Fusarium oxysporum and its biocontrol


Author for correspondence:
C. Alabouvette
Tel: +33 03 80 69 30 41
Fax: +33 03 80 69 32 26
Email: alabouvette@dijon.inra.fr


Fusarium oxysporum is well represented among the rhizosphere microflora. While all strains exist saprophytically, some are well-known for inducing wilt or root rots on plants whereas others are considered as nonpathogenic. Several methods based on phenotypic and genetic traits have been developed to characterize F. oxysporum strains. Results showed the great diversity affecting the soil-borne populations of F. oxysporum. In suppressive soils, interactions between pathogenic and nonpathogenic strains result in the control of the disease. Therefore nonpathogenic strains are developed as biocontrol agents. The nonpathogenic F. oxysporum strains show several modes of action contributing to their biocontrol capacity. They are able to compete for nutrients in the soil, affecting the rate of chlamydospore germination of the pathogen. They can also compete for infection sites on the root, and can trigger plant defence reactions, inducing systemic resistance. These mechanisms are more or less important depending on the strain. The nonpathogenic F. oxysporum are easy to mass produce and formulate, but application conditions for biocontrol efficacy under field conditions have still to be determined.


The species Fusarium oxysporum is well represented among the communities of soilborne fungi, in every type of soil all over the world (Burgess, 1981). This species is also considered a normal constituent of the fungal communities in the rhizosphere of plants (Gordon & Martyn, 1997). All strains of F. oxysporum are saprophytic and able to grow and survive for long periods on organic matter in soil and in the rhizosphere of many plant species (Garrett, 1970). Moreover, some strains of F. oxysporum are pathogenic to different plant species; they penetrate into the roots inducing either root-rots or tracheomycosis when they invade the vascular system. Many other strains can penetrate roots, but do not invade the vascular system or cause disease (Olivain & Alabouvette, 1997). The wilt-inducing strains of F. oxysporum are responsible for severe damage on many economically important plant species. Fusarium wilt pathogens show a high level of host specificity and, based on the plant species and plant cultivars they can infect, they are classified into more than 120 formae speciales and races (Armstrong & Armstrong, 1981). Management of Fusarium wilt is mainly through chemical soil fumigation and resistant cultivars. The broad-spectrum biocides used to fumigate soil before planting, particularly methyl bromide, are environmentally damaging. The most cost effective, environmentally safe method of control is the use of resistant cultivars, when these are available. For example, all the varieties of tomato grown in glasshouses for fresh fruit production are resistant to the common races of F. oxysporum f. sp. lycopersici. Breeding for resistance can be very difficult when no dominant gene is known (e.g. carnation, cyclamen, flax) or if the host is dioecious (palm trees). In addition, new races of the pathogen can develop which overcome host resistance. The difficulty in controlling Fusarium wilt has stimulated research in biological control of Fusarium wilt independently of the recent concern for environmental protection. The aim of this review is to present methods available for the characterization of F. oxysporum, whether pathogenic or nonpathogenic, to describe the different modes of action by which nonpathogenic strains can inhibit pathogenic strains, and to discuss how this knowledge can be applied to biocontrol of Fusarium diseases.

