Soil physicochemical characteristics
To conduct a controlled-environment trial, non-fertilized and naturally N-depleted soil was collected (at 0–10 cm depth) from the Shenton Park Field Station (− 31.948583, 115.793917), Shenton Park, Western Australia (WA). The soil was air-dried and sieved through a 2-mm sieve to remove large debris and root fragments. The soil physicochemical properties were determined according to the methods described by Rayment and Lyons (2011): texture loamy sand (sand: 81.6%; clay: 12.5%; silt: 5.8%); pH (CaCl2) 5.3; NH4+-N 3 mg kg−1; NO3−-N 1 mg kg−1; Colwell P 5 mg kg−1; Colwell potassium (K) 17 mg kg−1; sulfur (S, potassium chloride extraction) 2.3 mg kg−1; and organic C 11.6 g kg−1.
Fungal inoculum production
The A. occidentalis AB15 was used for the present study, which was isolated and purified from a fresh fruiting body collected from Jarrahdale (WA, Australia) as previously described (Kariman et al. 2020). Vermiculite (grade 2, 2–4 mm) was mixed with peat moss (5:1 v/v) to produce a substrate for hyphal inoculum production. Two hundred milliliters of the vermiculite-peat moss substrate was placed into each 500-mL polycarbonate jar and autoclaved at 121 °C for 15 min. Subsequently, all jars received 125 mL of a glucose-based liquid growth medium (Lambilliotte et al. 2004) and were autoclaved again (121 °C, 15 min). Jars were placed under a laminar flow (aseptic conditions) to cool down, and each jar was inoculated with 10 hyphal plugs (5 mm in diameter) taken from the 6-week-old fungal colonies growing on potato dextrose agar (PDA) plates at 20 °C. The inoculated jars were then closed, and gently shaken to mix the hyphal plugs with the substrate, and the lids were left slightly loose to allow air exchange. The jars were incubated at 20 °C for 10 weeks to produce the hyphal inoculum.
Isolation of free-living diazotrophs from soil
Two different soil samples were used to isolate the free-living diazotrophs, including the soil used for our controlled-environment trial, and a soil sample collected from Jarrahdale, WA (− 32.318397, 116.042577). The Jarrahdale soil physicochemical properties were determined following the methods described by Rayment and Lyons (2011): texture sandy loam (sand: 57.5%; clay: 20.8%; silt: 21.6%); pH (CaCl2) 5.5; NO3−-N 4 mg kg−1; Colwell P 6 mg kg−1; Colwell K 152 mg kg−1; S (potassium chloride extraction) 6.4 mg kg−1; and organic C 40 g kg−1. A Jarrahdale soil subsample (10 g) was added to 500-mL Erlenmeyer flasks containing 100 mL of sterile deionized (DI) water and shaken at 150 rpm for 30 min. Then, serial dilutions of 10−3 to 10−5 were prepared in sterile DI water, and 0.1 mL aliquots of each serial dilution were inoculated onto plates containing 25 mL of Burk’s N-free medium (sucrose: 20 g L−1; K2HPO4: 0.8 g L−1; KH2PO4: 0.2 g L−1; MgSO4·7H2O: 0.2 g L−1; CaCl2·2H2O: 90 mg L−1; FeCl3: 1.45 mg L−1; Na2MoO4·2H2O: 0.25 mg L−1; agar: 15 g L−1; pH: 7.0). The inoculated plates were incubated at 25 °C for 3–5 days. The growing colonies were considered diazotrophs and were re-inoculated onto Burk’s N-free medium plates to confirm their diazotrophic activity. After 3–5 days of growth at 25 °C, morphologically different isolates were selected, named, and maintained on nutrient agar (NA; g L−1, yeast extract: 3, peptone: 5, NaCl: 5, agar: 15, pH: 6.8) plates for short-term storage at 4 °C, or in nutrient broth (NB; same composition as NA, but without agar) medium amended with 20% (v/v) glycerol for long-term storage (− 80 °C).
