For the design of specific tracing tools, the following approach was used. First, the diversity of indigenous Pseudomonas rhizoplane isolates was determined by Matrix-Assisted Laser Desorption Ionization (MALDI) Time of Flight (TOF) mass spectrometry (MS) (Mulet et al., 2012). Then, representative isolates were sequenced and primer/probes specific for DSMZ 13134 were designed and cross-tested while an existing qPCR assay for Pf153 was further elaborated. This resulted in two qPCR assays enabling the specific tracing of the two strains in root and rhizosphere extracts of inoculated maize plants in multiplex assays. The details are given below.

The pot experiment compared the persistence of the two Pseudomonas strains in different soils and assessed their effect on maize performance. For this, maize of the organic variety Colisee (KWS Saat, Germany) was inoculated with P. fluorescens Pf153, and Pseudomonas sp. DSMZ 13134 (Table 1) and grown in the soils DOK-M, Buus and Le Caron (Table 3) together with a non-inoculated control. Each treatment was replicated eight times. Half of the replicates were harvested after 4 weeks and the second half after 8 weeks. The effect of seed inoculations on maize biomass, height, and nutrient uptake (N and P), as well as their influence on root colonization by native AMF was assessed. The soils DOK-M, Buus and Le Caron were chosen due to their contrasting parameters, such as pH, phosphorus content and management. The DOK-M soil belongs to the 36-year-old “DOK” system comparison trial (Mäder et al., 2000, 2002). This field was conventionally managed with only the addition of mineral fertilizers. The Buus soil originates from a field that has been organically managed (IFOAM, 2006; Demeter International e.V., 2012) for the past 30 years, while the Le Caron soil is derived from a 5-year-long grass/clover ley that was converted to a conventionally managed, no-tillage field in 2011.

Residuals of the data were tested for normality using the Shapiro-Wilk test and the following transformations were applied accordingly: arcsin for root length colonization, log10 for total Pseudomonas CFU and (log10)⁻¹ for CFU of inoculated Pseudomonas sp. DSMZ 13134 and P. fluorescens Pf153. Two-way analyses of variance (ANOVA) were conducted with bacterial strain, soil, and their interaction as factors. If the interaction was significant, interpretations were made by testing the effect of each strain separately. In case of significant model effects, pair-wise mean comparisons using Tukey's HSD post-hoc test between the soil treatments were conducted. Any difference mentioned was found to be significant at p < 0.05. Statistical analysis was performed with RStudio Inc. (Version 0.97.551, © 2009–2012).

A 827-bp-long dnaX fragment from Pseudomonas fluorescens Pf153, Pseudomonas sp. DSMZ 13134, Pseudomonas jessenii RU47 and from the rhizoplane isolates Pseudomonas sp. 4_12, 4_25, 4_35, 11_3, 11_8, 11_11, and 11_14 was submitted to GenBank under accession numbers KP247587 to KP247596.

The newly developed qPCR Taqman assays, based on a SCAR fragment for strain Pf153 and the dnaX gene for strain DSMZ 13134, allow a quantitative tracing of the two strains in the root and rhizoplane of maize.

Complete specificity of the tracing tool for Pf153 was confirmed (Table 4). As expected from the MALDI-TOF cluster diagram (Figure 1) and the in silico sequence comparison of the dnaX gene, the tracing tool for strain DSMZ 13134 showed a strong cross-reactivity with Pseudomonas protegens CHA0, Pseudomonas protegens Pf-5 and Pseudomonas chlororaphis LMG 1245. More than 10,000 times weaker amplification was observed with the isolates 18_2, 18_15 and 18_26 from the Buus soil. Nevertheless, no amplification was found to occur in the environmental samples of non-inoculated plants grown in the three soils used for the experiment as well as in seven additional soils (Table 4). All samples showed specific amplification of the internal standard ACMV with comparable quantification cycle values, which indicated that the DNA losses and/or the actions of PCR inhibitors were similar for all the DNA templates.

