Arbuscular Mycorrhizal Symbiosis Enables Efficient Phosphorus Uptake in Sorghum Accessions With Contrasting Root Traits

ABSTRACT

Phosphorus (P), an essential nutrient, is apparently unavailable to plants due to strong sorption in soils. Plants with shallow root systems and high surface area exhibit high P acquisition efficiency (PAE). Arbuscular mycorrhizal fungal (AMF) symbiosis can also enhance PAE. However, whether AMF symbiosis will equally benefit crop accessions with contrasting root traits is less known. We selected sorghum accessions that varied in root traits to evaluate P uptake strategies and assessed changes in root traits, acid‐phosphatase activity, primary and specialised metabolome in the presence of AMF, and under limited and stratified P availability. Our results revealed that regardless of the inherent accession differences in root traits, all accessions had higher shoot P and biomass with AMF inoculation. AMF‐inoculated plants had lower specific root length, higher hyphal length and acid phosphatase activity than the non‐inoculated control, indicating that plants can enhance PAE with AMF, irrespective of inherent accession differences. The AMF induced similar changes in root metabolome, where AMF‐inoculated plants had higher organic acids and specialised metabolites necessary for a functional symbiosis. Our results emphasise the critical role of AMF in efficient P uptake regardless of inherent root traits, which should be considered while selecting crop accessions for improved PAE.

Summary statement

Arbuscular mycorrhizal fungi (AMF) enhanced shoot biomass and phosphorus uptake in sorghum accessions with contrasting root traits. AMF shifted plant phosphorus acquisition from root‐based ‘do‐it‐yourself’ strategies to ‘outsourcing’ via AMF, independent of inherent root traits. These findings highlight the potential for improving P‐uptake in diverse crops through optimising AMF symbiosis.

1. Introduction

Phosphorus (P), an essential plant macronutrient, is present in the soil mainly as organic forms or strongly sorbed onto minerals, making P apparently unavailable for plant uptake (McConnell et al. ²⁰²⁰). Moreover, P is non‐uniformly distributed in soil due to the low mobility of inorganic phosphate. Therefore, plants adopt various P acquisition strategies to solubilise and mine P from the soil. These strategies include increased soil exploration by modifying the root traits, including architecture and morphology, and establishing a symbiotic association with mycorrhizal fungi that can effectively capture P through the extensive hyphal network (Campos et al. 2018; Wen et al. ²⁰¹⁹; Mei et al. ²⁰²⁴). Based on the root economic spectrum of resource acquisition, the above strategies of P uptake can be classified as the ‘do‐it‐yourself’ strategy, which is the plant‐mediated P uptake (e.g., root morphological changes) and the ‘outsourcing’ strategy, which is the mycorrhizal mediated P uptake (Bergmann et al. 2020). The preference of plants to choose either the ‘do‐it‐yourself’ or ‘outsourcing’ strategy depends on the availability and spatial distribution of P and the cost‐benefit ratio of each of these contrasting strategies in acquiring P (Lambers et al. 2015; Vance et al. ²⁰⁰³; Wen et al. ²⁰¹⁹). However, it is less known whether crop genotypes that vary widely in their inherent root traits would preferentially choose one strategy over another or utilise a combination of these strategies, particularly in the presence of arbuscular mycorrhizal fungi (AMF) and under limited and stratified P availability.

Under P‐limiting conditions, plant P efficiency depends on increasing the internal P use efficiency (PUE) or enhancing P acquisition efficiency (PAE; Chen et al. ²⁰²³). In crop plants, increased PAE depends on improved root foraging strategies, including changes in root system architecture (RSA) and morphology, which help plants acquire P under limiting conditions. The modulation of RSA is highly species/genotypic‐specific under P limitation (Gamuyao et al. 2012; Hufnagel et al. ²⁰¹⁴). For example, several genotypes of crop plants, including grasses (e.g., sorghum, maize) and legumes (e.g., common bean, chickpea), are classified as P‐efficient and P‐inefficient based on PAE (Lynch 2011). The production of a shallow root system that can efficiently uptake P from topsoil is a characteristic of P‐efficient genotypes compared to P‐inefficient genotypes with deep root systems (Lynch 2019). The shallow root system is particularly important, as most of the P is present in the topsoil due to the turnover of plant residues in the topsoil together with the low diffusion coefficient of P (Walbridge et al. 1991; Lynch and Brown ²⁰⁰¹). The P‐efficient genotypes also adapt to low P by changing root biomass, number of lateral roots, and root hairs (Zhu et al. 2005; Gahoonia et al. ²⁰⁰⁷; Liao et al. ²⁰⁰¹; Magalhães et al. ²⁰¹¹). For example, under limiting soil P conditions, P‐efficient accessions have greater plasticity to produce higher root biomass with increased lateral roots and root hairs than P‐inefficient accessions (Liu et al. 2023; Hufnagel et al. ²⁰²⁴). Since the higher production of roots and root hairs for greater soil exploration incurs a considerable cost to the plant, the P‐efficient genotypes compensate for this cost by producing metabolically less demanding roots (Lynch 2019). Plants with high PAE may also have enhanced P mining strategies, such as the secretion of carboxylates (eg., citrate, malate, oxalate) that can solubilise P from metal complexes and enzymes such as phosphatases that can mineralise P from organic sources (Richardson et al. 2011). Although the above strategies differ between P‐efficient and inefficient accessions, the presence of AMF that helps plants in P uptake through the mycorrhizal pathway can alter these strategies. The AMF hyphae increase the volume of soil explored beyond the root depletion zone and provide up to 70%–80% of plant P requirement under low soil P availability (Smith and Read 2008) by utilising 4%–20% of the host plant's daily photosynthetic production (Walder and Van Der Heijden ²⁰¹⁵). The hyphal foraging and uptake of nutrients through AMF circumvent the need for extensive root foraging and increased root exudates (Campos et al. 2018). However, whether the AMF differentially alters these strategies in P‐efficient and P‐inefficient accessions is surprisingly less investigated.

