Isolation and characterization of rhizosphere bacteria with potential for biological control of weeds in vineyards

Authors

  • R.D. Flores-Vargas,

    1. Centre for Rhizobium Studies, School of Biological Sciences and Biotechnology, Division of Science and Engineering, Murdoch University, WA, Australia
    Search for more papers by this author
  • G.W. O'Hara

    1. Centre for Rhizobium Studies, School of Biological Sciences and Biotechnology, Division of Science and Engineering, Murdoch University, WA, Australia
    Search for more papers by this author

Ruben D. Flores-Vargas, Centre for Rhizobium Studies, School of Biological Sciences and Biotechnology, Division of Science and Engineering, Murdoch University, Perth, WA 6150, Australia.
E-mail: fruben@murdoch.edu.au

Abstract

Aims:  Deleterious rhizosphere inhabiting bacteria (DRB) have potential to suppress plant growth. This project focuses on the isolation of DRB with potential for development as commercial products for weed control.

Methods and Results:  Bacteria were isolated from the rhizosphere, rhizoplane, and endorhizosphere of seedlings and mature plants of wild radish (Raphanus raphanistrum), annual ryegrass (Lolium rigidum) and capeweed (Arctotheca calendula) growing in vineyards in the Swan Valley, Western Australia. A majority (81·5%) of the 442 strains was obtained from either rhizospheres or rhizoplanes. Rapid screening techniques were developed to evaluate in the laboratory and glasshouse the effects of bacteria on plants. Strains were screened in the glasshouse for deleterious effects on annual ryegrass, wild radish, grapevine rootlings (Vitis vinifera) and the legume cover crop subterranean clover (Trifolium subterraneum). Three strains were identified using the Biolog system and 16S rRNA gene sequencing as two strains of Pseudomonas fluorescens (WSM3455 and WSM3456) and one strain of Alcaligenes xylosoxidans (WSM3457). One of the P. fluorescens (WSM3455) strain produced hydrogen cyanide, an inhibitor of plant roots and a broad-spectrum antimicrobial compound.

Conclusions:  Three strains specifically inhibited wild radish but had no significant deleterious effects on either grapevine rootlings or subterranean clover.

Significance and Impact of the Study:  This study suggested manipulation of the weed seedling rhizosphere using identified DRB as a potential biocontrol agent for wild radish.

Introduction

In recent years some weed populations have developed multiple herbicide resistances, causing concerns throughout the world (Elliott et al. 1996; Culliney 2005) and particularly in Australia (Crump et al. 1999; Neve et al. 2004; Vila-Aiub et al. 2005). Winter growing annual weeds such as wild radish (Raphanus raphanistrum L.), annual ryegrass (Lolium rigidum Guad.) and capeweed (Arctotheca calendula L.) emerge during autumn and early winter, and subsequently cause a significant problem in many vineyards in Western Australia. When these weeds grow under the grapevine canopy they are often controlled using herbicides because of the difficulties with their removal by either mowing or cultivation without damage to the grapevine plants.

Public concerns over soil and water contamination by herbicides has increased interest in safer control methods such as biological control using living organisms to manage pests and weeds (McLean and Evans 1996; Ying and Williams 1999, 2000; Alavanja et al. 2004; Shepard et al. 2004).

One advantage of using microorganisms (bioherbicides) for weed control is that these can be more selective than herbicides in the weed species they affect (Bolton and Elliott 1989). In addition, the potential for bioherbicides to control weeds by inhibition provides a second advantage over herbicides because there is a decreased chance of bioherbicide resistance developing in the target weeds due to the multiple mechanisms involved. Furthermore, if the microorganism used as a bioherbicide has evolved in association with the target plant host then strains of the microorganism could be available to overcome any resistance as it develops in the host (Crump et al. 1999).