Production of diazotrophs inoculum
To produce the inoculum of a given isolate, a loop of bacterial colonies from NA plates was transferred to Erlenmeyer flasks containing 50 mL of autoclaved NB medium, and incubated on a shaker (100 rpm) at 25 °C for 48 h. The bacterial cells were subsequently pelleted by centrifugation (1968 g for 10 min). The supernatant was discarded, and the bacterial cells were resuspended in 50 mL of 20 mM MgSO4 by manually inverting the tubes five times. The number of colony-forming units (CFU) per milliliter was counted on NA plates and adjusted to 3 × 108 CFU/mL for inoculation. To prepare the diazotrophs consortium, 25 mL aliquots of each diazotroph inoculum were pooled together; hence, 1-mL consortium contained 7.5 × 107 CFU of each diazotroph isolate.
Molecular identification of the diazotrophs
Four diazotroph isolates were selected for the controlled-environment trial, including SP5, SP12, and SP15 (from the Shenton Park soil) and J7 (from the Jarrahdale soil). For each isolate, DNA was extracted from pure bacterial cells growing on NA plates following a method described by Cenis (1992). The V4 region of the 16S rRNA gene was amplified using the 515F/806R primer pair (Liu et al. 2007), and amplicons were sequenced on an Illumina MiSeq platform (see below for the sequencing details and bioinformatics analysis).
The controlled-environment trial was carried out in a completely randomized design with four replications. Four inoculation treatments included control (no added microbes: only containing soil indigenous microbes), diazotrophs (a consortium of four free-living diazotroph isolates in 20 mM MgSO4), FM (hyphal inoculum of A. occidentalis), and dual (co-inoculation with A. occidentalis and diazotrophs consortium). To equalize the amount of nutrients/organic matter across treatments, heat-sterilized fungal inoculum (for control and diazotrophs treatments; added before sowing) or heat-sterilized diazotrophs inoculum (for control and FM treatments; added during sowing) was also added to the soil in respective treatments.
To prepare the inoculation treatments, the living or sterilized hyphal inoculums were thoroughly mixed with soil (1:10 v/v, equivalent to 70 mL of inoculum per kg soil) within clean plastic bags, and 2 kg of the prepared mixtures was placed into pots. Prior to sowing, an aqueous KNO3 solution was added to the soil (33 mg N kg−1) to assist with the early establishment of plants; there was no additional N input during the growth period in order to have N deficiency conditions. Pots also received optimal rates of all other nutrients (except P, to ensure a more effective symbiosis) via the addition of the following basal nutrient salts (in mg kg−1) to the soil as solutions: K2SO4: 125; MgSO4·7H2O: 81.2; CaCl2·2H2O: 146.7; MnSO4·H2O: 9.2; CuSO4·5H2O: 2.1; ZnSO4·7H2O: 8.8; Na2MoO4·2H2O: 0.2; and H3BO3: 0.7.
Wheat seeds (cv. Mace; developed by the USDA-ARS and the Nebraska Agricultural Experiment Station) were surface-sterilized as described previously (Kariman et al. 2020), and submerged in DI water for 3 h to imbibe. The imbibed seeds were drained and incubated in the dark at 4 °C overnight to break any possible dormancy and achieve uniform germination. Ten seeds were sown per pot (at 2 cm depth), and each seed received 1 mL of the diazotrophs consortium (in diazotrophs and dual treatments) or autoclaved diazotrophs consortium (in control and FM treatments). All pots were covered with sterile plastic beads (3–4 mm in diameter, 35 g per pot) to minimize cross contamination and reduce evaporation. Seedlings were thinned to six seedlings/pot 1 week after sowing. To assure an effective diazotroph inoculation, seedlings were re-inoculated with 1 mL of the living or sterilized diazotrophs consortium (added to the soil around each seedling) 2 weeks after planting. Plants were grown in controlled-environment growth chambers at 12/12 h light/dark and 20/15 °C day/night temperature. During the growth period, pots were watered to field capacity (14% volumetric water content).
