Is resource allocation and grain yield of rice altered by inoculation with arbuscular mycorrhizal fungi?

Abstract

Aims

Our study quantified the combined effects of fertilization and inoculation with arbuscular mycorrhizal fungi (AMF) on grain yield and allocation of biomass and nutrients in field-grown rice (Oryza sativa L.).

Methods

A two-factor experiment was conducted at a field site in northeast of China (in Shuangcheng, Heilongjiang Province, Songhua River basin): six nitrogen–phosphorus–potassium fertilizer levels were provided (0, 20, 40, 60, 80 and 100% of the local norm of fertilizer supply), with or without inoculation with Glomus mosseae. At maturity, we quantified the percentage of root length colonization by AMF, grain yield, shoot:root ratios, shoot N and P contents and nutrients allocated to panicles, leaves and stems.

Important Findings

As expected, inoculation resulted in greatly increased AMF colonization, which in turn led to higher shoot:root ratios and greater shoot N contents. Shoot:root ratios of inoculated rice increased with increasing fertilization while there was a significant interaction between fertilization and inoculation on shoot:root ratio. Additionally, AMF inoculation increased panicle:shoot ratios, panicle N:shoot N ratios and panicle P:shoot P ratios, especially in plants grown at low fertilizer levels. Importantly, inoculated rice exhibited higher grain yield, with the maximum improvement (near 62%) at the lower fertilizer end. Our results showed that (i) AMF-inoculated plants conform to the functional equilibrium theory, albeit to a reduced extent compared to non-inoculated plants and (ii) AMF inoculation resulted in greater allocation of shoot biomass to panicles and increased grain yield by stimulating N and P redistribution to panicles.

arbuscular mycorrhizal fungi, resource allocation, rice, grain yield, fertilizer

INTRODUCTION

The way plants respond to variations in nutrient availability strongly influences their ability to grow and compete in natural and managed habitats (Benner and Bazzaz 1985; López-Bucio et al. 2003), with shifts in resource allocation under low and high nutrient supply being crucial (Reynolds and D’Antonio 1996). Underpinning such shifts are changes in the relative investment of whole-plant biomass into shoots and roots and/or relative allocation to reproductive and vegetative structures (Schmid and Weiner 1993; Vannoordwijk and Dewilligen 1987; Veresoglou et al. 2012). The pattern of biomass allocation determines investment in organs involved in photosynthesis, reproduction and nutrient uptake. Given this, shifts in biomass allocation can have a profound effect on the ability of plants to capture the resources necessary for plant growth. The functional equilibrium theory (FET), also referred to as the ‘balanced growth hypothesis’ (Shipley and Meziane 2002) or the ‘optimal partitioning theory’ (Kleczewski et al. 2010; McConnaughay and Coleman 1999), proposes that once soil resources [e.g. nitrogen (N) or phosphorus (P)] become limited, plants will allocate more biomass to roots rather than to shoots; conversely, plants should invest less in roots when nutrient limitation is removed, allowing for greater investment in above-ground resources (Vannoordwijk and Dewilligen 1987). The FET has been supported by a number of studies where the root mass ratio (i.e. ratio of root biomass to whole-plant biomass) decreased with increasing nutrient availability (Kleczewski et al. 2011; Reynolds and D’Antonio 1996; Sims et al. 2012).

In terrestrial ecosystems, arbuscular mycorrhiza (AM) represents a near ubiquitous symbiosis between plant roots and arbuscular mycorrhizal fungi (AMF) (Smith and Read 2008). AMF may provide host plants with a range of benefits including improved access to P (Karandashov and Bucher 2005) and N (Cruz et al. 2007; Govindarajulu et al. 2005; Hodge and Fitter 2010; Hodge et al. 2001), receiving in return photosynthetically fixed carbon. In addition, AMF can influence patterns of biomass allocation within host plants. A recent meta-analysis showed that AMF colonization resulted in reallocation of biomass towards shoots (and away from roots); there was, however, no such reallocation effect in host plants exposed to stresses such as drought, heavy metal contamination or acid precipitation (Veresoglou et al. 2012). Importantly, little is known about how AMF impacts biomass allocation to shoots and roots in host plants exposed to other types of stress including nutrient deficiency. In addition, it is not clear whether AMF-inoculated plants conform to the FET when exposed to altered nutrient conditions.

The relationship between shoots and roots put forward by the FET offers no explanation of the actual functional relationships characterizing biomass allocation to both vegetative and reproductive organs. Leaves supply the entire plant with photosynthate, while stems function as transport organs and provide physical support to leaves and flowers (Cui et al. 2003). Investment in reproductive organs such as panicles represents a major sink for carbon and nutrients, as well as determining the economic yield in cereal crops. Given this, it is necessary to subdivide shoots into leaves, stems and panicles when assessing how treatments affect biomass allocation and plant performance (Poorter and Nagel 2000; Poorter et al. 2012). Importantly, the relative amount of biomass present in the various organs is not fixed but varies across environments. Therefore, plants under different environmental conditions can balance resource allocation to these above-ground organs in a way that satisfies their physiological activities and functions. Past studies have reported that resource allocation to leaves increases relative to roots with increasing fertilization (Poorter and Nagel 2000; Shipley and Meziane 2002). In addition, root and stem traits were reported to be closely aligned (Fortunel et al. 2012), suggesting that stems should be allocated less resources with increasing nutrient availability. Most research in this area, however, has focused on environmental impacts on allocation to roots, stems and leaves during vegetative growth (Poorter et al. 2012), but not during the reproductive stage.

