Variability of populations of Fusarium oxysporum f.sp. melonis

A study was undertaken to investigate the variability among isolates of Fusarium oxysporum f.sp. melonis with regard to their pathogenetic relationship with the host. 36 F.o.m. isolates were submitted to pathogenicity tests to determine the race to which they belonged and the disease incidence, using a set of differential genotypes and the susceptible “Supermarket”, respectively. For each isolate, a non-linear regression was estimated according to the logistic model and, afterwards, the day on which disease incidence reached 10, 50 and 90% of affected plants (D10, D50 and D90) was calculated. To obtain groups of homogeneous isolates for the development of disease incidence, a cluster analysis was performed using D10, D50 and D90 as separating variables. To find the probability that an isolate belonged to the group assigned by the cluster analysis, a discriminant analysis was performed, which, moreover, allowed us to identify the most important variables in the separation of isolates. The mean progress of disease incidence in each cluster of isolates was then calculated as a logistic regression analysis. Finally, the degree of similarity between the two clusters of F.o.m. isolates (the former based on the physiological races, the latter on the disease incidence) was evaluated with the Pearson similarity correlation coefficient. The statistical approach adopted allowed us to ascertain that significant differences exist among the F.o.m. isolates studied, as regards their rapidity in causing disease symptoms in susceptible hosts, and a significant degree of similarity does not exist between the two clusters of isolates.