Introduction

Several studies have demonstrated the suppression of soil-borne pathogens by composts derived from agro-industrial wastes including olive processing by-products, spent mushroom substrates, and grape/winery wastes (Noble and Coventry 2005; Termorshuizen et al. 2007; Ntougias et al. 2008).

Suppressiveness of composts to plant diseases has been attributed to complex interactions between abiotic characteristics and/or microbial populations and plants, involving microbial competition for nutrient substrates and ecological niches, antibiosis, microbial production of lytic enzymes, fatty acid degradation, parasitism, changes in nutrient availability and host-mediated induction of resistance (Lorito et al. 1993; Erhart et al. 1999; Hoitink and Boehm 1999; McKellar and Nelson 2003; Kavroulakis et al. 2005; Borrero et al. 2006; Perez-Piqueres et al. 2006). Synthesis and excretion of simple bioactive compounds (amino acids, organic acids, vitamins), plant growth hormones and chelators alter physiological conditions related to plant growth and may also affect the enhanced plant tolerance to diseases conveyed by certain composts (Van Loon 2007).

As a consequence, the capacity of composts to suppress plant diseases is highly variable, even when using similar composted materials and application rates (Erhart et al. 1999; Bonanomi et al. 2007; Noble and Coventry 2005; Termorshuizen et al. 2007). Inoculation of mature composts with efficient biocontrol agents could potentially improve the efficiency and the spectrum of disease suppression (Zhang et al. 1998; Trillas et al. 2006; Siddiqui et al. 2008) but this route has not been systematically explored.

In contrast to the general suppression of Oomycetes pathogens, which appears to be related to the proliferation and activity of broad microbial consortia (Hoitink et al. 1996), specific microbial agents appear to be responsible for the efficiency of compost amendments on the control of Eumycetes, such as Fusarium and Rhizoctonia pathogens (Hoitink and Boehm 1999; Weller et al. 2002).

Plant chitinases and β-glucanases inhibit fungal growth (Mauch et al. 1988; Arlorio et al. 1992) and have also been shown to inhibit plant colonization and to reduce the plant disease incidence caused by pathogenic fungi (Broglie et al. 1991; Vierheilig et al. 1993). Microorganisms expressing these activities have also been shown to effectively inhibit fungal growth and to lower disease incidence caused by soil-borne pathogens (Chet et al. 1990; Jung et al. 2003; Nagarajkumar et al. 2004). Microbial hydrogen cyanide (HCN) production is another characteristic related to suppression of plant pathogens (Voisard et al. 1989), and efficient producers of both HCN and lytic enzymes appear to be competent biological control agents (Nagarajkumar et al. 2004)

In this study we made use of three compost media derived from olive mill, winery and Agaricus bisporus agro-industrial waste materials, which were shown to be highly suppressive to the tomato pathogens Phytophthora nicotianae and Fusarium oxysporum f.sp. radicis-lycopersici (FORL) (Ntougias et al. 2008). All three composts are rich in chitin rather than cellulose and we hypothesized that this common property could lead to the proliferation of chitinase and β-glucanase bacterial producers. The objectives of this work were i) to isolate from these media novel rhizospheric and root-associated/endophytic (RAE) bacterial strains as potential biocontrol agents; ii) to test the hypothesis of whether these compost media could be selective for biocontrol agents showing lytic properties against chitin and β-glucans; iii) to identify effective biocontrol agents by evaluating their in vitro antagonistic activity against major soil-borne pathogens and their suppressiveness to tomato crown and root rot disease caused by FORL, and iv) to investigate whether cell-free culture extracts from the most effective biocontrol agents can suppress plant disease.

Materials and methods

Plant pathogens

Plant root pathogen FORL Jarvis & Shoemaker (strain CBS 101587) was obtained from the Centraal Bureau voor Schimmelcultures, Baarn, The Netherlands; Fusarium oxysporum f.sp. raphani Kendr. & Snyder was kindly donated by CMJ Pieterse (University of Utrecht, The Netherlands), while Phytophthora nicotianae Breda de Haan (strain IK-23), Phytophthora cinnamomi Rands and Rhizoctonia solani were provided by the Benaki Phytopathological Institute, Athens, Greece.

