Introduction

Oilseed rape (OSR) is a major oil-crop in temperate regions of Europe1. Two of the most important diseases of this crop are Phoma stem canker disease (syn. blackleg disease) caused by the ascomycetous fungus Phoma lingam (asexual form; sexual forms Leptosphaeria maculans and L. biglobosa of the L. maculans complex) and Sclerotinia stem rot caused by the ascomycete Sclerotinia sclerotiorum. During the growth cycle of winter OSR, fungi of the L. maculans complex produce sexual ascospores on colonised stubble in autumn, which infect cotyledons and secondary leaves and cause lesions; from there the pathogen spreads systemically and enters the stem, causing the typical blackleg symptoms in summer2. Infection can also spread aerially via asexual pycnospores3. Phoma stem canker (blackleg disease) causes huge economic losses in OSR4. Resistance breeding is often specific and overcome by new virulent races5. The causal agent of Sclerotinia stem rot, S. sclerotiorum, persists in soil as black resting structures called sclerotia, that can either produce soilborne mycelium or airborne ascospores for infection of its host6; the wide host range hampers control by crop rotation and development of resistant cultivars is still in its infancy6. Thus, extensive application of fungicides is necessary in oilseed rape to control both diseases6,7. Oilseed rape does also have high nutrient demands and high doses of (synthetic) fertilisers are required8.

Due to the environmental and societal demands to reduce agrochemical inputs, plant beneficial microorganisms are intensively researched as an alternative to synthetic pesticides and fertilisers, also for brassicaceous crops such as oilseed rape9. Endophytes promote plant growth by various physiological mechanisms such as increasing nutrient availability for the plant via siderophore production, phosphorus (P) solubilisation and nitrogen (N) fixation, modulating plant growth and stress responses via phytohormone production, and controlling pathogens via production of antimicrobial compounds10. Endophytes that inhabit seeds and systemically colonise the growing plant from there lend themselves to an application as seed treatment11. Such treatments that remain effective during a longer period of the OSR growth cycle would remove the need for fine-tuned spray timing, which is a pivotal and critical issue in chemical control of Phoma stem canker7 and Sclerotinia stem rot6.

In OSR, endophyte communities of culturable bacteria12,13 and fungi14 have been characterised, with the application as plant growth promoters in contaminated soil13, and biocontrol agents12,14,15,16 in mind. While there is an extensive focus on the use of endophytes as biocontrol agents, the potential influence of plant pathogens on endophyte communities has been underexplored so far. Not only can endophytes influence the growth of plant pathogens, but also plant pathogens may change the environment for endophytes17 and thus have the potential to alter their community structure. Cultivar resistance against Verticillium wilt affected the community structure of bacterial endophytes in oilseed rape12. However, to our knowledge, no comparable studies exist for the two other major diseases in OSR, Sclerotinia and Phoma stem canker.

With the work presented here, we start to close this gap. The two major aims of our study were to assess the effect of Phoma stem canker infection on bacterial and fungal endophyte communities in oilseed rape and to explore their potential for plant growth promotion and biocontrol. Similar to the previous studies cited above, we focused on culturable endophytes for potential industrial production as beneficial inocula. Isolates were characterised in vitro for their plant growth promoting properties. A selection of bacterial and fungal isolates representing the major genera was tested in planta for plant growth promotion of OSR under nutrient limiting conditions and for control of the two major pathogens Phoma stem canker and Sclerotinia stem rot. We hypothesised that both bacterial and fungal communities in plants with symptoms of Phoma stem canker would differ from communities in healthy plants. We also expected that microbes from the endophyte community in symptomless plants in particular would provide protection to OSR plants from Phoma stem canker and/or Sclerotinia white rot.

Results

Bacterial endophyte communities

Culturable bacterial endophyte communities of healthy and diseased OSR plants were distinct at the species level (Venn diagram in Fig. 1) but relatively similar at the genus level (Fig. 2A,B). Out of several diversity indices calculated in the study Shannon–Wiener, Gini–Simpson and Pielou evenness indices showed no difference in richness and evenness of bacterial communities isolated from healthy and diseased plants, but Margalef diversity index and Total taxonomic distinctness clearly indicated a higher diversity and total taxonomic breadth of an assemblage in diseased plants (Table 1). This probably arose from several genera (Achromobacter, Pectobacterium and Sphingobacterium) being isolated only from diseased plants, albeit in very low numbers. In both healthy and diseased plants, Gammaproteobacteria and Bacilli dominated, with the former contributing more than 85% of all isolates (Fig. 2a,b). With a share around 37%, Pseudomonas was the prevailing genus in both healthy and diseased OSR. The second largest group were Enterobacteriales represented by the genera Enterobacter and Serratia (Fig. 2b). Xanthomonadales with the genus Stenotrophomonas were also present in a greater proportion. Bacilli, which made up 9.5–15% of isolates, were largely represented by the genus Bacillus, and to a lesser extent by the genus Staphylococcus (Fig. 2a,b). While the majority of Pseudomonas isolates originated from roots, a larger proportion of the Serratia and Bacillus isolates were isolated from stems (Fig. 2a,b).

Figure 1
figure 1

General outline of the study and Venn diagram of the isolated culturable endophytes from symptomless oilseed rape (H, healthy) and oilseed rape showing symptoms of Phoma Stem Canker (D, diseased).

Full size image
Figure 2
figure 2

Community composition of the cultivable bacterial endophytic community associated with symptomless oilseed rape (A) and oilseed rape showing symptoms of Phoma Stem Canker (B). The total number of isolates in each group is shown in the central circle. The chart shows percentages by phyla, classes, orders, and genera progressively from the central to the outermost annuli. N.D. not determined. On the genus level, the abundances of isolates from roots and shoots are presented separately, with the additional letters “R” for roots, and “S” for shoots. Systematic classification according to49.