Using soil microorganisms to control weeds in vineyards is an alternative method to herbicides that may reduce grape and wine production costs, decrease dependence on chemical herbicides and increase the use of environmentally sound practices. One group of microorganisms overlooked for their potential as biological control agents of weeds are the deleterious rhizosphere inhabiting bacteria (DRB), characterized as nonparasitic rhizobacteria (exopathogens) colonizing plant root surfaces and being able to suppress plant growth (Kremer and Kennedy 1996). Many DRB are plant specific (Suslow and Schroth 1982; Elliott and Lynch 1985; Cherrington and Elliott 1987). Their presence on weeds and their potential as biological control agents has only recently been investigated. DRB with potential as biological control agents have been reported on downy brome (Bromus tectorum L.) occurring in winter wheat fields (Cherrington and Elliott 1987; Schippers et al. 1987; Kennedy et al. 1991) and on several broadleaf weed seedlings (Elliott and Lynch 1985). Biological control of downy brome in winter wheat by Pseudomonas spp. isolated from downy brome roots has been demonstrated under field conditions (Kennedy et al. 1991). Two strains of Pseudomonas spp. consistently reduced density, growth and seed production of downy brome but did not affect density of winter wheat. The grain yield of winter wheat was significantly increased and attributed to the growth suppressive effects of the applied bacteria on downy brome, which allowed the wheat to be more competitive (Kennedy et al. 1991). A study carried by Adam and Zdor (2001) demonstrated that certain bacteria isolated from velvetleaf (Abutilon theophrasti Medik.) are potentially useful in suppressing weed growth. Norman et al. (1994) also evaluated rhizobacteria and their phytotoxins as weed control agents in cranberry vines. A major group of rhizobacteria with potential for biological control is the pseudomonad-like bacteria. A secondary metabolite produced commonly by rhizosphere-inhabiting pseudomonads is hydrogen cyanide (HCN), a gas known to inhibit plant metabolism and root growth (Adam and Zdor 2001; Kremer and Souissi 2001).

The objectives of the research described here were to isolate and characterize potential DRB from three weed species, wild radish, annual ryegrass and capeweed, frequently found growing under grapevine canopies in Western Australia, and then investigate the effects of these bacteria on the growth of these weeds, grapevine plants (Vitis vinifera L.) and subterranean clover (Trifolium subterranean L.), a legume purposely grown between grapevine rows as a cover crop. The aim of this study was to obtain bacteria that inhibit the weeds, but not grapevine plants and subterranean clover, for potential development of an effective bacteria-based weed management control method for vineyards.

Material and Methods

Isolation of bacteria from the rhizosphere, rhizoplane and endorhizosphere of weeds

Seedlings and mature plants of wild radish, annual ryegrass, and capeweed were collected during October 2000 from Henley Park Vineyard, Jane Brook Vineyard and Lamont Vineyard in the Swan Valley Western Australia (31° 50′S; 116° 00′E). Weed plants for isolation of bacteria from the rhizosphere (the region of soil under the influence of the root), rhizoplane (surface of the root) and endorhizosphere (intercellular spaces within root tissues) were collected between grapevine rows and under the canopy of grapevine plants within a row. Three plants of each species were collected from each vineyard and stored in sterile plastic bags at 4°C until processing in the laboratory. Standard microbiological methods were used to isolate bacteria from the rhizosphere, rhizoplane and endorhizosphere (Colins and Lyne 1980; Bakker and Schipper 1987). Bacterial suspensions were separately diluted in 0·1% (w/v) peptone water (1 : 40) and then plated onto two selective media, PseudoselTM (Sigma, Munich, Germany) and Sands and Rovira medium (Sands and Rovira 1970). Sands and Rovira and Pseudosel are highly selective media useful for isolating pseudomonads, and two nonselective media, tryptic soy agar (TSA) and nutrient agar (NA) were used to isolate rhizobacterial strains. After incubation at 28°C for 48 h isolated colonies were sub-cultured onto nonselective media using morphological characteristics to distinguish different strains. Strains were purified in culture, and stored cryogenically at −80°C (Rovira and Davey 1974; Kloepper and Schroth 1981; Kennedy et al. 1991; Adam and Zdor 2001).