Plant growth and nutritional analyses
Plants were harvested after 7 weeks of growth. Shoots were cut 1 cm above the soil surface, and were dried in an oven (70 °C for 72 h). Roots were separated from the bulk soil and washed over a 2-mm sieve to remove debris and the adhering soil particles. The root total fresh weight was measured for all samples, and each root system was subsequently split into two subsamples, one of which was weighed and oven-dried (70 °C for 72 h) to be used for dry weight calculations, and the other subsample was stored in 50% ethanol (v/v) for root AM colonization measurements.
Soil from each pot was homogenized manually, and two subsamples (about 50 g each) were subsequently placed in zip-lock bags, one of which was immediately placed in foam-insulated container containing dry ice (− 78 °C) and stored at − 20 °C to be used for microbial analysis. The other soil sample was air-dried and used for mineral and total N analyses.
Oven-dried shoot samples were ground to a fine powder, and a measured amount (195–205 mg, the exact weight recorded for nutrient content calculations) was digested in a mixture of nitric and perchloric acid (4:1 v/v). The sample digests were used to determine the concentration of P, K, Ca, Mg, S, Zn, Fe, and Mn using inductively coupled plasma optical emission spectrometry (ICP‐OES; Optima 5300 DV; PerkinElmer). Shoot N content was determined using a combustion analyzer (Elementar Vario Macro, Hanau, Germany).
Determination of soil N forms
For soil extractable NH4+ and NO3− content determination, soil samples were extracted using 0.5 M K2SO4 and the mineral N fractions were quantified spectrophotometrically (Joergensen and Brookes 1990; Rayment and Lyons 2011). To determine soil total N, the air-dried soil samples were ground to a fine powder, and total N was measured using the combustion analyzer.
Root colonization by indigenous AM fungi
Root subsamples were cleared in 10% w/v KOH and stained as previously described (Kariman et al. 2012), and the percentage of root AM colonization was determined following the gridline intersect method (Giovannetti and Mosse 1980) by counting at least 250 intersects per sample.
Soil DNA extraction, amplicon sequencing, and bioinformatics analysis
Soil subsamples (0.3 g) were used to extract DNA from four biological replicates per treatment (n = 16) using a Qiagen PowerSoil Kit; the manufacturer’s protocol was followed throughout the process with the sole modification of re-loading the final elution buffer onto a filter column in order to maximize the DNA recovery yield. The V4 region of the bacterial 16S rRNA gene was targeted for amplicon sequencing using the 505F/806R primer pair (Liu et al. 2007), which was carried out at the Australian Genome Research Facility (Melbourne, Australia) using an Illumina MiSeq v2 platform (250 PE chemistry).
Sequence analysis was performed using the QIIME2 pipeline (Bolyen et al. 2019). In total, 2.88 million reads were generated for the V4 region, out of which about 580,000 reads (~ 20%) were retained after quality filtering steps such as read overlap detection, de-noising, and chimera filtering. Amplicon single variants (ASVs) were selected using the DADA2 plugin after trimming the first and last five nucleotides of each sequence. Taxonomic assignment was carried out using the Scikit-learn algorithm within the QIIME2 pipeline. Chloroplast- and mitochondria-like sequences as well as low-abundance ASVs were discarded (< 20 hits, representing 15% of the dataset). Due to the unbalanced read count among samples, a rarefaction step (set at 19,276) was performed prior to the statistical analyses. The Bray–Curtis dissimilarity matrix was employed to generate the non-metric multidimensional scaling (NMDS) plot based on the relative microbial abundance using the “metaMDS” function of the “vegan” package in R (Oksanen et al. 2019). Similarity percentage analysis (SIMPER, implemented in the “vegan” R package) was also performed to quantify dissimilarity between groups and better explain the observed clustering (Clarke 1993). Shannon diversity was calculated using the function “Diversity” in the “vegan” package and visualized using the ggplot2 package (Wickham et al. 2020).
