The effects of arbuscular mycorrhizal fungi and root interaction on the competition between Trifolium repens and Lolium perenne

Experimental design

We employed a full factorial design that consisted of combinations of five plant-competition treatments, two levels of AMF inoculation (+AMF and −AMF) and two levels of root interaction treatments (+Root and −Root). Individuals of T. repens (T) and L. perenne (L) were planted in the mixed ratios: 8:0, 6:2, 4:4, 2:6 and 0:8 (T8L0, T6L2, T4L4, T2L6 and T0L8, respectively) to simulate competition between these two species. There were 20 treatments in total with five replicates per treatment. To investigate the amount of nitrogen fixed by T. repens and potentially transferred to other plants, additional three replicates were added for T8L0, T6L2, T4L4 and T2L6 in both +AMF and −AMF treatments. We established 124 microcosms under greenhouse conditions in cylindrical containers (Fig. 1). Microcosm without root interaction was achieved by eight separated plant growth pillars in which plant individuals were planted respectively (Fig. 1A). Pillar was made of 25 μm nylon mesh, which allowed arbuscular mycorrhizal mycelia access but not roots. Each pillar contained soil mixture of 188 g. The eight plant growth pillars were evenly located close to the container wall. In microcosm with root interaction, eight plant individuals were planted together and evenly spaced into the space between the container and root restriction pillar, which contained soil mixture of 1,504 g (188 g *8). Thus, the total growth space for the roots of the eight plants was the same in microcosms without root interaction and in microcosms with root interaction. The methods only tested the root physical interactions (root interaction in the following text) as roots also interact through their exudates, which cannot be avoided to go through the mesh we used in the experiment.

Soil and inoculum preparation

Each container was 15 cm in height and 19 cm in diameter and was filled with 5.88 kg sterilized (25 kGy γ-irradiation) soil mixture of field soil, sand and grass peat (5:4:1 v/v/v). Soil used in the experiment was collected from Pailou experimental station of Nanjing Agricultural University (118.78°E, 22.04°N). The characteristics of the soil were measured, which contained 5.60% soil water content, 7.70 g/kg soil organic C, 18.49 mg/kg available P, 36.24 mg/kg nitrate (NO₃^–), 19.61 mg/kg ammonium (NH₄⁺), 0.83 g/kg total nitrogen and pH of 7.42.

The five AMF species used were Claroideoglomus etunicatum, Glomus tortuosum, Rhizophagus intraradices, Glomus versiforme, Glomus aggregatum. Soil inoculation with AMF strains purchased from Beijing Academy of Agriculture and Forestry, China. AMF spores contained in soil inoculation were propagated in autoclaved (121 °C for 120 min) substrate (sand/soil, 1:2) with maize for two successive propagation cycles (three months for each cycle). The soil of five AMF species were mixed as inoculants with spore density of 48/g.

AMF inoculation and plant growth conditions

In microcosms without root interaction, each plant growth pillar received 6 g of AMF inoculants for +AMF treatment and 6 g of autoclaved inoculants (121 °C for 120 min) for −AMF treatment. In microcosms with root interaction, the soil mixture in the growth space between the container and root restriction pillar was inoculated with 48 g AMF inoculum for +AMF treatment and 48 g autoclaved inoculants (121 °C for 120 min) for −AMF treatment. To compensate other lost soil microbe because of sterilization, 80 ml filtrate from fresh field soil mud excluding AMF by 25 μm mesh was applied to each microcosm. Because the root of T. repens is easily colonized by rhizobium in the field, 24 ml (OD₆₀₀ = 0.90) mixed rhizobium inoculants (Rhizobium sp. WYCCWR R10051 and Rhizobium sp. WYCCWR R10062) isolated from T. repens was added to each microcosm. Rhizobium inoculants were kindly provided by Dr. Zhang Junjie in Zhengzhou University of Light Industry (Zhang et al., 2016).

We used commercial seeds of T. repens (Trade name: Haifa) and L. perenne (Trade name: Maidi). The two varieties were widely used in south of China for pasture cultivation. The seeds were sterilized with 10% H₂O₂ for 10 min, washed with sterilized water and germinated in wet filter paper in lighting incubator (20 °C) for two to four days. And then seedlings were transplanted into each microcosms. Three seedlings of T. repens or L. perenne were transplanted to each plant growth pillar in the microcosms without root interaction. We alternated T. repens with L. perenne in the T4L4 treatment. Three seedlings of T. repens or L. perenne were randomly located in one growth pillar and another three seedlings of T. repens or L. perenne were transplanted into the opposite pillar in T2L6 or T6L2 treatment, respectively. Seedlings of T. repens or L. perenne were spaced in the same way in microcosms with root interaction.