AMF have also been shown to affect allocation to reproductive versus vegetative structures within shoots. In Andropogon gerardii, e.g. greater biomass was allocated to reproductive rather than vegetative organs in AMF colonized plants (Miller et al. 2002). Moreover, AMF was reported to increase the number of flower buds in Calendula officinalis, suggesting a positive effect on plant reproduction (Zaller et al. 2011). Importantly, the promotion of reproduction by AMF was observed in both sexual and asexual plants (Varga 2013; Zaller et al. 2011). Besides reproductive organs, AMF play a key role in promoting leaf and stem growth. For instance, AMF were found to increase leaf area (Soares et al. 2012), number (Novoa et al. 2010) and biomass (Altoe et al. 2008; Cartmill et al. 2008). Similar increases have been observed for stems, with AMF increasing stem diameter and biomass for a range of host species, including sweet passion fruit, genipap, citrus and grapes seedlings (Filho et al. 2012; Schreiner and Pinkerton 2008; Soares et al. 2012; Wu et al. 2011). However, little information is available on how AMF affect biomass allocation within shoots, indicated by panicle:shoot ratios, leaf:shoot ratios and stem:shoot ratios.

As a result of the strong influence that host nutrient status has on the AMF, use of fertilizers has a major impact on the relationship between host plant and AMF (Gosling et al. 2006). For example, reduced AMF colonization of roots (in multiple crop species) and lower AMF spore density have been observed as soil P fertilization increased (Kahiluoto et al. 2000). Moreover, Artemisia vulgaris were reported to exhibit high levels of AMF colonization in N-deficient soils (Blanke et al. 2005). Importantly, earlier research on giving mesic to semi-arid grasslands also reported decreased allocation to AMF structures directly involved in nutrient acquisition under conditions of rising nutrient availability (Johnson et al. 2003). However, whether the impact of AMF on biomass allocation within shoots and grain yield is affected by nutrient availability is not clear.

In our study, we investigated the effects of AMF inoculation and fertilization on allocation of whole-plant biomass to shoots and roots and of shoot biomass to panicles, leaves and stems. The effects of AMF inoculation and fertilization on grain yield were also assessed. We chose rice (Oryza sativa L.) for our study, reflecting its importance for global food supplies and the need to improve rice yield while minimizing the use of inorganic fertilizers. Potentially beneficial effects of AMF inoculation for optimizing rice production systems have received much attention in the scientific literature. A range of positive responses has been observed even under anaerobic conditions (Purakayastha and Chhonkar 2001; Wangiyana et al. 2006), including an increase in plant size, tissue nutrient concentrations (Solaiman and Hirata 1996, 1997a) and photosynthetic rates (Black et al. 2000). Conclusions on the influence of AMF inoculation on rice grain yield are conflicting, however, with other studies reporting yield increasing, declining or remaining constant compared with non-inoculated plants (Li et al. 2011; Secilia and Bagyaraj 1994; Solaiman and Hirata 1995, 1997b, 1998; Wangiyana et al. 2006). Therefore, more information about the effects of AMF on rice grain yield is required to improve rice production via AMF inoculation.

Our study tested the following hypotheses:

 

(H1) Given that AMF improves N and/or P uptake (Hodge and Fitter 2010; Smith and Read 2008), a positive effect of AMF inoculation on shoot:root ratios is expected. Moreover, we hypothesize that rice inoculated with AMF will conform to the FET just as non-inoculated rice does.

(H2) AMF inoculation, by altering N and (or) P redistribution within shoots, will result in an increase in shoot biomass allocated to panicles, but a decrease to leaves and stems.

(H3) Based on H2, a positive effect of AMF inoculation on rice grain yield is expected. We also expect that this positive effect will decrease with increasing nutrient availability.

MATERIALS AND METHODS

Site description and experimental design

The experiment site was located adjacent to the Yinla River in Shuangcheng, Heilongjiang province, Songhua River basin (45°13.82′N, 126°22.61′E), a typical region for rice production in China. The experiment took place during the summer (May–October) of 2011. Mean annual temperature is 4.3°C, while the annual temperature range (comparing the average temperature of the hottest and coldest month) is 42.2°C. The frost-free period lasts 135 days and annual rainfall is near 500mm. The hydromorphic paddy soil contains 26.3 g·kg⁻¹ of organic matter, 125.3 mg·kg⁻¹ of hydrolysable N, 160.3 mg·kg⁻¹ of available P (Bray No. 2) and 17.6 mg·kg⁻¹ of available potassium (K) and the pH of the soil was 6.0 (in distilled water) (Zhang et al. 2012). Soil analysis methods used in this study are described in a soil analysis manual (Sparks et al. 1996).

Our study sought to assess the effect of AMF inoculation on rice under conventional paddy-field conditions. Ideally, such an experiment would involve comparing inoculated plants with plants grown on sterile soil. While the scale of the experiment precluded sterilization of soil in control plots, preliminary measurements done in the year before the current experiments revealed that the baseline abundance of AMF at the field site was low, with root length colonization (RLC) of non-inoculated plants being ca. 2–3% (see Supplementary Material for additional details), consistent with other flooding rice systems where 3–5% of RLC was monitored (Wangiyana et al. 2006). Past studies have used plots with low colonization (e.g. 2–3%) to study the effect of AMF inoculation on Zn uptake (Purakayastha and Chhonkar 2001). Here, we adopted the same approach to assess the effect of AMF inoculation on rice growth, biomass allocation and yield.

The experimental design consisted of three replicate blocks (each block being ca. 380 m²) distributed over a 1400 m² area. Within each replicate block were located six main plots (i.e. for the six fertilization treatments), each of which was 5×5 m (i.e. 25 m²). Split plots (each 1 m²) embedded within each main plot were used to facilitate two inoculation treatments [i.e. added inoculum (+m) and non-inoculated (−m)]; there was one +m and one −m split plot within each main plot. Each plot, main and split, had one water flow entry (from the adjacent Yinla River) and one water exit.

To prevent the movement of surface and ground water among neighbouring plots, a vertical geo-membrane (extending 50cm above and belowground) was placed around the perimeter of each main and split plot (within the replicate blocks), with the distance between any two main plots being at least 2 m. To avoid the influence of neighbouring plots, samples were only collected from the centre of the split plots. As the roots of rice are usually considered to be distributed within the upper 30cm of soil depth, with the majority within 15cm (Yang et al. 2004), it was unlikely that subsurface movement of nutrients would have occurred, as the geo-membrane extended to 50cm depth. Furthermore, the high level of soil compaction at the field site would have further reduced the possibility of sub-surface nutrient movement.