The high spatial and temporal heterogeneity in soil nutrient availability is another critical factor driving plant decisions about root placement and proliferation (Hodge 2004). Due to its low mobility in soil, P availability is highly heterogeneous and stratified (Lynch 2019). Moreover, in agricultural soils, the legacy effect of the decadal application of excessive fertilisers and manures results in P stratification, a high P concentration near the soil surface (Rowe et al. 2016). Plants that encounter a nutrient‐rich patch preferentially allocate resources to produce a higher number of lateral roots to exploit the nutrients in the patch. In addition to this morphological plasticity, plants also exhibit high physiological plasticity where the roots can produce carboxylates to solubilise and uptake P from the patch (Neumann and Römheld 1999). The different crop genotypes could vary in their root morphological and physiological plasticity, resulting in differential P uptake from concentrated patches in soil (Zhu et al. 2005). The symbiosis with AMF can help crop genotypes in P uptake from nutrient patches and be more cost‐effective than roots if the cost for hyphal production is less than that for roots and the hyphae can better access the nutrients than the roots (Cui and Caldwell 1996). Under stratified P availability, it is less known whether plants would ‘outsource’ P acquisition by forming a symbiosis with AMF or would adopt a ‘do‐it‐yourself’ strategy by producing more roots to acquire P from patches. Unravelling the link between the different P acquisition strategies (root mediated vs. AMF symbiosis) in relatively P‐efficient and P‐inefficient crop accessions under stratified or uniform P availability will help us formulate efficient P management strategies in agroecosystems. This is particularly critical in the face of fertiliser P‐induced environmental issues, including eutrophication.

To establish and maintain a functional AMF symbiosis, both plant and AMF partners must engage in a molecular dialogue (Wang et al. 2023; Fiorilli et al. ²⁰²⁴) involving several primary and specialised (secondary) plant metabolic pathways (Kaur et al. 2022). Thus, in addition to changes in root morphology under limited soil P, colonisation of roots by AMF can alter another vital root trait, such as the primary and specialised metabolome. Plants colonised with AMF have been found to alter both primary and specialised metabolites for a functional symbiotic association (Lohse et al. 2005; Kaur et al. ²⁰²²). Among the specialised metabolites, flavonoids are of particular interest as they function as signalling molecules in plant‐AMF symbiosis (Kaur and Suseela 2020). However, the phytometabolome of relatively P‐efficient and P‐inefficient accessions under limited and stratified P availability and in the presence of AMF is less known. Understanding the changes in the root primary and specialised metabolome can provide deeper insights into whether AMF can modulate plant metabolome similarly in crop genotypes that differ in root traits for the maintenance of a functional symbiotic association and efficient P uptake.

The main questions that we addressed in this study are‐ (i) is there a tradeoff between AMF colonisation and root proliferation in relatively P‐efficient and P‐inefficient accessions in a P‐limiting environment? (ii) under stratified or uniform distribution of P, does the relatively P‐efficient and P‐inefficient accessions prefer ‘do‐it‐yourself’ or ‘outsourcing strategy’? We hypothesised that relatively P‐inefficient accessions would rely more on AMF for P acquisition than the relatively P‐efficient accessions with improved root traits. We also hypothesised that the above difference in P acquisition strategies would be more pronounced in uniform P availability conditions than in stratified P availability since most plants can increase lateral roots and utilise the nutrient patch efficiently under stratified or concentrated P availability. We predicted that the difference in the P acquisition strategies of different accessions under AMF and P treatments would be reflected in the root morphology and metabolome. To test these hypotheses, we chose sorghum as a model species since it is grown for food, feed, and fibre and sorghum accessions vary widely in their root traits and PAE (Gladman et al. 2022). Previous studies have established that the adaptation of sorghum to low P is mainly due to increased PAE through improved root traits such as increased root surface area and root length that were highly correlated with grain yield under low P availability (Hufnagel et al. 2014, ²⁰²⁴; Bernardino et al. ²⁰¹⁹). Here, we used two accessions each of Sorghum bicolour with higher root surface area and shallow roots (PI‐533752, PI‐597964; hereafter P‐efficient) and those with lower surface area and deep roots (PI‐561073, PI‐629034; hereafter P‐inefficient). The average root surface area and the length of fine roots of P‐efficient accessions (PI‐533752, PI‐597964) were 55%–60% higher than the P‐inefficient accessions (PI‐561073, PI‐629034) under field conditions resulting in higher P uptake and grain yield than P‐inefficient accessions (Liu et al. 2023; Hufnagel et al. ²⁰¹⁴, ²⁰²⁴). We further tested the rooting depth of the selected accessions in a preliminary greenhouse study (details in methods).

2. Materials and Methods

2.1. Greenhouse Experiments

We set up a preliminary greenhouse experiment at Clemson University (greenhouse parameters are similar to that in the main experiment described below) to test the rooting depth of two P‐efficient (PI‐533752 [SC103], PI‐597964 [SC1319]) and two P‐inefficient accessions (PI‐561073 [BTX635], PI‐629034 [RTX437]) by supplying uniform low P (iron phosphate) and uniform high P (potassium phosphate) in autoclaved sand media. Evaluating rooting depth is important as shallow root depth is efficient for topsoil P foraging than accessions with deep roots. We used 15 × 60 cm (diameter x length) plastic sleeve‐lined PVC columns to grow each accession (four replicates per accession, a total of 32 PVC columns). The transparent plastic sleeve was made using a four‐mil thick plastic sheet with the help of a bag sealer. The long PVC column was used to avoid root bounding and it allowed deeper rooting. After 40 days, plants were harvested by placing the PVC cylinder horizontally and pulling out the plastic sleeve entirely intact out of the cylinder. We then cut the plastic sleeve slowly and carefully washed away the sand media by holding the root to observe the root distribution with depth. Root images that captured the entire root depth were used in the ImageJ software (version 1.54i 03, Schneider et al. ²⁰¹²) to measure the rooting depth. The plant dry shoot biomass, shoot P content, and root morphology were obtained using the method mentioned in the main experiment.