Characterization of F. oxysporum and diversity among strains

Because telomorphic stages of most Fusaria are unknown, Fusarium taxonomy has been based on morphological characteristics of the anamorph, including the size and shape of macroconidia, the presence or absence of microconidia and chlamydospores, colony colour, and conidiophore structure (Windels, 1992). The difficulty in delineating species based on these features is evidenced by the many different systems that have been proposed, recognizing anywhere from 30 to 101 species (Booth, 1971; Gerlach & Nirenberg, 1982; Nelson et al., 1983). Many of these taxonomic schemes group the species into sections. Based on morphological criteria, it is sometimes difficult to distinguish F. oxysporum from several other species belonging to the sections Elegans and Liseola. To further complicate the picture, plant pathogenic, saprophytic and biocontrol strains of F. oxysporum are morphologically indistinguishable. As mentioned above, pathogenic F. oxysporum is very host specific attacking only one or a few species of plants, and in many cases, attacking only certain cultivars of that plant. The specificity for a particular host and for cultivars of that host is designated, respectively, as formae speciales and race of the pathogen. These pathogenic fungi are morphologically indistinguishable from each other as well as from nonpathogens. Puhalla (1985) proposed a system of classification of strains of F. oxysporum, based on their vegetative compatibility and described a method based on pairing nitrate nonutilizing mutants to determine the vegetative compatibility group (VCG) of each strain. Kistler et al. (1998) note the usefulness of VCG and proposed a system for standardization of numbering of VCG. Some formae speciales correspond to a single VCG, while others include several VCGs. In a recent review, Katan (1999) recognized 59 VCGs for 38 formae speciales. Among nonpathogenic populations, many isolates are single member VCGs and some are even self-incompatible (Gordon & Okamoto, 1992; Kondo et al., 1997; Steinberg et al., 1997). Therefore, although useful, the determination of VCG cannot be used as a universal tool to identify formae speciales or nonpathogenic isolates.

Given the shortcomings of classical taxonomic characters for delineating species and subgeneric groupings of Fusarium, researchers have turned to molecular tools to provide information needed for a taxonomic framework for species identification, as well as to elucidate the evolutionary relationships among species. Sequences of the translation elongation factor EF-1α and the mitochondrial small subunit (mtSSU) ribosomal RNA genes have been valuable in distinguishing species and origins of Fusaria (Baayen et al. 2000; O’Donnell et al., 2000; Baayen et al., 2001; Skovgaard et al., 2001). DNA sequences of UTP-ammonia ligase, trichothecene 3-O-acetyltransferase, and a putative reductase (O’Donnell et al., 2000) and nitrate reductase, phosphate permease (Skovgaard et al., 2001) have also been used successfully to distinguish Fusarium species. Sequences of the β-tubulin region have been useful to distinguish some Fusaria (O’Donnell et al., 2000) but not others (O’Donnell et al., 1998). While DNA sequences of the ITS regions are very useful in distinguishing species in many eukaryotic organisms, they have not been very informative for Fusarium (O’Donnell & Cigelnik, 1997). Similarly, sequencing the calmodulin region did not help resolve the origins of F. oxysporum f. sp. cubense (O’Donnell et al., 1998). Rosewich et al. (1999) used nuclear restriction fragnment length polymorphism (RFLP) and VCG to determine that F. oxysporum f. sp. radicis-lycopersici migrated from Florida to Europe. Assigbetse et al. (1994) were able to use random amplified fragment length polymorphisms (RAPD) to differentiate races of Fusarium oxysporum f. sp. vasinfectum on cotton.

While these molecular tools have successfully identified pathogenic strains and in some cases even races of the pathogen, determining pathogenicity still relies largely on bioassays. Forma specialis is determined by testing the fungus for pathogenicity on various plants species, while race is determined by testing for pathogenicity on cultivars of a single plant species. Although bioassays are very useful in verifying pathogenicity, they cannot establish that a strain is nonpathogenic. Because more than 120 formae speciales and races have been described, it would be necessary to inoculate the unidentified strains to an endless number of different plant species and cultivars. This is obviously not possible, especially when a large collection of soil isolates must be characterized. Because neither the molecular nor the biological methods can be used to characterize the nonpathogenicity of strains, it must be stressed that the so-called ‘nonpathogenic’ strains of F. oxysporum are strains that failed to induce disease on a limited number of plant species to which they have been inoculated.