Inoculum and plant material

In our study, the AMF isolate was Glomus mosseae HDSF1, which was provided by Professor Fuqiang Song and has been deposited in China General Microbiological Culture Collection Center (CGMCC) (Beijing), with the deposition number being CGMCC No. 3012.

A mixture of soil, sand, vermiculite, root segments, hyphae and spores of G. mosseae was used as inoculum. The number of spores in the inoculum was 33–35g⁻¹ and percentage of RLC was 74.8% for G. mosseae. The inoculum was developed in pots by cultivating AMF on roots of Trifolium repens L. in autoclave sterilized growth medium (soil diluted with sand and vermiculite with volume ratio of soil, sand and vermiculite 2:5:3). After 3 months of growth in a controlled-environment room, the shoots were discarded and the inoculum mixture in the pots was air dried, homogenized and refrigerated at 4°C for 1–2 months before use.

Wetland rice (O. sativa L.) was planted in a greenhouse (91% of seeds were germinated after 3 days at 27°C in distilled water). The nursery beds were established in plastic boxes (58cm long × 28cm wide × 3cm height) using air-dried soil. Twenty identical nursery boxes were set up to provide plants for subsequent transplanting into the field (10 for each inoculated and non-inoculated treatment). The bedding soil in each box was fertilized with 160mg N as urea, 800mg P as calcium magnesium phosphate fertilizer and 160mg potassium as KCl. For the inoculated treatment (+m), 250g of root inoculum was layered on top of the air-dried soil, then 50g of rice seeds followed by 1kg of air-dried soil above the seeds. For the non-inoculated treatment (−m), 250g of sterilized growth medium (as above, volume ratio of soil, sand and vermiculite was 2:5:3) was added to each nursery bed.

Seedlings from the nursery beds were transplanted into the field site 6 weeks after sowing. Seedlings for two inoculation treatments, inoculated (+m) and non-inoculated (−m), were transplanted at the same density (three seedlings per hill). The spacing in the rows was 13cm while that between the rows was 30cm. Inoculated and non-inoculated plants were transplanted into their respective split plots; in addition, non-inoculated plants were transplanted into the remaining areas of each main plot.

Fertilization treatments

Six levels of fertilizer (labelled as F0, F20, F40, F60, F80 and F100) were provided: representing 0, 20, 40, 60, 80 and 100% of the local norm of fertilizer (Table 1). Three forms of fertilizer were applied in combination to all the transplanted rice: (i) nitrogen–phosphorus–potassium (NPK) compound fertilizer (NH3, P₂O₃ and K₂O in ratio 16:17:12), (ii) urea and (iii) ammonium sulphate. Fertilizers were applied in two ways: (i) ‘basal dressing’ before transplanting and (ii) ‘top dressing’ during the field growing period (Table 1). Fertilizers were applied to wet soil and no drainage was allowed when applied in standing water. The basal dressing was NPK compound fertilizer. The first and third top dressings were ammonium sulphate and added on the 1st day (DAT 1: 1 June 2010) and 28th day (DAT 28: 28 June 2010) after transplantation. The second top dressing was urea, added on the 11th day (DAT 11: 11 June 2010) after transplantation (Table 1).

Table 1:

differing doses of fertilizer added to each plot.

Fertilizer levels	Basal dressing (g m−2)	Top dressing (g m−2)
		DAT 1	DAT 11	DAT 28
F100	50.0	6.0	12.5	10.0
F80	40.0	4.8	10.0	8.0
F60	30.0	3.6	7.5	6.0
F40	20.0	2.4	5.0	4.0
F20	10.0	1.2	2.5	2.0
F0	0.0	0.0	0.0	0.0

There was no difference in fertilization for inoculated and non-inoculated rice. Abbreviations: basal dressing = nitrogen–phosphorus–potassium (NPK) compound fertilizer (NH₃, P₂O₃ and K₂O in ratio 16:17:12); DAT = day after transplanting; F0, F20, F40, F60, F80 and F100 = provided with 0, 20, 40, 60, 80 and 100% of the local norm of nutrient input; top dressing = ammonium sulphate on DAT 1 (June 1) and DAT 28 (June 28) and urea on DAT 11 (June 11).

Water management

In our study, rice was irrigated with surface water sourced from the nearby Lalin River. Before rice seedlings were transplanted, the paddy fields were irrigated with sufficient water to cover the soil with 1–3cm depth of water. Subsequently, an overlying water layer of 3–5cm was maintained during the whole growing season, except for two aeration periods, which were induced by draining of the main and split plots. The first aeration (to control tiller number and algae growth) lasted 7 days (21–28 June 2011, i.e. 21–28 days after transplanting). The second was carried out 34 days before harvest (September 1–October 3, i.e. 93–125 days after transplanting) to facilitate mechanized combine harvesting.

Root colonization analysis

AMF colonization was assessed at plant maturity. Fine roots were cut into 1.5cm segments and mixed thoroughly; a 0.25g (fresh weight) subsample was randomly taken for determination of root colonization. These subsamples were cleared with 10% KOH by heating in a beaker filled with water kept at a rolling boil for 30min. Thereafter, samples were neutralized with 0.2% HCl and stained with 0.2% acid fuchsin (Aladdin Industrial Corporation, Shanghai, China) for 30min at 90°C. Colonization by AMF was estimated by quantifying RLC under a dissecting microscope (Eclipse E 200; Nikon Instruments, Beijing, China) using the grid line intersects method (Giovannetti and Mosse 1980).