Compost material used

Three composts, derived from agro-industrial residues and by-products, i.e. grape-marc wastes (GM), extracted olive press cake (EPC, press cake remaining after the extraction of residual kernel and pulp oil from olive press cake), olive tree leaves (OL), olive mill wastewaters (OMW) and spent mushroom compost from the cultivation of Agaricus species (SMC) were prepared on site at the Institute of Kalamata and have been previously described (Ntougias et al. 2008). The composts were labelled GM+EPC, OL+OMW and SMC, in reference to the raw materials used to make the respective composting blends. All three composts had demonstrated high levels of suppressiveness to disease caused by P. nicotianae and FORL in tomato plants, as previously reported (Ntougias et al. 2008).

Isolation of antagonistic bacteria

Sphagnum peat, (Sunu Kura, Seda joint-Stock company), treated with 20 g l−1 CaCO3, was mixed with each of the composts at a ratio of 1:1 w w−1, in order to obtain RAE and rhizospheric bacteria. Tomato plants (Solanum lycopersicon cv. ACE55) were grown on the peat–compost mixtures in 300 cm3 volume pots. Seeds were surface-sterilized in 2.5% NaOCl and three seeds were directly sown in each pot. The pots were placed in a growth chamber at 65% relative humidity (RH) (light: 16 h, 25°C; dark: 8 h, 20°C) and irrigated daily. Plants were grown for 1 month.

To isolate rhizospheric bacteria, plants were removed carefully from the pots and 1 g (wet weight) of compost-based potting medium, loosely attached to plant roots, was put in 25-ml glass test tubes containing 9.5 ml of sterile 8.5 g l−1 NaCl solution (Ntougias et al. 2004). The mixture was shaken thoroughly, tenfold dilution series were prepared and 0.1 ml of the suspensions was sequentially aliquoted onto isolation media. For the isolation of RAE candidates, tomato plant roots were surface-sterilized with 2.5% NaOCl for 10 min and extensively washed with sterilized tap water. Roots were partitioned and placed on solid isolation media, i.e. Luria-Bertani (LB) and King’s B (KB) supplemented with the antifungal compound cycloheximide (100 μg ml−1). All plates were incubated at 25°C and bacterial isolates were obtained for up to 7 days. Since we have not performed further microscopic analysis to verify the ingression of these bacteria inside the root tissues, these isolates represent either true endophytes or bacteria tightly attached to the root surface.

In vitro evaluation of the antagonistic activity of isolated bacteria against soil-borne fungi

The RAE and the rhizospheric bacteria isolated were evaluated for their in vitro antagonistic activity against FORL, F. oxysporum f.sp. raphani, P. cinnamomi, P. nicotianae and R. solani. Bacterial isolates were incubated in broth medium (LB or KB based on the medium from which they were isolated) at 25°C. Fungal pathogens were grown on potato dextrose agar (PDA). A 6-mm plug from the leading edge of a 5-days-old culture of each fungal pathogen tested was placed at the centre of a 9-cm diameter Petri dish containing PDA. Aliquots of 0.1 ml of each growing bacterial culture were placed at approximately 3 mm on the edges at three sides of the plate (five replicate plates per each biocontrol agent–pathogen combination). Escherichia coli strain DH5α (no antagonistic activity) was used as negative control. Isolates that formed zones of inhibition (haloes without mycelial growth or distorted hyphae) over 2 mm were selected.

In vitro detection of secreted potential antifungal metabolites

For the detection of secreted bacterial β-glucanase and chitinase, 1.7% agar plates containing 0.1% lichenan (Walsh et al. 1995) and 0.1% colloidal chitin (Garbeva et al. 2004) were used, respectively. The plates were inspected for enzymatic degradation indicated by a clearing zone following 5 days of growth at 25°C. HCN was detected by cyanide indicator paper (Castric 1975). The E.coli strain DH5α was used as negative control.

