2.1 Soil, Arbuscular Mycorrhizal Fungi, and Plant Preparation

The soil used for this experiment was a low-nutrient clay loam from the Gawler River region of Adelaide, South Australia. It was sieved to < 2 mm and sterilised twice by autoclaving, with > 24 h intervals between autoclave runs. The sterilised soil was mixed in a 1:1 w/w ratio with autoclaved fine beach sand, resulting in a sandy loam with a pH of 7.7, a plant-available (Colwell) P concentration of 5.5 mg P kg^–1, and a nitrate (2 M KCl) concentration of 6.9 mg kg^–1.

Free-draining pots with a volume of 0.8 L were used, each holding 1.1 kg of the sand/soil mix. The soil volume was chosen based on a previous aerobic rice experiment (Nguyen et al. 2025), to ensure sufficient AM colonisation of the roots as well as generation of grain sample, while conforming to the mass requirements of the precision irrigation platform. The small soil volume may have influenced plant–fungal interactions and resource acquisition by the roots versus the AM fungi. All pots received N in the form of urea (powder) mixed through the soil at a rate of 125 mg urea kg^–1 soil, before planting. This equates to 150 kg urea ha^–1, in line with industry recommendations for these rice varieties (Proud et al. 2023). The AM fungal-inoculated plants were provided with approx. 1000 spores of Rhizophagus irregularis using a commercial product (StartUp Ultra, MicrobeSmart, Adelaide, Australia) were added to the soil and mixed thoroughly. The viability of the spores in the R. irregularis product was confirmed through an assessment of root length colonisation on each plant. The non-mycorrhizal control plants had no AM fungal inoculum added to the soil. The High P plants were given P at a rate of 25 mg P kg^–1 soil, equating to 30 kg P ha^–1, in the form of KH₂PO₄ solution mixed thoroughly through the soil before planting. The Low P plants were starved of any P for the first 19 days, and at the onset of P deficiency symptoms were given 10 mg P kg^–1 soil, equating to 12 kg P ha^–1, as KH₂PO₄ solution applied to the soil.

The four Oryza sativa (rice) varieties are all japonica varieties bred for Australian production and are commonly grown under flooded soil conditions in the temperate rice-growing regions of New South Wales, Australia. Varieties selected for this experiment were: Sherpa, Reiziq, Topaz, and Viand. Seeds of the four varieties were surface-sterilised, then germinated on a moist paper towel in a plastic compartmented box in the dark for 48 h, before being moved onto the lab bench under lights to continue germination for a further 72 h.

2.2 Plant Growth Conditions and Experimental Design

Pre-germinated seeds with radicle and cotyledon emerged, were planted into the prepared soils and placed in one of two controlled environment rooms (CERs). The conditions in the CERs were: 70% humidity, 28°C/18°C during the day/night, with a daylength of 14 h with LED lighting. The CER was a specialised Phenospex “DroughtSpotter” at the Australian Plant Phenomics Network's facility in Adelaide, Australia, which enables automated precision irrigation and collection of gravimetric data (Cousins et al. 2021). The experiment utilised a split-split-unit design with five replicate blocks, comprising three blocks in CER1 and two blocks in CER2. Each block comprised 32 single-plant pots residing on 32 DroughtSpotter load cells. Each block was contained within an area of six Lanes by six Positions and was divided into two main units, which were located on either the north or south sides of the CER. Each main unit was subdivided into two subunits of eight pots contained within three Lanes by three Positions. The two Water treatments were randomised to the main units within a block, the two Phosphorus treatments were randomised to the two subunits within a main unit, and the eight combinations of the two Mycorrhiza treatments and the four Variety treatments were randomised to the eight pots within a subunit. The design was generated and randomised using Brien (2020b), a package for the R statistical computing environment (R Core Team 2023).

Each pot was placed on an independent balance and mass data was collected for each pot every 30 min. This system allows for estimation of plant transpiration, and the calculation of plant water use efficiency. The plants were watered twice daily (5:45 and 14:00) using the automated system to a nominated mass, which equated to either 60% (water-limited) or 80% (well-watered) field capacity of the soil. To estimate water loss via evaporation, there were three plant-free pots of soil for each watering treatment. Plants stayed on the DroughtSpotter automated watering system until they had completed ripening. There were five biological replicates of each of the 32 Variety × Phosphorus × Mycorrhiza × Water treatments, with a total of 160 pots across two CERs.

During the growth period just after tillering finished, plants displayed visual symptoms of N deficiency, so were given N in the form of NH₄NO₃ at 42 DAP (10 mg) and 62 DAP (20 mg). From the stem elongation stage, plants were also given 10 mL of full-strength Long-Ashton solution (P omitted) on a weekly basis.