Role of nonpathogenic F. oxysporum in soil suppressiveness to Fusarium wilts

The existence of soils that naturally limit the incidence of Fusarium wilts has been recognized for more than a century. During the 1960s, studies by Stover (1962) and Stotzky & Martin (1963) were devoted to the role of abiotic factors, mainly clays, in relation to the reduction of disease incidence in the so-called ‘long-life’ soils in banana plantations in Central America. Later, starting with the studies of Smith & Snyder (1971) more attention was given to the role of the biological factors, especially to the saprophytic microflora including nonpathogenic Fusarium spp. Toussoun (1975) stated that soil suppressive to Fusarium wilts supported large populations of nonpathogenic species of Fusarium spp. Similarly, the suppressive soil from Châteaurenard harboured high populations of F. oxysporum and F. solani (Louvet et al., 1976). A form of Koch's postulates confirmed the involvement of nonpathogenic populations of Fusarium spp. in the mechanism of soil suppressiveness. Indeed, suppressiveness was destroyed by heat treatment at a temperature > 55°, that eliminated the thermo-sensitive microflora including the Fusarium spp., and suppressiveness was restored by artificial introduction of strains of nonpathogenic F. oxysporum or F. solani in the heat-treated soil (Rouxel et al., 1979). Since then, numerous studies clearly identify a role for nonpathogenic Fusarium spp. in suppressive soils from different areas in the world (Schneider, 1984; Tamietti & Alabouvette, 1986; Paulitz et al., 1987; Tamietti & Pramotton, 1990; Larkin et al., 1993a; Larkin et al., 1996). Strains of F. oxysporum and F. solani were much more efficient in establishing suppressiveness in soil than other species of Fusarium. Moreover, effective biocontrol strains of nonpathogenic F. oxysporum have been isolated from the stems of healthy plants (Ogawa & Komada, 1984; Postma & Rattink, 1992). Therefore, most of the research dealing with biocontrol of Fusarium diseases has focused on the modes of action of the nonpathogenic strains of F. oxysporum.

Modes of action of nonpathogenic F. oxysporum

A thorough understanding of the mechanisms of action is needed to maximize consistency and efficacy of biocontrol. The mechanisms of action associated with nonpathogenic F. oxysporum can be divided into two broad categories: direct antagonism of the nonpathogenic strains to the pathogen and indirect antagonism mediated through the host plant.

Direct antagonism

Generally speaking, mechanisms of direct microbial antagonism include parasitism, antibiosis and competition. Thus far, there is no evidence of either parasitism or antibiosis among strains of F. oxysporum, but data from many studies support the role of competition. Competition can be divided into saprophytic competition for nutrients in the soil and rhizosphere, and competition for infection sites on and in the root.

Competitive interactions in the soil and rhizosphere In the absence of any evidence of antibiosis between nonpathogenic and pathogenic strains of F. oxysporum, the hypothesis of trophic interactions was proposed to explain the role of nonpathogenic F. oxysporum in the mechanisms of soil suppressiveness. More precisely, the hypothesis of competition for carbon sources was proposed based on the fact that a single addition of glucose to a suppressive soil was sufficient to make the soil conducive (Louvet et al., 1976). Couteaudier & Alabouvette (1990) demonstrated the validity of the hypothesis of competition for carbon between strains of F. oxysporum by comparing the growth kinetics of a small collection of strains of F. oxysporum introduced into a sterilized soil amended with various amounts of glucose. Modelling of the growth curve (Couteaudier & Steinberg, 1990) enabled calculation of the growth rate and the yield coefficient (i.e. the number of propagules formed per unit of glucose consumed) for each strain. Results showed a great diversity among the seven strains compared. The yield coefficient varied from 1 × 106 to 8 × 106 propagules formed per mg of glucose consumed. Six of these strains were then confronted with a 7th strain, the pathogenic strain F.o. f.sp. lini (Foln3) resistant to benomyl. Each strain was introduced into sterilized soil in combination with the pathogenic strain Foln3 at five different inoculum ratios. By following the kinetics of growth of each strain in mixture it was possible to calculate a ‘competitiveness index’ for each strain. These indices ranged from 1.3 to 3.5, indicating a large diversity in the ability of these six strains to compete in soil with the pathogenic strain F.o. f.sp. lini. Lemanceau et al. (1993) have confirmed, in vitro, that carbon was the major nutrient for which a pathogenic strain of F.o. f.sp. dianthi was competing in soil-less culture with the biocontrol agent Fo47. These results were confirmed by Larkin & Fravel (1999) who demonstrated that isolate Fo47 significantly inhibited pathogen chlamydospore germination in soil at glucose concentrations of 0.2 mg g−1 soil and greater. In addition, germ tube growth also was significantly reduced in soil containing Fo47 compared with untreated soil. By contrast, the biocontrol isolate F. oxysporum CS-20 had no effect on germination or germ tube development of the pathogen. Competition for nutrients has also been shown to be involved in the mode of action of other isolates of nonpathogenic F. oxysporum such as strain 618.12 (Postma & Rattink, 1992) and strains C5 and C14 (Mandeel & Baker, 1991).