The main experiment was set up in a greenhouse at Clemson University in August 2021. The day and night temperatures were 28°C and 23°C, respectively, with supplemental lighting (350–400 W m⁻²) providing 16 h per day length. We used a completely randomised design with a factorial combination of the above four sorghum accessions (two P‐efficient: PI‐533752, PI‐597964, and two P‐inefficient: PI‐561073, PI‐629034), two iron phosphate treatments (stratified vs. uniform), and two AMF treatments (AMF‐inoculated andvs. non‐inoculated control). The treatments are abbreviated as SAMF, SNAM, UAMF and UNAM (S: stratified treatment; U: uniform treatment; AMF: AMF inoculated treatment; NAM: non‐inoculated control treatment; Supporting Information Figure S1). The AMF‐inoculated treatment included a mixture of Rhizophagus intraradices, Rhizophagus clarus, Claroideoglomus Claroideum and Claroideoglomus etunicatum (INVAM; Morton ²⁰⁰⁰). We mixed 5 mL of the sand‐based inoculum of each individual species, and 20 ml of this mixture was applied as the AMF inoculum in the inoculated treatments, while an equal amount of autoclaved sand was added to the non‐inoculated treatments. We used iron phosphate (FePO₄), which is a sparingly water‐soluble form of P [FePO₄; K_sp (solubility product constant) = 1.3 × 10⁻²²], to simulate limiting soil P conditions. The experiment was set up in 12.04 L tree pots (39.37 cm height and 22.86 cm diameter; Steuwe & Sons, Tangent, OR, USA). We used autoclaved river sand (~17 kg per pot) as the substrate. In the stratified P treatment, we first filled 27 cm of the pot with autoclaved sand and then added iron phosphate as a band at a 7.62 cm patch of sand (Supporting Information Figure S1). We applied 20 mL of AMF mixed inoculum to the AMF inoculated pots at the same above‐mentioned 7.62 cm sand patch in both uniform and stratified P treatments. This band application of AMF was followed to ensure high spore density in both stratified and uniform treatments (Supporting Information Figure S1) to ensure maximum colonisation (Kaur et al. 2022). In the uniform P treatment, iron phosphate was uniformly applied throughout the sand (34.62 cm) except the top 4.44 cm. The top 4.44 cm of all treatment pots were filled with autoclaved sand devoid of AMF inoculum or FePO₄ to ensure no contamination of AMF between control and AMF treatments and to avoid disturbing the iron phosphate band while watering the plant. Each treatment had eight replicates with 128 pots (4 accessions x 2 AMF x 2 P treatments x 8 replicates). Plants were supplied with full strength‐P Hoagland solution as per Kaur et al. (2022) with slight modifications. 1.5 L full strength‐P Hoagland solution was applied at sowing, followed by 400 mL full strength‐P Hoagland solution every 4 days till harvest. Plants were supplied with approximately 300 mL DI water daily (split application in the morning and evening) to avoid excessive drainage and disturbance to phosphorus treatments.

2.2. Plant Growth Parameters

Plant carbon dioxide (CO₂) assimilation and stomatal conductance were measured after 28 days of sowing on a fully opened young leaf with a portable gas exchange system (LiCOR 6400‐XTR, Li‐COR, Lincoln, NE, USA). Measurements were taken from 8:30 a.m. to 12:00 p.m. at a photosynthetic photon flux density of 1000 µmol m⁻² s⁻¹ and an atmospheric CO₂ of 400 ppm. Plants were destructively harvested after 30 days of sowing and separated into shoots and roots. The soil sticking to the roots was gently removed immediately after harvest. The rhizosphere soil from the 7.62 cm stratified area was collected and stored at −80°C for enzyme analysis and hyphal length assessment. Fully opened young leaf was collected from each plant and stored at −80°C for metabolomics analysis. The shoots were dried in the oven at 60°C for 48 h to measure the dry shoot weight. The total P content in the shoot was analysed by digestion using a wet ashing procedure and inductively coupled plasma mass spectrometry (ICP‐MS). The shoot tissue P concentration (ppm) was converted to percent tissue P and then normalised with shoot dry weight to obtain tissue phosphorus content (mg P per plant).

2.3. Roots Traits and Mycorrhizal Parameters

The roots were gently washed with distilled water (DI water) to remove the remaining sand from the roots. The roots were separated into lateral roots (≤ 0.4 mm diameter), secondary roots (0.5–1.4 mm) and primary roots (> 1.4 mm). Fresh root biomass was recorded for each replicate after washing the root and gently patting it dry. The dry root biomass was not recorded, as sub‐samples of fresh roots were used in different analyses. A subsample of the lateral roots was stored in 70% ethanol for determining AMF percent colonisation, and another subsample was rapidly frozen on dry ice upon harvest and stored at −80°C for metabolomics analysis. The remaining roots were stored at 4°C for root morphology analysis with WinRHIZO software. We selected the lateral roots for root morphology as they have important absorptive functions and accommodate AMF. The lateral roots from each treatment were scanned using a desktop scanner. The images were then processed using WinRHIZO software (Regent Instruments Inc., Québec, Canada). After scanning, the roots were dried and weighed separately. Using the dry weight and data from WinRHIZO, including length, diameter and volume, we calculated the specific root length (SRL; m g⁻¹) and root tissue density (RTD; mg cm⁻³) of lateral roots (Chen et al. 2013). To assess the AMF percent colonisation, the subsample of lateral roots stored in 70% Ethanol was cleaned with 10% KOH and stained with Trypan blue dye as per INVAM (Morton 2000). The percent root colonisation was calculated by observing the presence/absence of AMF structures (vesicles, arbuscules, hyphae) in at least 100 intersections of the slide under a compound microscope (Kaur et al. 2022). The mycorrhizal growth response (MGR) was calculated as follows.