Bacterial strains were designated with CRS (Centre for Rhizobium Studies). The three strains identified were lodged in the Western Soil Microbiology (WSM) collection at the CRS Germplasm (Murdoch University).

Laboratory screening for inhibitory effects of bacteria and metabolites on germination and growth of weed seedlings

Seeds of annual ryegrass and wild radish were surface sterilized by immersion in 3·25% (v/v) sodium hypochlorite (NaOCl) for 1 min, followed by 70% (v/v) ethanol for 1 min, rinsed five times in sterile distilled water and blotted on sterilized filter paper. The effectiveness of surface sterilization was assessed as described by Gealy et al. (1996). Cultures of each strain, grown for 1 day at 28°C in glucose minimum salt medium (Brown and Dilworth 1975), were centrifuged at 60 000 (g) for 10 min, and 2 ml of supernatant was added to the surface of 0·9% (w/v) water agar plates. Fifteen surface-sterilized seeds of each weed species were then placed on each plate and incubated in the dark at 20°C for 5 days. Controls were inoculated with 2 ml of sterile medium. Each strain was tested in four replicates. After 5 days, the seedlings were removed and root lengths measured.

Glasshouse screening of bacteria for inhibitory effects on weed seedlings

Surface sterilized seeds of annual ryegrass and wild radish were germinated for 2 days, on 0·9% (w/v) water agar and four seedlings were planted into a 110 mm pot containing a pasteurized mixture of yellow sand and washed river sand (>1 mm). Bacterial cultures were grown as described for the laboratory screening. The four seedlings in each pot were inoculated with a 0·5 ml of suspension (OD600 = 0·9) of either an individual strain or a combination of strains.

None of the strains screened in the glasshouse flocculated in the culture enabling consistent application of the uniform inocula. After inoculation a 1 cm layer of sterilized plastic beads was placed on the surface of each pot to reduce evaporation and airborne contamination. Plants were grown in a temperature-controlled glasshouse at 25°C, and watered every second day through a PVC tube with nutrient solution containing 0·3% (w/v) KNO3. There were four replicates of each treatment. After 6 weeks the plants were harvested, roots were washed free of sand, and shoot and root lengths measured. Shoots were separated from roots, oven dried at 60°C for 1 week and dry weights of shoots and roots were recorded. The experiment was repeated three times.

Screening for production of secondary metabolites

Three strains (WSM3455, WSM3456, and WSM3457) were characterized for their ability to synthesize HCN on TSA plates supplemented with glycine (Bakker and Schippers 1987).

Characterization and identification of strains

Phenotypic characterization

Strains that inhibited the target weed plants under laboratory conditions were characterized using standard microbiological methods for Gram-reaction, motility, catalase and oxidase production (Prescot et al. 1993). Three strains (WSM3455, WSM3456, and WSM3457) were characterized using the Biolog system (Microlog Version 3·20), which is based on the differential utilization of a large number of organic compounds (Bochner 1989). The strains were identified on the basis of their patterns of utilization of 95 substrates using the Biolog Microlog software (Biolog, Hayward, CA, USA).