Plant harvesting

At plant maturity (early October), three hills were sampled randomly from each split plot within a given replicate block. As there were three replicate blocks, in total, we harvested nine hills per fertilization/inoculation treatment combination. The shoot portion at each hill was cut off and panicles, leaves and stems were separated. To establish the soil depth from which roots needed to be harvested, we collected several intact root/soil cores (one per hill, centred on the stems) from the main plots outside the split plot areas. Root/soil material was collected down to a depth of 30cm; thereafter, roots from the upper 15cm were separated from those of the lower 15cm. Roots were washed and dried. We found that 82–84% of roots were top 15cm. Given this, we subsequently restricted our sampling of roots from the treatment split plots to intact from the upper 15cm, with each core being 10×15cm (width × length). Sampled cores were transported to a nearby laboratory, where all roots were sieved using several layers of mesh screen. Sieved roots were then washed with tap water to remove any remaining soil. Roots were separated into coarse (>1.0mm diameter) and fine (<1.0mm diameter) categories. One half of fine roots were used to determine root colonization, while the other half and the coarse roots were used to calculate the dry mass of the sub-sample. The panicles, leaves and stems were washed and then dried with roots (including intra-radical hyphae) in an oven at 70°C for 15min, then at 105°C for 48h. Grain yield was recorded on an air-dry basis including husk. The dry mass values were then used to calculate the following parameters: shoot biomass (panicles + leaves + stems), root biomass (fine roots including intra-radical hyphae and coarse roots), plant biomass (shoot biomass + root biomass), shoot:root ratio, panicle:shoot ratio (proportion of shoot biomass in panicles), leaf:shoot ratio (proportion of shoot biomass in leaves) and stem:shoot ratio (proportion of shoot biomass in stems).

Chemical composition

Samples were ground in a ball mill and nitrogen concentration ([N]) was determined using an elemental analyser (Thermo Scientific FlashEA 1112). For determination of phosphorus concentration ([P]), the ground samples were firstly mineralized with HCl (36%), HNO₃ (67%) and HF (49%) acids (5:2:2, volume) in a Microwave Accelerated Reaction System (CEM MARS-5), then dissolved in 1ml HNO₃ (1%) and diluted to 50ml with ultrapure water and analysed for [P], using inductively coupled plasma-atomic emission spectrometry (Perkin Elmer Optima 5300DV). [N] and [P] (combined with data on tissue biomass, where appropriate) were then used to calculate N:P ratios and the N and P content of panicles, leaves and stems.

Statistical analysis

Two-way analysis of variance (ANOVA) was used to distinguish main and interactive effects of AMF inoculation and fertilization on plant resource allocation traits. One-way ANOVA was conducted on traits for which there was a significant fertilization × AMF inoculation interaction term. We calculated Pearson’s correlation coefficients (r) to establish correlations among traits. Statistical analyses were performed in SPSS 21.0 (SPSS Inc., Chicago, IL, USA).

RESULTS

AMF colonization

Supplementary Figure S1 shows RLC values plotted against fertilizer level. Table 2 shows that there was a significant overall main effect of AMF inoculation on RLC; the overall RLC average was 2.5±0.3% in non-inoculated (−m) rice, while RLC of inoculated (+m) rice ranged from 12.4 to 19.5%. While there was also a significant overall main effect of fertilization on RLC (Table 2), the response to fertilization differed between +m and −m plants, as demonstrated by the significant interaction term (Table 2). For +m rice, RLC decreased significantly with rising fertilization, while fertilization did not affect RLC in −m rice (Table 3 and Supplementary Figure S1). This further demonstrated that the impact of inoculation was more pronounced at low than that at high fertilizer levels. Consequently, inoculation improved colonization by AMF in the roots of rice, but this improvement was reduced by fertilization.

Table 2:

results of two-way ANOVAs assessing the main and interactive effects of fertilization and AMF inoculation on RLC, several biomass allocation parameters and grain yield.

	F	I	F × I
RLC	19.221***	926.026***	20.374**
Plant biomass	554.030***	158.048***	5.641*
Shoot biomass	51.338**	501.283***	5.304*
Root biomass	163.862***	531.350***	67.037**
Shoot:root ratio	13.949*	16.344*	3.090*
Panicle:shoot ratio	6.684*	92.682**	3.731*
Leaf:shoot ratio	1.752	26.258**	1.385
Stem:shoot ratio	4.257*	67.379**	4.421*
Grain yield	772.512***	520.149***	33.101**

Table shows Fisher values and significance (*P < 0.05; **P < 0.01; ***P < 0.001) of main and interactive effects. Abbreviations: F = fertilization (df = 5); I = inoculation (df = 1); F × I = interaction between fertilization and inoculation (df = 5).

Table 3:

Fisher values and significance of simple main effect tests for traits for which there were significant interactions between fertilization and AMF inoculation observed in Table 2

Factors	Levels	RLC	Plant biomass	Shoot biomass	Root biomass	Shoot:root ratio	Panicle:shoot ratio	Stem:shoot ratio	Grain yield
F	+m	39.242***	321.960***	691.349***	110.988***	4.462**	1.652	40.564***	441.475***
	−m	0.353	237.712***	815.238***	119.911***	12.576***	8.563***	176.862***	711.776***
I	F0	962.713***	67.072***	2.074	406.835***	19.651***	58.977***	343.534***	287.714***
	F20	811.075***	34.261***	1.885	190.439***	8.370**	17.043***	372.971***	253.426***
	F40	704.175***	47.071***	3.306	252.324***	3.596	7.952**	149.282***	259.830***
	F60	772.160***	6.633*	5.684*	10.528**	0.117	8.627**	105.010***	64.608***
	F80	589.835***	4.682*	9.599**	1.143	0.042	5.745*	65.282***	62.667***
	F100	323.937***	26.533***	55.308***	5.266*	0.018	16.593***	201.320***	9.775**

Fisher values in the first and second line indicated the effect of fertilization (F) on traits of inoculated plants (+m) and non-inoculated plants (−m), respectively, while Fisher values in lines from the third to the eighth showed the effect of inoculation (I) under each fertilizer level. Abbreviations: F = fertilization (df₁ = 5, df₂ = 48); I = inoculation (df₁ = 1, df₂ = 16).

*P < 0.05, **P < 0.01, ***P < 0.001.

