Carbon for nutrient exchange between arbuscular mycorrhizal fungi and wheat varies according to cultivar and changes in atmospheric carbon dioxide concentration

Abstract Arbuscular mycorrhizal fungi (AMF) form symbioses with most crops, potentially improving their nutrient assimilation and growth. The effects of cultivar and atmospheric CO2 concentration ([CO2]) on wheat–AMF carbon‐for‐nutrient exchange remain critical knowledge gaps in the exploitation of AMF for future sustainable agricultural practices within the context of global climate change. We used stable and radioisotope tracers (15N, 33P, 14C) to quantify AMF‐mediated nutrient uptake and fungal acquisition of plant carbon in three wheat (Triticum aestivum L.) cultivars. We grew plants under current ambient (440 ppm) and projected future atmospheric CO2 concentrations (800 ppm). We found significant 15N transfer from fungus to plant in all cultivars, and cultivar‐specific differences in total N content. There was a trend for reduced N uptake under elevated atmospheric [CO2]. Similarly, 33P uptake via AMF was affected by cultivar and atmospheric [CO2]. Total P uptake varied significantly among wheat cultivars and was greater at the future than current atmospheric [CO2]. We found limited evidence of cultivar or atmospheric [CO2] effects on plant‐fixed carbon transfer to the mycorrhizal fungi. Our results suggest that AMF will continue to provide a route for nutrient uptake by wheat in the future, despite predicted rises in atmospheric [CO2]. Consideration should therefore be paid to cultivar‐specific AMF receptivity and function in the development of climate smart germplasm for the future.

