Most soils inhibit fungal germination and growth to a certain extent, a phenomenon known as soil fungistasis. Previous observations have implicated microorganisms as the causal agents of fungistasis, with their action mediated either by available carbon limitation (nutrient deprivation hypothesis) or production of antifungal compounds (antibiosis hypothesis). To obtain evidence for either of these hypotheses, we measured soil respiration and microbial numbers (as indicators of nutrient stress) and bacterial community composition (as an indicator of potential differences in the composition of antifungal components) during the development of fungistasis. This was done for two fungistatic dune soils in which fungistasis was initially fully or partly relieved by partial sterilization treatment or nutrient addition. Fungistasis development was measured as restriction of the ability of the fungi Chaetomium globosum, Fusarium culmorum, Fusarium oxysporum, and Trichoderma harzianum to colonize soils. Fungistasis did not always reappear after soil treatments despite intense competition for carbon, suggesting that microbial community composition is important in the development of fungistasis. Both microbial community analysis and in vitro antagonism tests indicated that the presence of pseudomonads might be essential for the development of fungistasis. Overall, the results lend support to the antibiosis hypothesis.

We released genetically modified Pseudomonas putida WCS358r into the rhizospheres of wheat plants. The two genetically modified derivatives, genetically modified microorganism (GMM) 2 and GMM 8, carried the phz biosynthetic gene locus of strain P. fluorescens 2-79 and constitutively produced the antifungal compound phenazine-1-carboxylic acid (PCA). In the springs of 1997 and 1998 we sowed wheat seeds treated with either GMM 2, GMM 8, or WCS358r (approximately 10 7 CFU per seed), and measured the numbers, composition, and activities of the rhizosphere microbial populations. During both growing seasons, all three bacterial strains decreased from 10 7 CFU per g of rhizosphere sample to below the limit of detection (10 2 CFU per g) 1 month after harvest of the wheat plants. The phz genes were stably maintained, and PCA was detected in rhizosphere extracts of GMM-treated plants. In 1997, but not in 1998, fungal numbers in the rhizosphere, quantified on 2% malt extract agar (total filamentous fungi) and on Komada's medium (mainly Fusarium spp.), were transiently suppressed in GMM 8-treated plants. We also analyzed the effects of the GMMs on the rhizosphere fungi by using amplified ribosomal DNA restriction analysis. Introduction of any of the three bacterial strains transiently changed the composition of the rhizosphere fungal microflora. However, in both 1997 and 1998, GMM-induced effects were distinct from those of WCS358r and lasted for 40 days in 1997 and for 89 days after sowing in 1998, whereas effects induced by WCS358r were detectable for 12 (1997) or 40 (1998) days. None of the strains affected the metabolic activity of the soil microbial population (substrate-induced respiration), soil nitrification potential, cellulose decomposition, plant height, or plant yield. The results indicate that application of GMMs engineered to have improved antifungal activity can exert nontarget effects on the natural fungal microflora.