Biomass production and allocation to shoots and roots

Table 2 shows that inoculation had a significant effect on all biomass production and allocation traits, with plant biomass, shoot biomass and root biomass being significantly lower for inoculated rice (+m) than non-inoculated (−m) rice; by contrast, greater shoot:root ratios were associated with AMF inoculation (Table 3 and Fig. 1). There was also a significant overall effect of fertilization on these traits (Table 2); increasing fertilization was associated with a significant increase in plant and shoot biomass (Table 3 and Fig. 1a and b). Significant interaction terms between fertilization and AMF inoculation were observed for traits of plant biomass, shoot biomass, root biomass and shoot:root ratio (Table 2); in +m rice, there was a significant increase in root biomass as fertilizer levels increased, whereas increasing fertilization resulted in a decline in root biomass in −m rice (Table 3 and Fig. 1c).

Figure 1:

impact of fertilization and inoculation on plant biomass production and allocation to shoots and roots. F0, F20, F40, F60, F80 and F100, provided with 0, 20, 40, 60, 80 and 100% of the local norm of nutrient input. Values shown are averages, calculated using nine replicates for each treatment (±SD of the mean).

To explore whether or not the effects of AMF inoculation on shoot:root ratios shown in Fig. 1 were influenced by treatment-induced differences in plant size at final harvest, we plotted biomass allocation patterns (−m and +m plants) against final plant biomass. In plots of shoot:root ratios against final plant biomass, there were two separate trend lines for +m rice and −m rice (Supplementary Figure S2a), suggesting that shoot:root ratios of +m and −m rice were indeed different, irrespective of plant size at a final harvest (albeit with differences being more pronounced in smaller plants). Collectively, these results suggest that while AMF inoculation has a slight negative effect on biomass production, it markedly increased biomass allocation to shoots, with the positive effect on shoot:root ratios being strongly reduced by high nutrient availability, both in small and larger plants.

Biomass allocation within shoots

Inoculation significantly increased panicle:shoot ratios, with those increases being accompanied by decreases in stem:shoot ratios at all fertilizer levels and leaf:shoot ratio at higher fertilizer levels (Tables 2 and 3; Fig. 2). Importantly, the effect of inoculation on panicle:shoot and stem:shoot ratios (but not leaf:shoot ratios) was dependent on the level of fertilization (i.e. there was a significant interaction term; Table 2), with the positive and negative effects of inoculation on panicle:shoot and stem:shoot ratios, respectively, being greater under low fertilizer levels than at high fertilizer levels (Table 3 and Fig. 2). Thus, AMF inoculation increased the relative investment of shoot biomass into panicles, with the stimulatory effect being inhibited by high fertilization.

Figure 2:

impact of fertilization and inoculation on biomass allocation within shoots. F0, F20, F40, F60, F80, and F100, provided with 0, 20, 40, 60, 80 and 100% of local norm of nutrient input. Values shown are averages, calculated using nine replicates for each treatment (±SD of the mean).

As was the case for shoot:root ratios, plots of panicle:shoot ratios, leaf:shoot ratios and stem:shoot ratios against final plant biomass showed that there were two separate trend lines between +m rice and −m rice (Supplementary Figure S2b–d); this demonstrates that the effect of inoculation on these biomass allocation parameters occurred irrespective of plant size at final harvest.

Taken together, the above results highlight that AMF inoculation markedly alters biomass allocation (not only between shoots and roots but also among panicles, leaves and stems within shoots) in mature rice harvested at the end of the growing season. In addition, the impact of inoculation on biomass allocation varied largely with fertilization level for some traits (e.g. shoot:root ratio) but less so for others (e.g. shoot biomass).

Shoot chemistry

To gain insights into whether fertilization and inoculation impact on the nitrogen and phosphorus content of individual organs of rice, we quantified the nitrogen and phosphorus concentrations of each organ (expressed as [N] and [P], respectively). Neither fertilization nor AMF inoculation had a consistent effect on panicle [N], leaf [N] or stem [N] (Supplementary Table S2); panicles [N] of the two sets of plants (i.e. +m and −m plants) were around 11–12mg g⁻¹ while those of leaves and stems were near 6mg g⁻¹. By contrast, inoculation and fertilization did influence the total amount of N in shoots, when expressed as nutrient content per hill. Specifically, AMF inoculation led to slightly higher shoot N content (Supplementary Tables S2 and S3; Fig. 3a), while the amount of shoot N increased with increasing fertilization, in both +m and −m rice (Supplementary Table S2). This indicated that the absence of changes in the mass-based concentration of N in any of the shoot tissues (but with variations in total amount of shoot N per hill) likely reflected treatment-mediated changes in total shoot biomass (Fig. 1b), combined with changes in biomass allocation within shoots (Fig. 2).

Figure 3:

impact of fertilization and inoculation on the total amount of shoot N and P (mg nutrient per hill). F0, F20, F40, F60, F80 and F100, provided with 0, 20, 40, 60, 80 and 100% of the local norm of nutrient input. Values shown are averages, calculated using three replicates for each treatment (±SD of the mean).

As was the case with [N], neither panicle [P] and leaf [P] nor stem [P] responded to either fertilization or AMF inoculation (Supplementary Tables S2 and S3). Panicle [P] and leaf [P] of the two sets of plants (i.e. +m and −m plants) were around 1.8mg g⁻¹ while those of stems were near 0.8mg g⁻¹. Additionally, shoot P contents per hill of the two sets of plants increased with increasing fertilization, while the effect of AMF inoculation was not significant (Supplementary Table S2 and Fig. 3b), suggesting that the stimulatory effect of inoculation on shoot P content was subtle.

To estimate whether the rice were N-limited or P-limited, we calculated N:P ratio for each organ and found that neither fertilization nor AMF inoculation led to significant differences in N:P ratios of panicles, leaves and stems. N:P ratios were 6.0, 3.5 and 8.0 in panicles, leaves and stems, respectively.

There were significant positive relationships between shoot:root ratios and the total amount of shoot N and P (Supplementary Table S4). Although the N and P concentrations of individual parts of shoots were not influenced, AMF inoculation markedly increased the amount of total shoot N and P under lower fertilizer levels. Thus, associated with an inoculation-mediated increase in the allocation of biomass to shoots (and to panicles within shoots) was a concomitant increase in N and P allocation to shoots.