The seedlings were grown under glasshouse conditions at 50–70% relative humidity, a temperature regime of 20–25 °C during day and night and natural lighting. Each microcosm was watered weekly, and soil water content was corrected to 15.4% (field capacity) every two weeks. One week later, the strongest seedlings were left and the other two were removed from each pillar of the microcosms without root interaction and the corresponding location in microcosms with root interaction. We randomized the location of all microcosms every two weeks.

Harvesting and measurements

Shoots were harvested nine, 13 and 16 weeks after planting to simulate mowing/grazing and reduce possible shoot competition for light. We failed to determine nitrogen transfer among plant individuals because of technical problems. Samples in these microcosms were added to each harvest. At the time of harvest at nine and 13 weeks, shoots were mowed to 5 cm height, sorted by species, oven dried at 65 °C for 72 h and weighed. During the final harvest all plants were destructively harvested, separated into shoots and roots and sorted by species. Roots of one plant individual of T. repens or L. perenne in each microcosm were frozen at −20 °C for determining AMF root colonization. Shoots and the left roots were oven dried at 65 °C for 72 h and weighed. We estimated the shoot dry matter for each species by pooling the three harvests. We assumed that plant individual of the same species in each microcosm had the same root dry matter. Thus, the dry matter value for roots used for determining colonization was estimated by the average root biomass of the same species. Shoot N content was examined using alkaline-hydrolysable diffusion method with Kjeldahl apparatus (Kjeltec Analyzer Unit 8400; FOSS, Hillerod, Sweden) (Bremner, Smith & Tarrant, 1996). Shoot P content was measured using an inductively coupled plasma emission spectrometer (TJA IRIS Advantage/1000 Radial ICAP Spectrometer; Thermo Jarrell Ash, Franklin, MA, USA), following digesting the plant tissue with trace-metal-grade nitric and perchloric acid and diluting in 100 ml of double-distilled water.

Frozen roots were cut into approximately 1 cm fragments. The root samples were cleared in 10% (w/v) KOH at 90 °C in a water bath for 60 min, and then washed and stained with 0.05% (w/v) Trypan blue. Thirty root segments from each sample were examined microscopically to assess AMF root colonization (Trouvelot, Kough & Gianinazzi-Pearson, 1986).

Statistical analysis

Competitive interactions between T. repens and L. perenne were calculated as the relative shoot biomass per individual (RY_ind) by the equation: RY_ind = O_ij/M_ij, where O_ij is the shoot biomass per individual of plant species i grown in mixed planting microcosm of root interaction × AMF treatment combination species j, and M_ij is the mean shoot biomass per individual of plant species i grown in monoculture microcosm of root interaction × AMF treatment combination species j (de Wit, 1960; Wagg et al., 2011).

Relative yield totals (RYTs) were calculated by the equation: RYTs = RY₁ + RY₂, where RY₁ or RY₂ are shoot biomass of plant species 1 or 2 in mixture divided by the shoot biomass in the monoculture of root interaction × AMF treatment combination. We used RYTs to estimate overyielding in grass–clover mixtures. Values greater than 1 indicate overyielding that a greater biomass production was observed in mixtures than the average of the two species in monoculture (de Wit, 1960; Wagg et al., 2011).

A general linear model (GLM) was used to determine the effect of planting ratio, root interaction and their interactions on AMF root colonization in the +AMF treatment. GLM was also applied to analyze the effects of AMF inoculation, planting ratio, root interaction and their interactions on shoot and root biomass, RY_ind, RYT, shoot N and P content. AMF inoculation, planting ratio and root interaction were treated as the fixed factors in these analyses. The data were log-transformed prior to analysis to ensure normality and homogeneity by using Shapiro–Wilk test.

As AMF inoculation and its interaction with other factors did not significantly alter any measurements (Tables S1–S3), +AMF and −AMF were treated as replicates in the following analysis. We used GLM to test the main factor of planting ratio, root interaction and their interaction on each measurement. Tukey’s simultaneous test were used to compare the variations between +Root and −Root treatment, and the level of significance was p < 0.05. LSD multiple-range test was applied to analyses the least significance differences among planting ratio treatments. All statistical analyses were performed using SAS Version 8.0 (SAS Institute Inc., Cary, NC, USA), and all figures were made by using Sigma plot 12.0 (Systat Software Inc., San Jose, CA, USA).