16S rRNA gene sequence analysis

The 16S rRNA gene analysis for WSM3455, WSM3456, and WSM3457 was carried out using the method described by Yanagi and Yamasato (1993). The 16S rRNA gene of each isolate was amplified using PCR primers SeraF (5′ GATTGAACGCTGGCGGCAGG 3′), and SeraR (5′ CTTCACCCCAGTCATGAATC 3′), resulting in amplification products of ∼1·5 kb. The PCR reactions were done in a final volume of 25 μl comprizing 4 mM MgCl2, 200 mM dNTPs, 10 pmol of primers, 1× PCR buffer, and 0·5–1·0 U Taq DNA polymerase. PCR cycling conditions were as an initial denaturing at 94°C for 3 min, followed by 35 cycles of denaturing at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s. Both strands of the product were sequenced using an Applied Biosystems 377 DNA sequencer. The 16S rRNA sequences were analysed using the gapped BLASTn http://www.ncbi.nlm.nih.gov. search algorithm and aligned to Pseudomonas fluorescens (strain CHAO AJ 278812·1, strain ATCC AF 094726·1 and strain MM-B16 AY 196702·1).

Glasshouse screening of strains on grapevine plants

Grapevine cuttings were prepared for rooting as follows: grapevine cuttings (cv. Cabernet) approximately 150 mm long (three buds) were sterilized in 3% (v/v) NaOCl for 10 min, washed five times with sterile distilled water and blotted on sterile filter paper. The cuttings were placed in containers (400 × 650 × 250 mm) with a pasteurized mixture of yellow sand and washed river sand (1 : 1). High humidity for cuttings was obtained by covering the containers with plastic sheet (as recommended by Vinitech nursery, Western Australia). After 4 weeks grapevine plants with leaves and roots were transplanted into 150 mm diameter pots containing the pasteurized sand mixture. The grapevine plants were inoculated with 10 ml of suspension (OD600 = 0·9) of a single strain. A 1 cm layer of sterilized plastic beads was placed on the surface of each pot. Plants were grown in the temperature controlled glasshouse at 25°C and watered every second day through a PVC tube with nutrient solution containing 0·3% (w/v) KNO3. The experimental design was completely randomized with four replicates and repeated three times. After 8 weeks the plants were harvested, examined for disease symptoms, roots were washed free of sand, and shoot and root lengths measured. Leaves were separated from roots, leaf area was measured, leaves, and roots were oven dried at 60°C for 1 week and dry mass was recorded.

Glasshouse screening of strains on subterranean clover

Surface sterilized scarified seeds of subterranean clover were germinated for 2 days on 0·9% (w/v) water agar and eight seedlings were planted into a 150 mm pot containing pasteurized mixture of yellow sand and washed river sand. Seedlings of subterranean clover were inoculated with a peat suspension (2 ml per plant) of commercial strain of rhizobium (Rhizobium leguminosarum bv trifolii Frank 1889 WSM409). One-day old cultures of strains grown at 28°C on glucose minimal salt medium were used to inoculate the seedlings. Each seedling was inoculated with 1 ml of suspension. Plants were grown in the glasshouse as described above. The experimental design was completely randomized with four replicates and repeated three times. After 6 weeks the plants were harvested, examined for diseases symptoms roots were washed and scored for nodulation according a nodule scoring system (Beck et al. 1993) and shoot and root lengths measured. Leaves were separated from roots, oven dried at 60°C for 1 week and dry mass of leaves and roots were recorded.

Data analysis

Data from laboratory and glasshouse experiments were analysed separately for each experiment using software of the GenStat® for Windows® eighth edition developed by VSN International Ltd., UK. A linear mixed model was fitted to each measurement. Data from wild radish and ryegrass were analysed separately. Effects of time, control vs strains, and species and position on the plant from which the bacterial strains were isolated, were fitted as fixed effects. Other treatment effects were included in the random model, which also included time by treatment interactions and allowed for different experimental variance at each time. Residual plots were examined for outlying data and to check whether the assumptions made were valid. All measurements of mass were log transformed prior to analysis as residual variance increased with mass.

Results

Isolation of rhizobacteria from target weed species

A total of 442 bacterial strains were obtained from the rhizospheres, rhizoplanes, and endorhizospheres of seedlings and mature plants of wild radish, annual ryegrass and capeweed sampled from three vineyards in the Swan Valley, Western Australia during October/November 2000. A majority (81·5%) of the strains were obtained from either rhizospheres or rhizoplanes of the weed plants, while only 18·5% of strains originated from the endorhizospheres of these weed species (Table 1). All the strains were relatively fast growing with most producing single colonies after overnight incubation at 28°C on the media used in this study.