In planta evaluation of the antagonistic activity of RAE and rhizospheric isolates against FORL

To evaluate the bacterial antagonistic activity in planta, conidial suspensions of FORL (105 conidia cm−3 of potting medium) and cell suspensions (106 ml−1) of each bacterial isolate were prepared as follows: FORL was grown on potato dextrose broth (PDB) at 25°C in the dark, for 5 days. Following removal of mycelial fragments by sieving, conidia were recovered by centrifugation at 5,000 g, counted and resuspended in the appropriate volume of 0.85% w/v NaCl. Τhe pre-selected bacterial strains were incubated in broth medium (LB or KB) at 25°C. Cells were harvested by centrifugation at 5,000 g for 10 min, counted on the basis of standard growth curves and resuspended in the appropriate volume of 0.85% w/v NaCl.

Seeds of tomato plants (cv. ACE55) were surface-sterilized in 2.5% NaOCl, pre-germinated for two days on moist paper and then sown into 15-cm diameter pots (five seeds per pot), each containing 300 cm3 of peat, blended with NPK + micronutrients fertilizer (5.8% NO −3 -N, 3.9% NH +4 -N, 10.6% urea-N, 20% P2O5, 20% K2O, 0.5% MgSO4, 0.03% Fe ethylene diamine tetraacetic acid (EDTA), 0.02% Zn-EDTA, 0.02% Mn-EDTA, 0.01% Cu-EDTA and 0.02% B) to a concentration of 0.8 g l−1. The pots were placed in a growth chamber at 65% RH (light: 16 h, 25°C; dark: 8 h, 20°C) and watered to the initial weight on alternate days. Once a week the plants were fertilized with a balanced nutrient solution including micronutrients.

Potting media were inoculated with the bacterial strains 1 week after the sowing of tomato seeds. One ml of bacterial cell suspension (106 ml−1) was applied for each growing plant. Pots planted with tomato plants in the absence of bacterial inoculum were used as control. The plant pathogen FORL was applied at 105 conidia per cm3 of potting mixture at the stage of the first true leaf emergence (approximately 2 weeks after sowing). Plants were scored daily either as healthy (no or scarce disease symptoms) or as diseased (severe wilting or dead plants). Five replications (pots) per treatment were included for the in planta evaluation of the RAE and rhizospheric bacteria tested. The experiment was performed twice for all isolates.

Molecular characterization of the bacterial antagonists: 16S rRNA gene sequencing and phylogenetic analysis

DNA from pelleted cells of the selected bacterial isolates was extracted as described by Wilson (1992). The nearly full-length 16S rRNA gene was amplified using primers Afor (5′-GGAGAGTTAGATCTTGGCTCAG-3′) (positions 6–27 by E. coli numbering) and Crev (5′-AGAAAGGAGGTGATCCAGCC-3′) (positions 1542–1525 by E. coli numbering) (Ntougias et al. 2006). PCR reactions were performed in a PTC-200 thermocycler (MJ Research Inc., USA), with a denaturation step of 2 min at 94°C, followed by 35 cycles of 30 s denaturation at 94°C, 30 s primer annealing at 55°C and 1 min polymerization at 72°C. The PCR was completed by 5 min DNA chain extension at 72°C. The amplified fragments were cloned into the pGEM-T Easy Vector (Promega, WI, USA) and transformed into E. coli JM-109 competent cells. For each isolate, three independent clones were grown in LB and plasmid DNA was extracted using the “NucleoSpin plasmid Quick Pure” kit (Macherey-Nagel, Germany). PCR sequencing was performed at the Institute of Molecular Biology and Biotechnology (IMBB), Heraklion, Greece, as previously described (Ntougias et al. 2006).

The 16S rRNA gene sequences were assembled using the CAP3 programme (http://pbil.univ-lyon1.fr/cap3.php). Similarity searches were performed using “BLASTN” (http://www.ncbi.nlm.nih.gov/BLAST/) and “Seqmatch” (Ribosomal Database Project II—http://rdp.cme.msu.edu/seqmatch/seqmatch_intro.jsp), and the sequences of their closest relatives identified in both searches were used for further phylogenetic analysis. Alignment was carried out using the “ClustalW Submission Form” (http://www.ebi.ac.uk/clustalw/), evolutionary distances were calculated using the method of Jukes and Cantor (1969) and topology was inferred using the “neighbour-joining” method (Saitou and Nei 1987) based on bootstrap analysis of 1,000 replicates. A phylogenetic tree was constructed using TREECON for Windows (Version 1.3b) (Van de Peer and De Wachter 1993).