Competitive interactions on the root surface and in plant tissues Competition also occurs on root surfaces. Mandeel & Baker (1991) postulated that the root surface had a finite number of infection sites that could be protected by increasing the inoculum density of the nonpathogenic strain. Using GUS-transformed strains, Olivain & Alabouvette (1999) clearly showed that both a pathogenic and a nonpathogenic strain were able to actively colonise the surface of the tomato root. Both were able to penetrate the epidermal cells and colonise the upper layers of cortical cells. The plant reacted to this fungal invasion by expressing defence reactions (wall thickenings, intracellular plugging) that were more intense in the case of the nonpathogen. Benhamou & Garand (2001) observed the same defence response in transformed pea root confronted with Fo47. As a result, these defence reactions always prevented the nonpathogen to reach the stele, although the pathogenic strain grew quickly towards the vessels, which were invaded. These observations are concordant with the hypothesis of competition between strain of F. oxysporum for infection sites at the root surface, and for root tissue colonisation. Indeed, both strains colonized the same spots at the root surface and showed great similarities in the colonisation process of the root. These observations are also in agreement with previous results obtained by Eparvier & Alabouvette (1994) who utilized another approach to demonstrate that pathogenic and nonpathogenic strains were competing for root colonisation.

Postma & Luttikholt (1996) considered the hypothesis of direct competition between two strains of F. oxysporum in the vessels of the host plant. They compared the growth in the stele of carnation of a pathogenic strain of F. oxysporum f. sp. dianthi and of several nonpathogenic strains after artificial inoculation of these strains, alone or in combination into vessels of the plant. They showed that some nonpathogenic strains were able to reduce the stem colonisation by the pathogen resulting in a decrease of disease severity. Locally induced resistance or direct competition between strains within the vessels could cause this disease suppressive effect after mixed inoculation into the stem. These observations are in agreement with the results of Ogawa & Komada (1984) who selected a nonpathogenic strain of F. oxysporum able to control Fusarium wilt of sweet potato when it was introduced into the stem of the plant. Taken together, these results support the existence of competition between pathogenic and nonpathogenic F. oxysporum, not only for infection sites at the root surface, but also inside plant tissues.

The competitive ability of a nonpathogenic strain partly determines its capacity to establish in soil and in the plant rhizosphere, and is probably involved in its capability to colonise the root surface. Nagao et al. (1990) demonstrated that different strains have different capacities to colonise heat-treated soil. Moreover, when flax was grown in soil fully colonised by the nonpathogenic strains, root colonisation was also drastically different. There was no correlation between the population density of the biocontrol strains in soil and their capacity to effectively colonise the roots.

Saprophytic colonisation of soil depends not only on the fungal strain but also on biotic and abiotic soil characteristics. The ecological parameters affecting the biotic component of these strains have seldom been studied. Colonisation of the root surface and root tissues probably depends not only on the fungal strain but also on the plant species and plant cultivar. The compatibility between strains of nonpathogenic F. oxysporum and the plant species or plant cultivar has not been thoroughly investigated. Hervas et al. (1997) studied the interactions between the genotype of chickpea, inoculum concentrations of the pathogen, and application of a strain of nonpathogenic F. oxysporum. The plant cultivar can also influence the Fusarium population in soil. Larkin et al. (1993b) found that the watermelon cultivar ‘Crimson Sweet’ created its own suppressive soil via its root exudates, which increased populations of beneficial F. oxysporum while other watermelon cultivars did not. Clearly, many factors must be considered to maximize biocontrol efficacy (Hervas et al., 1997).