$$\frac{\text{MGR}(\%)=\text{SDW}(\text{AM}\text{fungi}-\text{inoculated}\text{treatment})–\text{mean}\text{SDW}(\text{Non}-\text{inoculated}\text{treatment})}{\text{mean}\text{SDW}(\text{Non}-\text{inoculated}\text{treatment})}\times 100,$$

where SDW is the shoot dry weight. For all the treatments, we also calculated a 95% confidence interval to assess the significance of MGR (%) of inoculated treatments from the non‐inoculated treatment, and when the interval did not overlap with zero, the treatment was considered significant (Kaur et al. 2022). We replaced SDW with P content (mg per shoot) in the above equation to estimate MPR (%). We also assessed the extra‐radical hyphal length as per Cornejo and Aponte (2020) with slight modifications. Briefly, 3 g of soil (from the stratified area: Supporting Information Figure S1) was shaken with 20 mL of 32.7 mM sodium hexametaphosphate solution for 60 min. This mixture was then passed through a series of 250, 53 and 20 μM sieves, and the material from the 20 and 53 μM sieves was transferred to a 50 mL tube and stained with 0.005% Trypan blue dye in lactic acid. The stained hyphae were passed through a 20 μM sieve, washed, and suspended in tap water. This hyphal suspension was then passed through a gridded nitrocellulose membrane. Hyphal counts were made using a microscope at 400x by performing a zigzag scan across the entire grid and counting the intersections between the grid and stained hyphae. The intersections were converted to hyphal length using the formula proposed by Newman (1966).

2.4. Acid Phosphatase Enzyme Assay

We used 5 g of soil (from the stratified area: Supporting Information Figure S1) in 250 mL sodium acetate buffer (50 mM) to determine the activity of acid phosphatase enzyme using 4‐Methylumbelliferyl phosphate and 4‐Methylumbelliferone (MUB; 10 μM) as a specific substrate and a standard, respectively. The assays were conducted in black 96‐well plates, and the amount of methylumbelliferone produced after 2 h of incubation at 25°C was measured by using a microplate‐fluorometer (Excitation‐365 nm, Emission‐450 nm; Saiya‐Cork et al. ²⁰⁰²). The emission measured by fluorometer (blanks, standards, sample controls and samples) was used to calculate the enzyme activity (DeForest ²⁰⁰⁹).

2.5. Metabolomics Analysis

We used lateral roots of all accessions and leaves of one P‐efficient accession for primary metabolite analysis. The lateral roots from one P‐efficient (PI‐533752) and one P‐inefficient (PI‐561073) accession were used for specialised metabolite analysis. We did not select all accessions for the primary metabolite analysis of the leaves and specialised metabolite analysis of roots since there was no accessional difference regarding growth parameters such as shoot biomass and P content. The leaf tissues for the analysis were excised from sorghum leaves using a cork borer, avoiding the midrib. The tissues were extracted using the method described by Kaur et al. (2022) and Xia et al. (2022). Details of extraction are provided in Supporting Information Methods S1.

To analyse the primary metabolites, we transferred 50 µL of the water‐methanol phase from each sample to vials with glass inserts, spiked with 15 µL of internal standard (50 ppm myristic acid + 20 ppm ribitol) and dried under vacuum. The dried samples were methoximated with 20 µL Methoxyamine hydrochloride (40 mg mL⁻¹ in pyridine) and derivatized by 80 µL MSTFA + 1% TMCS with 5 ppm alkanes before analysing on a gas chromatography‐mass spectrometer (GC‐MS). A standard mixture of 52 compounds of primary metabolites was derivatized similarly and analysed along with the samples. Details of GC‐MS parameters are provided in Methods S2.

GC Data processing‐ GC‐MS files for all samples were processed using MS‐Dial (4.70) for peak detection, deconvolution, identification and alignment. The GC‐MS raw files (CDF format) were converted to analysis base file (ABF) format to input to MS‐DIAL. The peak picking was done based on the quantifier ion m/z after deconvolution. Compound annotation was based on mass spectral fingerprints and retention‐index (RI) matches by comparing the Kovats RI built on n‐alkanes standards (C10‐C30) with a database (MS‐DIAL metabolomics MSP) and/or literature. Sugars were confirmed by running separate external standards. The peak area was used as the response.

To analyse specialised metabolites, 50 µL methanol extract per root sample was spiked with 50 µL internal standard (1 ppm resveratrol, 99 atom % 13 C, in methanol). The samples were analysed in Ultimate 3000 high‐performance liquid chromatography (UHPLC; Thermo Scientific, Waltham, MA, USA) coupled to an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific) equipped with an electrospray ionisation source operated in positive polarity (Suseela et al. 2020; Kaur et al. ²⁰²²). Details of LC‐MS and orbitrap parameters are provided in Supporting Information Methods S3.

The data from orbitrap was analysed using a node‐based workflow of Compound Discoverer 3.3. Multiple adducts of the same compound were grouped to minimise false positives. The tentative compound identities were established by comparing the accurate mass (error < 5 ppm) and fragmentation pattern of the mass features with various online mass spectral libraries (MoNA, mzCloud, Flavonoid Search, Mass Bank, HMDB). Compound IDs were also confirmed by compound class prediction of CANOPUS (Dührkop et al. 2021) and CSI: Finger ID from the SIRIUS v4.4. Details of MZmine and CANOPUS analysis are provided in Supporting Information Methods S4. In the specialised metabolites, we focused specifically on flavonoids, which are involved in different aspects of plant‐AMF symbiosis.

