1CSIRO Land and Water, PMB 2 Glen Osmond 5064, South Australia; 2Department of Crop Protection, University of Adelaide, PMB 1 Glen Osmond 5064, South Australia; 3Department of Soil Science, University of Adelaide, PMB 1 Glen Osmond 5064, South Australia

Background and objectives
Pseudomonas corrugata strain 2140R (2140R) reduces take-all disease of wheat, caused by Gaeumannomyces graminis var. tritici (Ggt), inhibits Ggt in vitro, and produces many new variant phenotypes. Variants are detected by altered colony morphology on a medium containing triphenyl tetrazolium chloride (TZCA). Common in vitro variants have lost the ability inhibit Ggt and to reduce take-all disease.

Bacteria are known to change phenotype and the production of variant phenotypes is a common laboratory phenomenon. It is not known to what extent phenotype plasticity occurs in vivo. If it does occur, it could have an impact on (1) the rhizosphere populations of disease antagonists introduced as biocontrol agents, or (2) disease control by the introduced bacteria. Strain 2140R was used as a model Pseudomonas disease antagonist to investigate changes in the population structure and control of take-all disease by an biocontrol agent introduced after culture on wheat roots.

Materials and methods
P. corrugata 2140R (rifampicin resistant) was cultured on healthy or take-all diseased wheat plants grown in a nonsterile field soil-sand mixture in pots. Four 2140R-wheat cultures contained no added pathogen and four contained added Ggt. Populations of 2140R were reisolated from healthy roots or diseased root lesions and reinoculated onto successive cycles of wheat (eight cycles of 10-14 weeks). Populations of 2140R were reisolated on nutrient agar containing rifampicin (100 mg/l).

Single 2140R colonies with the wild type and variant colony morphologies were isolated, purified, and assayed for in vitro inhibition of Ggt, Rhizoctonia solani and Pythium irregulare, control of take-all disease (pot assay) and GC-FAME profiles (as used for taxonomic identification). Wild type and variant colony types were compared with the ancestral 2140R isolate stored in 15% glycerol at -70C.

Results and conclusions
Variant 2140R colony types were detected after one cycle of wheat, but were rare. All 2140R populations contained variant colony types after four cycles, and the proportion of variant types increased from 20% to 100% after eight cycles. Both wild type and variant colony types isolated after four cycles on wheat roots differed from the ancestral 2140R isolate in the inhibition of wheat pathogens and GC-FAME profiles. GC-FAME profiles of the wild-type colonies were as divergent from the ancestral 2140R as the variant colony types (Euclidian distance of 8 to 9), with some identified as different species. After eight cycles on wheat roots, some isolates were significantly different (P=0.05) from each other in control of take-all disease. Isolates with wild-type colony morphology generally had an increased ability to control take-all disease, whereas in variant types this generally decreased, but this was not significantly (P=0.05) different from the ancestral 2140R isolate.

This work shows that a population of the disease antagonist, P. corrugata 2140R, started initially from a single isolate, can become a mixed population of more than one phenotype after as early as one 10-week cycle on wheat roots. Variant colony types can eventually dominate the population in some pot cultures. Wild-type colonies isolated after a number of cycles can diverge from the ancestral type as much as colony morphology variants. By evolutionary theory, the persistence and multiplication of new phenotypes depends on the selection pressure exerted on the population by biotic and abiotic conditions. It is hypothesized that other soil bacteria can also change phenotype in the root-soil environment. This may have implications for variable effectiveness of biocontrol agents. It may also have implications for the development of disease suppressive soils where selection for a disease-suppressive microbial community could be either from within the existing soil community or from new phenotypes which may arise.