AMF inoculation effect

AMF inoculation and its interaction with other factors did not significantly alter any measurements in our study (all p > 0.05; Tables S1–S3). The average mycorrhizal root colonization was less than 0.5% in the −AMF treatment, while AMF successfully colonized both T. repens and L. perenne in the +AMF treatment by 39% and 21%, respectively. As Fig. S1 shown, mycorrhizal root colonization of T. repens was not affected by either root interaction, planting ratio or their interactions (p > 0.05). Mycorrhizal root colonization of L. perenne was affected by planting ratio (p < 0.001). Mycorrhizal root colonization of L. perenne increased significantly along with the increasing proportion of T. repens in mixtures.

Above- and belowground biomass of T. repens and L. perenne

The aboveground biomass of T. repens and L. perenne were all significantly affected by root interaction, planting ratio and their interactions, as well as belowground biomass (p < 0.05; Fig. 2). The exception was total aboveground biomass, which was not affected by root interaction (p > 0.05). When the planting ratios ranged from T2:L6 to T4:L4 and T6:L2, the aboveground biomass of T. repens with root interaction respectively decreased from 47%, 46% to 42% in comparison with no root interaction (Tukey’s simultaneous test, p < 0.05; Fig. 2A). The aboveground biomass of L. perenne increased by 10% and 30% at the planting ratio of T4L4 and T6L2 (Tukey’s simultaneous test, p < 0.05) and did not alter aboveground biomass in monoculture and the planting ratio of T2L6 (Tukey’s simultaneous test, p > 0.05; Fig. 2B). Root interaction significantly decreased total aboveground biomass at the planting ratio of T4L4 and T8L0 (Tukey’s simultaneous test, p < 0.05; Fig. 2C). Total aboveground biomass peaked at the planting ratio of T4L4 (Fig. 2C).

The belowground biomass of T. repens decreased with root interaction in the mixed planting (Tukey’s simultaneous test, p < 0.05; Fig. 2D), which was similar to the responses of aboveground biomass (Fig. 2A). In monoculture, though the presence of root interaction increased belowground biomass of T. repens by 21% (Tukey’s simultaneous test, p < 0.05; Fig. 2D), the aboveground biomass of T. repens was decreased by 12% (Tukey’s simultaneous test, p < 0.05) with root interaction, indicating that the root interaction in intraspecies decreased the aboveground biomass of T. repens. In contrast, the belowground biomass of L. perenne significantly increased with root interaction in the mixed planting (Tukey’s simultaneous test, p < 0.05; Fig. 2E). In monoculture, the presence of root interaction increased belowground biomass of L. perenne by 64% (Tukey’s simultaneous test, p < 0.05; Fig. 2E), while the aboveground biomass of L. perenne was not altered with root interaction (Fig. 2B), indicating that the intraspecific root interaction did not affect the aboveground biomass of L. perenne. Root interaction significantly increased total aboveground biomass at the planting ratio of T0L8 and T2L6 (Tukey’s simultaneous test, p < 0.05; Fig. 2F)

Relative yields

The relative yield per individual (RY_ind) of T. repens and L. perenne were strongly influenced by root interaction, planting ratio and their interactions (Tukey’s simultaneous test, all p < 0.05; Figs. 3A and 3B). In mixtures with L. perenne, the RY_ind of T. repens was depressed below its RY_ind in monoculture by 27% in the absence of root interaction (Fig. 3A). The presence of root interaction reduced the RY_ind of T. repens by 54% compared with that in monoculture (Fig. 3A), suggesting that competitive pressure by L. perenne root decrease the growth performance of T. repens.

In mixtures with T. repens, the RY_ind of L. perenne was promoted above its RY_ind in monoculture by 65% on average in the absence of root interaction (Fig. 3B). The presence of root interaction largely enhanced the RY_ind of L. perenne by 101% on average compared with that in monoculture (Fig. 3B), suggesting that L. perenne benefited from interspecies root interaction. In addition, the RY_ind of L. perenne increased along with the proportion of T. repens in mixture (p < 0.0001; Fig. 3B). The positive effect of root interaction on the RY_ind of L. perenne was also raised with the increase in the proportion of T. repens in mixture (Root interaction × Planting ratio, p < 0.0001; Fig. 3B). Planting ratio heavily affected the RYT (p < 0.0001; Fig. 4). RYT peaked at the 4:4 mixture of T. repens and L. perenne.

N and P content

Shoot N content of T. repens were not affected by root interaction, planting ratio or their interactions (Table S3). Shoot N content of L. perenne significantly decreased when its planting ratio changed from monocultures to mixtures in any ratio with T. repens (LSD, p < 0.05; Table 1). Shoot P content of both T. repens and L. perenne were not significantly altered by root interaction, planting ratio or their interactions (Table S3).