2.6. Statistical Analysis

The effect of treatments on plant biomass, shoot P content, photosynthesis, SRL, RTD, hyphal length and acid phosphatase activity were analysed using a three‐way ANOVA with accessions, AMF treatments, and phosphorus treatments as the fixed factors. Data were log or square root transformed when necessary to satisfy the assumptions of normality and variance. Individual treatment differences were examined using Tukey's honestly significant difference (HSD) multiple comparisons test. In statistical analysis, when the interaction effects (AMF x P stratification treatment) were not significant, and only the main effect of treatments (AMF or P treatment) was significant, we provided only the main effect graphs and avoided the interaction effect graphs. SigmaPlot v.14 (Systat Software Inc., Chicago, IL) was used for statistical analysis and graphical interpretation. For both primary and specialised metabolites, peak areas were used as a measure of metabolite abundance. The multivariate analysis of metabolites (heatmaps and PCA) was conducted using Metaboanalyst v5.0 (Pang et al. 2021), where peak areas were normalised using log transformation and autoscaling. For primary metabolite analysis, peak areas of all annotated metabolites were first normalised with ribitol (internal standard), and significant metabolites were then identified using a three‐way ANOVA followed by Tukey's HSD post‐hoc test.

3. Results

3.1. AMF Treatments Altered Plant Growth Parameters

The P‐efficient and P‐inefficient sorghum accessions we selected had different rooting depths, as observed from the preliminary greenhouse experiment. PI 629034 (P‐inefficient) had higher rooting depth than both P‐efficient accessions, regardless of the forms of P application (p = 0.043; Supporting Information Figures S2, S3). The specific root length (SRL) of lateral roots did not vary with accessions or P treatment. As expected, plants subjected to potassium phosphate had higher shoot biomass and shoot P than those supplied with iron phosphate (p < 0.001). These accessions, when subjected to different AMF (AMF vs. non‐AMF) and P treatments (stratified vs. uniform) in the main greenhouse experiment, did not show any accessional difference regarding different growth parameters. We observed that the dry shoot biomass of all sorghum accessions in the P treatments varied by AMF treatments (p < 0.05; Figure 1a). Plants subjected to SAMF had the highest dry shoot biomass, followed by UAMF, SNAM and UNAM. Plants in SAMF had an 83% increase in shoot biomass compared to SNAM and a 90% increase in UAMF compared to UNAM. The photosynthesis measured in terms of CO₂ assimilation rate depended on the interaction of phosphorus and AMF treatments (p < 0.05; Supporting Information Figure S4a). The SAMF and UAMF treatments had higher CO₂ assimilation rates, followed by SNAM and UNAM treatments. Stomatal conductance for all the accessions varied with AMF treatments (p < 0.001; Supporting Information Figure S4b), where AMF had higher stomatal conductance than NAM, irrespective of phosphorus treatments and accessions. The shoot P content varied by the main effects of AMF and P treatments where AMF and stratified had 87% and 40% higher tissue P content (mg g⁻¹ shoot) than NAM and uniform treatments, respectively (p < 0.001; Figure 1b,c). The results indicated the effect of AMF in enhancing plant dry biomass and tissue P content in all accessions.

Figure 1.

Dry Shoot biomass (a), and shoot tissue phosphorus content (b and c). Values in (a) represent mean ± SE (n = 8) and in (b and c) represent mean ± SE (n = 3). Bars with different letters indicate a difference (Tukey's honestly significant difference [HSD]) between treatments. AMF, mycorrhiza inoculated treatment; NAM, non‐inoculated control treatment; S, stratified phosphorus application; SAMF, stratified AMF treatment; SNAM, stratified NAM treatment; U, uniform phosphorus application; UAMF, uniform AMF treatment; UNAM, uniform NAM treatment.

3.2. Root Parameters Reflected ‘Outsourcing’ Versus ‘Do It Yourself’ Strategies

The different P‐acquisition strategies of sorghum were reflected in the various root parameters. The total fresh root biomass of all sorghum accessions varied with the main effect of AMF (p < 0.001; Figure 2a) and phosphorus treatments (p < 0.001; Figure 2b). Plants subjected to AMF and stratified treatments had 78% and 51% higher fresh root biomass than NAM and uniform treatments, respectively. Total root biomass also varied by the main effect of sorghum accessions where PI 533752 and PI 561073 had higher root biomass than PI 629034 (p < 0.001; Figure 2c). The fresh root biomass in the stratified area varied by an interaction of AMF and phosphorus treatments (p < 0.05; Figure 2d). In the stratified area, independent of sorghum accessions, SAMF treatment had 83% higher fresh root biomass than SNAM, and UAMF had 88% higher fresh root biomass than the UNAM treatment. The SRL and RTD of lateral roots varied by the main effect of AMF and P treatments. SRL was higher in NAM (38%; p < 0.001; Figure 2e) and uniform P treatment (13%; p < 0.05; Figure 2f) compared to AMF and stratified P treatment, respectively. RTD of lateral roots had an inverse trend compared to SRL. RTD was higher in AMF (p < 0.001; Figure 2g) and stratified P treatment (p < 0.05; Figure 2h) compared to NAM and uniform P treatment, respectively. The increase in SRL reflects the ‘do‐it‐yourself’ strategy of the plant for P acquisition which was observed in the NAM treatment. This trend was reversed in AMF‐inoculated treatments where the roots were shorter and thicker (lower SRL and higher RTD), indicating an outsourcing strategy.

Figure 2.

Total fresh root biomass (a–c), fresh root biomass‐stratified area (d), specific root length (SRL) of lateral roots (e and f), and root tissue density (RTD) of lateral roots (g and h). Values in (a–d) represent mean ± SE (n = 8) and in (e–h) represent mean ± SE (n = 4). Bars with different letters indicate a difference (Tukey's honestly significant difference [HSD]) between the treatments. Key as in Figure 1. [Color figure can be viewed at wileyonlinelibrary.com]