Table 1.  Rhizobacterial isolates obtained from weed plants (wild radish, annual ryegrass and capeweed) from different vineyards in the Swan Valley, Western Australia, Australia
Name of the vineyardRhizobacteria source
Wild radish (Raphanus raphanistrum)Annual ryegrass (Lolium rigidum)Cape weed (Arctotheca calendula)Total number of isolates
Number of rhizobacterial isolatesNumber of rhizobacterial isolatesNumber of rhizobacterial isolates
RsRpEndoRsRpEndoRsRpEndo
  1. Rs, rhizosphere; Rp, rhizoplane; Endo endorhizosphere.

Henley Park, WA23181219191022208151
Jane Brook, WA2123623141123186145
Lamont, WA18201024179182010146
Total number of isolates626128665030635824442

Inhibitory effects of rhizobacteria on growth of weed seedlings under laboratory conditions

A total of 125 randomly selected rhizobacterial strains were screened on agar plates under laboratory conditions to investigate deleterious effects on the seedlings of wild radish and annual ryegrass. Approximately 59% (74) of strains significantly reduced the root length of wild radish and/or annual ryegrass seedlings compared to the control plants grown on agar alone (data not shown).

Characterization of selected rhizobacteria

The cultural and morphological characteristics of the 74 rhizobacterial strains with deleterious effects on wild radish and annual ryegrass were recorded (data not shown). Gram-positive bacteria comprized only 13·5% of these strains. All 74 strains were catalase and oxidase positive, and all strains, except CRS71, CRS72, CRS73, and CRS74, were motile in freshly grown broth cultures.

Inhibitory effects of rhizobacteria on growth of weed seedlings under glasshouse conditions

Many of the rhizobacterial strains that reduced root growth in the laboratory bioassays did not show deleterious effects on the development of shoots and roots of annual ryegrass and wild radish under glasshouse conditions. However, 19 strains had an array of effects on length and dry weight of shoots and roots of weeds. Fourteen of the strains (CRS1, CRS2, CRS3, CRS4, CRS5, CRS6, CRS7, CRS8, CRS9, CRS10, CRS11, CRS13, CRS15, and CRS17) inhibited growth of both wild radish and annual ryegrass seedlings. Five strains (CRS12, CRS14, CRS16, CRS18, and CRS19) inhibited only wild radish (Table 2).

Table 2.  Deleterious effects of rhizobacterial strains on wild radish and annual ryegrass under glasshouse conditions
Rhizobacterial strainsTarget weed plants
Wild radishRyegrass
Length† (cm)Mass† (g)Length† (cm)Mass† (g)
ShootsRootsShootsRootsShootsRootsShootsRoots
  1. †The values are means based on 12 samples.

  2. ‡Mass of wild radish and annual ryegrass roots required a logarithmic transformation prior to analysis.

  3. §Treatment means (retransformed means).