2.3 Harvest and Sample Analysis

Once plants had finished ripening, a soil core of 15 mm diameter was taken from each pot to excavate a sub-sample of fresh roots for the determination of AM fungal colonisation. Roots were carefully washed from the soil core and placed into 70% ethanol for > 24 h for fixing. They were then placed into a 10% potassium hydroxide solution at room temperature for 7 days for clearing. The cleared roots were stained using a 5% ink in vinegar solution at 60°C for 10 min, and de-stained in a 1% vinegar solution for 24 h before being placed in a 50% glycerol solution for storage and quantification of percentage root length colonised by AM fungi under a stereomicroscope using the gridline intersect method (Giovannetti and Mosse 1980).

Destructive plant harvests were staggered, as different varieties and watering treatments matured at different rates. Plants were harvested between 134 and 188 DAP when they had completed the ripening phase, as follows. Shoots were cut at the soil surface and weighed for shoot fresh weight, then placed in a paper bag and dried at 50°C for 48 h. Then, the sample was weighed again for shoot dry weight before the grain was separated from the rest of the aboveground biomass and weighed. The grain sample was homogenised to a fine powder using a Geno/Grinder 2010 (SPEX SamplePrep), and a weighed sub-sample was digested using nitric acid and hydrogen peroxide before the diluted digest was analysed by ICP-OES (Avio 200 ICP Optical Emission Spectrometer) for P concentration.

2.4 Statistical Analysis and Equations

The total water use (total WU, L) was calculated as the sum of the raw daily water use values for DAP 4 to 133 inclusive. Daily water use (WUR mL day^–1) was computed for each plant over the period DAP 4 to DAP 133 inclusive (130 days) using the Weight and Water data in the DroughtSpotter data log. In most cases, the single-day WUR was computed for the 24-h period ending at 14:00, coinciding with the second, afternoon daily watering. Exploratory smoothing of the WUR was performed using function traitSmooth from package growthPheno (Brien 2020c). Smoothed WUR (sWUR, mL day^–1) was then produced by directly smoothing the WUR using P-splines with the smoothing penalty set to three. The sWUR for each of the DAP intervals 14–20, 20–25, 25–30, 30–35, 35–40, 40–45, 45–50, 50–70, 70–90, 90–110 and 110–130 was calculated as the mean of the sWUR values in the interval.

The vector $${{\boldsymbol{\beta }}}_{{\rm{s}}}$$ consists of just the Block effects vector $${{\boldsymbol{\beta }}}_{{\rm{B}}}$$ that allows for Block differences. The vector $${{\bf{u}}}^{{\rm{\top }}}$$ of large-scale spatial effects is partitioned as $$\left[{{\bf{u}}}_{{\rm{B}}:{\rm{M}}}^{{\rm{\top }}}{{\bf{u}}}_{{\rm{B}}:{\rm{M}}:{\rm{s}}}^{{\rm{\top }}}\right]$$ for differences between MainUnits within Blocks and SubUnits within MainUnits, respectively.

The residual effects are assumed to be normally distributed with variances that differ between the combinations of the Phosphorus and Water levels. In addition to this maximal model, three simpler variance models were fitted: (i) different Phosphorus variances only; (ii) different Water variances only; and (iii) a single variance. The Akaike Information Criterion (AIC) was used to select the variance model that gave the best fit while taking into account the parsimony of the model.

In the case of the AM colonisation trait, the analysis was restricted to the AM-inoculated group; consequently, all terms involving Mycorrhiza were dropped from the model.

Residual-versus-fitted values plots and normal probability plots of the residuals for each trait were inspected to check that the assumptions underlying their analyses were met. Diagnostic residual plots for all traits were satisfactory, indicating that the selected models appear to be appropriate.

Wald F-statistics were used to test the significance ($$\alpha =0.05$$) of the Water, Phosphorus, Variety and Mycorrhiza effects. Testing began with the four-way interaction. If this was not significant, tests of the three-way interactions were conducted. Tests were then conducted for each two-way interaction which did not occur within any significant three-way interaction. Finally, tests were conducted for the main effect of each factor which did not occur in a significant interaction, if any. Based on these tests, a chosen model that includes the significant effects (p ≤ 0.05) was identified for each trait (see Supporting Information S1: Tables 1 and 2). EMMs that conform to the chosen model were obtained. Least significant differences for $$\alpha =0.05$$ [LSD(5%)] were calculated to determine the significance of pairwise differences between the EMMs.

3.1 Shoot and Grain Biomass

Greater water availability (FC80) led to significantly greater (p ≤ 0.05) shoot biomass accumulation in all four rice varieties, the percentage increase ranging from 23.5% (Sherpa) to 32.0% (Reiziq) (Figure 1a; Supporting Information S1: Table 1). At low water availability (FC60), Reiziq produced more shoot biomass (12.17 g) than the other varieties (ranging from 10.54 to 11.26 g). There was also an effect of the Phosphorus treatment, where the plants in the High P treatment accumulated 10.8% more shoot biomass than the ones in the Low P treatment.