3.3. Mycorrhizal Parameters and Phosphatase Activity Indicated an Outsourcing Strategy

The mycorrhizal parameters and phosphatase enzyme activity further indicated an outsourcing strategy by plants in the presence of AMF. The lateral roots of AMF inoculated treatments had 80% AMF root colonisation compared to NAM. We observed no mycorrhizal structures in NAM (p < 0.001; Supporting Information Figure S5), confirming no cross‐contamination of AMF in NAM. The AMF‐inoculated treatments had positive MGR and MPR, indicating that all accessions benefited from AMF in stratified and uniform P treatments. The MGR and MPR varied by an interaction of phosphorus treatments and accessions (p < 0.001; Supporting Information Figure S6a, p < 0.05; Supporting Information Figure S6b). Although dry biomass was higher in SAMF compared to UAMF, the MGR was higher in three of the accessions under uniform treatment. This is due to the lowest mean biomass in UNAM compared to all other treatments (Figure 1a). The P‐inefficient accession, PI‐561073, had the highest MGR and MPR in the UAMF treatment compared to all accessions in other treatments. PI‐533752 (P‐efficient accession) had a higher MGR in the UAMF treatment compared to the SAMF treatment. The hyphal length varied with the interaction of phosphorus and AMF treatments (p < 0.05; Figures 3a and Supporting Information Figure S7). SAMF treatment had the highest hyphal length, followed by UAMF treatment and control (NAM treatments). Plants inoculated with AMF had higher acid phosphatase activity than the plants in the non‐inoculated (NAM) treatment (p < 0.05; Figure 3b), indicating that under natural field settings, AMF enhancing acid phosphatase can help in the mineralisation of organic P and plant P uptake.

Figure 3.

Hyphal length (a) and acid phosphatase activity (b) in the rhizosphere soil from the stratified area. Values in (a and b) represent mean ± SE (n = 5). Bars with different letters indicate a difference (Tukey's honestly significant difference [HSD]) between the treatments. Key as in Figure 1. [Color figure can be viewed at wileyonlinelibrary.com]

3.4. Root Metabolite Profile Varied With AMF Treatments

The primary and specialised metabolites varied with AMF treatments. Based on the predefined deconvolution parameters, GC‐MS analysis positively identified 29 primary metabolites in leaves and roots (Supporting Information Table S1). Organic acids in roots varied by the main effects of AMF and P treatments where the organic acids were higher in the AMF treatment compared to the NAM treatment (p < 0.001; Figure 4a) and the stratified treatment compared to the uniform treatment (p < 0.001; Figure 4b). The roots of UNAM treatments had a higher abundance of amino acids followed by SNAM treatment than both SAMF and UAMF treatments (p < 0.05; Figure 4c). The primary metabolites clustered based on AMF and NAM treatments irrespective of accessions and P stratification (Supporting Information Figure S8). Amino acids such as aspartic acid, asparagine, glutamine, tryptophan and serine were significantly abundant in the NAM treatment (Supporting Information Figure S9). Alanine, glutamic acid, isoleucine, valine and tyrosine were abundant in the AMF treatments. Asparagine and aspartic acid were highly abundant in uniform treatments compared to stratified treatments. All organic acids (malic acid, citric acid, shikimic acid, succinic acid and aconitic acid) were higher in the AMF treatments than NAM treatments (Supporting Information Figure S9). Sugars had a contrasting effect in mycorrhizal and non‐mycorrhizal treatments. Fructose, glucose and ribose were significantly higher in AMF treatments. Sucrose, maltose, xylose and galactose were significantly higher in NAM treatments (Supporting Information Figure S9). Total organic acids for leaf metabolites varied with the interaction of AMF and P treatments (Figure 4e). Organic acids were highest in leaves of plants under SAMF and UAMF treatment, followed by UNAM and SNAM treatment (p = 0.05; Figure 4d). Total amino acids in leaves varied with AMF treatment where amino acids were higher in plants under NAM than AMF‐inoculated treatments (p < 0.001; Figure 4e).

Figure 4.

Total organic acids (a and b) and amino acids (c) in lateral roots of all accessions, and total organic acids (d) and amino acids (e) in leaves of PI‐561073. Values in (a–c) represent mean ± SE (n = 5) and in (d and e) represent mean ± SE (n = 4). Bars with different letters indicate a difference (Tukey's honestly significant difference [HSD]) between treatments. Key as in Figure 1. [Color figure can be viewed at wileyonlinelibrary.com]

After accounting for the multiple adduct formation in Compound Discoverer 3.3, we detected 2698 metabolic features of specialised metabolites. The specialised metabolic features expressed in different treatments were grouped into four distinct clusters (Figure 5). Cluster I represented metabolic features higher in all AMF inoculated treatments, and Cluster II represented higher in control (NAM) treatment. Clusters III and IV represented metabolic features inherent to PI‐561073 (P‐inefficient) and PI‐533752 (P‐efficient) accessions, respectively. The mass features of specialised metabolites detected were classified by CANOPUS into 13 superclasses belonging to kingdom organic compounds (Figure 6). Out of these mass features, 490 (17.71%) were upregulated, and 343 (12.39%) were downregulated in AMF treatments compared to NAM (Figure 6). This difference in compound classes was mainly contributed by lipids and lipid‐like molecules (6.58% upregulated in AMF) and organic acids and derivatives (7.48% upregulated in AMF). Interestingly, AMF inoculation resulted in the downregulation of only lignans, neolignans and related compound classes.

Figure 5.

Heatmap and hierarchical clustering analysis of 2698 mass features of specialised metabolites (log‐transformed and auto‐scaled) of PI‐533752 (P‐efficient) and PI‐561073 (P‐inefficient) accessions under various phosphorus and AMF treatments. Each column represents a replicate of a treatment, and each row represents a metabolic feature (the metabolic features are not provided in the figure since there are 2698 features). Blue to red on the scale represents an increase in the abundance of a metabolic feature. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 6.