Full size image
Table 1 Diversity indices of culturable bacterial and fungal communities in healthy (symptomless) and diseased (with symptoms of Phoma stem canker) OSR assessed at the genus level. Total taxonomy distinctness was calculated according to Clarke and Warwick68.
Full size table

With MALDI-TOF scores mostly between 2.0 and 2.3, the genus ID of most isolates was certain, whereas the species ID was probable, especially in Pseudomonas isolates (Supplementary Table S1). The largest proportion of pseudomonads (around 85% in diseased plants and 50% in healthy plants) belonged to the P. fluorescens clade (P. antarctica, P. extremorientalis, P. fluorescens, P. grimontii, P. libanensis, P. rhodesiae, P. synxantha, P. tolaasi, P. veronii) within the wider P. fluorescens complex18. Among members of the P. fluorescens clade, P. tolaasi was prevalent in the roots of healthy plants, whereas the other species prevailed in diseased plants (Supplementary Table S1). In healthy plants, the P. corrugata clade (P. brassicacearum, P. kilonensis, P. thivervalensis) contributed 50% of Pseudomonas isolates. Single isolates of the P. koreenenis clade, the P. pyrolytica clade and of P. viridiflava were isolated from diseased plants (Supplementary Table S1). Remarkably, P. viridiflava was only isolated from shoots. Nearly all Bacillus isolates were identified as B. cereus. Only one Bacillus isolate from the roots of diseased plants was identified as B. mycoides, another one from diseased shoots as B. subtilis. Within other genera, 100% of respective isolates were identified as Enterobacter amnigenus, Erwinia percinia, and Serratia proteamaculans, respectively (Supplementary Table S1).

Fungal endophyte communities

Diversity indices of culturable fungal communities were similar in healthy and diseased plants (i.e. showing comparable genera richness and evenness) except for the Total taxonomic distinctness index which clearly indicated a broader taxonomic span of endophytic fungal genera in Phoma symptomatic plants (Table 1). Most genera had distinct morphotypes and only few (Cladosporium, Leptosphaeria, Plectosphaerella) were morphologically variable (Supplementary MS Excel file 1). Nearly all fungal isolates had 99–100% sequence similarity to the closest identified relatives, allowing their taxonomic resolution to the species level (Genbank accession numbers are listed in the Supplementary MS Excel file 1). Contrary to bacterial endophyte communities, fungal endophyte communities were profoundly and qualitatively different in healthy plants and plants affected by Phoma stem canker, and taxonomic groups detected in healthy and diseased OSR were mostly represented by different genera (Fig. 3). Only Ascomycetes were recovered from healthy plants. In the fungal community isolated from diseased plants, this phylum largely prevailed as well, but 17% of isolates were Basidiomycetes, represented by the genera Vishniacozyma and Holtermaniella (formerly Cryptococcus) isolated from shoots, and Bjerkandera/Thanatephorus (anamorph of the latter: Rhizoctonia) isolated from roots (Fig. 3a). Among the Ascomycetes, Dothidiomycetes were the prevailing group in both diseased and healthy plants; unexpectedly, fungi with 99–100% sequence homology to one causal agent of Phoma stem canker (Leptosphaeria biglobosa, L. maculans species complex) could be isolated in large numbers from the shoots of both diseased and healthy plants (20% and 34.5% of isolates, respectively). The only other genus that could be found in both healthy and diseased plants was Plectosphaerella (Sordariomycetes, Glomerellales). Isolates of this genus were morphologically diverse (see Supplementary MS Excel File 1). While the large majority of closely related sequences in GenBank is classified as Plectosphaerella, identification by intergenic spacer sequencing does not allow a clear differentiation from Monographella (Sordariomycetes, Xylariales, 99–100% sequence homology to members of both genera). All other fungal genera were exclusively isolated from either healthy or diseased plants. In healthy plants, Cladosporium (Capnodiales) was a dominating Dothideomycete (29% of fungal isolations) that was isolated from shoots as well as from roots (Fig. 3a). Isolates of this genus also exhibited significant morphological diversity (Supplementary MS Excel file 1). Epicoccum was also isolated in relatively large numbers in healthy plants; other species of Dothideomycetes detected were Leptosphaerulina, Torula and isolates related to Periconiella (94% sequence homology). In diseased plants, isolated Dothideomycetes belonged to the genera Alternaria, Ramularia and Peltaster (Fig. 3b). Among other Ascomycete classes, Botrytis (Helotiales) was only found in healthy plants whereas Fusarium (Sordariomycetes) was only isolated from diseased plants. While they cannot be distinguished from Sclerotinia by their intergenic spacer region sequence, all Botrytis isolates could be identified morphologically by their typical sporulation.

Figure 3
figure 3

Community composition of the cultivable fungal endophytic community associated with healthy oilseed rape (a) and oilseed rape showing symptoms of Phoma Stem Canker (b). The total number of isolates in each group is shown in the central circle. The chart shows percentages by phyla, classes, orders, and genera progressively from the central to the outermost annuli. On the genus level, the abundances of isolates from roots and shoots are presented separately, with the additional letters “R” for roots and “S” for shoots. Systematic classification according NCBI Genebank Database.

Full size image

Plant growth promoting traits of bacterial and fungal endophytes determined in vitro

Siderophore production and the ability to solubilise P was particularly widespread among Serratia isolates and Pseudomonas, especially those of the P. fluorescens-clade, but less frequent among isolates of the genera Stenotrophomonas, Enterobacter/Erwinia and Bacillus (Supplementary Table S1). All tested bacteria and fungi had antioxidative activity (Supplementary Tables S1 and S2). Phytohormone production was also common among bacterial and fungal endophytes. Production of the auxin IAA was generally higher than production of cytokinins (iP, IPR) in both bacteria and fungi. IAA-production of fungi was generally at least one order of magnitude higher than that of bacteria. Bacteria mostly produced iP but no iPR (Supplementary Table S1), whereas iPR was prevailing in fungi among the cytokinin compounds measured (Supplementary Table S2). High cytokinin production was particularly found among isolates of the genera Stenotrophomonas, Bacillus and Cladosporium. Only two of all tested fungal isolates (Penicillium chrysogenum RHR13 and Epicoccum sp. RHR15) produced very high levels of the mycotoxins roquefortin C and meleagrin. While P. chrysogenum is well known to produce both of these mycotoxins69, the production of either roquefortine C or meleagrin by Epicoccum sp. has yet not been reported to our knowledge.

Neither PCA analysis (Fig. 4) nor CCA cluster analysis (Supplementary Fig. S2) based on PGP trait profiles separated genera of endophytic bacteria (Fig. 4A, Fig. S2A) or fungi (Fig. 4B, Fig. S2B) into distinctive clusters. The origin of isolates (shoot vs. root, diseased vs. healthy) had no influence on PGP trait profiles of endophytic bacteria and fungi at all (Fig. 4, Supplementary Fig. S2).

Figure 4
figure 4

Ordination PCA diagram of bacterial (A) and fungal (B) isolates. First two axes explained 66% (A) and 79% (B) variation. Colours indicate taxonomic grouping (See Table S1 and S2). The two characters behind the underscore (bacteria) or initial letter “R” (fungi) in the isolates code are related to the origin of the isolate: R, root; S, shoot/stem; H, healthy = without Phoma stem canker symptoms; D, diseased = with symptoms of Phoma stem canker.