  4. Strains different to the control are in boldface.

Control1·43053·187−4·718‡ (0·009)§−4·800 (0·008)3·4053·418−4·849 (0·008)−4·416 (0·012)
WSM34560·75041·8455·257(0·005)5·893(0·003)3·3783·402−4·860 (0·008)6·556(0·001)
WSM34570·68022·3865·243(0·005)5·926(0·003)3·3923·2455·109(0·006)−4·697 (0·009)
CRS31·35492·615−4·846 (0·008)−4·906 (0·007)2·4982·5435·390(0·004)−4·707 (0·009)
CRS40·61422·4265·228(0·005)5·674(0·003)2·3802·5145·194(0·006)4·915(0·007)
CRS51·13513·127−4·850 (0·008)5·284(0·005)2·6032·3165·599(0·004)−4·452 (0·012)
CRS61·22602·8534·957(0·007)−4·991 (0·007)2·4772·367−5·036 (0·006)5·226(0·005)
CRS71·21602·7905·028(0·006)−4·680 (0·009)2·1182·1675·143(0·006)−4·736 (0·009)
CRS80·68402·1315·329(0·005)5·588(0·004)2·2162·7015·420(0·004)−4·609 (0·010)
CRS90·78582·1265·357(0·005)5·584(0·004)3·4253·3125·153(0·006)5·593(0·004)
CRS101·22762·7444·994(0·007)−4·923 (0·007)2·2882·1395·140(0·006)4·958(0·007)
WSM34550·71251·9755·558(0·004)5·804(0·003)3·1063·521−4·715 (0·009)−4·320 (0·013)
CRS120·76032·3155·121(0·006)5·709(0·004)3·3633·331−4·849 (0·008)−4·501 (0·011)
CRS130·70822·1165·586(0·004)5·544(0·004)3·4613·291−4·823 (0·008)5·923(0·003)
CRS140·77512·3395·645(0·004)5·586(0·004)3·4893·477−4·806 (0·008)−4·447 (0·012)
CRS151·26692·9254·941(0·007)−4·994 (0·007)2·6492·6815·226(0·005)−4·615 (0·010)
CRS160·98902·3505·070(0·006)−4·802 (0·008)3·2053·217−4·979 (0·007)−4·528 (0·011)
CRS170·97022·1074·996(0·007)5·613(0·004)2·6312·685−4·945 (0·007)−4·687 (0·009)
CRS180·86212·2685·195(0·006)5·503(0·004)3·3813·450−4·798 (0·008)−4·328 (0·013)
CRS190·84762·0605·107(0·006)5·419(0·004)3·3973·221−4·879 (0·008)−4·581 (0·010)

Deleterious effects of three rhizobacterial strains on wild radish

Three of the rhizobacterial strains randomly selected (WSM3455, WSM3456, and WSM3457) were further screened for deleterious effects on wild radish plants. Strains WSM3456 and WSM3457 were originally isolated from the rhizoplane of ryegrass whereas WSM3455 was originally isolated from the rhizosphere of wild radish. Strains WSM3455 and WSM3456 significantly reduced dry mass of shoots and roots of wild radish (P = 0·001). Although strain WSM3457 reduced dry mass of roots of wild radish the difference was not significant at the 5% level of significance (Fig. 1a). There was no significant reduction in the length of shoots and roots (data not shown). However, a range of foliar symptoms was observed in wild radish grown in the glasshouse experiments when inoculated with the three strains. The symptoms varied from general growth retardation to various types of leaf chlorosis. Lateral root development was poor in wild radish inoculated with these strains.

Figure 1.

Effect of rhizobacterial strains (WSM3455, WSM3456, and WSM3457) on the growth of (a) wild radish plants, (b) vine plants and (c) subterranean clover plants. Control plants were inoculated with sterile medium alone. (bsl00022) Shoots and (bsl00036) Roots.

Screening of deleterious strains for HCN production

Rhizobacterial strains (WSM3455, WSM3456, and WSM3457) were tested for ability to synthesize HCN, an inhibitor of plant roots, and broad-spectrum antimicrobial compound. The change in colour of the medium inoculated with strain WSM3455 indicated this strain produced HCN as a secondary metabolite (data not shown).

Identification of rhizobacteria

The three rhizobacterial strains were identified in detail using the Biolog system (Microlog Version 3·20). Two of the strains (WSM3455 and WSM3456), were identified as P. fluorescens and the third strain (WSM3457) was identified as Alcaligenes xylosoxidans. The identity of two strains, WSM3455 and WSM3456, were confirmed by 16S rRNA gene sequence analysis. The 16S rRNA gene sequence of these two strains exhibited 99% sequence identity to the sequence of strain CHAO AJ 278812·1, strain ATCC AF 094726·1, and strain MM-B16 AY 196702·1 of P. fluorescens.