The sensitivity of the two TaqMan assays was measured with 10-fold dilution series of the plasmid constructs holding the respective target sequences, as well as with bacterial cells (DSMZ 13134 and Pf153) that were serially diluted and spiked into a non-inoculated rhizoplane matrix. The detection limit for target DNA contained in the plasmid constructs was as few as 4 copies and 23 copies per μl reaction for Pf153 and DSMZ 13134, respectively. When bacterial cells were spiked into a rhizoplane matrix, the detection limits were 1.4 × 10² CFU and 1.3 × 10⁴ CFU per gram root fresh weight for strain Pf153 and strain DSMZ 13134, respectively.

The tracing tools for strains Pf153 and DSMZ 13134 were used to measure the bacteria's persistence in three contrasting soils and in two different compartments, the root and the rhizoplane (Figure 3).

Both strains were able to colonize the root and the rhizoplane of 4-week-old maize plants, but only strain Pf153 was still detectable after 8 weeks of growth (data not shown for persistence after 8 weeks). The persistence of strain Pf153 was influenced by the soil (p < 0.05; Figure 3) but not by the compartment or by the interaction between soil and compartment (p > 0.1 and p > 0.05, respectively). Population size of Pf153 in the root and the rhizoplane together was the highest for plants grown in the organic soil Buus (3.6 × 10⁶ CFU), followed by DOK-M (1.1 × 10⁶ CFU) and Le Caron (3.7 × 10⁵ CFU–in each case per gram root fresh weight). In contrast to strain Pf153, persistence of strain DSMZ 13134 was influenced by the compartment (p < 0.001) but neither by the soil and nor by the interaction of the two (p > 0.1). Colonization across all soils after 4 weeks was about 10 times higher in the root compared to the rhizoplane for strain DSMZ 13134.

After 8 weeks, population size of strain Pf153 was reduced about 3-fold for soils DOK-M and Le Caron and about 7-fold for soil Buus compared to week 4 (data not shown for week 8). As after 4 weeks, the persistence was higher for plants grown in soil Buus (4.8 × 10⁵ CFU), followed by DOK-M (3.0 × 10⁵ CFU) and Le Caron (1.2 × 10⁵ CFU–in each case per gram root fresh weight).

The response of maize upon the successful colonization up to week 4 (strain DSMZ 13134) and week 8 (strain Pf153) was analyzed. Strain Pf153 did not influence positively shoot biomass at any time or in any of the soils compared to non-inoculated plants (Figure 4). On the contrary, a 23% reduction of root biomass in the soil Buus compared to the non-inoculated plants was measured (p < 0.05) and a significant decrease in shoot biomass compared to plants inoculated with DSMZ 13134. Interestingly, strain DSMZ 13134 increased shoot biomass in the acidic and P-poor soil Le Caron by 62% compared to non-inoculated plants (p < 0.01; Figure 4) after 8 weeks. In this treatment, plant height was significantly increased already after 2 weeks of plant growth with, at week 8, a 21% increase in height compared to the non-inoculated plants (Figure 5). A stimulating growth effect on height, but not on biomass, was measured after 2, 6, and 7 weeks in soil Buus (p < 0.01).

None of the strains improved shoot nitrogen or phosphorus concentration of plants grown in any of the three soils (Supplementary Material 1, Figure 1). Nevertheless, the shoot P content of plants inoculated with DSMZ 13134 in the Buus soil was higher than in the other treatments. Shoot P and N concentrations of plants grown in Le Caron and inoculated with DSMZ 13134, where plant biomass was increased, were significantly reduced (Figure 4 and Supplementary Material 1, Figure 1).

Colonization by native arbuscular mycorrhizal fungi was not influenced neither by strain Pf153, known for its production of antifungal compounds (Fuchs, 1993; Fuchs et al., 2000) nor by strain DSMZ 13134, a known mycorrhizal helper bacterium (Yusran et al., 2009) in any of the soils (p > 0.1; Table 5). On average, plants grown in the Buus soil showed distinctly higher colonization by arbuscular mycorrhizal fungi (70%) compared to plants grown in either DOK-M (44%) or Le Caron soil (49%) (p < 0.001).