Due to the short length of the targeted amplicons, fine taxonomic resolution is rarely achieved for soil microorganisms. Here, two diazotrophs (SP5 and SP12) were taxonomically identified as the same Arthrobacter species, despite presenting different morphology (Fig. S1). A phylogenetic approach was taken to differentiate these two isolates. The 16S rRNA V4 region sequences obtained from the pure cultures of these two diazotroph isolates were analyzed separately following the same methodology as described above. A total of 331,000 quality reads were binned into 10 ASVs, all of which were identified as an unknown Arthrobacter species. The ASV sequences were extracted and placed into a comprehensive phylogenetic reference tree (Hug et al. 2016). Reference sequences were aligned using the INFERNAL, and tree construction was performed using the RAxML algorithm according to the GTRGAMMA substitution model. The ASV sequences of SP5 and SP12 isolates were concatenated with the reference sequences and aligned as described above. Reference package and the alignment were then fed into the “pplacer” (Matsen et al. 2010) with the flag “-keep-at-most” set to 1. Phylogenetic placements were then visualized and annotated in IToL v5 (Letunic and Bork 2019). All 16S rRNA gene sequences were deposited in the European Nucleotide Archive (accession number: PRJEB46901).
C source utilization test
An in vitro experiment was conducted to explore the fungal capacity to utilize different C sources. There were nine treatments: PDA (as a universal/standard fungal growth medium), vermiculite-peat moss mixture (VPM) (3:1 v/v), VPM + glucose, VPM + fructose, VPM + sucrose, and five lignocellulosic substrates including organic compost (made of tree-trimming green waste, Richgro Pty Ltd, WA), pine sawdust shavings (Pinus radiata, softwood), Tasmanian blue gum sawdust (Eucalyptus globulus, hardwood), wheat (Triticum aestivum) straw, and lupin hay (narrow-leafed lupin, Lupinus angustifolius). Wheat straw and lupin hay substrates were pulverized into small pieces (˂ 10 mm) prior to the in vitro trial. The PDA plates were prepared (39 g medium L−1). To study the fungal capacity to utilize simple sugars, 200 mL of the VPM medium was supplemented with 125 mL of DI water (the VPM treatment; no added sugar), or 125 mL of 85 mM glucose or fructose, and autoclaved (121 °C, 15 min). To avoid possible sucrose hydrolysis during the autoclaving, the sucrose solution (85 mM) was filter-sterilized (0.2 µm, Filtropur, Sarstedt, Germany) and added to the VPM medium after autoclaving (125 mL of solution for 200 mL of medium). The lignocellulosic substrates were soaked in DI water for 12 h, drained for 1 h, and autoclaved (121 °C, 15 min).
For all treatments, 25 mL substrate was placed in each sterile Petri plate, and there were five replicates per treatment. All plates were inoculated with a triangle hyphal plug (7 × 7 × 7 mm) that was taken from the actively growing (4-week-old) colonies of A. occidentalis on PDA plates. The hyphal plugs were placed on the media upside down, i.e., the mycelia were in direct contact with the medium. Inoculated plates were incubated at 20 °C in the dark for 4 weeks. Substrates were examined visually and also under a dissecting microscope. The mycelial colonization of the medium was considered a capacity to utilize the respective C source, and the colony diameter was measured.
The controlled-environment plant experiment was conducted in a completely randomized design with four replicates. The in vitro C source utilization test was carried out with five replicates. The homogeneity of variances, normality of the residuals, and independence of samples were considered during the data analysis. To detect significant differences among treatments, analysis of variance (ANOVA) was performed followed by the Fisher’s protected least significant difference (LSD) test at a probability level of 95% (p ≤ 0.05). PERMANOVA was performed using the function “adonis” within the “vegan” package (in R) to detect significant differences among microbial clusters in the NMDS plot.