Full size image

Representatives of the bacterial genera Achromobacter, Enterobacter, Pseudomonas, Serratia and Stenotrophomonas, exhibiting all or most plant growth promoting traits (importance weighed in the following order: ACC-deaminase > siderophore production = P-solubilisation > IAA > iP = iPR > antioxidative activity AA = AOQ > NH3 production) and preferentially isolated from Phoma symptomless oilseed rape, were selected for pot experiments with OSR. For the selection of fungal isolates a similar approach was used: weighed importance of the traits: IAA > iP = iPR > antioxidative activity AA = AOQ, preferential origin from Phoma symptomless plants, ease of spore production based on our observation and literature records. Representatives of the genera Cladosporium, Epicoccum, Leptosphaeria, Penicillium, Periconiella, Plectosphaerella, Ramularia, Torula and few isolates from Tremellales (related to Cryptococcus) were selected for pot experiments with OSR (in the case of Sclerotinia rot test only Cladosporium isolates were tested).

Growth promotion of oilseed rape in pot experiments

Although the large majority of selected bacterial strains solubilised P in vitro, none of them promoted the growth of oilseed rape in a P-limited soil significantly (Supplementary Table S3). Red leaf discolouration in all treatments after 102 days of growth except for the control Cp with normal P-fertilisation and a significant increase of growth and pod yield in this control indicated that insufficient P-supply was limiting the growth. Plants were fertilised after 123 days of growth in order to alleviate P-limitation and to detect growth promotion caused by mechanisms other than P-solubilisation, but no significant effects could be found at the end of the experiment. Only tendencies to increase plant dry weight were observed in plants treated with Achromobacter piechaudii AP_RD9, Pseudomonas chloraphis/veronii PCh_RH1, Stenotrophomonas sp. SS_RD24, Enterobacter amnigenus EA_RH18 and Serratia proteamaculans SP_RH21. Also, cumulative height tended to be increased in plants treated with Pseudomonas sp. PG_RD39 (P. fluorescens clade), and E. amnigenus EA_RH18 (Supplementary Table S3). Isolates A. piechaudii AP_RD9, P. chloraphis/veronii PCh_RH1, Pseudomonas sp. PG_RD39 and Stenotrophomonas sp. SS_RD24 were selected to be tested as a growth promoting combined inoculum (IG) in a field experiment (see sub-chapter below); E. amnigenus EA_RH18 was not included as it showed a tendency to increase Phoma severity (Fig. 5). The strains finally included the combined growth promoting inoculum IG did not show any mutual antagonism in vitro (Fig. S3).

Figure 5
figure 5

Effect of selected bacterial endophytes on Phoma disease (Leptosphaeria maculans, anamorph Phoma lingam) in a cotyledon assay on oilseed rape seedlings. Data are presented as means ± SE; AP = Achromobacter piechaudii, PS =  Pseudomonas sp., EA = Enterobacter amnigenus, SP = Serratia proteamaculans, SS = Stenotrophomonas sp.,  C = negative (healthy) control without Phoma inoculation and without bacterial inoculation, C + = positive (diseased) control with Phoma inoculation and without bacterial inoculation, CF = fungicide treated control, Polyversum = commercial preparation of Pythium oligandrum. There was a significant treatment effect on Phoma severity in the directly treated cotyledon (leaf A, P = 0.0003) and on Phoma symptoms after seed treatment only (leaf B, P < 0.0005). Treatments with the same letters are not significantly different according to Unequal N HSD test. Where Dunnettʼs test showed a significant treatment effect compared to the control C + the significance of this treatment effect is shown as asterisks in superscript.

Full size image

None of the tested fungal endophytes increased growth and pod yield of OSR (Supplementary Table S4).

Control of Phoma lingam in planta (cotyledon assay)

In the cotyledon assay, some plants did not emerge or showed non-specific stress symptoms (yellowing and wilting of leaves) and thus could not be included in the analysis. Due to this reduction in the number of replicates (minimum N = 3 as compared to N ≥ 5 in other experiments) and due to a large variation in lesion sizes, only trends were visible in the cotyledon assay. None of the tested bacteria was comparable to the fungicide treatment (Fig. 5). Lesion size tended even to be increased after treatment with some endophytic isolates, most notably with E. amnigenus EA_RH18 or S. proteamaculans SP_SH2. For S. proteamaculans SP_SH2 and some other isolates, this trend was apparent not only in the directly sprayed leaf but also in the non-treated second cotyledon that could have been influenced only by a systemic effect of the seed treatment. Seedlings treated with Pseudomonas sp. PA_RD1A (P. fluorescens clade) had the lowest lesion score (3.4 vs 5.7 of the non-treated control; Table 1) but only when the leaves were directly sprayed. Generally, there were no positive systemic effects of endophytes when these were applied as seed treatment (Fig. 5).

Control of Sclerotinia sclerotiorum in a pot experiment

Amended to soil, S. sclerotiorum caused damping off in OSR seedlings. First symptoms were a girdled and discoloured hypocotyl, followed by wilting and finally death of the seedling; symptom development was often accompanied by visible outgrowth of typical white mycelium. The isolates Pseudomonas sp. PA_RD1A (P. fluorescens clade) and Serratia. proteamaculans SP_RH21 significantly reduced incidence (P = 0.001 and P = 0.005, respectively; Fig. 6) after 27 days of growth. A. piechaudi AP_RD9 and Stenotrophomonas sp. SS_RD24 showed a weaker but still significant effect (P = 0.014 and P = 0.028, respectively). Only S. proteamaculans SP_RH21 was able to control Sclerotinia significantly (P = 0.002) over a longer period of 68 days (Fig. 6). None of the tested Cladosporium isolates had any significant control effect, whether applied as spores (RHR7), or as mycelial inoculum. Pseudomonas sp. PA_RD1A had a negative effect on OSR seedlings emergence even before visible symptoms of Sclerotinia infection became apparent (14 d growth; P = 0,009; Table S5). The most effective isolate S. proteamaculans SP_RH21 was selected to be tested as a protective inoculum (IP) in a field experiment.