The qPCR tools developed for this study have been successfully applied for quantitative tracing of the two bacterial strains in environmental samples. The pre-characterization of native Pseudomonas related species with a MALDI-TOF MS approach has been instrumental to determine against which strains our tracing tools had to be cross-checked for absence of amplification. It has also helped to determine which native and reference strains should be included in the alignment of the target gene sequence for the design of strain specific markers. Von Felten et al. (2010) who established a SYBR Green qPCR assay on the same SCAR fragment in order to quantify Pf153 in the rhizosphere of maize, found a detection limit of 4 × 10³ CFU per g rhizosphere. While the detection limit for that same strain with our method is only slightly improved (1.4 × 10² CFU per g root fresh weight), it allows a faster sample throughput with no need to assess dissociation temperature curves of the PCR products and it can be used in multiplex assays. The newly developed tracing tool for strain DSMZ 13134 is less sensitive compared to the tool developed for Pf153, having a 100 times higher detection limit. Still, the detection limit is comparable with other studies using qPCR tracing tools for environmental root and soil samples (Ibekwe and Grieve, 2003; Höppener-Ogawa et al., 2007; Savazzini et al., 2008). However, since the tracing tool for Pseudomonas sp. DSMZ 13134 is not entirely strain-specific, it is essential to include non-inoculated control samples in any assay it is used. Whether there is a high or negligible probability of cross-reacting with indigenous P. protegens or P. chlororaphis type bacteria present in soils, remains to be elucidated although in our soils such bacteria were not present or found in less than 2% of the cases for the taxa Pseudomonas chlororaphis.

Introduced bio-inoculants need to colonize the roots in sufficient numbers in order to express their plant growth promoting potential (Bull et al., 1991; Chin-A-Woeng et al., 2000). The data provided by the newly developed tools give further insight into the complex influences of physical, chemical and biological soil properties. In general, there was a trend toward better persistence of the inoculated strains in the Buus soil, which was statistically significant for the strain Pf153. The soil Buus was collected from an organically managed field site. Its crop rotation comprises 50% grass-clover, which increases soil organic carbon and microbial biomass. Therefore, the conditions for resource acquisition were better in that soil even though the data on the abundance of the indigenous microbiota (as assessed by the number of 16S copies per gram of soil and the population size of the native Pseudomonas, see Table 3) indicate a higher potential for competition with native populations. The high level of root colonization by AMF in this soil, which can be explained by its organic management (Brundrett, 1991; Jansa et al., 2002; Oehl et al., 2004), could have further facilitated the colonization by the strain Pf153 (Mayo et al., 1986; Meyer and Linderman, 1986; Barea et al., 1998). Studies on the influence of soil parameters on PGPR persistence are relatively scarce. Nevertheless, a positive correlation between the percentages of clay, total nitrogen, organic matter and water holding capacity and the survival of Azospirillium brasilense was found in a large screening of 23 soils (Bashan et al., 1995). In the present study, the latter three parameters (data for water holding capacity and total nitrogen not shown) were higher in the soil Buus, compared to DOK-M and Le Caron. The higher abundance of DSMZ 13134 in the root compared to the rhizoplane compartment shows that this strain attaches very efficiently or can even enter root crevices. This is in line with a study that demonstrated its preferred colonization site on the root surface of barley, and similarly to our results, colonization decreased after 3 weeks (Buddrus-Schiemann et al., 2010). On the other hand, Pf153 was also confirmed to be a good colonizer of maize (Von Felten et al., 2010). Its ability to produce cyanide could have conferred it a competitive advantage against the native microbiota (Fuchs, 1993; Compant et al., 2010). These results about persistence show that an active and rich soil like the Buus soil will not prevent root and rhizoplane colonization by PGPR. Nevertheless, more research is needed to define more precisely the set of soil conditions that enable a good persistence of introduced bio-inoculants.