Indirect antagonism: induction of systemic resistance

It is well established that preinoculation of a plant with an incompatible strain of F. oxysporum (either a nonpathogenic strain or a pathogenic strain belonging to another forma specialis) results in the mitigation of symptoms when the plant is later inoculated with a compatible pathogen (Matta, 1989). This phenomenon is considered as an expression of induced systemic resistance, a general plant defence response to microbial infection or various stresses. Induced systemic resistance (ISR) has been extensively studied, since it could explain the disease control provided by nonpathogenic strains of F. oxysporum. Biles & Martyn (1989) were the first to attribute to ISR the control of Fusarium wilt of watermelon achieved by several strains of nonpathogenic F. oxysporum. Many investigators have used a split root method to study ISR in Fusarium (Biles & Martyn, 1989; Mandeel & Baker, 1991; Kroon et al., 1992; Olivain et al., 1995; Fuchs et al., 1997; Larkin & Fravel, 1999). Papers reported experiments where a nonpathogenic strain applied to some roots of a host plant can delay symptom expression induced by the pathogen separately applied to other roots or directly into the stem of the plant. Since there is no direct interaction between the two microorganisms, the observed disease reduction is attributed to increased plant defence reactions in response to root colonisation by the nonpathogenic strain. Indeed, it has been clearly established that most of the nonpathogenic F. oxysporum actively colonise at least the upper layers of root cells as shown in Fig. 1 for Fo47 and CS-20. These observations are correlated with the fact that both Fo47 and CS-20 are able to induce ISR in some plant species.

Figure 1.

Light micrographs of transverse semi-thin sections of Flax roots stained with toluidine blue O and methylene blue O: (a) and (b), roots inoculated with Fusarium oxysporum CS20, (c) and (d) with F. oxysporum Fo47. Hyphae (H) of the fungi are observed at the root surface, in the epidermis (EC) and hypodermis (HC) cells. Colonised cells showed defence reactions limiting the growth of the fungi. The cortex and the stele did not appear colonised. CC, cortical cells; XV, xylem vessels. Magnification: (a) and (c) × 200; (b) and (d) × 600.

When competition is the main mode of action, typically the population of the biocontrol fungus must be at least as large, if not larger, than that of the pathogen population in order to achieve control; whereas, when ISR is the main mode of action, control can often be achieved when the pathogen population is much greater than that of the biocontrol fungus. Larkin & Fravel (1999) significantly reduced wilt incidence in tomato when the pathogen population was up to 1000 times greater than that of strain CS-20. By contrast, strain Fo47, which functions mainly through competition, is only effective when it is introduced at concentrations 10–100 times higher than the pathogen concentration.

ISR is correlated with both enzymatic changes in the plant often leading to induction of the physical barriers discussed above (Benhamou & Garand, 2001). Tamietti et al. (1993) found increased activity of several plant enzymes related to plant defence reactions in tomato plants transplanted in sterilized soil infested with a strain of nonpathogenic F. oxysporum. Fuchs et al. (1997) attributed the biocontrol activity of the nonpathogenic strain Fo47 to induced resistance in tomato, correlated with an increased activity of chitinase, β 1–3 glucanase and β 1–4 glucosidase. Duijff et al. (1998) showed that the nonpathogenic strain Fo47 although not very effective in inducing systemic resistance in tomato induced an increase of PR proteins. Recorbet et al. (1998) showed an overall increased activity of constitutive glycosidase isoforms in response to infection by F. o. f. sp. lycopersici that did not occur in roots colonised with nonpathogenic strains. These contrasted results all obtained with strain Fo47 applied to tomato demonstrate that the biochemical response of the plant is not clearly understood and must be accurately described, before this system can be compared to other plant–pathogen models where the cascade of biochemical events is better known.