Figure 6
figure 6

Effect of seed treatment with endophytes on incidence of Sclerotinia disease in oilseed rape seedlings. Data are presented as means ± SE; AP, Achromobacter piechaudii; Ps.,  Pseudomonas sp.; EA, Enterobacter amnigenus; SP, Serratia proteamaculans; SS, Stenotrophomonas sp.; Cs_HR7s, Cladosporium sp. HR7 applied as spores; Cs_HR7, Cladosporium sp. HR7 applied as mycelium; bC−,  negative (healthy) control without Sclerotinia inoculation and without bacterial inoculation; bC+, positive (pathogen-inoculated) control in Sclerotinia infested soil without bacterial inoculation; fC−, fungal negative (healthy) control without Sclerotinia inoculation and fC+, fungal positive control (with pathogen). bC− and bC+ are the corresponding controls for all bacterial treatments and Cs-HR7s, while fC− and fC+ are the corresponding controls for all fungal endophyte treatments applied as mycelium (Cs_). There was a significant treatment effect on incidence of Sclerotinia after 27 days and 68 days growth (P < 0.0001). Treatments with the same letters are not significantly different according to Unequal N HSD test (inoculation with cell or spore suspension) or according to Kruskal–Wallis Test. Where Dunnettʼs test showed a significant treatment effect compared to the control C + the significance of this treatment effect is shown as asterisks (* P < 0.05, ** P < 0.01, ***P < 0.001) in superscript.

Full size image

Effect of seed treatments with bacterial endophytes on OSR yield, growth and disease incidence under commercial field conditions (Poříčí, Czech Republic)

Except for a highly significant (P = 0.00002 and P = 0.00033, respectively) reduction of stress symptoms (red discolouration of leaf margins) by both IP and IG inocula in early spring (March 2017; Table 2) none of the seed treatments with bacterial endophytes affected growth, yield and diseases of OSR significantly under commercial field conditions. Yet, some trends were observable. Despite its efficacy against Sclerotinia in the pot experiment S. proteamaculans SP_RH21 (treatment IP) failed to control the disease in the field (Table 2). Albeit not statistically significant, there was a trend towards increased yield (+ 15%) and reduced severity of Phoma stem canker (-42%) after seed treatment with the IG consortium consisting of A. piechaudii AP_RD9, Pseudomonas sp. PCh_RH1, and PG_RD39, and Stenotrophomonas sp. SS_RD24. Fungicide treatment significantly lowered disease incidence of Sclerotinia stem rot and Alternaria on stems (Table 2).

Table 2 Effect of bacterial endophytes applied as a seed coat on winter oilseed rape yield, Disease incidence of Phoma stem canker (Phoma lingam), Sclerotinia stem rot (Sclerotinia sclerotiorum), and Alternaria stem infection and on oilseed rape overwintering under commercial field conditions (Poříčí, Czech Republic).
Full size table

Discussion

In most studies on microbial communities that rely on cultivation12,14 the number of isolated strains is limited due to the time effort needed for isolation and isolate characterisation. The small sample size of 48/65 bacterial isolates and 35/29 fungal isolates from healthy/diseased plants, respectively, in this study means that the relative abundance of bacterial species needs to exceed 5–6% and the relative abundance of fungal species needs to be greater than 8–10% to be detected with 95% probability19. Thus, only dominant genera were securely detected in our study. Small sample size means a further risk that observed differences, especially quantitative shifts, might be based on stochastic events13. This was especially apparent for bacteria, where the considerable difference in Margalef index and Total taxonomic distinctness may be due to genera detected in very low numbers (Achromobacter, Pectobacter and Sphingobacterium). In addition, differences in bacterial communities manifested themselves mainly at the species level, where identification via MALDI-TOP was uncertain but only probable for most isolates. On the contrary, fungal communities in healthy and diseases plants differed not only quantitatively but also qualitatively in their dominant groups. Such profound differences, as also observed in poplar under contrasting environmental conditions20, appear unlikely to be due to stochastic variation alone, but indicate a strong effect of plant disease on endophyte communities. Though leaves were sampled in one of the symptomless plants next to stems while only stems were sampled in diseased plants it is improbable that this difference had a significant influence on the composition of microbial communities in this study. Neither bacterial nor fungal communities isolated from the aboveground parts of healthy plants were richer in isolated taxa compared to diseased plants as would be expected from the inclusion of another organ (leaves). Furthermore, the differences in fungal genera between diseased and healthy plants were also profound in the roots and not only in the aboveground plant parts.

Unexpectedly, isolates with 99–100% sequence homology to the plant pathogen Leptoshaeria biglobosa could be isolated in high frequency from both apparently healthy plants as well as plants with blackleg symptoms. As a causal agent of Phoma stem canker, L. biglobosa has been found to prevail over L. maculans in Eastern European countries21 and causes major damage in this region, especially in later growth stages during summer (also the time of our sampling)22. Previously it has been considered less aggressive than L. maculans, albeit also dependent on external conditions23. Its isolation from apparently healthy OSR plants is an indication that it also can be present asymptotically in OSR. Some isolates of the species may even promote OSR growth14. Moreover, strong interspecies sequence homology makes secure identification of the species not possible based on intergenic spacer sequences alone. There is also strong possibility that several Leptosphaeria species were present in our samples as Leptoshaeria isolates recovered in this study showed considerable morphological diversity when grown on MEA Agar (Supplementary Excel file 1). Interestingly, two other sequences with 99–100% sequence homology to our isolates were described as stemming from Pithomyces, a Phoma isolate causing leaf spot in Japanese horseradish (KR998327.1 and KR998328.1, unpublished).

The other fungal group that could be isolated from both asymptotic and symptom-bearing plants plants in this study, Plectosphaerella/Monographella, was previously isolated from the rhizosphere of OSR16,24. While a majority of closest related sequences in Genbank were characterised as Plectosphaerella, a large number is described as Monographella25. Although these two genera belong to different orders of Ascomycetes (Glomerellales vs. Xylariales), intergenic spacer and ribosomal sequences show nearly 100% homology, indicating that their taxonomy still needs to be resolved.

Of the fungal isolates exclusively originating from either plants with blackleg symptoms or symptomless plants, a larger proportion were (potentially) plant pathogenic species, i.e. Botrytis, Fusarium and Alternaria. The observation that Botrytis (as well as Cladosporium and Epicoccum) were isolated in larger numbers, but only from symptomless plants indicates their potential suppression in diseased plants. Leptosphaeria maculans increases the level of some glucosinolates in OSR17. Even if this does not affect the pathogen itself, it may inhibit other fungi. Other examples of mutual suppression have been found among the endophytic fungal community of OSR where, among other (potential) plant pathogens, L. biglobosa inhibited S. sclerotiorum, an ascomycete closely related to Botrytis cinerea 14. Conversely, Alternaria and Fusarium, (as well as Basidiomycetes of the former Cryptococcus group), which were only present in symptom-bearing plants here, might have profited from a change in the micro-environment, such as the release of nutrients in necrotic tissue. Alternaria, Epicoccum, Fusarium and Penicillium were also isolated from the endosphere of OSR in China whereas other less frequent species were not14. On the other hand, we could not detect Chaetomium, Clonostachys and Periconia which were dominant in their study14.