Greater water availability increased grain biomass by a mean 93.9% (Figure 1b), but grain biomass showed different patterns to shoot biomass in terms of the effect of AM fungal inoculation; inoculation with R. irregularis resulted in greater GDW in Topaz (increase of 28.4%) and Reiziq (increase of 26.4%), but not in Sherpa or Viand. Also, the rice varieties were affected differently by soil P availability; at Low P, the Topaz plants produced significantly more grain (3.31 g) than the other three varieties (ranged from 2.33 to 2.62 g), but at High P, the rice varieties did not differ in their GDW (within a Water and Mycorrhiza treatment). Within a variety, there was no effect of increased soil P availability on GDW.

3.2 Root AM Fungal Colonisation

Colonisation of roots by R. irregularis ranged from 7.1% to 24.8% in the roots of AM-inoculated plants (Supplementary Figure 1). The non-inoculated roots displayed no colonisation apart from a small amount (2.7% root length colonised) observed in one plant. There was a main effect of Phosphorus on AM colonisation, whereby the roots grown at Low P were more highly colonised than those grown at High P.

3.3 Plant Water Use and gWUE

Longitudinal water use followed a similar pattern over time for all four rice varieties (Figure 2a). The High P plants all followed a similar trend with significantly greater water usage in the first 35 days than that of the Low P plants. Water use peaked at different DAP depending on the P treatment applied, with the High P plants peaking at 30–35 DAP (at 84.93 mL day^–1) and the Low P plants peaking at 45–50 DAP (at 81.95 mL day^–1) and decreasing from that point. Plant water use converged in all treatments between 70 and 90 DAP (at 63.0 mL day^–1) and remained that way until the conclusion of water use data collection at 130 DAP (Figure 2b).

The cumulative water use of each plant was calculated from 4 DAP when the watering treatments were implemented, to 130 DAP, before when the first plants were destructively harvested (Supporting Information S1: Figure 2). The plants that had more water available (80% FC) used 22.1% more water than the 60% FC plants, and plants that had more P available (High P) used 10.3% more water than the Low P plants. Furthermore, Sherpa and Viand had greater total water usage (7.02 and 6.96 mL day^–1) than did Topaz and Reiziq (6.59 and 6.77 mL day^–1).

Water use efficiency of grain (gWUE) was calculated as total applied water divided by grain dry weight for each plant, respectively. Overall, gWUE was 14.4% greater in the plants that were inoculated with R. irregularis than those that were not inoculated (Figure 3a). The gWUE was also 57.3% greater with more water available (80% FC vs 60% FC). The effect of Variety on gWUE was interactive with soil P availability whereby the gWUE of Topaz at Low P (0.52 g L^–1) was greater than the other three varieties at Low P (ranging from 0.36 to 0.39 g L^–1), and greater than Reiziq and Sherpa grown at High P.

3.4 Grain Phosphorus Nutrition and PAE

The uptake of P into the grain was affected by the three-way interaction between Phosphorus, Mycorrhiza, and Variety, whereby the Reiziq plants (at High P; 42.8% increase) and the Topaz plants (at both Low and High P; 20.4% and 35.1% increase) had greater grain P contents when inoculated with AM fungi than when they were not inoculated (Supporting Information S1: Figure 3).

Phosphorus acquisition efficiency (PAE) was calculated as grain P content divided by the amount of P applied to that plant and multiplied by 100, where a value of 50% indicates that the plant partitioned half of the P applied to the soil, to the grain. Previous work has identified that majority of P taken up by plants is partitioned to the grain rather than shoots in these Australian rice varieties (T.D. Nguyen, pers. comm.). PAE was greater in the AM-inoculated Topaz and Reiziq plants than the non-inoculated controls (23.7% and 24.6% increase), but not significantly different in Sherpa or Viand (Figure 3b). PAE was 65.1% higher in the Low P plants than High P, and at High P, all varieties displayed similar PAE, but at Low P, Topaz had greater PAE (mean 49.0% PAE) than the other three varieties (ranged from 37.0% to 42.4% PAE).

4.1 AM Fungi Can Potentially Improve Rice Water and Nutrient Use Efficiency in Aerobic Cultivation Systems

In this study, inoculation with R. irregularis enhanced both gWUE and PAE in two of the four varieties grown. The benefit of AM fungal inoculation was particularly pronounced in Topaz and Reiziq, both of which are high-yielding varieties commonly grown under flooded conditions in Australia.