Finally, when the main mode of action of a nonpathogenic strain is induction of systemic resistance, it is obvious that this phenomenon implies the physiological state of the plant and fluctuating environmental conditions may affect the ability of the plant to express its resistance to the pathogen, induced by the nonpathogenic F. oxysporum.

Complementary modes of action

It must be emphasised that the different modes of action described above do not exclude each other. On the contrary, the same nonpathogenic strain can express several modes of action either simultaneously or at different times. This is the case for the strain Fo47 for which several teams have reported the involvement of competition for nutrients in soil, competition for root colonisation and induced systemic resistance. This is also the case for the nonpathogenic isolates C5 and C14 isolated by Mandeel & Baker (1991) and 618–12 (Postma & Rattink, 1992). One might expect that a strain expressing several modes of action would be more efficient and provide a more consistent control than a strain having a single mode of action.

Nonpathogenic F. oxysporum as plant protection products

While knowledge of the interactions between pathogenic and nonpathogenic strains of F. oxysporum and the modes of action of the biocontrol agents is necessary, it is not sufficient to predict the conditions under which biocontrol can be applied and will be effective. Thus, there is a need for applied research specifically dealing with development of biocontrol methods.

Discovery and development of effective strains of nonpathogenic F. oxysporum

The first step in developing a biocontrol method is screening for an effective strain. The success of all subsequent steps depends on how well this first step is accomplished and some screening methods will only identify biocontrol organisms with a particular mode of action. Many studies dealing with nonpathogenic F. oxysporum have proven that not all the nonpathogenic strains are effective in controlling Fusarium wilts. Since there is currently no known genetic marker to identify biocontrol strains, the only available and reliable method to screen for efficient strains is a bioassay in which the potential biocontrol agents are confronted with the pathogen in the presence of the host-plant and disease incidence or severity is monitored. While very labour-intensive and time-consuming, the advantage of such a bioassay is that it takes into account most of the possible interactions among microorganisms, and between microorganisms and the plant, that can lead to biological control. In general, the closer the screening method is to the production system, the greater the chances are for success. For example, because tomato is a transplant crop in Florida, Larkin & Fravel (1999) chose to screen it for biocontrol agents by drenching tomato seedlings with potential biocontrol agents and then transplanting these into field soil infested with the pathogen.

One of the dilemmas in screening is the question of how many organisms to screen. Is the level of control observed in initial screening the best that can be achieved, or if one knew more about the organism, could a higher level of control be achieved? Most nonpathogenic Fusaria provide control that is at least slightly better than the disease control treatment. Thus, tests must be repeated several times under different environmental conditions (soil type, temperature, moisture), and on different cultivars to determine reliability of the control. Similarly, tests are needed to determine the inoculum density relationships between the pathogen and biocontrol agent. In addition to providing information on how much biocontrol agent is needed to achieve disease control, inoculum density tests often provide the first clues about the mechanism of action, because, as discussed above, competitors usually need to be present in populations greater than that of the pathogen to achieve control while those that induce resistance can achieve control when present in populations lower than that of the pathogen. Larkin & Fravel (1999), studying the dose–response relationships between three formae speciales of pathogenic F. oxysporum and three nonpathogenic strains, have proposed several mathematical models to analyse these relationships and calculate an ‘effective biocontrol dose’. For example, strain CS-20 was able to control the disease at an inoculum density as low as 100–500 propagules per g of soil, even in the presence of a high inoculum density of the pathogen (1 × 105); whereas efficacy of Fo47 required population densities substantially greater than that of the pathogen. These results corroborate previous studies showing that, under experimental conditions, the maximum efficacy of Fo47 will be reached with a concentration 100 times higher than that of the pathogen (Alabouvette et al., 1993). These high inoculum densities of the nonpathogen necessary to achieve biological control would be difficult to reach under agricultural production conditions, which is why, until now, the use of nonpathogenic F. oxysporum has been mostly limited to crops cultivated in potting mixtures or soil-less systems.