Studies on the culturable endophytic bacterial community of OSR12,13,26, including this study, also differed in their outcome, with different genera being predominantly isolated. These differences in outcome may be due to different growth stages (young plants vs. flowering vs. post-flowering stage), cultivar choice and environmental conditions (geographical region) and due to differences in sampling size. Not only diseases, as observed here, but also disease resistance (against Verticillium wilt) changed bacterial endophyte communities in OSR12. We found the endophyte community of OSR being largely comprised of the genera Pseudomonas, Stenotrophomonas, Enterobacteriales (Serratia and Enterobacter) and Bacillus which were also frequently isolated as endophytes of Brassicaceae in numerous other studies9. This spectrum of bacterial genera is remarkably similar to that of the bacterial community antagonistic to the plant pathogenic fungus Verticillium dahliae in the OSR rhizosphere15,27, with only small differences. Such similarity of the community of antifungal strains in the rhizosphere and the endophytic flora of OSR might suggest that these genera colonise both compartments and comprise a large proportion of the culturable endophytic community in OSR, but that their antifungal traits are not necessarily expressed in planta, as they were equally abundant in healthy plants and in plants displaying blackleg symptoms. Nevertheless, it has to be borne in mind that cultivation-based methods of community profiling favour these three groups over others and cultivation-independent approaches reveal much more diverse endophytic community28. This was also the case in OSR, where cultivation independent methods unveiled a much wider range of systematic groups among root-associated bacteria29,30 than studies confined to culturable bacteria but other factors come into play here as well, as these authors did not differentiate between the rhizoplane and the endosphere.

Strong P solubilisation and other plant growth promoting traits in vitro did not translate into plant growth promotion in planta under P-limiting conditions. Especially the assay with tricalcium phosphate (Pikovskaya’s agar31) employed here tends to overestimate the P-solubilising capability of isolates, since most P-complexes in soil are harder to dissolve32,33. It is notable, however, that isolates with strong siderophore production and P-solubilising capability detected in vitro showed the strongest trend to increase growth in the pot experiment under P-limiting conditions. They also alleviated stress symptoms in the field and tended to increase yield in the field when applied as combined inoculum. Consortia of plant growth promoting microorganisms have been shown to be more effective than the single strains34 probably due to complementarity of their traits and their different ecological optima. This was confirmed for OSR in another study, where growth promotion by single strains of Pseudomonas, Serratia and Enterobacter isolates was only significant when the strains were applied together as one combined inoculum in the field35. In a similar way, our consortium consisting of Achromobacter, Pseudomonas and Serratia strains showed a tendency to decrease disease incidence of Phoma in the field while the single isolates were not effective against Phoma stem canker in the cotyledon assay.

Unexpectedly, isolates arising from symptomless plants were not more effective against fungal pathogens than bacteria isolated from plants with Phoma stem canker, and some of the most antagonistic bacteria against Phoma and Sclerotinia diseases in our in planta assays were isolated from diseased plants. Hardly any isolates effective against P. lingam could be detected in the cotyledon assay, possibly because the employed method of mycelial inoculation is more invasive and creates greater disease pressure than inoculation with pycnidiospores used by other authors23,65,66. We used in planta tests with seedlings rather than in vitro assays for the screening of antagonist activity as the latter are not always correlated with in planta performance36, and do not detect the mechanism of resistance induction, which is particularly common in the genera Pseudomonas 37 and Serratia38. Bacteria of both genera have been shown to be effective against Sclerotinia in OSR previously39,40. An apoplastic isolate of P. viridiflava effectively supressed growth of three OSR pathogens (L. maculans, S. sclerotiniorum and Xanthomonas campestris) in dual culture-assays70. The same isolate also restricted propagation of X. campestris and reduced necrotic lesions caused by S. sclerotiniorum in OSR leaves and promoted the growth of OSR. The protective effect of the endophyte was probably due to the induction of host resistance via salicylic acid and jasmonic acid signalling pathways and/or production of antimicrobial compounds70. Several other bacteria controlled S. sclerotiniorum with most of them belonging to the genera Bacillus and Pseudomonas. Protection mechanisms comprised increased levels of hydrolytic enzymes such as chitinase and β-1,3-glucanase, increased expression of pathogenesis-related proteins and production of antifungal compounds71,72,74. While effective plant growth promoters have been identified within the genus Achromobacter42,43,44, fewer reports exists on successful biocontrol using this genus45,46. In OSR, Achromobacter has been found to promote plant growth by improving nutrient uptake47. Next to the above mentioned bacteria Serratia plymuthica HRO-48 was successfully developed as seed treatment for OSR against Verticillium dahlia 11 and Phoma lingam41. To our knowledge, the species Serratia proteamaculans has not yet been reported as a biocontrol agent in OSR, but an isolate of this species controlled Alternaria in tomato38.

Our strain of S. proteamaculans that was most effective against Sclerotinia in the seedling assay failed in the field, possibly because the seedling assay does not mimic natural infection conditions in OSR closely, where the spread via ascospores germinating on senescent petals is the prevailing infection pathway6. Systemic colonisation and a long time of survival in the plant is needed if strains applied as seed treatments are to prevent infection via this pathway, where infection occurs in the flowering stage. Alternatively, foliar application is an efficient way of introducing bacteria providing biological control of fungal diseases in oilseed rape but in such a case exact timing of the application within the time window of petal infection by S. sclerotiniorum ascospores is crucial72. In addition, the imbalanced design of the fungicide application in our field experiment (two out of three control plots but only one out of three IG and IP plots were treated with fungicide) might have putatively lowered the difference in yield and disease incidence between the control and the bacterial treatments IP or IG.

In contrast to a large number of bacterial endophytes providing biological control against OSR diseases the number of efficient fungal strains mentioned in literature is limited to a few, corresponding also to the output of this study where several bacterial strains but no fungal strain exhibited fungal disease suppression. Among fungi, Gliocladium catenulatum and Acremonium alternatum protected OSR from clubroot caused by Plasmodiophora brassicae73,75 and Ulocladium atrum and Coniothyrium minitans reduced incidence and severity of Sclerotinia stem rot76. However, the profound differences in fungal communities in Phoma-symptomless and symptomatic plants found here raise hope that more fungal endophytes antagonistic to fungal diseases might be found.

In conclusion, we have shown considerable differences in fungal and, to a lesser extent, also in bacterial endophyte communities in P. lingam-symptomatic and symptomless OSR plants. While the scope of the study could provide only an introductory characterisation of their plant-growth promoting and biocontrol properties, it yielded bacterial strains that are worthy of further study (Achromobacter piechaudii AP_RD9, Pseudomonas chloraphis/veronii PCh_RH1, P. grimontii/fluorescens PG_RD39, and Stenotrophomonas sp. SS_RD24). In future experiments, it would be desirable to improve the formulations of the strains48, to optimise their application technique and to monitor their long-term survival in the plant.

