1Cooperative Research Centre for Tropical Plant Pathology, The University of Queensland, St Lucia, Queensland 4072, Australia; 2Horticulture Institute, DPI, Indooroopilly, Queensland 4068, Australia; 3Farming Systems Institute, DPI, Indooroopilly, Queensland 4068, Australia; 41012 Plum Lane, Davis, California 95616, USA

Background and objectives Bananas and plantains are major staples in many developing countries, and dessert bananas of the Cavendish subgroup are significant foreign exchange earners in many countries. Although the importance of the diseases and pests that affect banana and plantain varies depending on the region, Fusarium wilt is regarded as one of the most significant threats to banana production. The incitant of Fusarium wilt in banana, Fusarium oxysporum f. ;sp. cubense (Foc), has been classified into four races based on virulence to host cultivars in the field (race 1, Gros Michel; race 2, Bluggoe; race 3, Heliconia species and race 4, Cavendish).

Understanding genetic diversity within Foc is a necessary component in the selection or breeding of banana cultivars with durable resistance to Fusarium wilt, as potential cultivars ideally should have resistance to all variants of the pathogen. Characterization of Foc has indicated that significant genetic diversity exists within the pathogen, at least 20 vegetative compatibility groups (VCGs) have been described. The aims of this work were to determine (1) the genetic variation among isolates within each VCG and between different VCGs of the pathogen, (2) the genetic variation among races of Foc, and (3) the centre of origin of the pathogen.

Materials and methods
Genetic variation among 342 isolates of Foc representing races 1, 2, 3 and 4, and the 20 known VCGs was analysed using modified DNA amplification fingerprinting. Isolates which did not belong to any of the presently recognised VCGs of Foc were also examined, including race 3 isolates from Heliconia species and also isolates from symptomatic wild banana species growing in the jungle.

Results and conclusions
There was little or no genetic variation among isolates within each VCG independent of host or geographical origin. A total of 33 different genotypes were identified within Foc; 19 genotypes were distinguished among the isolates which belonged in the 20 known VCGs and 14 new genotypes were identified among the isolates which did not belong to any of the existing VCGs. Nine clonal lineages were differentiated within Foc. Five of these lineages each contained numerous closely related VCGs and genotypes and the remaining four lineages each contained a single genotype. The lineages within Foc were as different from each other as they were to other formae speciales of F. ;oxysporum.

There are two hypotheses for the origin of Foc. The first hypothesis proposed that the pathogen co-evolved with edible bananas and their wild diploid progenitors in Asia and has been distributed to other countries in infected banana rhizomes and attached soil. The second hypothesis is that Foc evolved independently from local populations in different countries to attack an introduced host plant. The genetic isolation and limited geographical distribution of several of the lineages of Foc indicates that they have probably arisen independently outside the centre of origin of host. If the co-evolution hypothesis is correct, it is expected that there will be greatest diversity within Foc at the centre of origin. In this study, greatest diversity was identified among isolates of Foc from Indonesia, Malaysia and the Philippines. DNA fingerprinting analysis of isolates found wilting plants in a wild population of Musa acuminata ssp. malaccensis in Malaysia provides further evidence for the co-evolution hypothesis. The co-evolution hypothesis has important implications in the selection of banana cultivars with resistance to Fusarium wilt, as resistant genotypes are most likely to be present in regions where there is great diversity within the host and the pathogen.