Species-specific synergistic effects of two plant growth—promoting microbes on green roof plant biomass and photosynthetic efficiency

Rhizophagus irregularis, an arbuscular mycorrhizal fungus, and Bacillus amyloliquefaciens, a bacterium, are microorganisms that promote plant growth. They associate with plant roots and facilitate nutrient absorption by their hosts, increase resistance against pathogens and pests, and regulate plant growth through phytohormones. In this study, eight local plant species in Finland (Antennaria dioica, Campanula rotundifolia, Fragaria vesca, Geranium sanguineum, Lotus corniculatus, Thymus serpyllum, Trifolium repens, and Viola tricolor) were inoculated with R. irregularis and/or B. amyloliquefaciens in autoclaved substrates to evaluate the plant growth−promoting effects of different plant/microbe combinations under controlled conditions. The eight plant species were inoculated with R. irregularis, B. amyloliquefaciens, or both microbes or were not inoculated as a control. The impact of the microbes on the plants was evaluated by measuring dry shoot weight, colonization rate by the arbuscular mycorrhizal fungus, bacterial population density, and chlorophyll fluorescence using a plant phenotyping facility. Under dual inoculation conditions, B. amyloliquefaciens acted as a “mycorrhiza helper bacterium” to facilitate arbuscular mycorrhizal fungus colonization in all tested plants. In contrast, R. irregularis did not demonstrate reciprocal facilitation of the population density of B. amyloliquefaciens. Dual inoculation with B. amyloliquefaciens and R. irregularis resulted in the greatest increase in shoot weight and photosynthetic efficiency in T. repens and F. vesca.


Introduction
Countless microorganisms reside and propagate in the rhizosphere where plant roots and soil meet. Some of the microbes have neutral or lethal effects on the growth and survival of plants, whereas others support their host plants via various mechanisms [1]. Because  R. irregularis and B. amyloliquefaciens could be achieved in plants suitable to be grown on green roofs, it would optimize ecosystem services provided by green roofs to urban areas via plant growth enhancement.
In this study, we determined whether synergistic interactions occur between R. irregularis and B. amyloliquefaciens by setting up greenhouse experiments under controlled conditions using eight local plant species from Finland as hosts (Antennaria dioica, Campanula rotundifolia, Fragaria vesca, Geranium sanguineum, Lotus corniculatus, Thymus serpyllum, Trifolium repens, and Viola tricolor). A few of these plants, such as T. serpyllum, T. repens, and L. croniculatus [46][47][48], are also stress resistant and hence potentially more suitable for green roofs. Four treatments were conducted for each plant species: inoculation with (i) R. irregularis, (ii) B. amyloliquefaciens, (iii) co-inoculation with both microbes, and (iv) no microbial amendment (control). The aim was to find out which plant-microbe combinations promote the physiological performance of plants. Shoot biomass and the photosynthetic efficiency were chosen as indicators of physiological performance, as they directly reflect plant physiology and are relatively easy to measure.

