1Technological Education Institute of Heraklion, 71110 Heraklion, Crete, Greece; 2Department of Agriculture, MacRobert Building, University of Aberdeen, Aberdeen AB24 5UA, Scotland, UK; 3Institute of Marine Biology of Crete, PO Box 2214, 71003 Heraklion, Crete, Greece

Background and objectives
Control of grey mould (Botrytis cinerea) has become increasingly more difficult due to the predomination of resistant strains of the pathogen to the most effective fungicides. Currently attention is being given worldwide to the introduction of biological control agents (BCAs) for the control of this disease, with reports on several potential antagonists [1]. However, the inconsistency of the effect of BCAs and their moderate efficacy is still a major problem. In this study Bacillus brevis and Bacillus subtilis antagonists were tested against grey mould of tomato grown in plastic greenhouses in Crete.

Materials and methods
Experiments were carried out in a plastic greenhouse where 432 tomato plants were grown. Six treatments, notably B. brevis Nagano wild-type strain, which produces gramicidin S and a biosurfactant; the gramicidin S-negative strain E1, which produces the biosurfactant alone; a strain of B. subtilis, which shows in vitro antagonism against B. cinerea; full-strength fungicides iprodione+dichlofluanid; and two controls, were included in the 1995 growing season. In the 1997 growing season B. subtilis was replaced with the fungicide pyrimethanil. Treatments were arranged in a randomized block design (6x6). Disease incidence was assessed at 7-10-day intervals in terms of number of infected sites on tomato leaves. The sensitivity of B. cinerea to the fungicides was regularly monitored. Data obtained were subjected to GLM repeated measures ANOVA. Average disease reduction was studied in each year with linear regression analysis. The confidence interval approach was used to compare model parameters between years.

Results and conclusions
The epidemic was most severe in 1997 following a sigmoid growth pattern, achieving a maximum absolute rate (infection point). In 1995 the epidemic was exponential and its maximum level was at 45% of that of 1997 when recorded at the same time period. The 1995 data indicated that B. brevis E1 best controlled B. cinerea, reducing infection to almost the level of fungicides. B. brevis WT was nearly as effective as E1 but inferior to fungicides. B. subtilis was least effective. The fungicide treatments reduced disease incidence significantly compared to the untreated plots.

In 1997 the efficacy of B. brevis E1 in most records was effective and differed significantly from control, but it was inferior to the fungicide treatments. B. brevis WT was only effective in two out of eight records. The fungicides reduced disease incidence in all records and they differed significantly from control plots and the BCA plots. The average disease reduction obtained by B. brevis E1 was 34.31.5% (1995) and 28.31.9% (1997), and the difference between years was significant at P< 0.05.

Observations reported indicated that B. brevis E1 might influence periods of leaf surface wetness dependent on the prevailing environmental parameters. It might have been expected that B. brevis WT would have been more effective than B. brevis E1 since there is added production of gramicidin S which inhibits conidial germination of B. cinerea [1]. This should have given added protection especially under conditions of high relative humidity where the biosurfactant is expected to be ineffective. Since this was not the case, other factors and parameters must be important. The inferior effectiveness of E1 in 1997 compared to 1995 was partly due to more adverse environmental conditions in 1997 for the performance of B. brevis E1 which at the same time would be favourable to the pathogen. The lower effectiveness of the fungicides in 1995 was due to the presence of strains with low resistance to the pathogen. No resistance was detected to fungicide treatment in 1997.

The constant level of disease reduction, despite the variability of the epidemics over the 2 years, and even when infection point is achieved, indicates that B. brevis E1 is a potential BCA.

1. Seddon B, Edwards SG, Markellou E, Malathrakis NE, 1997. In Gange AC, Brown VK, eds, Multitrophic Interactions in Terrestrial Systems. Blackwell Science, pp. 5-25.