypan blue staining method was used to confirm whether the pleosporalean fungi colonize the inner roots endophytically (Padamsee et al., 2016). Briefly, the roots were cleaned and fixed in 50% ethanol for 24 h. Then the samples were rinsed three times with sterilized deionized water and then soaked into 5% potassium hydroxide (KOH) for 2–3 h in a water bath at 90°C. Subsequently, the roots were acidified by 2% lactic acid for 1–2 min followed by strained with 0.05% (w/v) trypan blue (a mixture of 1:1:1 lactic acid/glycerol/distilled water) for 10 h. Finally, the roots were placed into 50% glycerin for 24 h to de-stain. The hyphal structures in roots were viewed with light microscope (Zeiss, Axio Scope A1) using differential interference contrast (DIC) illumination and an AxioCam MRc5 camera and Zen software.

As shown in Table 1, a preliminary BLAST analysis of ITS and LSU sequences from the 27 fungal strains isolated from halophyte roots, seeds, and rhizosphere showed that they comprised a diverse group of pleosporalean taxa.

To identify species level taxonomy and determine the relationships among our isolates and known pleosporalean species, we constructed a phylogeny using a multi-locus DNA sequence dataset (SSU, LSU, and tef1) from 89 taxa. Hysterium angustatum and H. pulicare (Hysteriales, Dothideomycetes) served as outgroup taxa (Figure 2). The analysis indicates that the pleosporalean fungi obtained in our work belong to seven distinct families including Phaeosphaeriaceae, Pleosporaceae, Didymellaceae, Sporormiaceae, Didymosphaeriaceae, Massarinaceae, and Lentitheciaceae, and were distributed across two suborders (Pleosporineae and Massarineae). Most isolates belong to the families Phaeosphaeriaceae, Pleosporineae, and Didymosphaeriaceae. It should be especially noted that, nine isolates isolated from three host plants and two compartments (rhizosphere and endosphere) formed a strongly supported clade sister to Phaeosphaeriaceae and Pleosporaceae, suggesting their potentially taxonomic novelty. Furthermore, LSU and/or ITS sequences of some isolates (such as SYC-R-4, JP-R-6, DF-R-9, JP-R-44, and LW-R-2) did not match any known species, and formed several strongly supported separate clades. This suggests that they are potentially new genera and that likely many more taxa to be discovered in this order. The three individual gene trees also yielded very similar topologies (Supplementary Figure S1).

All isolates investigated utilize both inorganic and organic N in liquid cultures. While final strain biomass was variable at different N sources (Figure 3F), a total of 21 isolates preferred organic N over inorganic N (Supplementary Figure S2, P < 0.05). Isolates of the suborder Pleosporineae produced the highest biomass on organic N. Strains LW-7 (Lentitheciaceae sp.) and LW-R-2 (Epicoccum sp.) accumulated the greatest biomass when NH₄⁺ was the sole N-source, while the remaining four strains of R20 (Phaeosphaeriopsis sp.), R(2015)-25 (Curvularia sp.), LW-R-3 (Paraconiothyrium sp.), and DW-R-4 (Paraconiothyrium sp.) produced the greatest biomass on nitrate.

A plant–fungi co-cultivation system was established to explore potential beneficial effects of pleosporalean fungi on plant growth. In view of the importance of the type of nitrogen in plant performance and productivity, we investigated whether the nitrogen source would influence the outcome of plant–fungal interactions (Upson et al., 2009; Newsham, 2011). Our preliminary data established that none of the isolates cause disease symptoms in rice seedlings (data not shown). As shown in Figure 4, plants co-cultured with fungi did not have any positive effects on plant biomass, and in some cases even negatively affected the growth when inorganic nitrogen sources (NH₄NO₃ and KNO₃) were used. While the root length of rice seedlings treated with CYTC-R-5 and plant height of JP-R-44 treated seedlings were enhanced (multiple t-tests comparison, df = 26; P < 0.05) (Figure 5), yet no significant differences in plant biomass were observed (Figure 4).

When organic nitrogen was provided, almost all isolates increased plant biomass production (Figure 4). Isolates DF-R-3 (Phaeosphaeriaceae sp.), DF-R-7 (Phaeosphaeriaceae sp.), DW-R-4 (Paraconiothyrium sp.), R(2015)-29 (Alternaria sp.), and R(2015)-4 (Alternaria sp.) significantly promoted seedling growth (multiple t-tests comparison, df = 24; P < 0.0001). Also, 12 other groups clearly improved plant development (multiple t-tests comparison, df = 24; P < 0.05). The data also showed that fungal induced plant biomass increases are mainly due to the accumulation of above-ground biomass as indicated by increased plant height (multiple t-tests comparison, df = 24; P < 0.05) and not root biomass (Figure 5).

The endophytic colonization of rice roots by the pleosporalean fungi was further investigated microscopically. Six isolates [RS-A-34, RS-C-28, RS-A-88, R(2015)-25, DW-R-4 and R20] (Figures 6A–F) successfully colonized the inner root tissue as shown by the distribution of an abundant number of hyphae in the cortex and epidermis cells, both inter- and intracellularly. Melanized microsclerotia-like structures were formed upon inoculation with the isolate RS-A-34 (Figure 6A), showing the typical characters of DSEs. Root cells were filled with microsclerotia-like structures. Although inoculation of the isolates RS-C-28, RS-A-88, R(2015)-25 did not form the typical microsclerotia-like structures in roots, initiation and development of chlamydospore-like structures have likely occurred. The remaining isolates colonized rice roots only on the surface, because no fungal structures were found inside root cells. Roots inoculated with the isolates DW-R-1 and R(2015)-29 became brown to dark, but no hyphae were seen in roots.