An ever-increasing human population (Gerland et al., 2014), depletion of natural resources such as rock phosphate (Cordell, Drangert, & White, 2009) and rising energy prices are making fertilizer and pesticide production unsustainable. In the context of global climate change, future food security is far from assured (Godfray et al., 2010).
Associating with AMF may confer further benefits on host plants beyond improving access to soil nutrients, such as improving plant growth, water uptake (Ruiz-Lozano et al., 2016) and priming of host plant defence responses (Cameron, Neal, Wees, & Ton, 2013), leading to increased tolerance and/or resistance to pests and diseases (Berdeni et al., 2018;Jung, Martinez-Medina, Lopez-Raez, & Pozo, 2012).
Taking consideration of AMF in widescale agricultural management decisions requires changes in current practice, although it has been argued that sufficient data corroborating the nutritional benefit of AMF in agricultural crops to warrant these shifts are currently lacking Ryan & Graham, 2018). A prevailing assertion is that cereals are generally negatively or neutrally affected by AMF colonization ; the fungi are assumed to offer little nutritional benefit to plants selectively bred for fine and dense root architecture optimized for nutrientacquisition efficiency, especially under high-nutrient environments Wen et al., 2019;Zheng et al., 2018). Despite two meta-analyses suggesting an overall benefit of AMF to crop nutrient uptake and grain yield (Lekberg & Koide, 2005; Lehmann, Zheng, You, & Rillig, 2019), a sceptical view remains in the literature with regard to the utility of AMF in modern and future agriculture (e.g. Ryan & Graham, 2018).
Atmospheric CO 2 concentrations ([CO 2 ]) have increased rapidly because of anthropogenic activities since preindustrial times, from 280 ppm in 1750 to concentrations in excess of 400 ppm today (Meinshausen et al., 2011). Climate model projections suggest that atmospheric [CO 2 ] will continue to rise, potentially reaching 800 ppm atmospheric [CO 2 ] by the end of the century (Meinshausen et al., 2011) if steps to curb emissions are not taken. The 'carbon fertilisation effect' is responsible for increased rates of carbon fixation under elevated atmospheric [CO 2 ] (hereafter eCO 2 ), especially among C 3 species in temperate zones (Ainsworth & Long, 2005;McGrath & Lobell, 2013;O'Leary et al., 2015) which include some of the world's most economically and socially important plants. As photosynthesis is not currently carbon-limited at ambient atmospheric [CO 2 ] (hereafter aCO 2 ; Fitzgerald et al., 2016), plants grown at eCO 2 generally show reduced photorespiratory losses and increased net photosynthetic rates. The extent to which increasing atmospheric [CO 2 ] will impact crop-AMF associations remains unclear (Cotton, 2018 (Alberton, Kuyper, & Gorissen, 2005;Drigo et al., 2013;Field et al., 2012;Treseder, 2004). Furthermore, recent evidence even suggests that AMF carbon acquisition from host plants might directly increase rates of carbon fixation (Gavito, Jakobsen, Mikkelsen, & Mora, 2019), potentially by ameliorating end-product inhibition of photosynthesis (Arp, 1991). Greater C acquisition by AMF may enable further hyphal proliferation through soil and thus increase their assimilation of mineral nutrients and subsequently increase transfer to host plants. However, whether this hypothetical positive feedback is realized in AMF-plant symbioses is not clearly supported by the available data (Cotton, 2018).
The nature and extent of atmospheric [CO 2 ] effects on AMF are complex (Cotton, 2018). Increased plant N uptake via AMF under eCO 2 has been demonstrated both in wild grasses, such as Avena fatua (Cheng et al., 2012) and in domesticated crop plants, including wheat Triticum aestivum L. (Zhu, Song, Liu, & Liu, 2016). In contrast, AMF-mediated P uptake in vascular plants appears to be less affected by changes in atmospheric [CO 2 ]. Mycorrhizal P uptake was not increased by eCO 2 in Pisum sativum (Gavito, Bruhn, & Jakobsen, 2002;Gavito, Schweiger, & Jakobsen, 2003), Medicago truncatula or Brachypodium distachyon (Jakobsen et al., 2016). Similarly, Plantago lanceolata showed decreased 33 P acquisition via AMF per unit of plant-fixed carbon allocated to the fungi in eCO 2 conditions (Field et al., 2012). Host plant genotype must also be considered when investigating the effect of environmental perturbation on symbiotic functioning between crops and AMF; intraspecific diversity is an important driver of variation in these interactions (Johnson, Martin, et al., 2015). As a result of intensive crop breeding to promote various economically important traits, modern crop cultivars vary in their receptiveness to colonization by AMF (Lehnert, Serfling, Enders, Friedt, & Ordon, 2017;Lehnert, Serfling, Friedt, & Ordon, 2018) and therefore potentially also vary in carbon-for-nutrient exchange between symbiotic partners in both aCO 2 and eCO 2 atmospheric conditions.
Here we address the critical research question, "How do eCO 2 and plant host genotype affect carbon-for-nutrient exchange between wheat and arbuscular mycorrhizas?" Using 15 N, 33 P and 14 C isotope tracers across three modern wheat (T. aestivum L.) cultivars, we determined (a) the extent to which AMF contribute to assimilation of N and P from soil, and (b) the extent to which wheat transfers C to extraradical mycelia of their fungal symbionts in three modern wheat (T. aestivum L.) cultivars at aCO 2 (440 ppm) and eCO 2 (800 ppm), to simulate the predicted increase in atmospheric [CO 2 ] over the next 80 years (Meinshausen et al., 2011). Specifically, we tested the hypotheses that (a) AMF would acquire greater amounts of plant-fixed C under future climate eCO 2 scenarios, and (b) increased C allocation would increase transfer and assimilation of 15 N and 33 P tracers from the AMF to the plant across all cultivars tested.