Production, formulation and delivery Having selected an efficient strain and determined basic information on its biology and ecology, it is necessary to mass-produce and formulate it in such a way that it can be stored and applied easily. The efficacy of a biological control agent greatly depends on the quality of the inoculant, which is function of both production and formulation (Lewis, 1991; Whipps, 1997). F. oxysporum is easily grown in liquid or solid fermentation. Both processes are being used to produce the strains Fo47 and CS-20 for large-scale experiments. Because of their role in survival, chlamydospores are considered the most desirable propagules for formulation, and production methods generally focus on increasing the percentage of chlamydospores. From studies in other biocontrol systems, we know that production methods that yield the greatest number of propagules in the shortest period of time do not necessarily yield the most efficacious propagules (Hebbar et al., 1997; Hebbar et al., 1998). Thus, bioassays are necessary to verify the ‘quality’ of the inoculum. Similarly, viability after storage does not necessarily guarantee biocontrol efficacy and a bioassay is needed to confirm bicontrol ability. Development of a biochemical assay obviating the need for these bioassays would be very useful.

The strain Fo47 can be produced by solid-state fermentation either in a sterilized peat or in calcinated clay enriched with an appropriate nutrient solution. In peat, no matter what the initial concentration of inoculum, strain Fo47 will reach, at the plateau, a concentration > 1 × 107 propagules g−1. Both the peat and the calcinated clay provide a carrier for the inoculum; there is no need for further formulation. This inoculant can be stored at 4°C or even at room temperature without losing its density, or its activity for up to 18 months (C. Olivain et al. unpublished). Efficacy of the inoculant can be improved by choice of a specific food base that will favour the growth of the biocontrol agent after release. Steinberg et al. (1997) compared the population kinetics and the biological efficacy of several formulations of Fo47. A formulation made of microgranules enriched with a food base provided a better survival and a better bio-control efficacy than the traditional talcum formulation used in the laboratory.

Because most industrial microbial fermentation is currently done in liquid, liquid fermentation is generally preferred. One disadvantage of liquid fermentation is that unless the final formulation is liquid, it is necessary to separate the fungal biomass from the fermentation liquid before formulation. Biomass can be separated by drying, filtration or centrifugation. Hebbar et al. (1997), working with a mycoherbicidal strain of F. oxysporum, increased the percentage of chlamydospores in liquid fermentation by use of selected substrates, such as soy hull fibre, and by lowering the dissolved oxygen concentration. Methods developed for this mycoherbicide have been successfully used with strain CS-20. Strain Fo47 has been produced in a 400 l fermentor, in an appropriate growth medium with control of O2 supply and pH (Durand et al., 1989). The propagules, mainly ‘bud cells’, are mixed with talcum, which is then dried for 48 h at 18–20°. This inoculant can be stored at 4°C for long periods, since the inoculum concentration was only decreased by 23% and 35% after 1 and 2 yr, respectively (Alabouvette & Couteaudier, 1992).

Successful biocontrol depends on having the biocontrol agent delivered to the right place, at the right time, in the appropriate physiological state. In addition to these considerations, application must be compatible with the production system. For example, because fresh market tomatoes are a transplant crop, Larkin & Fravel (1998) applied strain CS-20 as a drench while the plants were in the glasshouse before transplanting to the field. To protect cyclamen from Fusarium wilt, Fo47 is introduced in the potting mixture used to grow the plant. Thus it colonises the substrate before accidental introduction of the pathogen.

Future research

Additional research is needed in several areas including: field studies and integration into production systems; genetics; mechanisms of action; risk assessment; and genetic improvement of biocontrol agents.