Methods

Isolation of bacterial and fungal endophytes

Endophytes of OSR were isolated by a modified dilution-to-extinction method50. OSR plants (cv. ‘Viking’ and ’Catonic’) at growth stage of seed ripening (BBCH > 8151) were harvested in mid July 2013 from a research site of SELGEN a.s. near Chlumec n. Cidlinou, Czech Republic (N50:7.7552, E 15:29.3410, 218 m above sea level) at experimental plots not treated with pesticides. In the plots many plants had developed symptoms of Phoma stem canker, Verticillium wilt, Alternaria blight and powdery mildew (Erysiphe cruciferarum) while only few plants with Sclerotinia stem rot and light leaf spots caused by Cylindrosporium sp. were present. We harvested plants with no visible disease symptoms (except for slight presence of Alternaria sp. that was ubiquitous) and separately those with symptoms of Phoma stem canker. Three plants per variant (symptomless /healthy/, symptomatic /diseased/) were pooled and separated into roots and shoots (shoots consisted of stems only as leaves already deteriorated at this growth stage except of one plant from the symptomless variant where few leaves were present as well). Any adhering soil and debris was removed by washing with tap water. Subsequent isolation of endophytic bacteria and fungi from plant material (shoots and roots separately) was done as previously described20,52. Briefly, plant material was surface sterilised for 7 min (2.2% active chlorine with addition of 0.1% Tween 20), transferred to a 70% ethanol solution for 2 min, rinsed four times with sterile deionised water, and cut to small pieces (< 0.5 cm). After thorough mixing of the cut material under sterile conditions approximately 2 g (fresh weight) were homogenised (Ultra-Turrax IKA-T10, max speed, 1 min) in 20 mL of sterile 12.5 mM potassium phosphate buffer (pH 7.1). The homogenate was filtered (polyamide sieve Uhelon 13S, porosity 609 μm), diluted (5 to 50 times for fungi), and plated on 96-well plates with agar media (25 μl medium per well). Fungi were isolated on both synthetic nutrient poor agar (SNA) and nutrient rich malt extract agar (MEA: 2% malt extract, 0.2% meat peptone), each amended with the mixture of antibiotics (50 mg L−1 tetracycline, 80 mg L−1 streptomycin sulfate, and 60 mg L−1 penicillin) to suppress bacterial growth. Twenty-five microliters of the filtered homogenate was applied per well and six plates per given sample and growth medium were used (576 wells in total) to isolate the endophytic fungi. In the case of bacteria, the homogenate was diluted 10−1 to 10−4 and plated on Petri dishes (diameter 9 cm) with tryptic soy agar (5% TSA, 50% TSA) amended with cycloheximide (100 mg L−1). Bacteria were incubated at 20 °C for 48 h while fungi were grown in dark at 20 °C up to 4 weeks. Fungi displaying different morphologies were re-streaked on new plates to obtain clean axenic cultures. Isolated fungi were maintained on SNA and 1:10 strength MEA.

Identification of bacteria and fungi

Bacterial isolates were identified by MALDI-TOF MS53,54 using an Autoflex Speed MALDI-TOF/TOF mass spectrometer and MALDI Biotyper 3.1 software (Bruker Daltonik, Germany). Fungal isolates were morphotyped according to their colony appearance on malt extract agar (MEA; see Supplementary MS Excel file 1). DNA of at least two representatives of each morphotype was isolated55, and PCR-amplified for molecular identification based on their ribosomal DNA region56,57. Molecular identification and classification was based on the NCBI GenBank database58; a similarity of > 97% was used as a threshold for species identification. Higher taxonomic units were selected in the case of lower scoring. When the identity of the two selected representatives differed, all members of the morphogroup were molecularly identified. GenBank accession numbers of submitted fungal sequences are MN275840-MN275887.

Testing of in-vitro traits of bacteria and fungi

Phosphorus (P) solubilisation, siderophore excretion, ACC-deaminase, antioxidant activity (AA and AOQ = total antioxidative activity expressed in % of quenching activity) and NH3 production on peptone were detected as previously described31,59,60,61,62,77. For the measurement of phytohormones and antioxidants excreted into the culture medium, bacteria were grown in 20 ml LB in 50 ml Falcon tubes for 7 days at 28 °C; fungi were grown in 50 ml Falcon tubes in 15 ml malt extract broth (MEB; 2% malt extract; 0.2% meat peptone; pH 5.5) with 0.5% tryptophan for 10 days at room temperature on a shaker. Auxin indole-3-acetic acid (IAA), the cytokinins N6-(Δ2-isopentenyl) adenine (iP) and N6-(Δ2-isopentenyl) adenosine (iPR) in cultural filtrates were quantified by ultra-high performance liquid chromatography (Acquity UPLC HSS T3 column; 1.8 μm, 2.1 × 100 mm; Waters, Acquity, coupled with tandem mass spectrometry AB Sciex, Qtrap 5500; electrospray ionisation in positive mode). Blank sterile medium served as control63. The performance of the endophytic isolates in the in vitro tests was used for the selection step 2 (Fig. 1). The selection was based on following criteria: (1) Number of traits where the particular strain scored high; importance of the traits was weighed according the following order: ACC-deaminase > siderophore production = P-solubilization > IAA > iP = iPR > AA = AOQ > NH3 production for bacteria, IAA > iP = iPR > AA = AOQ for fungi, (2) Origin (we preferred isolates from symptomless plants over isolates from symptomatic plants), (3) Taxonomic placement of the isolates (we aimed at covering a wide taxonomic diversity in order to arrive to a broad group of strains putatively complementing each other in function and ecophysiological demands).

Mycotoxin assessment

Fungal isolates were grown on MEA enriched with 2% oilseed rape extract at 21 °C. Mycelial growth was quantitatively transferred into 250 ml Erlenmeyer flasks, mixed with 40 ml extraction solvent (acetonitrile : 1% formic acid in water, 50:50), homogenised using ultraturrax and shaken for 60 min. The supernatant was filtered into a 50 ml volumetric flask and filled up to the line with the extraction solvent (same as specified above). This extract was then analysed according the methodology previously published for quantitative determination of 57 mycotoxins (including Fusarium and Alternaria mycotoxins, ergot alkaloids, aflatoxins, fumonisins, and others)64. For separation and detection of the analytes, combination of ultra-high performance liquid chromatography (machine equipped with an Acquity UPLC HSS T3 column; 1.8 μm, 2.1 × 100 mm; Waters, Acquity), coupled with tandem mass spectrometry (AB Sciex, Qtrap 5500; electrospray ionisation in positive and negative mode) was used. For the quantification purposes, matrix-matched calibration standards of the mycotoxins (blanks, 0.05–1000 ng ml−1) were prepared using blank medium extracted by the procedure described for the samples.