Bar plot displaying the effect of AMF on the root secondary metabolite profile in the superclasses predicted by CANOPUS classification. The X‐axis represents various superclasses, and the Y‐axis represents the percentage of compounds upregulated (red) or downregulated (blue) (p < 0.05, FDR < 0.05, log2FC | > |1 | ) in AMF inoculated accessions with respect to controls (NAM). [Color figure can be viewed at wileyonlinelibrary.com]

We annotated 22 flavonoids from the specialised metabolite analysis of roots. Based on the heatmap, annotated flavonoids clustered into two groups (Figure 7a; Supporting Information Table S2). The flavonoids such as apigenin/genistein glycoside, daidzein glycoside, daidzein and formononetin were abundant in PI‐533752 (P‐efficient accession) and PI‐561073 (P‐inefficient accession) irrespective of AMF and P treatments. However, flavonoids such as luteolin/kaempferol, isoliquiritigenin/liquiritigenin and Naringenin were abundant in AMF‐inoculated treatments compared to NAM treatments, irrespective of accessions and P treatments. The PCA of identified flavonoids revealed that the accessions were separated along the PC 1 axis, which explained 44% of the variation in the data. The AMF‐inoculated treatments separated from NAM along the PC 2 axis, which explained 26% of the variation in the data (Figure 7b). Apart from flavonoids, we also identified blumenols, which were abundant in all AMF inoculated roots compared to NAM (Supporting Information Table S3; Figure S10).

Figure 7.

Heatmap and two‐way hierarchical clustering analysis (a) and principal component analysis (PCA) (b) of tentatively identified flavonoids in PI‐533752 (P‐efficient) and PI‐561073 (P‐inefficient) accessions under various phosphorus and AMF treatments. In the heatmap, the colour of each cell depicts the abundance of an individual metabolite feature. Blue to red on the scale represents an increase in the abundance of a metabolic feature. In PCA, data points represent biological replicates, and ellipses represent a 95% confidence interval. (Key as in Supporting Information Table S2: Luteo [Luteolin], Kae [Kaempferol], Apige [Apigenin], Gen [Genistein], Formo [Formononetin], Narin [Naringenin], Hespere [Hesperetin]). [Color figure can be viewed at wileyonlinelibrary.com]

4. Discussion

Our results revealed that the AMF treatments regulated the strategies for P uptake in the relatively P‐efficient and P‐inefficient accessions with inherently different root traits. All accessions inoculated with AMF had higher tissue P content and shoot biomass than NAM, facilitated by higher root biomass and extraradical hyphal length, indicating that AMF had a critical role in mediating plant P uptake and thus suggesting an ‘outsourcing’ strategy. The increase in root biomass could be partly contributed by AMF structures (arbuscules and hyphae) in the roots, as these roots were also shorter and thicker with lower SRL and higher RTD. In contrast, all accessions in the non‐inoculated treatment had longer and thinner roots (higher SRL and lower RTD), indicating that the plants had to adopt a ‘do it yourself’ strategy for P uptake in the absence of AMF. AMF also helped with P uptake in stratified and uniform P availability. This emphasises that irrespective of the inherent accessional differences in root traits contributing to differential PAE, the inoculation with AMF can potentially enhance soil P uptake. The root primary and specialised metabolite profiles also mirrored the above P uptake strategies. Schematic diagram 1.

Schematic diagram 1.

Figure Root strategies for phosphorus (P) uptake in P‐efficient and P‐inefficient sorghum accessions under AMF‐inoculated and non‐inoculated control (NAM) treatments. [Color figure can be viewed at wileyonlinelibrary.com]

4.1. AMF Alters Plant Strategy of ‘Do‐It‐Yourself’ to ‘Outsourcing’ for Efficient P Uptake

Plant roots that perform both absorptive and transport functions must fine‐tune their strategies for efficient resource acquisition under heterogeneous nutrient availability. The variation in root functional traits for resource acquisition is currently depicted by a two‐dimensional trait framework that includes the collaboration and conservation axes (Bergmann et al. 2020). The collaboration axis represents a gradient of strategies from ‘do‐it‐yourself’ to ‘outsourcing’ via mycorrhizal fungi. However, the significance of this collaboration axis has not been explored in relatively P‐efficient and P‐inefficient crop accessions where PAE is fundamentally defined based on the root traits. We hypothesised that P‐inefficient accessions would benefit more from AMF inoculation than P‐efficient accessions. However, our study revealed that AMF inoculation increased shoot biomass and P uptake in all accessions (Figure 1). In the AMF treatments, the higher RTD and lower SRL, (Figure 2), irrespective of accessions indicate that in AMF‐inoculated plants, the lateral roots were thicker due to the accommodation of AMF (Heck et al. 2016) and, shorter since the proliferation of fungal hyphae might have acted as a replacement for longer lateral roots, resulting in higher P uptake. A previous study found that the SRL of two barley genotypes was significantly reduced (10%–30%) under AMF treatment with a corresponding increase in shoot P content (Zhu et al. 2003). In general, a negative correlation is observed between the SRL of fine roots and mycorrhizal colonisation (Stiblikova et al. 2022; Wen et al. ²⁰¹⁹). Our study furthers this knowledge by revealing that the ‘outsourcing’ strategy via AMF was highly efficient under P‐limited conditions in both P‐efficient and P‐inefficient accessions, irrespective of the inherent PAE of the accessions. Although the accessions we selected were classified as relatively P‐efficient and P‐inefficient based on root characteristics, P uptake, and grain yield under field settings (Hufnagel et al. 2014), we did not observe this inherent difference in P uptake between these accessions in the non‐AMF control. This could be potentially due to the form of P source (iron phosphate in our study) and the harvest of the plant at the peak vegetative stage, as our study was conducted under greenhouse conditions.