Experimentation
Experiments were conducted in the Small Plant Unit of the National Plant Phenotyping Infrastructure (NaPPI) on the Viikki campus, University of Helsinki, Finland [49] (https://www. helsinki.fi/en/infrastructures/national-plant-phenotyping). NaPPI is an in-house facility that conducts high-throughput and high-precision plant phenotyping. It consists of programmable tools for weighing, watering, and imaging the plants-both for red-green-blue pictures (RGB Camera) and chlorophyll fluorescence (Fluorecam). These tools allow comprehensive analysis of plant growth and physiology. Light-emitting diode lamps were used to avoid heating the plants by illumination.
Seed of the plant species were purchased from Suomen Niittysiemen (Jyväskylä, Finland) and stored at 4˚C for 2 weeks before sowing. The inoculant products MYC4000 (4000 fungal spores of R. irregularis strain DAOM 181602 per gram) and Rhizocell (>10 9 CFU endospores of B. amyloliquefaciens strain FZB42 per gram) were provided by Lallemand Plant Care (Castelmaurou, France). MYC4000 and Rhizocell are powdered products. They were dissolved in distilled water according to the manufacturer's instructions before being sprayed onto the growth substrate.
Plant seeds were sown in autoclaved sand for germination. After 4 weeks, seedlings were individually transplanted into plastic pots (8 × 8 × 8 cm; VWR, Center Valley, PA, USA) filled with 450 cm 3 autoclaved soil from a conifer forest (provided by Kekkilä Oy, Vantaa, Finland), a soil volume that is typically used for rooting plants [50][51][52][53]. The small plant unit at NaPPI is designed to fit only this particular pot size, and it is suitable for plants up to 50 cm in height. The substrates used in this study were low in nutrients, and no fertilizer was applied to keep the seedlings smaller and to more readily observe the effects of the inoculants on plant growth [52,54,55]. A shallow soil depth was used, as is typical for green roofs. The soil properties were pH 6.4; organic matter, 5.6%; soluble P, 2.2 mg/kg; and soluble N, 0.4 mg/kg. For each plant species, 24 seedlings were divided into four treatments: six seedlings were inoculated with R. irregularis (R), six seedlings with B. amyloliquefaciens (B), six seedlings were co-inoculated with both microbes (R+B), and six seedlings remained as uninoculated controls. Altogether, 192 transplanted seedlings were maintained in the NaPPI facility for 7 weeks, during which the day/night length was set to 16 h/8 h. Light intensity was 172 μmol m −2 s −1 . Seedlings were irrigated daily to maintain a water-holding capacity of 65% [56]. Two independent, repeated NaPPI experiments were conducted in July 2015 and June 2016. During the first experiment, the NaPPI watering program was interrupted a few times.
Before sampling, each plant was measured for chlorophyll fluorescence activity with the Fluorecam in the NaPPI to determine the photosystem II photochemical capacity. This was calculated from the ratio of the variable fluorescence to the maximum chlorophyll fluorescence (Fv/Fm), which reflects the photosynthetic efficiency of tested plants [57]. The optimal value is 0.83 for most plant species, and a value lower than that indicates poor plant growth [58]. Plant shoots (including all aboveground biomass such as branches, leaves, and stems) were cut, desiccated at 70˚C in an oven for 48 h, and weighed to measure their dry biomass. Soil adhering to roots was collected from six different places in each pot and transferred to screw cap tubes. Each soil sample was mixed by vigorous shaking and stored at 4˚C. Roots were carefully brushed and stored in 70% ethanol at 4˚C.

Detection of R. irregularis in root samples
The classical AMF detection method includes root staining, microscope slide preparation, and AMF quantification [59]. After 7 weeks of cultivation in NaPPI, a root sample was collected from each of three random plants for each species and treatment. Roots were washed and stored in 70% ethanol until staining with Trypan Blue. The staining protocol was adjusted for each plant species [60] (Table 1). In brief, the roots were first soaked and softened in KOH solution, which made the staining process more effective. Then the roots were immersed in 1.5% hydrogen peroxide containing 5 ml/l ammonia (H 2 O 2 +NH 3 ) to remove the background color of the cells. Next, the roots were transferred into 1% HCl solution and held in Trypan Blue solution before being stored in clear glycerol (Table 1). Finally, the roots were mounted on microscope slides in polyvinyl-lacto-glycerol solution (10 ml/l water, 10 ml/l lactic acid, 1 ml/l glycerol, and 1.66 mg/l polyvinyl alcohol).
A modified gridline intersect method was applied for AMF quantification [61]. There was a fine crosshair pre-scored on the ocular, and intersections of root and the vertical crosshairs were made by vertically moving the slides. When the vertical hair intersected a root, AMF structure (arbuscule, vesicle, or hypha) that was intersected by the vertical hair was recorded. Observations were categorized into eight groups: hypha (H); arbuscule (A); vesicle (V); hypha +arbuscule (HA); hypha+vesicle (HV); arbuscule+vesicle (AV); hypha+arbuscule+vesicle (HAV); and no arbuscules, vesicles, or hyphae (negative, N). A total of 100 intersections were