| Wheat pregermination and AMF inoculation
Seeds of bread wheat (T. aestivum L., cv. 'Avalon', 'Cadenza', 'Skyfall'; RAGT Seeds, Cambridgeshire, UK) were surface sterilized using Cl 2 gas (Method S1) and incubated on moistened filter paper for 5 days to germinate. Avalon and Cadenza were selected as they are parent lines of a reference population currently used as a basis for improving European wheat germplasm , and Skyfall is currently among the United Kingdom's most commonly planted wheat cultivars. Healthy seedlings were selected and transferred to 1.5 L plant pots containing a 3:1 mix of agricultural top soil (collected on 7 December 2016 from Leeds University Farm; 53°52′30.1″N, 1°19′15.8″W) and heat-sterilized (120 min at <120°C) soft sand ( Figure S1).
To supplement the naturally occurring AMF inoculum in the field soil, an inoculum of the generalist mutualistic AMF species Rhizophagus irregularis (Kiers et al., 2011) was also added (Method S1).
Homogenized inoculum was added to the sterilized sand immediately prior to mixing with the soil, with each pot receiving 10 ml of the inoculum. Spore density was quantified at 1,300 ± 100 spores per ml, such that each plant was inoculated with an additional 13,000 ± 1,000 R. irregularis spores.

| Plant growth conditions
Plants were maintained in controlled environment growth cabinets (Snijder Labs) on a light cycle of 15 hr daytime (20°C and 70% humidity) and 9 hr night-time (at 15°C and 70% humidity). Daytime PAR, supplied by LED lighting was 225 µmol m −2 s −1 at canopy level.
CO 2 concentrations were 440 and 800 ppm. Atmospheric [CO 2 ] was monitored using a Vaisala sensor system (Vaisala), maintained throughout the addition of gaseous CO 2 . Plants were transferred between growth cabinets every 4 weeks to mitigate any cabinet effects. After 4 weeks, plants were given weekly doses of 40 ml of a low-P preparation (containing 25% of the original P quantity) of Long Ashton solution (Smith, Johnston, & Cornforth, 1983), prepared using the nitrate formulation (Table S1). Plants were watered with tap water, as required.

| 33 P and 15 N isotope tracing
Arbuscular mycorrhizal fungi-mediated N and P assimilation was quantified using an approach adapted from Johnson, Leake, and Read (2001) using mesh-walled cores, into which the 33 P and 15 N tracers were added. Briefly, each pot contained two mesh cores constructed from PVC tubing (length 80 mm, diameter 18 mm), with windows (approx. 50 mm × 12 mm) cut in each side ( Figure S2). These windows and the bottom of each core were covered in a 20 µm nylon mesh which prevents root access but permits hyphal growth into the core contents. Nylon mesh was attached to PVC cores using Tensol ® adhesive (Bostik Ltd). Two of the cores were filled with the same soil and sand substrate as the bulk soil, plus 3 g/L crushed basalt (particle size <1 mm), to act as a fungal 'bait' (Quirk et al., 2012). Each pot also contained a third mesh-windowed core, loosely packed with glass wool (Acros Organics) and then the top sealed with a SubaSeal ® (Perkin Elmer). This created an airtight septum through which gas sampling can be conducted with a hypodermic syringe, in order to measure belowground respiration throughout the course of the experiment.
To ensure only symbiotic fungal-mediated tracer movement was measured, one of the mesh-windowed soil cores in each pot was gently rotated immediately prior to isotope tracer additions, 10 weeks postplanting. This rotation severed the fungal connections between the plant and the core contents, preventing direct transfer of the isotope tracers to the host plants via extraradical mycorrhizal fungal mycelium. Core rotation was conducted every 48 hr until the end of the experiment. The second core in each pot remained static, thereby preserving the hyphal connections between the core contents and the host plant. After 10 weeks of growth, 100 µl labelling solution, containing 1 MBq 33 P (as H 3 33 PO 4 , specific activity = 111 TBq/mmol; Perkin Elmer) and 46.26 µg 15 N (as >98 atom% 15 NH 4 Cl; Sigma Aldrich) was introduced to each pot. Labelling solution was added via pierced capillary tubing running down the centre of the core to ensure even distribution of tracer within the core. In half of microcosms (n = 6 per cultivar), labelling solution was added to the static core, and in the remaining microcosms (n = 6 per cultivar), to the rotated core. Cores which did not receive tracer solution were given 100 µl autoclaved distilled H 2 O. By subtracting the quantity of isotope tracers detected in plants from pots with severed hyphal connections to the isotope core (rotated isotope core treatment) from those where the AMF mycelium remained intact (static isotope core treatment), we were able to account for movement of isotopes caused by dissolution and diffusion and/or alternative soil microbial nutrient cycling processes.