The rationale to inoculate plants with Pf153 and DSMZ 13134 is to improve biomass production and, to a lesser extent, nutrient acquisition also under P limiting conditions. Surprisingly, under our experimental conditions, strain Pf153 was not able to promote plant growth at all and in addition its increased colonization ability in soil Buus after 4 and 8 weeks was even associated with reduced plant biomass. The production of cyanide by Pf153, as reported in the literature (Fuchs et al., 2000), may have negatively affected plant growth directly, as it has been shown in other studies (Alstrom and Burns, 1989; Kremer and Souissi, 2001), as well as reduced the growth of indigenous beneficial microorganisms living in the maize rhizosphere and thereby negatively affected plant growth indirectly (Zdor, 2015). According to Kremer and Li (2003), the ability of cyanide producing rhizobacteria to affect other soil microorganisms is also related to the soil management with higher cyanide activity in soil receiving organic input and submitted to reduced tillage. This is the reason why many studies are now concentrating on side effects of PGPR inoculation, looking at their impact on the soil and rhizospheric microbial communities (Martinez-Viveros et al., 2010; Vacheron et al., 2013; Omirou et al., 2016; Schlaeppi et al., 2016). In contrast to Pf153, strain DSMZ 13134 was shown to significantly improve maize biomass in one of the studied soils. As hypothesized, this plant growth-promoting effect was observed in a soil with low soil P level (Le Caron). The other two soils had either a higher level of directly plant available phosphorus (DOK-M) or a higher level of phosphorus contained in the microbial pool (Buus). In these two soils, the inoculation could not further improve plant performance. The effectiveness of DSMZ 13134 in a nutrient-poor soil compared to a soil supplied with very high amounts of fertilizers was also found to be higher in a greenhouse experiment with barley (Fröhlich et al., 2012). In another study, strain DSMZ 13134 was able to increase tomato biomass and P content in a soil supplied with relatively high amounts of P (156 kg P ha⁻¹) (Yusran et al., 2009). Nevertheless, the effectiveness of DSMZ 13134 in the Le Caron soil supports the trend that PGPR are more effective in unfertile, stressed soils or in soils with poor microbial biomass and activity (Strigul and Kravchenko, 2006; Fliessbach et al., 2009; Mäder et al., 2011). The mechanisms by which DSMZ 13134 improved maize biomass in the Le Caron soil remain to be elucidated. Indeed, in none of the three soils root colonization by AMF was improved even though DMSZ 13134 has been shown to be a mycorrhizal helper strain (Yusran et al., 2009). In other studies, that same strain has shown ability to dissolve insoluble tricalcium phosphate in in vitro conditions (Fröhlich et al., 2012). It is unlikely that this property is of any importance in the acidic soil of Le Caron (where phosphate is more bound to iron and aluminum Gyaneshwar et al., 1998, 2002; Jones, 1998). In Buus, a soil with a more alkaline pH and higher soil organic carbon and higher microbial P content compared to Le Caron, DSMZ 13134 improved the maize plant P acquisition. This might be attributed to P mineralization from the organic P pool (as shown in Fröhlich et al., 2012) and/or to mobilization of phosphate from insoluble mineral phosphate provided by the added rockphosphate. Our results indicate therefore that the outcome of bacterial inoculations is highly dependent on soil parameters. In our case, we saw a plant growth promoting effect of DSMZ 13134 in the Le Caron soil, which had a low pH and a low level of plant available phosphorus.

In conclusion, the beneficial effect strain DSMZ 13134 had on maize plants in the acidic and poor soil was not associated with a better survival or persistence, despite the fact that this persistence is likely a prerequisite for the realization of its plant growth promotion potential. However, it was shown that strain Pf153 had a negative effect on plant biomass in one of the three soils, although its improved ability to colonize the roots in that same soil. Hence, the importance of long term persistence of an introduced PGPR strain on its effects on plant growth might be overestimated. The tools developed in this study will be useful for future monitoring of the persistence and spread of these strains in pot and field experiments. Future efforts should be made to elucidate the required colonization levels as well as the traits and mechanisms that are important for rhizosphere colonization and plant growth promotion by specific PGPR in order to understand the conditions under which they can be successfully applied to the soil.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.