Detection of B. amyloliquefaciens in soil samples
The primer pair BaG3F (5´-GTCGACCACTCTTGACGTTACGGTT-3´) and BaG4R (5´-CG ATCACTTCAAGATCGGCCACAG-3´), which amplifies a 94-bp fragment from gyrB was used to identify and quantify B. amyloliquefaciens in soil samples. Before quantitative PCR (qPCR), soil DNA samples were diluted to 5 ng/μl, and Rhizocell DNA was diluted with Milli-Q water at ratios of 1:1, 1:10, 1:100, 1:1000, and 1:10000. The Rhizocell DNA dilutions were used to produce a standard curve and calculate amplification efficiency for each qPCR run. qPCR reactions were run with the following program: 5 min at 95˚C; 45 cycles of 10 s at 95˚C, 10 s at 62˚C, and 10 s at 72˚C; and 5 min at 72˚C. B. amyloliquefaciens population densities from soil samples were calculated according to the standard curve equation [62] Population density of Bacillus amyloliquefaciens ¼ 10 ðCtÀ mÞ=À slope � n wt where "Ct" is the cycle threshold value from the qPCR; "slope" and "m" are the slope value and intercept value of the standard curve, respectively; "wt" is the weight of the soil from which the DNA was extracted; and "n" is the dilution ratio of each soil DNA sample.

Statistical analysis
Mean values for the plant shoot dry weight, photosynthetic efficiency, and AMF colonization rate were compared using least significant difference analysis (LSD 0.05 ,% confidence). Levels of significance for the treatments, host plant species, and their interactions were calculated by analysis of variance (ANOVA) using the SPSS software package (IBM SPSS Statistics 25, Armonk, NY, USA).

R. irregularis is enhanced in R+B treatments
Eight selected plant species were inoculated with R. irregularis or B. amyloliquefaciens, were co-inoculated with both microbes, or were untreated as controls. Two repeated NaPPI experiments were conducted to study the effect of colonization on host plants. In the first experiment, AMF structures were not detected in plants from any of the four treatments. In the second experiment, hyphae of AMF were detected in five of eight plants from treatment R, whereas arbuscules and vesicles were rare or absent in all tested plants. In the R+B treatment, AMF structures such as arbuscules and hyphae were observed in all eight plant species (Fig 1) and vesicles were observed in all tested plants except C. rotundifolia. Even though hyphae and arbuscules were detected in roots of C. rotundifolia, the colonization rate was low (3% and 1%, respectively), and no vesicles were observed (Table 2). According to the LSD 0.05 test, colonization by the specific fungal structures was significantly higher with the R+B treatment than with R alone (P < 0.05; Table 2). According to the AMF colonization response to co-inoculation with both microbes, the plant species could be divided into three groups: 1) plants that were not active R. irregularis host plants, regardless of the presence of B. amyloliquefaciens in the soil (C. rotundifolia); 2) plants that were R. irregularis host plants only when B. amyloliquefaciens was co-inoculated (L. corniculatus and T. repens); and 3) plants that were R. irregularis host plants when inoculated alone but became more efficiently colonized by R. irregularis when B. amyloliquefaciens was co-inoculated (G. sanguineum, F. vesca, V. tricolor, T. serpyllum, and A. dioica). The rates of AMF structures in roots followed the pattern hyphae > arbuscules > vesicles in all the plant species, i.e., hyphae were the most prevalent AMF structure ( Table 2).

B. amyloliquefaciens colonized all tested plant species
B. amyloliquefaciens was detected in varying amounts in the soil adhering to roots in nearly all plants inoculated with the bacteria. Extraction of the DNA from the soil and sequencing of the soil DNA confirmed that the gyrB sequence isolated from soil DNA matched that of the bacteria in the Rhizocell product. B. amyloliquefaciens was not detected by qPCR in the soil of nontreated controls nor in soil from R treatment. According to LSD 0.05 analysis in both experiments, no statistically significant difference (i.e., P < 0.05) in B. amyloliquefaciens population density was detected between the treatments B and R+B (Table 3).