| Plant-to-fungus carbon transfer
Two weeks after 33 P and 15 N tracer additions, plants were prepared for 14 CO 2 labelling, to allow movement of carbon from plant to fungus to be quantified. A 110 µl solution of NaH 14 CO 3 (Perkin Elmer) containing 1.0175 MBq 14 C (specific activity = 1.621 GBq/mmol) was added to a cuvette in each pot. The tops of all mesh-windowed cores were sealed using gas-tight rubber septa (SubaSeal) to minimize diffusion of 14 CO 2 into the cores. 14 CO 2 gas was liberated from the NaH 14 CO 3 by addition of 10% lactic acid, generating a 1.0175 MBq pulse of 14 CO 2 .
Samples of 1 ml above-ground gas and 1 ml below-ground gas (via the glass wool-filled core) were taken 1 hr after release of 14 CO 2 and every 4 hr thereafter to monitor the drawdown, respiration and flux of 14 C through the plant-AMF network. Gas samples were injected into gas-evacuated scintillation vials containing 10 ml Carbosorb ® (Perkin Elmer), a carbon-trapping compound. To this, 10 ml Permafluor scintillation cocktail (Perkin Elmer) was added, and 14 C content of each sample was quantified by liquid scintillation counting (Tricarb 3100TR scintillation counter; Perkin Elmer).
Pots were maintained under cabinet conditions until detection of maximum below-ground 14 C flux (20-22 hr after 14 CO 2 liberation) at which point 3 ml 2 M KOH was added to cuvettes within each microcosm to capture remaining gaseous 14 CO 2 .

| Harvest, sample preparation and analysis
All plant shoots, roots, bulk and core soil samples were separated, cleaned (roots only) and weighed before being immediately frozen and freeze-dried (Scanvac Cool-Safe freeze-dryer; LaboGeneApS) within 24 hr. Shoot, root and soil samples were homogenized and subsamples of core and bulk soils were collected for quantification of hyphal length density. Subsections of roots were separated before freezing for quantification of mycorrhizal colonization using acidified ink (Vierheilig, Coughlan, Wyss, & Piche, 1998 The volume of sample in the cuvette was made up to 3.8 ml and samples were kept in the dark for 45 min, after which absorbance was measured at 882 nm using a Jenway 6300 spectrophotometer (Cole-Palmer).

| Quantification of carbon-for-nutrient exchange between plants and AMF symbionts
Shoot and root 33 P content was quantified using aliquots of the digest product described above. About 1 ml aliquots of this digested product were mixed with 10 ml Emulsifier-Safe (Perkin Elmer) and 33 P Oxidiser; Perkin-Elmer).
Following the methods of Cameron, Johnson, Read, and Leake (2008), total C fixed by the plant and subsequently acquired by the fungus was calculated as a function of total CO 2 volume in the labelling chamber and the proportion of the 14 CO 2 which was fixed by wheat plants over the labelling period ( Figure S1). Comparing 14 C quantities in static versus rotated cores for each pot allows calculation of C acquisition by the fungi, controlling for 14 C detected due to root exudation or respiration, or alternative microbial carbon cycling processes.

| Statistics
Statistical analyses were carried out using 'R' statistical software, version 3.4.3. (R Core Team, 2017), implemented within the RStudio graphical user interface (RStudio Team, 2015). Data were tested by two-way ANOVA, where the cultivar and atmospheric [CO 2 ] were used as predictor variables. Where ANOVA gave p < .05 for the main effects, Tukey post hoc tests were used to identify statistical differences between groups. Prior to running analyses, data were tested for normality using Shapiro-Wilk test and by visual inspection of residual plots. Where data did not pass assumptions of normality and homogeneity of variance, data were log 10 transformed.
Following results from Akaike information criterion (AIC) testing which showed better model fit, data were log-transformed prior to statistical analysis.