In vitro inhibitory effect of culture extracts of Enterobacter sp. AR1.22 and Alcaligenes sp. AE1.16 on the growth of FORL

Cell-free culture extracts of biocontrol agents Enterobacter sp. AR1.22 and Alcaligenes sp. AE1.16 were obtained through centrifugation of 48-h bacterial cultures in LB, followed by filter-sterilization (0.2 μm filter, Nalgene, Rochester, NY). Inhibitory effects of culture extracts on the growth of plant pathogens were examined using 9-cm PDA plates, containing cell-free extract and deionized water in a ratio of 1:1 v/v. PDA plates (including half strength), in the absence of culture extracts, were used as controls. A 6-mm plug from the leading edge of a 5-days-old culture of FORL was placed in the centre of both extract-containing and control PDA plates. Cultures were incubated at 25°C and the radial growth was recorded daily. The experiment was performed twice.

Evaluation of Enterobacter sp. AR1.22 and Alcaligenes sp. AE1.16 culture extracts to suppress tomato crown and root rot disease caused by FORL

Cell-free extracts of Enterobacter sp. AR1.22 and Alcaligenes sp. AE1.16 were obtained, as described above. Potting media and tomato plants were prepared and grown, respectively, as previously described. Six experimental sets (each consisting of five replicate pots per treatment) were carried out to assess the effect of increasing concentrations of putative antifungal compounds, secreted by the bacterial isolates in the culture medium. Twenty ml of culture extract was applied to each pot once a week, with the first application being performed 1 week after sowing. The first experimental set received culture extract for 1 week only (i.e. week 2) and was irrigated with tap water thereafter. The second set received culture extract for two consecutive weeks (i.e. weeks 2 and 3) and tap water thereafter, whereas the third set received culture extract for three consecutive weeks (i.e. weeks 2, 3 and 4) and then tap water, et cetera. Controls were prepared by replacing culture extracts either with tap water or LB medium. The plant pathogen FORL was applied at 105 conidia per cm3 of potting mixture at the stage of the first true leaf emergence (2 weeks after sowing). Plants were scored daily either as healthy or as diseased.

Statistical analysis

Experimental data were analyzed using standard analysis of variance (ANOVA) followed by Duncan multiple comparison tests (a < 0.05). Standard errors were calculated for all mean values. Arcsine transformation was applied to all data expressed as percentages, before analysis.

Results

Isolation and in vitro evaluation of the antagonistic activity of bacterial strains against soil-borne fungal pathogens

Forty-four (35 rhizospheric and 9 RAE) out of the 329 bacterial isolates showed antagonistic activity against at least one of the fungal plant pathogens tested (Table 1). In vitro inhibition of all fungal pathogens tested was observed for 25 isolates (Table 1), whereas seven strains (AR1.15, BR1.1, BR1.3, BR1.6, BR1.10, AR3.4 and AR3.9) inhibited growth of FORL and F. oxysporum f.sp. raphani only; three strains (AR1.17, BR1.4 and BR1.11) inhibited growth of FORL, F. oxysporum f.sp. raphani and R. solani only; five strains (AR1.7, AR1.8, BR1.8, BR1.14 and BR2.4) inhibited growth of P. nicotianae, P. cinnamomi and R. solani only; and two strains (AR1.16 and AE1.16) inhibited growth of P. nicotianae and P. cinnamomi only. For the majority of the selected bacterial strains (26 out of 44), the inhibition of fungal pathogens was influenced, in a complex manner, by the growth medium used (LB or KB), depending on the biocontrol agent and the plant pathogen tested. However, pathogen inhibition was clearly favoured by the use of LB medium in the case of bacteria isolated from the SMC-derived compost and the effect was not observed when KB was used (Table 1). The majority of biocontrol agents selected were isolated from GM+EPC-based potting medium (26 strains in total, including one RAE), whereas many RAE (six strains) were isolated from plants growing on SMC-based amendments (Table 1).