Grain WUE can be increased through two means: reduced water use to achieve the same grain yield, or increased grain yield with no change to water uptake (Condon 2020). In this experiment, the water use data revealed that the mechanism underlying AM fungal-mediated gWUE was due to the increased grain biomass, rather than less water used by the mycorrhizal plants. We had hypothesised that AM fungal inoculation might lead to increased water use, given that fungi can directly transport water to host plants (Kakouridis et al. 2022), thereby extending the effective water uptake capacity of a root system (Ruiz-Lozano 2003). However, in our pot-based growing system, this effect may not have been observed due to the inherent limitations on root and hyphal growth, which restricted the ability of plants to explore additional soil resources. With greater soil volume in the pot, or in a field setting, the hypothesis of increased water use due to AM fungal colonisation of roots may be supported.

Meanwhile, the increased PAE observed in AM-colonised plants was driven by greater P uptake and translocation to the grain, especially when soil P availability was low (mean 41.5% efficiency at Low P compared to 25.1% in the High P plants). Many studies, including those using radioisotope tracing, have demonstrated the capacity of AM fungi to directly transport P from the soil to host plants (Watts-Williams 2022) including in rice (Yang et al. 2012). However, the amount of P transported is dependent both on the plant species in question (Smith, Smith, and Jakobsen 2004), and on the genotype of plant species (Sawers et al. 2017). Genotype-dependent differences in AM pathway P uptake could be a result of changes to the transcription or translation of mycorrhiza-specific phosphate transporter genes (OsPT11 in rice (Yang et al. 2012)), root architectural traits, or P acquisition or physiological use efficiencies. Although we did not isotopically trace P uptake in this experiment, we observed variety-dependent differences in AM fungal-mediated grain P content and PAE; generally, Topaz and Reiziq gained more P from AM colonisation than did Sherpa and Viand. The positive effects of AM fungi on P uptake and PAE in upland/aerobic rice production has also been demonstrated in the field (Maiti, Toppo, and Variar 2011; Maiti, Variar, and Singh 2011). The important takeaway points from this experiment were that the AM-inoculated plants experienced increased PAE even when water was limiting (60% FC), and that the extent of AM-mediated P uptake was dependent on the rice variety.

4.2 Greater Phosphorus Availability Promoted Plant Water Uptake, but Not Yield in Rice

Rice plants with more available soil P used more water in the first month of plant growth than those with less available P, regardless of whether the soil moisture was maintained at 60% or 80% field capacity. This suggests that the higher P availability helped to overcome the water limitation at 60% FC, at least in the initial phase of growth, and this was likely facilitated by more vigorous root proliferation and soil water and nitrogen capture (Palta and Watt 2009; van der Bom, Williams, and Bell 2020). However, beyond promoting early water uptake, soil P availability was not a determining factor in grain yield in these varieties. Typically, more applied P in agronomic systems leads to greater grain yield, but in this experiment, this was not necessarily the case, even though grain P uptake increased significantly from 4.5 to 6.9 mg plant^–1. Physiological P use efficiency (PPUE; the conversion of P taken up by the plant into grain biomass (Ahmad, Gill, and Qureshi 2001; Rose and Wissuwa 2012; Neto et al. 2016)) is a trait that differs between crop species and even genotypes, and is relatively low in modern varieties of rice (Ueda and Wissuwa 2022). The effect of AM fungi on PPUE has not been specifically tested to our knowledge, but we hypothesise that the symbiosis would not override the inherent PPUE capacity of the crop and genotype. Although AM fungi may increase PAE (uptake of P from the soil) (Campos et al. 2018) this may lead to reduced PPUE (how that “extra” P is utilised) depending on the plant species and genotype in question (Ueda and Wissuwa 2022).

4.3 Australian Commercial Rice Varieties Differed in Their Tolerance to Aerobic Soil Conditions

The four Australian commercial rice varieties studied here were bred for temperate conditions and for flooded or delayed permanent water management. Although much effort is currently being undertaken to develop new varieties for Australian conditions, including specifically for aerobic production (Vinarao et al. 2023; Gong et al. 2024), existing varieties are already being grown under rainfed conditions in subtropical agricultural regions of Australia (Northern Rivers, New South Wales).

Of the four varieties grown, Viand performed the best under water limitation, but this was dependent on available soil P being non-limiting (e.g., in the High P treatment only). Under P limitation, Topaz performed the best in terms of yield, gWUE and PAE, and it had the greatest quantifiable benefit from AM fungal inoculation in this study. In contrast, Reiziq appeared to be challenged both by water limitation and by P limitation, but also drew benefit from AM fungal inoculation especially at High P and water availability. These findings underscore the potential to optimise current commercial varieties for aerobic production by selecting those with greater compatibility with AM fungi and enhanced WUE and PAE.