Screening of rhizobacterial strains on grapevine plants

Results from the pot trial on grapevine rootlings showed the two strains WSM3455 and WSM3456 had no significant deleterious effects on the growth of grapevine plants compare to the control. Although strain WSM3457 slightly increased the dry mass of leaves and roots of grapevine plants the effect was not significant (Fig. 1b). In general the inoculated grapevine plants looked healthy and there was no evidence of disease symptoms.

Screening of rhizobacterial strains on subterranean clover

Strain WSM3456 had no significant effect on the accumulation of dry mass of shoots and roots of subterranean clover. Although the strains WSM3455 and WSM3457 had no significant effect on dry mass of subterranean clover shoots these strains significantly increased the dry mass of roots of subterranean clover (Fig. 1c). Observations of the nodulation of subterranean clover indicated that strains WSM3455, WSM3456, and WSM3457 increased nodulation on subterranean clover compared to the control plants (data not shown).

Discussion

The ability of DRB to inhibit the growth of various weed plants in different cropping systems is well documented (Bakker and Schippers 1987; Begonia and Kremer 1994; Gealy et al. 1996; Kremer and Kennedy 1996; Kremer and Souissi 2001). In this study neither fungi nor actinomycetes were investigated, and the focus was only on relatively fast-growing bacteria that formed visible colonies on agar plates within 24 h. Seventy four of the 125 strains screened for deleterious effects on weeds, significantly decreased the root growth of wild radish and ryegrass. The results reported confirm with earlier studies and indicate the great potential of many bacteria that inhabit plant rhizospheres to interfere with the growth of weed seedlings (Bakker and Schippers 1987; Begonia and Kremer 1994; Gealy et al. 1996; Adam and Zdor 2001).

Many of the 74 rhizobacterial isolates that showed a capacity to inhibit weeds in the laboratory screening shared a range of common characteristics and 86·5% of these strains were Gram-negative, positive for both catalase and oxidase production, and motile. These results complement and extend previous studies with rhizobacterial strains (Kremer et al. 1990).

Further screening of the 74 bacterial strains under environmental controlled glasshouse conditions obtained 19 rhizobacterial strains with deleterious effects on the growth of either wild radish or annual ryegrass, or on the growth of wild radish specifically. The results revealed that 73·7% of the rhizobacterial strains (14 strains) had deleterious effects on both wild radish and ryegrass and 26·3% (five strains) were specific for growth inhibition of wild radish. Analysis of the data indicated that 14 strains had detrimental effects on their respective host species of isolation. Seven of these strains were from wild radish, the other seven strains were from annual ryegrass.

Three of the rhizobacterial strains (WSM3455, WSM3456, and WSM3457) exhibited a variety of effects on wild radish, grapevine rootlings and subterranean clover plants when tested in glasshouse experiments. Strains WSM3455 and WSM3456 significantly inhibited the mass accumulation of shoots and roots of wild radish. Although strain WSM3457 reduced dry mass of roots of wild radish the difference was not significant. The accumulation of dry mass of wild radish was inhibited (53·2%) by WSM3455 (43·6%) by WSM3456, while lesser inhibition (29·3%) was caused by WSM3457, compared with the controls.

By contrast there was no significant reduction of dry mass of leaves and roots of grapevine plants inoculated with the three weed-inhibitory strains in comparison to the noninoculated control. In general all the inoculated grapevine plants were very healthy and none showed signs of disease symptoms.