Shoot weight of the plants
Differences in the plant shoot weight were studied in the two experiments by analyzing the fold change of the treated plants (R, B, and R+B) compared with untreated control plants. Plants were placed in two groups, namely those with consistent results (Fig 2) and those with less repeatable results (S1 Fig) obtained in the NaPPI experiments. The shoot weight of the four According to an ANOVA, there was a statistically significant interaction between the host plant species and treatment for shoot weight for all plant species (S1 Table, S2 Table).

Chlorophyll fluorescence
The four plant species that showed a consistent effect for shoot weight (Fig 2) also showed reproducible patterns in photosynthetic efficiency in the two NaPPI experiments (Fig 3). For C. rotundifolia and T. serpyllum, inoculation with the microbes showed little influence on photosynthetic efficiency. For T. repens and F. vesca, the Fv/Fm ratio was lowest in untreated control plants (for T. repens, the Fv/Fm ratio in the R treatment group and in the control was too low to be detected) and was higher after B and R+B treatments. Moreover, R+B treatment in T. repens and F. vesca produced the highest Fv/Fm ratio in at least one of the experiments (Fig  3). The Fv/Fm ratios for the four plant species that showed less-consistent results in the two experiments are presented in S2 Fig and S4 Table. The host plant species, treatment, and their interaction had significant effects on photosynthetic efficiency in all plant species (S3 Table, S4 Table).