| Elevated [CO 2 ] increases above-ground wheat growth and frequency of intraradical mycorrhizal structures
Plants grown under eCO 2 (800 ppm) had on average 14% greater shoot biomass than those grown in aCO 2 (440 ppm; Figure 1a;  (Table S3), driven by reduced arbuscule frequency at eCO 2 in Skyfall (Tukey: p < .001; Figure 2b). There is a trend towards greater vesicle abundance in wheat-AMF symbioses at 800 ppm than 440 ppm [CO 2 ] across cultivars (Figure 2c), although this is not statistically significant.

| Cultivar and aCO 2 drive differences in plant P and mycorrhizal-acquired 33 P
There are strong effects of cultivar and atmospheric [CO 2 ] effects on P content in shoots (F 5,70 = 38.96, p < .001; Figure 3a). P content in Cadenza shoots was 196% greater than in Avalon shoots, and 137% higher than in Skyfall shoots. Similarly, P concentration in Cadenza shoots was 186% higher than in Avalon shoots, and 153% higher than in Skyfall shoots. Cadenza plants grown at eCO 2 had the highest shoot P content and concentration of all cultivars for both atmospheric [CO 2 ] treatments (Figure 3a, Table S1).
Plant assimilation of fungal-acquired 33 P tracer in cultivars Avalon and Cadenza (content and concentration; Figure 3c,d; Table S3) was reduced in eCO 2 treatment, but slightly increased in Skyfall, although these trends were not statistically significant. There was high variability in 33 P tracer uptake by Skyfall, requiring log 10 transformation of the data to meet the assumptions of ANOVA. There were clear differences between cultivars in terms of 33 P acquisition via mycorrhizas. Combining data from eCO 2 and aCO 2 , Skyfall acquired 570 times more 33 P tracer than Avalon and 225 times more than Cadenza (Figure 3c,d).

| Cultivar-specific differences in plant-acquired N, but not mycorrhizal-acquired 15 N tracer
Elevated atmospheric [CO 2 ] significantly decreased shoot N content in Cadenza (Tukey: p < .001; Figure 4a; Table S3) but not in Avalon and Skyfall. Cultivars also showed significant variation in shoot N content ( Figure 4a; Table S3). Avalon shoots contained significantly  (Table S3).

| Carbon-for-nutrient transfer between wheat and AMF
Carbon for nutrient transfer between plants and AMF was tested using Spearman's rank correlation coefficient ( Figure S3) Significant variation in growth responses to colonization by AMF has previously been identified across cereal varieties (Hetrick, Wilson, & Cox, 1992;Lehnert et al., 2018;Watts-Williams et al., 2019). Such genotypic differences in growth resulting from AMF symbioses are likely to be linked not only to the receptivity to fungal colonization, but also to the physiological function of the AMF associations, particularly the degree to which the fungal symbionts represent a carbon sink (Walder et al., 2012) and nutrient source