N and P allocation within shoots

Panicle N:shoot N ratios (i.e. proportion of shoot N found in panicles) of +m rice were higher than those of −m rice (Supplementary Tables S2 and S3; Fig. 4a), while AMF inoculation had no significant effect on leaf N:shoot N ratios but decreased stem N:shoot N ratios (Supplementary Tables S2 and S3; Fig. 4b and c). There was no effect of fertilization on panicle N:shoot N ratios or on stem N:shoot N ratios. By contrast, fertilization decreased leaf N:shoot N ratios of +m rice but increased that of −m rice (Fig. 4b), confirming the significant interaction between fertilization and inoculation for these traits (Supplementary Table S2). Therefore, while inoculation impacted on leaf N:shoot N ratio and stem N:shoot N ratio differently, the overall trend was for inoculated rice to invest relatively more of the shoot N in panicles, especially at low fertilizer levels.

Figure 4:

impact of fertilization and inoculation on N and P allocation within shoots. F0, F20, F40, F60, F80 and F100, provided with 0, 20, 40, 60, 80 and 100% of the local norm of nutrient input. Values shown are averages, calculated using three replicates for each treatment (±SD of the mean).

Regarding investment patterns of P allocation, +m rice exhibited higher panicle P:shoot P ratios, but lower leaf P:shoot P ratios than −m treated rice across all fertilizer levels (Supplementary Tables S2 and S3; Fig. 4d and e). Stem P:shoot P ratios of +m rice were significantly lower than those of −m rice at low fertilizer levels (Fig. 4f; Supplementary Tables S2 and S3). The effect of fertilization and interaction between fertilization and inoculation were not significant (Supplementary Table S2).

Positive correlations were found between the relative investment of shoot biomass and nutrients (N and P) in panicles, whereas negative correlations were found between panicle:shoot ratios and several nutrient investment parameters (stem N:shoot N ratios, leaf P:shoot P ratios and stem P:shoot P ratios) (Supplementary Table S4). These correlations highlight the role that treatment-mediated changes in biomass allocation play in determining patterns of nutrient allocation within shoots of rice.

Collectively, the above results suggest that AMF inoculation strongly influences N and P allocation within shoots, with inoculated plants generally allocating a greater proportion of shoot N and shoot P to panicles (either all or some levels of fertilizer input); similarly, AMF inoculation increases investment of N in leaves in plants provided with low levels of fertilizer input. By contrast, AMF inoculation tends to reduce P investment in leaves, and both N and P investment in stems, particularly in rice provided with low inputs of fertilizer. Thus, associated with an increase in the allocation of shoot biomass to panicles was a concomitant increase in N and P allocation to panicles.

Grain yield

Given the increased investment of N and P in panicles brought by AMF inoculation (Fig. 4a and d), it was of interest to assess what effect AMF inoculation and fertilization had on grain yield. Both AMF inoculation and fertilization had a positive impact on grain yield and there was a significant interaction between these two factors (Tables 2 and 3; Fig. 5), with the effect of inoculation being greatest at low levels of fertilization. The maximum stimulatory effect of inoculation (when compared with −m rice) was ca. 62% in plants in the F0 treatment, and the minimum stimulation was 4.5% in F100-treated plants.

Figure 5:

impact of fertilization and inoculation on grain yield. F0, F20, F40, F60, F80 and F100, provided with 0, 20, 40, 60, 80 and 100% of the local norm of nutrient input. Values shown are averages, calculated using nine replicates for each treatment (±SD of the mean).

To assess what traits best predicted variations in grain yield, we calculated Pearson’s correlation coefficients (r) for bivariate relationships linking grain yield to a range of shoot traits (Supplementary Table S4). We found that grain yield was positively related to biomass allocation traits (shoot:root and panicle:shoot ratios), total amount of nutrients in shoot (shoot N and P per hill) and relative allocation of shoot N to panicles (panicle N:shoot N ratios) but negatively related to allocation of shoot N to stems (stem N:shoot N ratios) (Supplementary Table S4). Thus, by increasing overall biomass and nutrient investment in shoots, as well as investment of shoot biomass and nutrient in panicles rather than stems, yield was increased.

DISCUSSION

Building on past studies that have focussed on the effects of nutrient supply and AMF on plant growth and resource allocation (Miller et al. 2002; Yucel et al. 2009), our study investigated whether AMF inoculation influences allocation of resources to shoots and in turn to reproductive structures, and whether these effects of AMF inoculation differ among rice plants exposed to different levels of fertilizer. Our results highlight how in plants grown at low fertilizer levels, AMF inoculation markedly alters allocation of biomass and nutrients (i.e. nitrogen and phosphorus), not only between roots and shoots but also between vegetative and reproductive structures within shoots. Collectively, the results provided new insights into the applicability of the FET to AMF-inoculated rice, also highlighting the applied benefits of AMF inoculation in reducing the need for high levels of fertilizer in Chinese agriculture.

The level of AMF colonization depends on the availability of soil nutrients, particularly phosphorus (P) availability. At low P availability when plants are P-limited, plants typically allocate a greater fraction of available carbon to AMF, thus promoting AMF colonization (Johnson 2010). At high soil P availability, AMF receive less carbon from their plant host and are largely carbon limited; as a result, low colonization is expected (Treseder and Allen 2002). Given that available P in the soil in our rice paddy site was very high (160.3mg kg⁻¹), anticipated AMF colonization should have been low. Indeed, RLC by native AMF (i.e. not inoculated with G. mosseae HDSF1) was only 2–3% at our field site. This is consistent with results from other rice systems, where 3–5% of RLC was reported (Wangiyana et al. 2006). However, inoculation with our AMF isolate increased RLC to 12–19%, demonstrating that despite the high level of available P in our soils, we could markedly increase AMF colonization. Past studies have reported that AMF inoculation resulted in 25–36% RLC at field sites in Japan where soils contained 177.6mg kg⁻¹ (Solaiman and Hirata 1997b, 1998) or 187.5mg kg⁻¹ (Solaiman and Hirata 1996) of available soil P, demonstrating that colonization is possible at sites with high soil P availability, particularly when applying an AMF isolate used in our study. Clearly, further study is required to investigate why RLC values differ so markedly at sites such as ours (where soil P availability is high) when comparing roots developed in the presence of native AMF versus those inoculated with AMF isolates such as ours.