In planta testing for plant growth promotion

Pots were filled with 1.2 L of autoclaved substrate (7:3 autoclaved zeolite:sand mixture v:v; bulk density 1.17 g ml−1) amended with 11 g L−1 apatite resulting in a dose of 1200 mg insoluble P per pot. Fertiliser (Osmocote 11:10:18 nitrogen (N):phosphorus (P):potassium (K); 5–6 month release; Everris, The Netherlands) was added to the substrate (1.2 g per pot) resulting in a concentration of 50 mg available P, 130 mg N and 190 mg K per pot; a high P treatment received 5 g Osmocote per pot (260 mg P per pot). Pre-selected bacterial endophytic isolates based on the in vitro test performance were grown in Luria Bertani (LB) broth for 2 days. Seeds of oilseed rape ʻSherlockʼ were surface sterilised (30 s in 70% Ethanol, 15 min in 4.7 NaOCl, 3 × washing) and soaked in cell suspensions of the endophytic bacteria (re-suspended in 0.01 M MgSO4, target OD600nm = 0.5 resulting in 108–109 CFU ml−1); seeds for controls were soaked in sterile MgSO4. Four treated seeds were sown per pot and thinned to one plant per pot after emergence. Pre-selected fungal endophyte isolates based on the in vitro test performance were grown on filter paper laid on MEA and seeds were allowed to germinate on the mycelial mat and then planted into pots; seeds for controls were germinated on sterile filter paper laid on MEA. After 123 days of growth, plants were fertilised with a mixture of Kristalon fertiliser (AGRO CS a.s., Czech Republic) for tomatoes (N:P2O5:K2O 7.5:12:36 + 4.5 MgO + 10 SO3 + Trace elements (B Cu, Fe, Mn, Mo, Zn) and KH2PO4 dissolved in water (1.67 g Kristalon + 0.17 g KH2PO4 per L, pH 6.1). Four-hundred ml of this solution were added per pot, resulting in a final concentration of 0.05 g P, 0.05 g N and 0.22 g K (ratio 1:1:4.4), 0.018 g Mg and 0.027 g S per pot. Plants were harvested after approx. 190 days of growth and a number of growth parameters assessed (number of shoots and pods, maximum shoot height, cumulative shoot height, shoot, pot and root dry weight/all dry weights at 65 °C/together with the phenological phase/BBCH scale/).

In planta test for activity against Phoma leaf spot (Leptosphaeria maculans)

Antagonistic activity of the pre-selected bacterial endophytes against Phoma leaf spot was tested with a cotyledon assay on OSR seedlings23,65. Twelve mycelial plugs of Leptosphaeria maculans isolate ICBN3 (provided by Dr. Koopmann, University of Göttingen, Germany) grown on V8-agar were used to inoculate 100 ml liquid cultures. Culture medium was prepared as follows: 20 g L−1 oilseed rape leaves and roots were shredded in a mixer, boiled for 5 min, filtered, amended with 2% malt extract and 0.2% meat peptone, portioned into 300 ml flasks, amended with 3 mm filter paper pieces, and autoclaved; fungal cultures were grown for 3 weeks at 20 °C at 200 rpm before the overgrown filter paper pieces were retrieved for inoculation. Surface sterilised seeds of oilseed rape ʻSherlockʼ were treated with cell suspensions of the bacterial endophytes as described above and sown in 50 ml plastic falcon tubes filled with an autoclaved vermiculite:zeolite mixture amended with 0.15 g osmocote (NPK 12-11-17 + 2 MgO + trace elements fast release within 6 weeks); 2 ml of bacterial cell suspension were added to the substrate. After emergence, plants were thinned to 1–2 seedlings per tube and secondary leaves of plants were regularly removed. After full development of cotyledons (19 days plant age), one cotyledon was punctured with a needle on each half and the wounds treated with 10 µl of bacterial endophyte cell suspensions (resuspended in 0.01 M MgSO4, 108—109 CFU ml−1) to test a direct protective effect. The other cotyledon was left untreated to test for the systemic effect of endophytes (introduced via seed treatment). Additional controls were set up with a commercial fungicide (Horizon 250EW, Bayer Crop Science, active ingredient tebuconazole) and a commercial biocontrol agent (Polyversum; Biopreparáty Spol. s r. o., Czech Republic; active ingredient Pythium oligandrum) which were prepared according to manufacturer’s instructions and applied to wounded cotyledons. Two days later (21 days plant age) all cotyledons were punctured on each side with a needle and covered with filter papers overgrown with mycelium of Leptosphaeria maculans (non-inoculated controls C- were covered with sterile filter papers). Filter papers were attached to leaves with a drop of wax. Inoculated plants were kept in a moist chamber at 20 °C for two days and then grown in a growth chamber (16 h/8 h light/dark cycle, 21–23 °C) for further 14 days before lesion sizes were assessed (scale according Chèvre et al.66, modified: 0 = no lesion, 1 = lesion size 1.5 mm, 2 = lesion size 1.5 mm and chlorosis, 3 = lesion size 1.5–3 mm, 4 = lesion size 1.5–3 mm, chlorosis, 5 = lesion size 3.1–5 mm, 6 = lesion size 3.1–5 mm, chlorosis, 7 = lesion size 3.1–5 mm, severe chlorosis, 8 = lesion size > 5 mm, 9 = lesion size > 5 mm, chlorosis, 10 = lesion size > 5 cm, severe chlorosis, 11 10 = lesion size > 5 cm, severe chlorosis, withered or sporulation).