| Carbon outlay by wheat to AMF is unaffected by atmospheric [CO 2 ]
In our experiments, plant biomass increased in eCO 2 ( Figure 1b).
However, the C transferred to the extraradical mycelium, in terms of both total amounts, and per cent of recently fixed photosynthate, was not affected ( Figure 5). This suggests that transfer of  (Smith, Grace, & Smith, 2009). In our experiments, vesicle frequency did not differ between atmospheric [CO 2 ] treatments ( Figure 2c). Thus, it appears that there was no 'carbon fertilisation effect' of eCO 2 for wheat-associated AMF (Alberton et al., 2005). The lack of atmospheric [CO 2 ] response in terms of AMF C acquisition observed in our experiments runs counter to the trends observed in meta-analyses (Alberton et al., 2005;Treseder, 2004) and other experimental studies (Field et al., 2012). Intensive modern breeding programmes which have given rise to elite wheat cultivars such as those used in our experiments may be responsible for the lack of atmospheric F I G U R E 5 Total carbon transferred from wheat (Triticum aestivum L., cv. Avalon, Cadenza, Skyfall) to fungal mycelium during the course of 14 C labelling experiment (a), and per cent of carbon fixed during the labelling period which was recovered in the static core at harvest (%) (b). Plants were grown at ambient (440 ppm, white boxes) and elevated (800 ppm, grey boxes). Bars sharing letters are not significantly different, where p > .05 (ANOVA, Tukey post hoc test). Data were log 10 transformed where data assumptions were not met. N.S.D., not significantly different [CO 2 ] effect on AMF C acquisition. To maximize nutrient uptake efficiency in systems where fertilizer nutrients are applied in readily available forms (Good & Beatty, 2011), modern elite cereals are bred to have reduced root-to-shoot ratios compared to older cultivars (Siddique, Belford, & Tennant, 1990). Those cultivars with large root systems where nutrients are easily acquired could be viewed by breeders as C-inefficient, as C allocated to below-ground growth could be retained above-ground. To this end, the allocation of C to mycorrhizas and ERM may have been inadvertently selected against in the breeding of modern cereal cultivars. Alternatively, the apparent lack of atmospheric [CO 2 ] response observed here may be partly due to plant and fungal C allocation to AMF spores not being quantified in the present investigation; it is possible that under eCO 2 the AMF produced greater number of spores than in aCO 2 .
This would not have been quantified in our experiment given the relatively short 14 CO 2 labelling period, and might also account for a significant fraction of fungal C. In addition, AMF hyphal turnover is thought to be rapid (Staddon, Ramsey, Ostle, Ineson, & Fitter, 2003) and may represent a significant source of C input to soils (Godbold et al., 2006). Respiratory losses of hyphal-derived C would not be quantifiable in our experimental approach. How atmospheric [CO 2 ] affects hyphal turnover in AMF associated with crop plants remains to be determined.
The amounts of C allocated to AMF by the wheat cultivars in these experiments are similar to those recorded in comparable experiments with noncrop vascular plants (Field et al., 2012).
However, only a small fraction of the total C fixed during the experimental period by the various wheat cultivars here was allocated to their fungal mycelium (Figure 5b), regardless of the availability of C in the atmosphere. Adding 14 CO 2 to an enclosed system, such as the labelling chamber in our experiments, inevitably leads to an increased CO 2 concentration which would impact plant physiology.
However, the addition of 1.1 MBq of 14 CO 2 to our labelling chambers increased the concentration of atmospheric [CO 2 ] within the chambers by 1.24% in aCO 2 and 0.36% in eCO 2 treatments. This slight increase in atmospheric [CO 2 ] is unlikely to have elicited a substantial physiological response in the plants used in our experiment. Given that our plants were only able to fix and assimilate 14 CO 2 for one photoperiod, it is likely that the amount of C measured by the isotope tracing was not reflective of total plant carbon allocation to symbiotic fungi across the life cycle of the plant; this warrants further investigation. Despite this, our experiment provides valuable insights into the allocation of recently fixed C to fungal symbionts of wheat during a period of rapid plant growth and high nutrient demand.