A plant’s response to AMF inoculation is thought to be context dependent. Specifically, it has been found that AMF inoculation leads to an increase in growth in plants exposed to P limitation, but a decrease, or no net benefit, when exposed to N limitation (Hoeksema et al. 2010). In our study, N:P ratios <10 were observed in panicles, leaves and stems, which indicated that the two sets of plants (+m and −m) were likely N-limited (Güsewell 2004). Moreover, we found that the overall biomass of these N-limited plants was lower in plants inoculated with AMF compared to non-inoculated plants (Fig. 1a); by contrast, investment in panicles and grain yield were both enhanced in AMF-inoculated plants (Figs 2a and 5). How might we explain this apparent paradox? Reduced overall biomass in plants colonized by AMF has been reported previously (Johnson et al. 2008), potentially reflecting a sink demand for carbon by the developing hyphal network, which reduces the availability of carbon for investment in vegetative biomass. Under conditions of N deficiency, one might expect an even greater partitioning of photosynthate to the developing hyphal network (Franklin and Ågren 2002; Hoeksema et al. 2010), as AMF are known to play an important role in taking up not only phosphorus but also nitrogen (Hodge and Fitter 2010). If correct, this diversion of photosynthate to establishment of the AMF hyphal network presumably did not come at the cost of carbon, N or P investment in reproductive organs, as seed yield was markedly enhanced by AMF inoculation irrespective of the fertilizer input (Fig. 5).

When considering how AMF inoculation impacts biomass allocation, thought needs to be given to ontogenetic and plant-size-dependent shifts in allocation (Niklas and Enquist 2003), and the impacts such shifts can exert on comparisons made at a single time point. In our study, we compared allocation in fully mature plants at the time of grain harvest; at that point, marked differences in plant size were apparent at low levels of fertilizer input, with the inhibitory effect of low fertilizer on plant size differing between +m and −m plants (Fig. 1a). While such variations in plant size might impact the observed patterns in biomass and nutrient allocation, they do not negate our finding that AMF inoculation increases investment in reproductive tissues of rice at the point of grain harvesting. Moreover, in our study, both AMF inoculation and fertilization led to alterations in plant size (plant biomass) (Fig. 1a), and all traits related to biomass allocation responded to fertilization or inoculation, but not both (Figs S1c and 2). Furthermore, plots of each biomass allocation trait against plant biomass at harvest revealed differences between +m and −m plants at common total plant mass values, particularly in smaller plants at low fertilizer levels (Supplementary Figure S2). Therefore, in our experiment, the effect of AMF inoculation on biomass allocation was not a function of plant size, but a response to the limiting external resource. We conclude, therefore, that irrespective of the potential impact of plant size on biomass and nutrient allocation, AMF inoculation markedly alters investment patterns in rice, particularly at low levels of fertilizer input.

Our results suggest that AMF inoculation increased shoot:root ratios especially in plants exposed to nutrient deficiency (Hypothesis 1). Our data indicated that AMF inoculation played a more important role in altering shoot:root ratios by decreasing root biomass for plants provided with low levels of fertilizer than it did for well-fertilized plants (Fig. 1c and d). However, this was not consistent with a recent meta-analysis (Veresoglou et al. 2012), where AMF had no effect on biomass allocation to shoots and roots in plants subjective to abiotic stresses other than nutrient deficiency. In that meta-analysis, the various stresses considered were drought, heavy metal and acid precipitation. Therefore, part of the reason for this apparent discrepancy may be that the effect of AMF inoculation on shoot:root ratios differs depending on the stress factor present in the ecosystem under study. In the case of N deficiency, non-inoculated and inoculated roots are unlikely to exhibit the same efficiency in N uptake as AMF are known to have higher-affinity N transporters than plant roots (Harrison and Vanbuuren 1995; Lopez-Pedrosa et al. 2006). If the same was true for our plants, then under lower fertilizer levels, the inoculated roots would have taken up more N than non-inoculated roots. Support for this suggestion comes from the finding that inoculated rice exhibited higher total amount of shoot N than non-inoculated rice (Fig. 3a) and the positive correlation between shoot:root ratios and shoot N contents per hill (Supplementary Table S4).

As noted earlier, the hydromorphic paddy soil at our field site contained 160.3 mg·kg⁻¹ of available P—considered relatively high for rice production. While it is clear that this did not preclude colonization of roots by the G. mosseae HDSF1 isolate (as shown by the RLC values of up to 19%), it seems unlikely that the increased AMF colonization would have improved P uptake in the host plants, as past studies have reported little AMF-facilitated P uptake when soil P availability is high (Fageria and Santos 2002). Our results also indicated that high nutrient availability had a negative effect on the abundance of AMF (indicated by RLC, Supplementary Figure S1) (see also Liu et al. 2012; Treseder 2004). Thus, it seems unlikely that AMF inoculation would have enhanced P uptake in our study; given this, we suggest that the effects of inoculation on biomass allocation were likely mediated by other factors, such as enhanced uptake of N in our N-limited plants.

In a recent review, Johnson (2010) hypothesized that fertilization should enhance shoot:root ratios of both inoculated and non-inoculated plants, reflecting a shift to limitations in above-ground resource availability (e.g. competition for light) at high levels of nutrient availability. Given this, we anticipated that AMF-inoculated rice would conform to the FET (H1). In our experiment, the shoot:root ratio of both inoculated and non-inoculated rice was observed to increase with rising fertilization, while the slope of the shoot:root ratio ↔ fertilizer relationship was reduced by AMF inoculation (with an increased y-axis intercept; Fig. 1d) and there was significant interaction between fertilization and inoculation on shoot:root ratio. This suggests that the conformation of AMF-inoculated plants to the FET was weaker than that of non-inoculated plants.