Table 1 Screen for selection of potential bacterial biocontrol agents
Full size table

In vitro detection of secreted potential antifungal metabolites

Chitinase and β-glucanase (lichenanase) activity was detected in all bacterial strains exhibiting in vitro antagonistic activity against soil-borne plant pathogens, whereas only 14 strains were positive for HCN production (Table 1).

Evaluation of RAE and rhizospheric bacteria isolated for suppressiveness to tomato crown and root rot disease caused by FORL

In planta experiments showed that inoculation of tomato plants with the rhizospheric isolates AR1.2, AR1.5, AR1.22, BR1.4, AR2.4bii and BR2.1 resulted in a statistically significant reduction in plant disease incidence (20–60%), depending on the strain tested (Fig. 1a). The greatest in planta antagonistic activity against FORL was observed in potting media inoculated with the strain AR1.22 (40–54% decrease in disease severity) (Fig. 1a). Inoculation of potting media with the RAE strains revealed that strain AE1.16 reduced disease incidence by approximately 40–50%, while the remaining RAE did not cause statistically significant reduction in disease severity compared to the control potting medium with tomato plants in the absence of bacterial biocontrol agent (Fig. 1b).

Fig. 1
figure 1

a. In planta evaluation of rhizospheric bacteria selected in the present study against Fusarium oxysporum f.sp. radicis-lycopersici (FORL). Disease incidence is recorded as percentage of dead plants. Letters at the top of the bars indicate statistically significant differences (Duncan’s test, a < 0.05; n = 5). b.In planta evaluation of root-associated/ endophytic (RAE) bacteria selected in the present study against FORL. Disease incidence is recorded as percentage of dead plants. Letters at the top of the bars indicate statistically significant differences (Duncan’s test, a < 0.05; n = 5). Results from a representative experiment are shown

Full size image

Molecular characterization of the bacterial antagonists selected from in planta experiments

Seven bacterial strains that were identified as the most effective biocontrol agents were characterized at the species level. Their phylogenetic allocation and 16S rRNA gene sequence identities are presented in Fig. 2 and their description is presented in Table 2. Four rhizospheric strains were isolated from the GM+EPC-based compost medium. Three strains, namely AR1.2, AR1.5 and BR1.45 were highly related to Bacillus spp., whereas the closest relative of strain AR1.22 was Enterobacter ludwigii (99.7% identity in 16S rRNA gene) and the next closest species was E. cloacae. Bacterial strains AR2.4bii and BR2.1, isolated from OL+OMW-based compost medium, were affiliated with Serratia marcescens strains, one of which has been isolated from soybean rhizosphere and has been reported as antagonistic to Ralstonia solanacearum (GenBank direct submission AJ550467) However, isolates AR2.4bii and BR2.1 differed phylogenetically at subspecies/strain level (Table 2). The only RAE selected (strain AE1.16) was associated with Alcaligenes faecalis subsp. faecalis (99.7% identity in 16S rRNA gene).

Fig. 2
figure 2

Phylogenetic tree based on the 16S rRNA gene sequence of effective biocontrol agents (depicted in bold typeface), isolated from compost-based amendments (or their seeded plants) made by agro-industrial by-products abundant in the Mediterranean region. Evolutionary distances were calculated using the method of Jukes and Cantor (1969) and the topology was inferred using the “neighbor-joining” algorithm (Saitou and Nei 1987), based on a bootstrap analysis of 1,000 replicates (values on branches denote % of bootstrap support). Scale bar represents 0.02 inferred substitutions per nucleotide position

Full size image
Table 2 Closest phylogenetic relatives of most effective bacterial biocontrol agents identified in this study
Full size table

In vitro and in planta inhibitory effects of culture extracts produced by Enterobacter sp. AR1.22 and Alcaligenes sp. AE1.16

Culture extracts produced by the rhizospheric Enterobacter sp. AR1.22 isolate inhibited by 20% the radial growth of FORL, whereas no effect on the growth of the pathogen was observed with the culture extracts of the RAE Alcaligenes sp. AE1.16 isolate (Fig. 3).