| Cultivar-specific wheat nutrient gains via mycorrhizas
All cultivars assimilated 15 N and 33 P via their mycorrhizal symbionts, with the amounts of each tracer varying according to the cultivar. Skyfall assimilated the most mycorrhizal-acquired 33 P tracer compared to cv. Avalon and Cadenza (Figure 4c,d). This pattern of nutrient gain from AMF is not reflected in the total nutrient content or concentration of plant tissues across cultivars (Figure 4a,b).
Cadenza contains the most P, both fungal-and plant-acquired, in its above-ground tissues (Figure 4a,b) but it is cv. Skyfall that acquires the most 33 P tracer via AMF symbionts. This pattern may be reflective of variation in nutrient acquisition strategies across the cultivars tested. Cadenza has the greatest P concentration of above-ground tissues (Figure 4b), but lower AMF-assimilated tracer content ( Figure 4c) and concentration (Figure 4d) than other cultivars and thus appears to operate a more effective plant P assimilation pathway than cv. Skyfall, which appears to rely more heavily on the mycorrhizal pathway for nutrient acquisition (Smith, Smith, & Jakobsen, 2003;Smith et al., 2004). With the highest levels of AMF colonization ( Figure 2a) and extraradical mycelial density (Table S2), but the lowest AMF contribution to 33 P uptake (Figure 3c,d) and lowest above-ground dry mass (Figure 1a), it appears that Avalon forms a less nutritionally mutualistic interaction with AMF than the other two culitvars tested, potentially resulting in suppression of growth. This observation is unlikely a result of the AMF exerting an excessive carbon "drain" given that cv. Avalon does not allocate more C to its AMF than the other cultivars tested ( Figure 5), and that the percentage of C allocated to AMF by wheat is low compared to other plants (e.g. Field et al., 2012). Instead, it is possible that downregulation of plant phosphate transporters following AMF colonization may be partly responsible, and as a result, plant P uptake is reduced relative to the nonmycorrhizal counterpart (Li, Smith, Dickson, Holloway, & Smith, 2008). As we do not have nonmycorrhizal treatments to compare nutrient acquisition and growth in these cultivars against, it is not possible to determine whether AMF suppress growth of cv. Avalon but this certainly warrants further research.
Mycorrhiza-mediated uptake of 33 P and 15 N tracers was not significantly influenced by atmospheric [CO 2 ] in any of the cultivars tested (Figures 3c,d and 4c,d). This finding is counter to some modelling predictions (Bever, 2015) and some experimental data (Field et al., 2012) but is broadly in agreement with experiments conducted in Pisum (Gavito et al., 2002(Gavito et al., , 2003, Brachypodium and Medicago (Jakobsen et al., 2016) which also showed little effect of atmospheric [CO 2 ] on AMF-acquired plant nutrient assimilation. Increased total P content (i.e. plant-and mycorrhizal-acquired) at eCO 2 compared to aCO 2 treatment in shoots of cvs. Skyfall and Cadenza is counter to the general observation that P, like N, is usually relatively diluted in plant tissues at eCO 2 owing to increased plant biomass (Jakobsen et al., 2016). Increased P uptake at eCO 2 is not unprecedented, however (Campbell & Sage, 2002), it may be due to changes in root morphology (Nie, Lu, Bell, Raut, & Pendall, 2013). Our 33 P labelling suggests that the AMF were not responsible for this increased P uptake (Figure 4c,d).
Plant tissue N content and concentration may be reduced when plants are grown in eCO 2 conditions, as a result of increasing plant biomass (Cotrufo, Ineson, & Scott, 1998;Hogy & Fangmeier, 2008;Taub, Miller, & Allen, 2008). This trend is apparent in cv. Cadenza and Skyfall plants in our experiments, although not in Avalon (Figure 3a,b). The phenomenon of reduced N content of arable crops has potentially serious implications for the nutritional quality of grain and grain-based food products (Pleijel & Uddling, 2012 (Pleijel, Broberg, Hogy, & Uddling, 2019).
Our data support plant/cultivar identity as an important driver of mycorrhizal benefit to plant hosts (Field & Pressel, 2018;Klironomos, 2003;Walder & van der Heijden, 2015). However, our data do not support the notion that carbon for nutrient exchange between wheat and AMF are governed by a linear, 'reciprocal rewards' model of mutualism (Bever, 2015;Fellbaum et al., 2012;Kiers et al., 2011) as previously shown using Petri dish-based microcosm systems (e.g. Kiers et al., 2011)   . Perhaps surprisingly, root architecture traits may have limited effects on a plant's nutritional and growth response to mycorrhization (Maherali, 2014). Inter-and intraspecific functional diversity is also present in AMF species (Jones & Smith, 2004;Mensah et al., 2015;Munkvold et al., 2004;Watts-Williams et al., 2019).
By using unsterilized soil in our experiment, our experimental plants are likely to have been colonized by a mixed community of AMF, where the relative contributions of individual species or isolates cannot be ascertained. As AMF community structure is understood to impact symbiotic function (Frew, 2019;van der Heijden et al., 1998), this is of great potential agronomic interest. Understanding the role of genetic variability in plant-fungal interactions to the point where it can begin to help informing agriculture will likely prove to be a substantial, but ultimately worthwhile, undertaking (Johnson, Martin, et al., 2015). Metagenomic techniques should identify species and intraspecific diversity of the AMF present within field-crop plant roots, combined with functional studies to determine the role these fungi play in crop nutrient uptake or other non-nutritional beneficial roles. As illustrated by the present investigation, further factors to consider include the effects of abiotic factors on AMF community structure and diversity. Recent field-scale atmospheric [CO 2 ] manipulation has shown how CO 2 enrichment can affect AMF community composition (Cotton, Fitter, Miller, Dumbrell, & Helgason, 2015;Maček et al., 2019). How these atmospheric [CO 2 ]-driven community changes might influence the stoichiometry of carbon-for-nutrient exchange between symbionts in the field remains to be determined (Cotton, 2018).