4.2. Outsourcing the P Uptake via AMF Is Efficient Under Stratified and Uniform P Availability

In our study, we observed that the plants inoculated with AMF had higher biomass in both stratified and uniform P treatments compared to the respective NAM treatments, indicating that AMF helped to enhance biomass under stratified and uniform P availability (Figure 1). Overall, the plants utilised the ‘outsourcing’ strategy in the stratified and uniform treatments in the presence of AMF, indicating that outsourcing via AMF is potentially cost‐effective for nutrient uptake in all accessions. However, SAMF had 43% higher shoot biomass compared to UAMF (Figure 1a), indicating that AMF was highly efficient under concentrated P availability. This trend in dry shoot biomass also mirrored the fresh root biomass from the stratified area (Figure 2d), where SAMF had 52% higher fresh root biomass than UAMF. The higher root biomass in the stratified area in the presence of AMF (Figure 2d) suggests the potential of AMF in modulating root responses, resulting in efficient P uptake and shoot biomass. Moreover, the AMF extraradical hyphal length was also higher in the stratified treatment, contributing to higher P uptake and shoot biomass. A recent study with maize genotypes also reported that the mycorrhizal pathway was more cost‐effective to plants than the direct pathway under suboptimal soil P (Zhang et al. 2021) due to the increase in extraradical hyphal length at low P (Chu et al. 2020; Sawers et al. ²⁰¹⁷). Our results suggest that enhancing AMF symbiosis could help crops in efficient P uptake, particularly in agroecosystems where P availability is highly heterogeneous.

4.3. Primary and Specialised Metabolites Mirrored AMF Treatments

Symbiotic association with AMF induces changes in the host plant metabolome (Kaur and Suseela 2020). The root primary and specialised metabolites varied with AMF and NAM treatments, reflecting the changes in plant P uptake and growth parameters. Plant‐mycorrhizal symbiosis alters primary metabolites as plants provide carbon as sugars and fatty acids to AMF (Jiang et al. 2017). Compared to non‐mycorrhizal plants, plants with AM symbiosis exhibited higher photosynthetic capacity, which helps to increase the level of sugars in roots (Augé et al. 2016; Boldt et al. ²⁰¹¹). However, the carbon cost incurred in the AMF treatments was compensated by higher P uptake and an increase in dry shoot biomass in the AMF treatments compared to NAM. Among the primary metabolites, the organic acids in roots and leaves were more abundant in AMF‐inoculated treatments than in the NAM, irrespective of P treatments (Figure 4). This is because colonisation with AMF upregulates the tricarboxylic cycle in plants to produce more energy for supporting the AMF symbiosis (Rivero et al. 2015; Kaur and Suseela ²⁰²⁰). Moreover, in the AMF inoculated treatments, hyphal exudates can facilitate the uptake of soil P (Andrino et al. 2021). Plants under the P‐stress increase the production of organic acids to solubilise the unavailable P from the soil (Shen et al. 2011). Different organic acids vary in their P mobilisation, where malate and citrate are important compounds released by roots under P deficiency (Wang and Lambers 2020). The organic acids (Figure 4), including malic acid and citric acid were also higher in stratified versus uniform treatment, irrespective of AMF inoculation. This could be due to the presence of a hotspot of sparingly soluble iron phosphate in the stratified treatment, where the organic acids can mobilise a higher amount of P for uptake from iron phosphate. In contrast, the amino acids were most abundant in the uniform NAM treatment, followed by stratified NAM, compared to the AMF treatments (Figure 4), potentially due to P stress experienced by the plants in the NAM treatments. Asparagine had a higher abundance in NAM treatments compared to AMF treatments (Supporting Information Figure S10b). Under P‐stress, ethylene biosynthesis is upregulated, leading to plant root morphology changes (Lynch and Brown 1997), enabling the roots to increase soil exploration (Ma et al. 2003). The higher abundance of asparagine in NAM could be a P‐stress response, as the upregulation of ethylene biosynthesis leads to the production of hydrogen cyanide (Wang et al. 2002), which is further converted to asparagine.

The AMF colonisation increased the abundance of specialised metabolites potentially necessary to establish and maintain a functional symbiosis, which further benefits plants under abiotic stress (Kaur et al. 2022). Among the specialised metabolites, some flavonoids act as signalling molecules that help in spore germination, hyphal branching, fungal proliferation and arbuscule formation during AMF symbiosis, while others could have inhibitory effects on AMF, depending on the chemical group (Vierheilig et al. 1998; Chen et al. ²⁰²⁵). In our study, some flavonoids were upregulated in AMF treatments compared to NAM (Figure 7), including luteolin, which was shown to enhance the entry points and hence the colonisation rate of AMF (Scervino et al. 2007), and liquiritigenin (4,7‐dihydroxyflavonone) increased the spore germination of Glomus etunicatum (Tsai and Phillips 1991). Unlike flavonoids, blumenols, an important biomarker of AMF symbiosis, were abundant in all AMF treatments compared to NAM (Supporting Information Figure S10). Recently, blumenols were positively correlated with lipid accumulation and plant fitness (You et al. 2023). Our results also showed a greater abundance of lipids with AMF inoculation (Figure 6), as lipids form an essential source of carbon to AMF from the plant (Shi et al. 2023). Overall, the primary and specialised metabolites were reprogrammed by AMF inoculation and revealed a pattern that supported a functional symbiotic association that benefited all accessions.

5. Conclusion

Under P limitation, the RSA of sorghum is extensively remodelled, involving epigenetic and transcriptional changes to increase lateral root growth, root length and surface area for better P uptake (Gladman et al. 2022). Our study, using multiple crop accessions and several AMF species, provides further insights into how the presence of AMF modulates plant P uptake strategies similarly in sorghum accessions that vary in their root traits. Overall, our results revealed that irrespective of inherent root traits of sorghum accessions, the presence of AMF resulted in switching the phosphorus acquisition strategy of the plant from ‘do‐it‐yourself’ to ‘outsourcing’ via AMF. The outsourcing strategy could reduce the carbon cost of plants, thus aiding plants in acquiring higher biomass and shoot P content with the AMF hyphal‐mediated P uptake. Our results underscore that the role of AMF should be considered when selecting crop accessions for better P uptake. Our results also emphasise that enhancing the AMF symbiotic efficiency in agroecosystems can improve P uptake in crop accessions and reduce reliance on P fertilisers.

Supporting information

Supporting_Information_PEC_revised.

Data Availability Statement

The authors declare that the data supporting the findings of this study are available from the corresponding author upon request.