One explanation of the weaker conformation of AMF plants to the FET might be an underestimate of root biomass for +m plants at low fertilizer levels, due to our sampling method and the role of AMF inoculation in regulating root structure (Maillet et al. 2011). To estimate root biomass, we sampled intact cores (length: 15cm; width: 10cm; depth: 15cm) with rice plants in the centre of the cores—consequently, it is possible that parts of lateral roots growing outside this core were not collected, leading to underestimation in root biomass and then overestimation in shoot:root ratio. Additionally, past studies have observed that AMF increased the root length and specific root length (i.e. ratio of root length to root mass) of the host (Berta et al. 1995). If this had happened in our study, it may be that the underestimation in root biomass was exacerbated under low fertilizer levels as more roots grew beyond the cores. Support for this suggestion comes from the finding that inoculation resulted in an increase in the total amount of shoot N and P (per hill) (Fig. 3). It seems likely, therefore, that the AMF-inoculated plants do conform to the FET, albeit to a reduced extent compared to non-inoculated plants.

In contrast to Johnson (2010), other studies have reported results that suggest that AM colonized plants do not necessarily conform to the FET. For example, fertilization has been reported to decrease shoot:root ratios in colonized plants (Davies et al. 2005; Rabie 2005). Here, consideration needs to be given to the treatments used: in most studies, only two levels of nutrient availability were provided, with fertilization limited to addition of P alone (i.e. without alteration in the supply of other key nutrients such as N and K). Such studies have also tended to be of a small experimental scale (pot experiments in greenhouses). Our results, from field experiments involving a wider range of nutrient treatments and balanced fertilization (i.e. provision of a gradient of NPK) in combination with studies on non-inoculated plants (McCarthy and Enquist 2007; Poorter and Nagel 2000; Poorter et al. 2012; Vannoordwijk and Dewilligen 1987), provide support for the FET in terms of shoot–root relationship for both inoculated and non-inoculated plants, but with the degree of support for the FET greater in non-inoculated plants.

Some researchers have reported that AMF can have a positive impact on the reproduction of host plants (Miller et al. 2002). In view of this, we hypothesized that AMF inoculation would increase relative investment of shoot biomass in panicles (H2). In our study, as expected, AMF inoculation increased panicle:shoot ratios at all fertilizer levels (Fig. 2a). Similar results were reported previously in Japan (Solaiman and Hirata 1995). It has also been reported that in non-inoculated plants, the highest quantities of nutrients were observed in the panicles of rice (relative to other organs; Roy et al. 2012), indicating that nutrients allocation to panicles was of higher priority than to stems and leaves. Our results provided support for H2 with AMF inoculation increasing panicle:shoot ratios. We also found negative correlations between a number of biomass allocation ratios and nutrient allocation ratios (Supplementary Table S4). This supports the findings of earlier studies in rice, which reported that translocation of nutrients from shoots to grain was accelerated by AMF inoculation (Solaiman and Hirata 1997b, 1998).

The majority of studies that have quantified biomass allocation to reproductive structures focused on the harvest index (i.e. the proportion of the dry matter present in harvested structures of the crop relative to total dry matter production) (Dordas 2009). Harvest index has been found to decrease with increasing N availability (Li et al. 2012), increase with increasing P availability (Fageria and Santos 2002), but remain unchanged in plants grown in balanced nutrient availability (Tungate et al. 2002). However, AMF hyphae that are directly involved in nutrient acquisition and transportation to the host plant have been shown to decrease in abundance with rising nutrient availability (Johnson et al. 2003). This suggests that the benefit plants obtain from AMF, i.e. improvement in N and P availability, decreases with increasing fertilization. In our study, the positive effect of AMF inoculation on rice grain yield was highest (near 62% stimulation) at the lowest fertilizer level, then reduced as fertilization increased, with stimulation of yield being limited to ~4.5% at the highest level of fertilization (Fig. 5); this finding supports our hypothesis (H3). Similar results were reached in an earlier study where acerola seedlings displayed the maximum responsiveness in suboptimal nutrient conditions (Balota et al. 2011).

From our data, the positive effect of AMF inoculation on grain yield resulted from positive correlations between grain yield and shoot:root ratio and panicle:shoot ratio (Supplementary Table S4), which was associated with improved N accumulation (Fig. 3a) and redistribution of N and P within shoots (Fig. 4). In addition, our data also showed that increasing nutrient availability decreased the impact of AMF inoculation on rice grain yield (Fig. 5). The likely reason may be that the effect of AMF inoculation on N allocation within shoots declined at higher fertilizer levels (Fig. 4a–c).

CONCLUSIONS

Our study reports for the first time that field-grown rice inoculated with AMF conform to the FET, albeit to a reduced extent compared to non-inoculated plants. We also show that AMF inoculation results in an increase in allocation of plant biomass to shoots coupled to increasing N accumulation in shoots, especially under low fertilizer input. Importantly, our study further indicates that AMF inoculation increased allocation of N and P to panicles, resulting in markedly higher grain yield (≈62% increase relative to non-inoculated rice). Consequently, we suggest that AMF inoculation could be a powerful tool for improving rice grain yield, especially under low nutrient conditions. If true for rice production more widely, the next step will be ensuring that extensive field studies are conducted with the aim of developing region-specific high-quality AMF inoculum to ensure better exploitation of AMF under sustainable agriculture systems.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Journal of Plant Ecology online.

FUNDING

National Natural Science Foundation of China (51179041); Major Science and Technology Program for Water Pollution Control and Treatment (2013ZX07201007); National Creative Research Group from the National Natural Science Foundation of China (51121062); State Key Lab of Urban Water Resource and Environment, Harbin Institute of Technology, China (HIT) (2011TS07); Natural Science Foundation of Hei Longjiang Province, China (E201206); Special Fund for Science and Technology Innovation of Harbin (2012RFLXS026); Australian Research Council Centre of Excellence in Plant Energy Biology (CE140100008 to O.K.A.).

ACKNOWLEDGEMENTS

We would like to thank Fuqiang Song for the inoculum production and Xue Zhang, Guixiang Liu, Shengjie Fu, Xiaofeng Jiang and Yanan Xu for their contributions during the course of the study.

Conflict of interest statement. None declared.