None of the rhizobacterial strains tested had any detrimental effects on subterranean clover. Rather the two rhizobacterial strains WSM3455 and WSM3457, significantly increased accumulation of dry mass of roots of the subterranean clover plants. This increase in plant growth following inoculation with the rhizobacteria may be a consequence of the greater number of nodules developed on the roots of subterranean clover coinoculated either with WSM3455 or WSM3457 and Rh. leguminosarum bv trifolii WSM409. Alternatively WSM3455 and WSM3457 may be producing plant growth promoting metabolites. Earlier studies on plant growth promoting bacteria have reported that rhizobacteria are potential growth enhancers in different crops like potato, pearl millet, and sorghum (Lazarovitz and Novak 1987; Umesh et al. 1998; Raju et al. 1999).

Our results indicated that WSM3455 has the ability to produce HCN as a secondary metabolite. Albert and Anderson (1987) reported that cyanide producing rhizobacteria are involved in the reduction of plant development. Rhizobacteria are reported to reduce plant growth without obvious plant cell damage, an effect attributed to rhizobacterially produced metabolites being absorbed by roots (Cherrington and Elliott 1987; Begonia and Kremer 1994; Elliott et al. 1996; Kremer and Kennedy 1996).

In the present study, we found that the inhibition by these three rhizobacterial strains WSM3455, WSM3456, and WSM3457 was specific for the weed wild radish as there were no significant deleterious effects on growth of either on grapevine plants or subterranean clover plants.

The mechanisms involved in either the deleterious effects or the plant growth promoting effects of rhizobacteria are still not yet clearly understood. In this study three strains WSM3455, WSM3456, and WSM3457 appear to be DRB for wild radish and at the same time do not having any significant effect on the growth of grapevine plants. Furthermore, two of these strains WSM3455, WSM3457, are plant growth promoting rhizobacteria (PGPR) on subterranean clover plants. This finding that some of the rhizobacterial strains possess a range of diverse properties is consistent with the results of Michéet al. (2000) who reported that Azospirillum brasilense strains L4 prevented the germination of the parasitic weed Striga hermonthica (Del.) Benth. A. brasilense L4 promotes growth of sorghum [Sorghum bicolor (L.) Moench] (Bouillant et al. 1997). Further studies will be required to determine the specific mechanisms involved in these interactions. However, the isolation of rhizobacteria such as WSM3455 and WSM3457 provides opportunities for the development of inocula with capacity for targeted PGPR and DRB activity in agricultural, horticultural and viticultural systems.

Weed management with DRB does not depend on development of an endemic disease on established weeds. Rather the control strategy using rhizobacteria is to regulate the development of weeds before or coincident with the emergence of crop plants. Therefore, DRB do not necessarily eradicate the problem weeds, but significantly suppress early growth of weeds and allow the development of crop plants to effectively compete with weakened weed seedlings (Schroth and Hancock 1982; Kremer and Kennedy 1996). Our investigations indicate the potential for possible inhibition of the weed seedlings through introduction to their rhizosphere populations of specific DRB with identified detrimental activity. However, further investigations will be required to gain an understanding of inoculum responses in different cropping and soil systems, and to determine environment effects on responses to inocula. In addition the development of appropriate inoculum technologies for the targeted introduction of bacteria to weed rhizospheres will be necessary for the development of these DRB as a weed control agents in vineyards and other horticultural systems.

Acknowledgements

Collaboration with the Viticulture and Weed Science groups of the Department of Agriculture Western Australia and University of Extremadura, Spain is acknowledged. Authors thank Henley Park, Jane Brook and Lamont Vineyards for permission to collect weed samples, Vinitech nursery (Swan Valley) for providing the vine cuttings for this work, Mrs. J. Speijers (Senior Biometrician, Agriculture Western Australia) for helpful advice on statistical analysis of experimental data, Dr K Nandasena for assistance with 16S rRNA sequence analysis, Mr R Yates (Centre for Rhizobium Studies) for advice in glasshouse experiments and Dr M Perera (WA State Agricultural Biotechnology Centre) for useful discussions during manuscript preparation. This project was funded by the Grape and Wine Research and Development Corporation.

Ancillary