Fig. 3
figure 3

In vitro inhibitory effects of culture extracts produced by Enterobacter sp. AR1.22 and Alcaligenes sp. AE1.16 on the growth of the root pathogen Fusarium oxysporum f.sp. radicis-lycopersici (FORL). Letters at the top of the bars indicate statistically significant differences (Duncan’s test, a < 0.05; n = 4). Results from a representative experiment are shown

Full size image

In accordance, no inhibitory effect was observed in the disease severity caused by FORL in tomato plants, following successive weekly applications of Alcaligenes sp. AE1.16 culture extract, (data not shown). On the contrary, the cummulative application of Enterobacter sp. AR1.22 culture extracts to tomato plants had a significant effect on the reduction in disease severity caused by FORL. The reduction was positively related to the number of successive applications (Fig. 4). Thus, no difference in disease incidence was recorded in plants treated once or twice with culture extracts compared to control (water- or LB- treated) plants. Interestingly, extract applications carried out one and 2 weeks prior to plant inoculation with the pathogen (experimental sets 1 and 2; Fig. 4) conferred no protective effect on the plants; on the contrary, the application of extracellular compounds in the presence of the pathogen (experimental sets 4, 5 and 6) led to a 40% decrease in disease severity compared to control plants.

Fig. 4
figure 4

In planta inhibitory effects of Enterobacter sp. AR1.22 culture extracts, applied successively on a weekly basis for 1–6 weeks on the disease severity of the root pathogen Fusarium oxysporum f.sp. radicis-lycopersici (FORL). Disease incidence is recorded as a percentage of dead plants. Different letters shown for representative time interval points, indicate statistically significant differences (Duncan’s test, a < 0.05; n = 5). Results from a representative experiment are shown

Full size image

Discussion

In vitro testing of isolates from plant growth media is a common screen strategy for the selection of indigenous microbiota with biocontrol properties (Ntougias et al. 2004; Termorshuizen et al. 2007). The compost media used in the present study have been shown to drastically reduce disease incidence caused by FORL and P. nicotianae in tomato plants, and also leaf infection by Septoria lycopersici, which is an indication of the capability of the compost to induce systemic resistance in plants (Ntougias et al. 2008). The high percentage of isolates from the three compost media showing in vitro biocontrol capability depicts the plant protection properties that were reported for these composts.

In the present study, all bacterial strains, showing in vitro inhibition against plant pathogens, exhibited both chitinase and β-glucanase activity, clearly indicating a lytic action against fungal pathogens. In accordance, inhibition of plant pathogens has been reported to be more effective when glucanases and chitinases acted synergistically (Lim et al. 1991; Lorito et al. 1993). The three compost media used in the present study are rich in chitin rather than cellulose. It could be envisaged that this common property of the media favors the proliferation of chitinase and β-glucanase bacterial producers. The spent mushroom substrate is well-colonized and rich in fungal cell-wall material of A. bisporus, which has been successfully used as substrate for the production of chitinolytic enzymes (Dahiya et al. 2005); grape marc is particularly rich in yeasts (Streichsbier et al. 1982) and in grape skins that directly produce antifungal chitinases (Robinson et al. 1997; Fernandez-Caballero et al. 2009); olive mill wastewaters are particularly rich in yeast cell walls (Ben Sassi et al. 2006; Amaral et al. 2008) and favour the proliferation of basidiomycetous yeasts, when applied to environmental media (Karpouzas et al. 2009). Chitinase production has been shown to increase in the presence of chitin-rich cell walls of basidiomycetes (Bruce et al. 1995). Fungal and bacterial antagonists may not efficiently produce chitinases in the presence of simple sugars, such as glucose (Morrissey et al. 1976; de la Cruz et al. 1993; Gupta et al. 1995), which are produced by the degradation of cellulose-rich substrates. On the other hand, their activity is therefore expected to be favored in the mature composts used in this study. Important plant pathogens, such as Fusarium and Rhizoctonia species, are efficient cellulose degraders and may thrive on cellulose-rich substrates. Indeed, suppression of Rhizoctonia species has been related to compost media with low availability in cellulose and simple sugars (Chung et al. 1988; Hoitink et al. 1996).

HCN production, detected in 14 out of the 44 isolates showing in vitro suppressiveness, did not appear to be specifically related to the protection against one or more of the pathogens tested. In line with previous reports (Senthilkumar et al. 2009), no relation between HCN and the production of lytic enzymes was observed in this study.