Discussion
Mycorrhiza helper bacterium (MHB) is a generic name for bacteria that can stimulate the formation of a mycorrhizal association with host plants [63]. In our study, the colonization rate of R. irregularis among the eight tested plant species was improved by co-inoculation with B. amyloliquefaciens. Moreover, for some plant species (L. corniculatus and T. repens), successful colonization by R. irregularis could only happen when B. amyloliquefaciens was co-inoculated. We can infer that AMF colonization in some plant species was strongly dependent on the presence of a MHB (B. amyloliquefaciens in this case).
Our findings are consistent with the studies of Yusran et al. [64], which indicated that different species of MHB promote mycorrhization to different levels and that co-inoculation with a MHB and an AMF has a positive impact on biomass and phytopathogen control in tomato plants. Our results show that the colonization-promoting effects of B. amyloliquefaciens on R. irregularis can occur across a broad range of host plants suggesting that such a promoting effect is fungus specific, rather than host plant specific [65,66]. Many MHB species have been identified, and it is predicted that more are likely to be discovered [44,67]. Finding such beneficial combinations would be promising for sustainable and cost-effective urban greening, and potential combinations should be extensively tested for their reliability and consistency under field conditions before wider exploitation. The exact processes and compounds of microbes that are involved in stimulation of AMF colonization by MHB remain mostly unknown [68,69]. It does not necessarily require physical attachment but could happen through chemical signaling [63]. One possible mechanism behind these effects could be the release of gaseous volatiles, such as 2-methyl isoburneol, geosmin, and CO 2 , by the MHB that stimulates the growth of AMF. In addition to gaseous chemicals, other active metabolites produced by a MHB include vitamins, amino acids, and growth substances, any or all of which might directly stimulate the growth of an AMF [63,70].
The present study indicated that co-inoculation with R. irregularis and B. amyloliquefaciens led to increased plant biomass of at least T. repens and F. vesca in repeated experiments. A similar synergistic effect has been reported in a study that combined four Glomus species, namely G. aggregatum (Schenck and Smith), G. fasciculatum (Gerd & Trappe), G. intraradices, and G. mosseae (Nicolson & Gerd), with B. subtilis on Pelargonium graveolens (L'Her. ex Aiton). All four combinations (each of the four Glomus species inoculated separately with B. subtilis) resulted in synergistic effects to various degrees, as measured by increases in the plant dry weight of 25.7% to 74.3% as compared with that of control plants and of 7.8% to 14.5% as compared with inoculation with a single Glomus species [71].
We found that AMF colonization at a low level may not have any promoting effects on host plants. It is especially true if the internal hyphae are the only AMF structure present in the roots. However, little scientific data are available concerning the threshold level of arbuscules and vesicles needed for a significant promoting effect, and further studies on this topic are needed. C. rotundifolia hardly differed in biomass production between the two treatments R and R+B. A possible explanation could be that C. rotundifolia is highly tolerant to drought, soil pH, environmental variations, frost, and strong wind, so that it does not need symbiosis for further growth-promoting effects [72]. In contrast, Nuortila et al. [73] reported that, instead of increasing plant biomass, AMF colonization can significantly decrease the plant biomass of C. rotundifolia by 33% and can reduce flowers by 66%. Furthermore, in some specific AMF−host plant combinations, AMF can act as a "hitchhiker" that profits from the mycorrhizal-host network without returning benefits back to the host [74,75]. In the present study, T. serpyllum reached high levels of AMF colonization after the R+B treatment in the second experiment, but no significant difference in biomass production between R+B and B plants was observed. Similar results have been obtained using Thymus vulgaris (L.) as a host plant and G. mosseae as the AMF inoculant: a higher AMF colonization rate after AMF treatment did not increase plant biomass [76]. The phenomenon depends on the plant-AMF combinations, water availability, and soil types [77].
Enhanced photosynthetic efficiency is another benefit that AMF and plant growth−promoting rhizobacteria (PGPR) provide to some host plants, such as T. repens and F. vesca in the present study. A single inoculation with AMF or PGPR may significantly promote photosynthetic efficiency under suboptimal conditions such as high salinity and nutrient deficiency in the soil. For example, co-colonization by Glomus and Bacillus species enhances the photosynthetic efficiency of lettuce plants, compared with colonization by either of the two microbes. The extent of the synergistic effect is greatly dependent on the Glomus and Bacillus species combination [70].
Quantification of B. amyloliquefaciens revealed varying population densities between replicates in our study. Trevors et al. [78] also found large variation between replicates in their study of PGPR in the soil. It appears challenging to correlate PGPR population density with effects that promote plant growth [79], but the information would be useful, as beyond a certain PGPR population density no further enhancement may occur [79][80][81][82]. This scenario also seems possible in our study, as higher population densities of B. amyloliquefaciens did not lead to a higher biomass production or photosynthetic efficiency.
Our results suggest that neither a positive nor negative interaction was found between AMF colonization rate and the abundance of B. amyloliquefaciens in the soil in any of the eight tested plant species. These findings are supported by Kostoula et al. [83], who got similar results with Crithmum maritimum (L.) as a host plant and G. intraradices and B. amyloliquefaciens as inoculants. They found no significant effect of G. intraradices colonization on the population density of B. amyloliquefaciens in the soil. Likewise, Alam et al. [71] found no effect of four Glomus species on B. subtilis population density.
Interruptions in the NaPPI system during the first experiment made growth conditions unstable, which probably explains why the shoot weight and photosynthetic efficiency in half of the plant species were not reproducible across the two NaPPI experiments. Still, the other four plant species exhibited relatively consistent patterns in shoot weight and photosynthetic efficiency, suggesting that the plant species studied show different sensitivities to the stability of their growth environment.
In conclusion, for host plants T. repens and F. vesca, B. amyloliquefaciens could act as a MHB to facilitate colonization by R. irregularis, and dual inoculation with B. amyloliquefaciens and R. irregularis should result in higher shoot weight and photosynthetic efficiency. Such mycorrhiza-helping effects and dual-inoculation growth-promoting effects in other test plants were more conditional and dependent on the environment. Further field experiments to test these effects of co-infection on green roofs should focus on T. repens and F. vesca, as they showed consistent and promising results, and they are likely to produce better growth than other plants under such conditions. Supporting information S1 Table. Effect of microbial colonization and host species on shoot dry weight of four plant species whose results were consistent in the two NaPPI experiments.