In planta testing for activity against Sclerotinia

Sclerotinia sclerotiorum strain DBM 4244 (Collection of Yeasts and Industrial Microorganisms, Institute of Chemical Technology, Prague, Czech Republic) was inoculated onto OSR and re-isolated from diseased plants prior use. The experiment was set up in 200 ml pots filled with an autoclaved mixture of horticultural, peat-based substrate and sand (80:20, v:v) This autoclaved substrate was inoculated with S. sclerotiorum grown for 4 weeks on autoclaved millet (0.75% m:v) and pre-incubated in tightly closed plastic bags for further 4 weeks according El-Tarabily et al.67, modified). Bacterial endophytes were applied on surface-sterilised seeds as described above (~ 108 cfu ml−1 in 0.1 M MgSO4). The fungal strain Cladosporium strain RHR7 was applied twice: once as a spore suspension (107 spores ml−1 in 0.1 M MgSO4 with 10 µl Tween20) and the other time as a mycelial mat on which OSR seed germinated (analogously to other Cladosporium isolates). The reason behind was to compare the two methods of introducing the fungus. Seeds of OSR ‘Sherlock’ were soaked for 1–2 h in bacterial or fungal spore suspensions; seeds for controls were soaked in sterile 0.1 M MgSO4. Cladosporium isolates (RHR7, RHR8, RHR9, RHR14, RHS1, RHS5) were applied by allowing surface-sterilised seeds to germinate on the mycelial mat of an overgrown filter paper placed on MEA, as described above. Nine treated seeds were sown per pot; pots were placed in a warm greenhouse (20–30 °C). Emergence was assessed after 14 days. Additional millet inoculum grown for 14 days at 20 °C (1.6 g per pot) and inoculum grown on autoclaved oilseed rape leaves for 7 days at 20 °C (0.8 g overgrown leaves per pot) was distributed between the plants after 21 days of growth. Concomitantly, pots were moved to a cold greenhouse (Tmin 8 °C). The percentage of diseased and dead plants (girdled hypocotyl, white mycelial outgrowth at the hypocotyl, wilting) was assessed after 27 and 68 days of growth.

In vitro test for mutual inhibition of plant growth promising strains

To exclude the risk that PGPR strains (A. piechaudii AP_RD9, Pseudomonas spp. PCh_RH1, and PG_RD39, and Stenotrophomonas sp. SS_RD24) that we planned to apply in a combined inoculum in the field (see below) inhibited each other, potential mutual antagonism was tested in an in-vitro assay as follows. Cells from liquid cultures (grown overnight on tryptic soy medium at 24 °C) were centrifuged and washed once in phosphate buffered saline (PBS). Each of the strains was then diluted in PBS to 0.5 McF (150.106 CFU/ml, OD 0.125 for 550 nm) and evenly spread with a sterile cotton swab on tryptic soy agar (three replicate Petri dishes per strain) and allowed to dry. The other strains were then applied as droplets (5 µl) at a cell density of 5 McF (1500.106 CFU/ml, OD 1.25 for 550 nm) suspended in PBS onto this seeded lawn (3 replicate drops per Petri dish, 9 in total). After a growth period of 72 h at 24 °C, it was checked whether the outgrowing colonies produced clear inhibition zones in the surrounding bacterial lawn.

Field experiment in Poříčí, Czech Republic

The strains Achromobacter piechaudii AP_RD9, Pseudomonas chloraphis/veronii (PCh_RH1), P. grimontii/fluorescens (PG_RD39, P. fluorescens clade), and Stenotrophomonas sp. SS_RD24 were applied to OSR seeds as a combined inoculum (IG; for plant growth promotion) and the strain Serratia proteamaculans SP_RH21 was applied as a single inoculum (IP; for plant protection) as follows. Cultures growing in LB for 2 days were centrifuged and the cell pellet re-suspended in Ringer solution. For the combined inoculum, optical density at 600 nm of the cell suspensions was measured and pooled in proportions adjusted to a theoretical value of 2.5 for each isolate within the pooled suspension. Cell suspensions were lyophilised (Eco Fuel Laboratories, Prague, Czech Republic) leading to a final concentration of 8 × 108 cells g−1 lyophilisate for the combined inoculum IG, and 5 × 108 cells for the single inoculum IP. For incrustation of OSR seeds, these lyophilisates were mixed into an apatite – bentonite carrier (1:1 apatite : bentonite + 1.7% lyophilisate of IG or 0.9% lyophilisate of IP, respectively; all w:w). This incrustation mixture was added to seeds at the ratio 0.05:1 (w:w). The treatment resulted in 4 × 104 CFU per seed (4.26 µg lyophilisate per seed) for the inoculum IG, and 2 × 104 CFU per seed (2.24 µg lyophilisate per seed) for inoculum IP. Treated seeds were sown in a commercial oilseed rape field in Poříčí, Czech Republic (49.8464633 N, 14.6767358 E) in late August 2016 after winter wheat as a pre-crop. An imbalanced, factorial design with the factors bacterial treatment (controls, IP and IG) and fungicide treatment (+ F, − F) was used (Fig. 1). The field was fertilised with 30 t ha-1 cattle manure in autumn and with 300 kg ha−1 YaraBela Sulfan fertiliser (24% N + 6% S + 7% CaO) and 210 l ha−1 DAM 390 in spring. In March 2017, the number of plants per m2 and the percentage of plants with Phoma leaf spots and with reddish discolouration of leaf margins (indicative of stress) were counted. Three 1 m2-squares, spaced evenly across the length of each strip, were assessed per strip. On the 17th of May, + F plots were treated with fungicide (Acanto plus [picoxystrobin, cyproconazole], 1 L ha−1). An assessment of the incidence of Phoma stem canker, Sclerotinia stem rot an Alternaria stem symptoms was done on the 7th of July 2017 before the oilseed rape harvest. 60 plants across each strip were assessed for presence or absence of the diseases.

Statistical analyses

All plant experiments consisted of five replicates per treatment and 10 replicates for the control (non-treated in the experiment on plant growth promotion, C +  = control inoculated with pathogen, in the experiments on disease control) unless stated otherwise. Replicate pots were fully randomised during growth. Acquired data were checked for normal distribution (Shapiro–Wilk´s W test) and homoscedasticity (Levene´s test). and if necessary data were square root or common logarithm transformed before conducting Student’s t-test or Analysis of Variance followed by post-hoc Unequal N HSD test. Significant treatment effects in comparison to the non-treated control were detected using Dunnett’s test. When the assumption of homoscedasticity was not met, non-parametric alternatives (Mann–Whitney-test, Kruskal–Wallis Analysis of Variance) were used to test for significant differences among experimental groups. The field experiment was analysed in March 2017 using a nested factorial ANOVA with the random factor strip nested within seed treatment (factor 1) and fungicide application (factor 2). For disease assessment plants were assigned the number 0 for absence of disease, and 1 for presence of disease, and a nested ANOVA with the random factor strip nested within seed treatment and fungicide application was calculated. All statistical calculations were done with Statistica v. 12 (Dell Inc., Tulsa, USA). Biodiversity indices (Shannon–Wiener Diversity Index, Gini-Simpson's Index, Pielow evenness index, Margalef diversity coefficient and Total taxonomic distinctness68) were calculated based on the genera separately for bacteria and fungi.