The use of nonpathogenic strains of F. oxysporum to control Fusarium wilt has been reported for many crops including: banana (Gerlach et al., 1999), basil (Fravel & Larkin, 2002), carnation (Garibaldi et al., 1986), cucumber (Mandeel & Baker, 1991), cyclamen (Minuto et al., 1995), flax (Alabouvette et al., 1993), gladiolus (Magie, 1980), melon (Rouxel et al., 1979), tomato (Lemanceau & Alabouvette, 1991), spinach (Katzube et al., 1994), strawberry (Tezuka & Makino, 1991), watermelon (Larkin et al., 1996). Among these papers, only a few presented data that were obtained under commercial production conditions.

The greatest reason for lack of adoption of biological control is the lack of consistency of biological control performance based on the application of a strain of nonpathogenic F. oxysporum. In addition to field studies integrating biological control into commercial production systems, a more thorough understanding of the genetics, biology, and ecology of biocontrol agents, and their modes of action will enable optimal exploitation of these fungi for disease control. These issues are interrelated since molecular techniques can provide insight in each of these areas. Molecular genetics offers new tools for unravelling mechanisms and understanding genetic relationships among Fusaria. For example, an approach using transposon mutagenesis has recently been used to produce mutants of F.o.melonis affected in their pathogenicity (Migheli et al., 1999) and mutants of Fo47 showing either an increased or a decreased biocontrol capacity (Trouvelot et al., 2002).

Recently, C. Olivain et al. (unpublished) compared the early physiological responses of flax cells challenged with conidia of strains of pathogenic and nonpathogenic Fusarium oxysporum. The results observed with the pathogenic strain were typical of those described in the case of the compatible reaction and results observed with Fo47 were similar to those observed in the case of the incompatible reaction when a pathogen is interacting with a resistant cultivar. The nonpathogenic strain elicits early defence reactions restricting its growth into the root. These preliminary results open the way to new research in the field of the plant–microbial interactions leading to the identification of the biochemical pathways trigger by the nonpathogenic strains.

It is conceivable that understanding what is triggered in the induced resistance discussed above may lead to being able to trigger this in the absence of a biocontrol agent, thus avoiding the problems inherent in production, formulation and delivery of a living agent.

More research using new tools of molecular genetics is needed to determine the genetic relationships among pathogenic, biocontrol and saprophytic Fusaria, as well as to elucidate the genetic determinants of pathogenicity and biological control ability. Identification of genes involved in biological control should assist in making biological control more consistent and in optimization of control, identification of genes involved in biological control may also facilitate screening for new biocontrol agents or genetic improvement of current biocontrol agents.

One issue that has received little attention is the safety of releasing nonpathogenic F. oxysporum. Cook et al. (1996) list four areas of concern for the release of any biological control agents. These are: displacement of nontarget microorganisms; allergenicity to humans and other animals; toxigenicity to nontarget organisms; and pathogenicity to nontarget organisms. They point out that, based on experience, any adverse effects from biological control are likely to be short-term and can be eliminated by terminating use of the biocontrol agent, while agriculture as it is currently practiced has produced significant long-term adverse affects. Gullino et al. (1995) point out the need for data on which to base informed decisions about the risk involved in releasing nonpathogenic F. oxysporum. They found release of nonpathogenic F. oxysporum to have negligible nontarget effects with respect to persistence and survival, effect on indigenous microbial communities, and genetic stability and transfer. Because some Fusaria produce mycotoxins, it is important to establish that those used as biological control agents do not produce toxins. With F. oxysporum in particular, the concern arises as to whether the biocontrol agent is truly nonpathogenic, or whether it may be pathogenic on a species of plant on which it has not yet been tested. Further, given the lack of understanding about how new races of the pathogen arise, there is some concern from the public that the biocontrol agent could become pathogenic. Genetic data are needed to allay these fears. But progress in the development of nonpathogenic Fusaria as biological control agents needs a cooperative research effort in all these fields from basic understanding of the mechanisms to practical field studies of the environmental conditions influencing the efficacy of the control.