| Future perspectives
Our results, and those of other studies investigating mycorrhizal responses to eCO 2, must be contextualized with the likelihood that climate change will encompass shifts in multiple abiotic variables.
Factors such as N deposition, warming and drought are at least as important an influence on AMF as atmospheric [CO 2 ] (Kivlin, Emery, & Rudgers, 2013). Our data demonstrate that AMF will continue to provide N and P nutrition to their plant hosts under eCO 2 and that there is no evidence for significant C drain from the fungi. Whether these trends are seen following simultaneous perturbations of temperature, water availability and N deposition in crop plants is not clear, as experimental testing of such scenarios is lacking.
While AMF may not prove to be the silver bullet, 'sustainable saviours' for agricultural intensification (Thirkell et al., 2017), our experiments have demonstrated that AMF do have the potential to contribute to cereal nutrient assimilation. As such, AMF could have an important role to play in reducing application of N-and P-based fertilizers as part of a wider strategy for sustainable soil management. We echo calls for further field scale experimentation of the function of AMF in crop plants to determine what role, nutritional or otherwise, AMF might be playing in crop growth in situ (Lekberg & Helgason, 2018;Rillig et al., 2019). To date, very little work has been carried out on crop breeding to optimize mycorrhizal benefit.
Given the potential influence of AMF on plant nutrient uptake and growth (Klironomos, 2003) and their ubiquity in farm systems (Oehl, Laczko, Oberholzer, Jansa, & Egli, 2017;Sale et al., 2015) it appears remiss that AMF should not be considered in breeding programmes.
Recent steps have been taken to investigate the genetic basis for mycorrhizal colonization (Lehnert et al., 2017) as well as mycorrhizal "benefit" and drought response in wheat (Lehnert et al., 2018), while similar efforts in other crop species have been in progress for several years (De Vita et al., 2018;Galvan et al., 2011;Kaeppler et al., 2000).
Better understanding of the mechanisms underlying plant-microbial interactions remains important in the future-proofing and sustainable intensification of agriculture.

ACK N OWLED G EM ENTS
We thank Michael Charters, Ashleigh Elliott, Grace Hoysted and Bev Merry for assistance with plant harvesting and sample preparation, Richard Summers and RAGT for supply of seeds, and Heather Walker at the University of Sheffield for performing IRMS analysis. This research was funded by a BBSRC Translational Fellowship award (BB/ M026825/1) and a Rank Prize Funds New Lecturer Award to KJF.
We thank N8 Agrifood for support to KJF and TT. We thank two anonymous reviewers and the editor for their constructive comments during peer review.