Only seven out of the 44 potential biocontrol agents identified in vitro were also effective in planta and only one RAE biocontrol agent was positive to both in vitro and in planta assays. This is commonly observed in the complex rhizosphere environment and underlines the necessity for in planta experiments. Interestingly, 4 out of 7 of the in planta biocontrol agents identified in the present study belong to the genera Bacillus and Serratia. Gram-positive bacteria mainly of the genus Bacillus have been found to proliferate in media containing high chitinase-producing bacterial populations (Strohl 1997; Chae et al. 2006). Indeed, Chae et al. (2006) reported the dominance of Bacillus, Paenibacillus and Serratia spp. in chitin-containing rice compost. The role of S. marcescens strains in the biological control of several plant pathogens has been repeatedly reported: they are efficient producers of several chitinases (Brurberg et al. 1996) and their antagonistic properties are attributed to the excretion of lytic enzymes that degrade the cell walls of the pathogenic microorganisms (Ordentlich et al. 1988; Someya et al. 2005). The closest relatives of the biocontrol strain BR1.4 were Lysinibacillus sphaericus (formerly Bacillus sphaericus) strains, which are antagonistic to plant pathogens (Choudhary and Johri 2009). Besides, two biocontrol strains were related to B. thuringiensis, an organism known for its protective properties against plant pathogens beyond insect control (Raddadi et al. 2007). The isolated RAE Alkaligenes sp. AE1.16 also produces extracellular chitinases and glucanases. Previous studies, in which biocontrol strains related to the genus Alcaligenes had been isolated, report these agents as chitinase producers with a lytic mode of action (Vaidya et al. 2001). However, this strain exhibited biocontrol activity only when plants were inoculated with living bacterial cells, while culture extracts failed to inhibit fungal growth in vitro. This implies that the mechanisms employed by this strain to confer plant resistance may extend beyond antibiosis and could be related to the fact that AE1.16 resides most probably in the plant tissues rather than the rhizosphere. This, however, and the verification of the true endophytic nature of the isolate require further clarification.

Isolate AR1.22, the most effective rhizospheric biocontrol agent in the present study, was phylogenetically related to E. ludwigii and E. cloacae strains. E. ludwigii strain BNM 0357, isolated from the Lolium perenne rhizosphere, has recently been identified as an effective biocontrol agent inhibiting F. solani vegetative growth, through competition for nutrients and excretion of diffusible antifungal compounds (Shoebitz et al. 2009). Furthermore, the antagonistic properties of E. cloacae strains have been extensively reported (Roberts et al. 1994; van Dijk and Nelson 2000). The colonization of plant roots (Roberts et al. 1994) and the competition for fatty acids (van Dijk and Nelson 2000) have previously been proposed as the protective mechanism of E. cloacae strains. Our experiments on the in planta and in vitro effects of the Enterobacter sp. AR1.22 culture extract revealed a significant pathogen inhibition, the mechanism of which most probably relies on antibiosis. However, we cannot exclude the existence of a protection mechanism associated with root colonization. This antagonistic mode of action was dependent on the dose or the persistence of the inhibitory substances in the rhizosphere. Thus, when culture extract was applied for a long enough period of time, growth of the pathogen and its ingression into the plant tissues were restricted, leading to significant reduction in disease incidence (Fig. 4). On the contrary, the application of culture filtrates to the plants before the inoculation with the pathogen could not confer any level of resistance, indicating that the bacterial extracellular compounds per se most probably cannot induce plant defense responses. It cannot be excluded that, in the concurrent presence of the pathogen, plant responses could also be triggered, although this hypothesis requires further investigation. Nevertheless, the level of protection conferred by the extracellular compounds produced by Enterobacter sp. AR1.22 (>40%) could account for almost all the antagonistic activity of the strain observed when plants were inoculated with the bacterial cells. We have shown that Enterobacter sp. AR1.22 produces lytic extracellular enzymes in vitro, but it is important to investigate which other antifungal compounds may be produced by this highly antagonistic bacterial strain and whether the production of lytic enzymes takes indeed place in the rhizosphere.